Understanding Single-Firm Behavior: Remedies

This document is available in three formats: this web page (for browsing content), PDF (comparable to original document formatting), and WordPerfect.











10THURSDAY, MARCH 29, 2007

















1A P P E A R A N C E S








9Douglas Hilleboe, Federal Trade Commission

10Ed Eliasberg, U.S. Department of Justice




14Michael Cunningham, Red Hat, Inc.

15Renata B. Hesse, Wilson Sonsini

16Marina Lao, Seton Hall Law School

17William H. Page, University of Florida

18Howard A. Shelanski, UC Berkeley









1P R O C E E D I N G S

2 - - - - -

3MR. HILLEBOE: Good morning, everyone, thank you

4for coming. I'm Doug Hilleboe, attorney with the

5Federal Trade Commission, Office of the General Counsel,

6I'm going to be one of the moderators here today for

7this third session on remedies. My co-moderator is Ed

8Eliasberg, he's an attorney with the U.S. Department of

9Justice, Legal Policy Section of the Antitrust Division.

10Before we start, I need to go over a few

11housekeeping matters. As a courtesy to our speakers,

12please turn off your cell phones, Blackberries and other

13devices that make a noise, and I'll ask the speakers to

14do the same, they actually interfere with the

15microphones and we had a little problem with that.

16Second, the restrooms are located down the hall,

17through the double doors that you came through. Third,

18in the unlikely event that the building alarms go off,

19please proceed calmly and quickly, as instructed. If we

20must leave the building, take the stairway which is to

21the right, on Pennsylvania -- on the Pennsylvania side,

22and after leaving the building, follow the stream of FTC

23people and meet at the sculpture garden, which is across

24from the intersection of Constitution Avenue and 7th



1Also, we must enforce our rule that there's no

2questions or comments that come from the audience during

3the session. Thank you.

4We're honored today to have assembled a

5distinguished group of panelists that have agreed to

6offer their testimony in connection with this hearing on

7remedies in the face of technology change.

8Howard Shelanski is an associate dean and

9professor of law at the University of California,

10Berkeley, and the director of the Berkeley Center For

11Law and Technology.

12Renata Hesse is a partner at Wilson Sonsini

13Goodrich and Rosati, and formerly was a chief of the

14Networks and Technology Enforcement Section At the

15Antitrust Division.

16Michael Cunningham is general counsel at Red

17Hat, Inc.

18William Page is a Marshall M. Criser eminent

19scholar at the University of Florida's Levin College of


21And Marina Lao is a professor of law at Seton

22Hall Law School.

23We plan to hear from each of the speakers for

24about 15 minutes each and then take a ten-minute break

25and then we'll hear from the remaining speakers. We


1will then have the speakers comment upon what they've

2heard, and then have a moderated discussion among the

3speakers with Ed and I leading the discussion.

4Before starting, I would just like to state by

5way of introduction that many of the product markets in

6which the United States enjoys a comparative advantage,

7vis-a-vis the rest of the world, are fast-changing

8dynamic markets, including high technology markets.

9Some critics of the antitrust laws have claimed that the

10laws, including Section 2, are not nimble enough for

11effective use in these types of markets. Others

12disagree. We will explore this issue and others in this

13 session.

14Some commentators have suggested that the

15potential for error in antitrust enforcement may be

16 greater in these dynamic markets; however, other

17commentators have suggested that due to network effects

18and other possible factors, these markets may tend

19towards monopolization to a greater agree and therefore

20perhaps deserve particular antitrust scrutiny.

21We are interested to learn what these panelists

22believe about these and other issues, and their

23implications for antitrust enforcement in Section 2


25Before beginning with the speakers, my


1co-moderator, Ed Eliasberg has some words about the


3MR. ELIASBERG: Thank you, Doug. I very briefly

4on behalf of the Antitrust Division plan to welcome our

5panelists, thank you for coming and we look -- we're

6very much looking forward to hearing what you have to


8So, with that, Ed, let me turn back to you.

9MR. HILLEBOE: Thank you, Doug. Howard

10Shelanski is the Associate Dean and Professor of Law,

11Boalt Hall, University of California, Berkeley and the

12Director of the Berkeley Center for Law and Technology.

13From 1999 to 2000, he served as chief economist of the

14Federal Trade Commission -- Federal Communications

15Commission, excuse me, and from 1998 to 1999, he served

16as senior economist for the President's Council of

17Economic Advisors At the White House.


19MR. SHELANSKI: Thanks, Doug, and I appreciate

20the promotion. Well, I have a few main points that I

21want to make and the points that I am going to make I

22hope connect to what my co-panelists are going to say.

23We had a call a week ago and I just want to set

24up a few ideas here about the implications of the

25implementation of remedies for monopolization in a


1high-tech or technologically dynamic markets. And I

2think my main point, my overall point would be this:

3Remedies are hard in the best of circumstances, and I

4think they become more complicated in technologically

5dynamic settings, but I also think that innovation and

6the presence of ongoing innovation in a market may

7affect remedies in somewhat unpredictable ways, and may

8create opportunities along with the challenges.

9In particular, I think while innovation makes

10structural remedies more difficult, it may in some cases

11make conduct remedies particularly valuable. So, I

12think while innovative markets are cause for agencies

13and courts to be more cautious about remedies, I think

14innovation is not cause for systematic retreat from

15enforcement or from behavioral injunctions.

16So, let me explain a little bit why I think this

17is the case. You'll hear, and I think one often hears

18that structural remedies are preferable to conduct

19remedies or behavioral remedies in monopolization cases.

20But, there are some caveats to this. First I would say

21that structural remedies are not always available.

22Where a firm is so integrated that there are not obvious

23divisions, it's very hard to know how to implement a

24structural remedy. Just as a classic example, the

25District Court's second opinion in the United Shoe


1machinery case would be an example.

2The second caveat I would have is that

3structural remedies are not always easier than conduct

4or behavioral remedies, and in fact must often include

5some supporting behavioral remedies, and as an example,

6I would talk about the AT&T vertical divestiture that

7 had to be implemented by open access regulations

8enforced by the FCC and overseen by the District Court.

9And then, finally, I would say as a general

10caveat, the effectiveness of structural remedies in

11Section 2 cases is not assured and there's certainly

12quite a bit of debate of effectiveness historically over

13structural remedies. I'll give you a couple of

14examples. One early quotation, "In administering the

15antitrust acts, a number of great and powerful defenses

16against them have been dissolved. So far as is possible

17to judge the consuming public has not yet greatly

18profited by their dissolution." That's Judge Rose in

19United States against American Can in 1916.

20Okay, now, we haven't had a lot of experience in

21enforcing Section 2 by 1916, so maybe things have

22changed, at least some people disagree. Bob Crandell in

232003 writes, divestitures are "costly exercises in

24futility," but I would point you to the excellent work

25of John Baker and Greg Werden in 2003 providing some


1counter arguments. Just a way of saying effective

2remedies structurally offer no guarantee of success.

3Now, I think the structural remedies may

4actually be even harder in technologically dynamic

5markets, and let me offer a couple of reasons. First,

6where a firm or industry is driven by R&D, it may do no

7good to divest a given division or to leave a company in

8two without sending the R&D operations with the divested

9portions of the entity, but R&D operations are often,

10perhaps even likely, to be more integrated and

11 inter-dependent within the firm and not susceptible to

12clean lines of separation.

13The second reason why I think the presence of

14ongoing technological change may make structural

15remedies difficult is that even if divestiture is

16possible, high-tech firms may require more monitoring of

17conduct during after the divestiture, because key assets

18in such divestiture are likely to be intellectual

19property, IP that in some cases may provide joint uses,

20uses across the lines of the new or divested entities,

21disputes are likely to be offered over what items to

22transfer and whether all IP has been disclosed to the

23new entity.

24Moreover, because of the cooperative nature of

25research and development, and in production, in markets


1where product life cycles are short, some post

2divestiture monitoring of relationships between newly

3distinct entities may be needed because there may be a

4natural incentive to favor each other as business

5partners, and that was something that came up in the

6wake of the AT&T divestiture, for example.

7The third reason I think that fast technological

8change renders structural remedies more challenging is

9that firm and market structure may be less of an issue,

10in some technologically dynamic markets. To the extent

11that the so-called Schumpeterian School is correct, that

12dynamic markets often display competition that occurs

13sequentially, through periodic waves of creative

14destruction, rather than concurrently, through

15simultaneous production, divestitures may be less

16effective or necessary such markets, although this is

17probably more true for horizontal than for vertical


19Okay, and my final reason that structural

20remedies are tough in technologically dynamic markets,

21is that where network effects are at issue, structural

22issues might harm consumers by dissipating positive

23network externalities. The fact that it might have been

24better not to have monopoly in the first place does not

25always mean it is better to break up the monopoly later,


1and if such divestitures are to preserve network

2externalities, they may have to be accompanied by

3conduct remedies related to interconnection and

4interoperability, doing away with those clean properties

5of structural remedies.

6Okay, let me turn now to conduct remedies, talk

7a little bit about how they might work in high-tech

8markets. As a general matter, we often hear that

9conduct remedies are difficult, but there are some

10caveats here as well. Not all conduct remedies are

11created equal, and as many people have pointed out,

12negative prohibitions, thou shalt not have exclusive

13deals, for example, are probably easier to implement

14than affirmative obligations, thou shall deal with your

15rivals. In part because the negative prohibitions

16entail less involvement of courts or agencies in

17regulating terms of trade.

18The second caveat that I would add is that

19conduct remedies can have beneficial prospective impact,

20even if they cannot roll back illegally accumulated or

21prolonged market power. Some people say, look, conduct

22remedies are closing the barn doors after the cows are

23out, but if there are still some cows inside the barn,

24it's not a bad idea to shut the door.

25Third, even if a conduct remedy is ineffective


1or weak in a given case, I think conduct remedies can

2have important deterrent effects on others contemplating

3the illegal behavior, and it's -- in a point that's

4often made, some people say, if you can't be sure that

5your conduct remedy is going to be effective, why bring

6the case? Another reason to bring the case beyond

7deterrence is I think as we get more experience with

8different kinds of conduct, it can become clearer what

9is good and what is bad, and it enables agencies to move

10more quickly in subsequent cases, and perhaps get a

11remedy implemented while the harm is still able to be --

12to be nipped in the bud, so I would not let lack of a

13clearly successful conduct remedy -- I think one needs

14to be clearly articulable at the start of a case, but if

15you can't be sure it will be implemented in time or it

16will be successful in remedying the market power, there

17may be some reasons to go ahead with the case anyway in

18terms of establishing precedent and creating deterrence


20And finally, just an observation, I think that

21the effectiveness of conduct remedies are likely to --

22the effectiveness is likely to be tied to the precision

23with which one can define the cause of anticompetitive

24harm, and in some cases, this can be done quite clearly,

25and in those cases, I think behavioral injunctions can


1be quite effective.

2So, the overall lesson about conduct remedies, I

3think that it is right to be weary of behavioral

4remedies, particularly those in which the enjoined

5conduct has ambiguous welfare effects, or in which

6courts or agencies will have to become involved that

7were doing terms of trade, but in the right context,

8conduct remedies can work and can send valuable

9deterrent signals.

10 I would just say that inability to articulate a

11structural remedy therefore should not be decisive in

12whether or not to prosecute an argument that is

13sometimes heard.

14Okay. Well, I think that technologically

15dynamic markets create both challenges and opportunities

16for implementing conduct remedies. The first challenge

17is this: If one accepts that remedies may deter

18marginal innovation, and I'll assume for the moment that

19all innovation is good, because private returns are less

20than social returns to innovation. Let's just take that

21as a working assumption, it need not be true in all

22cases, but if one accepts that, and one accepts that

23remedies can marginally deter innovation, then the

24deterrence risk and the costs of such deterrence may be

25much greater in dynamic markets. It needn't be the


1case, but I think innovation deterrence becomes a more

2salient issue and a more salient concern in

3technologically dynamic markets.

4The second challenge is that in fast-changing

5markets, it is more likely than it is in more static

6settings that the conduct at issue in the case will be

7moot by the time antitrust liability is established.

8And in such cases, neither conduct nor structural

9remedies are likely to be effective, and perhaps

10something else like disgorgement might be called for if

11such a remedy can be created.

12But there are also opportunities in high

13technology settings, I think, for conduct remedies to be

14particularly effective. In some cases, technological

15dynamics can render conduct remedies effective where

16they would not be in more static markets.

17In some cases, monopoly once obtain may not be

18easily eroded, even if exclusionary or predatory conduct

19that contributed to that monopoly is stopped. Whether

20because of brand recognition, economies of scale, or

21customer switching costs, new entrants will be slow to

22appear or succeed, even when other barriers to entry,

23such as the exclusionary or predatory conduct at issue

24in the case, even when those barriers are eliminated,

25 you might not see competition arising.


1But I think where competition is more innovation

2based and where product life cycles are short, an

3injunction against the behavior that led to the

4establishment or maintenance of monopoly power may prove

5very effective, as it is the latter set of barriers,

6rather than any brand or economic advantage, that might

7have kept the incumbent dominant.

8As new waves of innovation come forward, how did

9they stop someone else from being the innovator who came

10in with the new product? Well, through the exclusionary

11or predatory conduct, and branded here and switching

12costs, other things like that, may be very, very

13different in the high-tech environment. So, merely

14eliminating the harmful conduct may open the door for

15new entry and the conduct or remedy, particularly

16negative injunctions, I think, can be very successful

17and very helpful.

18I would like to just raise an additional point

19about the overall question of whether or not the cycles

20of innovation move so quickly and the innovation process

21moves in such different a way from the standard

22competitive process that we should step back generally

23from antitrust enforcement, and this is an argument that

24one hears quite often.

25I think when one looks at the kinds of behavior


1that limit innovation, and that stop people -- that stop

2competitors from innovating, it's very unclear to me

3whether or not monopoly has anything particular to

4recognize it, nor is it clear to me that new waves of

5innovation are always going to be sufficiently powerful

6to overcome artificial barriers to entry like

7exclusionary -- exclusionary kinds of behavior like

8exclusive deals when it is a monopolist that has that

9exclusive deal, contractual terms that bar competitors'

10products from ever being used, tying that prevents

11consumers from ever having access to products.

12It's unclear to me no innovation will always be

13so great that it can overcome those barriers, those

14barriers can lead to slower product life cycles, and

15greatly harm consumers, and I think that there's a lot

16of evidence of benefits from antitrust enforcement in

17high-tech areas. And when one looks at the studies that

18have said there are no benefits to Section 2

19enforcement, or in a more nuance way, no benefits to

20Section 2 enforcement in technologically dynamic

21markets, there's a counterfactual, all of these papers

22acknowledge the counterfactual, and we can't tell what

23would have happened absent the antitrust enforcement, we

24can't tell what would have happened in other markets had

25there been antitrust enforcement, and then those


1arguments are sort of dismissed, tucked under the


3 I wouldn't dismiss them so easily. And, so, my

4overall argument would be, be very cautious, be very

5case-by-case in the application of Section 2 remedies in

6high-tech markets, I think structural remedies are

7likely to be harder to implement, but there may be good

8opportunities for conduct remedies to be very effective.



11MR. HILLEBOE: Thank you very much, Howard. Our

12next speaker, excuse me, is Renata Hesse, who is a

13partner at Wilson Sonsini Goodrich and Rosati. Prior to

14joining Wilson Sonsini, Renata served as the chief of

15the Networks and Technology Enforcement Section at the

16Antitrust Division and oversaw much of the division's

17technology litigation, including the Oracle/Peoplesoft

18and First Data/Concord matters. In addition, Renata

19worked extensively on both the American Airlines and the

20Microsoft case.


22MS. HESSE: Getting myself around is a little

23harder these days.

24So, Howard covered a lot of ground which I think

25fundamentally I agree with almost everything he said.


1In fact, I think I probably agree with everything he

2said, but wanted to pick up where he was leaving off,

3which was I think in talking about the notion that you

4shouldn't back away from Section 2 enforcement in high

5technology markets, and the main reason why I think

6that's true is that despite all of the innovation and

7the fast pace of change in those markets, there is an

8opportunity for durable market power to exist in them,

9and you do want to make sure that you're not overlooking

10that possibility and potentially addressing it.

11So, I wanted to start with just a few basic

12points about Section 2 remedies that I think are

13important, and some of these overlap with some of the

14things that Howard said and I'm sure that will happen as

15we go along down the line of speakers, but the first

16thing that I wanted to talk about is the importance of

17focusing on remedy early, and the main reason -- there

18are several reasons for that, but the biggest reason is

19that it helps you try to figure out what your goal is.

20What's the violation that you're really thinking about,

21what do you think has really happened that's harmful,

22and how can you address it? That isn't to say that if

23you can't come up with a perfect solution to the problem

24that you shouldn't go ahead and try and do something

25 about it.


1I think Howard is right that there's a good

2deterrent effect in enforcing the law, even if you're

3not 100 percent sure that the way that you think you can

4fix it will be successful, but I do think it will -- it

5helps you focus your investigation, and here again, I'm

6speaking as if I were a government lawyer, but focus

7your investigation and theories so that you can really

8figure out whether or not you've got a case that is

9worth allocating resources to, and pursuing.

10And I just think it gives you a much better

11sense of the definition of the harm that you're trying

12to alleviate.

13The second point is that I think when you start

14with thinking about remedy, or at least you think about

15remedy relatively early in the process, you can get a

16better sense for whether or not you actually can come up

17with a remedy that is really going to leave the

18marketplace in a better place than it was when you


20And I would sort of call this the first do no

21harm rule, and it is one of these things which you

22always need to bear in mind, which is that you don't

23always want to make things worse, you don't want to

24deter innovation or take an action in the marketplace

25which stifles productivity, and I think in technology


1markets, that's something that you really need to keep

2in mind.

3But if you were stepping back and thinking about

4that early, you can think about whether or not there are

5ways to achieve the goal that you want to achieve

6without having at least a large countervailing harmful


8The third point is related to the resource

9allocation point that I made. I think fundamentally

10it's just a basic responsibility that particularly

11government enforcers have to think about how you're

12going to fix the problem, and whether or not the problem

13is subject to a fix that's worth the investment of

14resources in not only the investigation and prosecution

15of the matter, but also the compliance and enforcement

16activities that will happen post judgment, and those

17are, I think, much more complicated when you're talking

18about conduct remedies and structural remedies, but,

19again, Howard correctly notes that when you do a

20structural remedy in these markets, very often there are

21going to be conduct remedies associated with it in any


23But I think you really do want to have in your

24mind whether or not the consumption of the resource is

25likely to result in some improvement to the competitive


1conditions in the marketplace.

2And then there's a fourth point which is that

3sort of the question of if you have a good idea of what

4you think the remedy that you want to put into place is,

5then I think you'll have a better idea of whether or not

6the -- again, the pursuit of the investigation or

7prosecution is worth while, and by that I mean that

8there are some kinds of Section 2 violations that are

9easier to remedy than others.

10So, one example might be you can think of

11exclusive dealing or vertical foreclosure, for example,

12where you have fairly easily identifiable concrete types

13of conduct that you can undo. I think monopoly

14maintenance, to a certain degree, monopoly acquisition

15cases are much harder.

16So, if you're in the situation where you're

17balancing these things out, and you've got a choice

18between two matters that you want to devote your

19resources to and one of them has a reasonably good

20likelihood of being able to be fixed, and the other is a

21little tougher, then you've got to figure out how to

22allocate your resources, then you might want to think

23about going towards the one that actually has a solution

24that you can identify and that you think will be likely

25to result in an improvement in the competitive



2And this just goes back to something that I

3think people often think about in the context of -- of

4the -- when you're trying to come up with a remedy, what

5is it that you're trying to achieve, are you looking at

6a monopoly that you believe has been illegally created

7and are you trying to undo that, or are you looking at

8conduct that has maintained a monopoly and are you

9trying to restore the conditions of the competitive

10marketplace to the pre-exclusionary conduct state? And

11depending on which of those two things you're looking

12at, you're going to have a pretty different, I think,

13idea about what's the right way to go about recommending

14the harm.

15The second thing I wanted to talk about was just

16the point that Howard started with, which is structural

17remedies and the general point that generally I think

18structural remedies should be preferred. I think it's

19clearly true that they are not always possible, and

20that's certainly more true in Section 2 cases than in

21other kinds of cases, but I wouldn't advise sort of

22ignoring them as possible ways of recommending harm,

23because I think they do have a number of benefits.

24One of the benefits is that developing a

25functional set of conduct restrictions that are likely


1to have a beneficial effect, without having this sort of

2countervailing, potentially negative effect on the

3marketplace is an extremely complicated and resource

4intensive process. It took a really long time to come

5up with the conduct restrictions that we developed in

6the Microsoft case, and I think, you know, you can --

7it's open for debate whether or not those were worked

8well or not well, but it took a long time to figure them

9out, and to just evaluate all the different

10possibilities and try to develop language that's

11concrete enough and understandable enough in a legal

12document for people to actually then be able to

13implement it and understand it and understand what the

14rules of the road are. It's just an inherently

15difficult process to do, and I think that isn't just

16Microsoft, that's any time when you're trying to come up

17with a set of conduct restrictions where you're dealing

18with complex technology.

19It's also hard to judge their success, I think,

20and that's also true in structural remedies, in some

21situations, but it's very hard to know when conduct

22restrictions have succeeded. I think you can know when

23they've failed, but I don't think you can know as easily

24when they've succeeded. How do you measure success with

25conduct restrictions?


1I think structural remedies generally eliminate,

2although not entirely, the need for ongoing enforcement

3in compliance activity, which also can be an extremely

4time consuming and resource intensive process. It can

5require, and this is something else I can talk about a

6little bit later, but it can require a lot of assistance

7from people who know more about technology and business

8and licensing and all these things that come up in

9technology markets work, and structural remedies tend to

10need a lot less of that.

11I think structural remedies are generally less

12easy to evade. It's pretty clear what you're supposed

13to do, and you've either done it or you haven't done it.

14You've either divested the plant or the asset or

15whatever it is, or you haven't. You know, there are

16issues associated with those kinds of things, whether or

17not you found an adequate buyer and all of those other

18sorts of issues, but at least there's a very clear line

19about what you are supposed to have done.

20I think they have a potentially greater

21deterrent effect, because they have the capability at

22least of really restructuring a business in a way that

23most businesses don't want to have happen. So, that can

24discourage people from engaging in conduct that folks

25think violates Section 2.


1And I think generally, again with some of the

2caveats that Howard laid out, they're more likely to

3work. The lines are clearer, and if you've actually

4proven a violation where you can support imposition of a

5structural remedy, I think the likelihood of that

6structural remedy having an effect is probably higher.

7So, those are some kind of basic points. A few

8points that are more directly connected, just to sort of

9the technology markets, and the first is, you know,

10everybody always talks about technology markets are fast

11changing and innovation changes everything, and as

12Howard said, sometimes people say, maybe you don't need

13to worry about them because they're just going to be

14self correcting. I tend not to agree with that latter

15viewpoint, for the reason that I started with, which is

16that it's clear that there's a possibility for the

17existence of durable market power in these markets, so I

18think just leaving them alone and hoping that the

19exclusionary conduct somehow magically stops and things

20correct themselves is not likely to lead to a lot of


22I do think that the fact that they can sometimes

23be slow and that the antitrust enforcement process can

24sometimes be slow is a down side in these markets, a

25greater down side in these markets than in other


1markets, because sometimes you feel like you get to the

2end and you're addressing the problem when it's actually

3a little bit too late.

4As a consequence, I think you need, when you're

5thinking about conduct remedies in technology markets,

6to be a little bit more flexible about how you think

7about them. And to address categories or types of

8conduct relating to types or categories of products or

9services as opposed to saying, well, this -- you did

10this particular thing with this particular kind of

11product, and you should do that -- you shouldn't do that

12anymore. This is the negative prohibition point versus

13an affirmative obligation point.

14If the conduct remedy is too narrowly focused,

15it runs the risk of being ineffective, and I think in

16most cases is likely to be ineffective, particularly,

17again, if you're talking about undoing some sort of harm

18that has occurred.

19You know, Microsoft is a simple example of this,

20the consent decree doesn't just talk about browsers,

21which was the primary focus of the case, but it talks

22about other products which were potential platform

23threats and has some construct restrictions in it that

24are designed to try to go after those particular -- or

25not go after them, but to try and make sure that the


1conduct relating to those other kind of potential

2platform threats were restrained.

3There's a possibility in technology markets that

4they should be of shorter duration. Again, Microsoft is

5another example, it was a five-year consent decree, it's

6now been extended in some pieces for longer than that,

7but I think there's a reasonable basis for at least

8looking at the question of whether or not you really

9need something to last ten, 20, some decrees in the past

10have lasted for hundreds of years, some of them very

11perpetual, and whether or not that makes sense

12particularly in the context of technology markets is I

13think something that people -- it's worth looking at.

14I also think if you're going to think about

15decrees of shorter durations, or remedies of shorter

16durations, that including some mechanism for revisiting

17that question before the term of the decree expires is a

18good idea. I think it's just these markets are

19inherently unpredictable, and given the complication of

20structuring conduct provisions in them, that giving

21yourself an opportunity to take a second look and having

22a standard for how you would be able to convince a court

23that you need to extend a decree in these kinds of

24markets is something that should be given some



1And the final point on this area is that I think

2conduct remedies in Section 2, Section 2 remedies in

3technology markets may need to be more forward looking,

4and this is a little slightly basically the same thing

5with a slightly different pitch on it, but you do have

6to think about what it is that you can predict about the

7marketplace and changes in the marketplace going forward

8and whether or not what you've devised in the context of

9the conduct remedy is adequate to address the changing

10technology in the marketplace.

11The last piece about technology markets that I

12think makes them different is that they're hard, and

13it's hard to understand them, and they're particularly

14hard for people who are not educated in technology.

15And, so, compliance monitoring enforcement can be a

16difficult thing to do.

17As a consequence, I think if you're looking at

18these markets and you're looking at behavioral

19restrictions, particularly ones that relate to licensing

20of intellectual property or access to technology or

21just, you know, you're requiring a company to stop doing

22a particular activity with a particular type of

23technology, that you really need to anticipate getting

24some technical help, and when I think of technical help

25in this context, I don't think just of software


1engineers or hardware engineers, but I also think of

2licensing expertise, business expertise, you know,

3trying to figure out whether a royalty ran is a

4difficult problem, and it's not a problem that most

5antitrust lawyers deal with on a day-to-day basis.

6And having the ability to have access to people

7who actually do that kind of work for a living, who know

8what particular types of technologies, what kinds of

9royalties particular types of technologies command is, I

10think, critical to the ability to actually do an

11adequate job of monitoring and enforcing compliance.

12Again, I started with sort of a more broad

13definition of technical assistance, but a narrow

14definition of technical assistance, which is just

15actually having somebody who knows how software code is

16written, and what to look for and how to evaluate

17whether or not something has been done in the code is

18very important. I think one of the really unusual and

19innovative things that was in the Microsoft decree was

20the technical committee provision, which allowed the

21Department of Justice and the states to have access to

22basically a full-time group of technical consultants who

23were hired to work for those people and the cost of

24which was borne and continues to be borne by Microsoft.

25I think it was an unusual idea, but it really


1has become, I think, a key component to the United

2States enforcement and monitoring, compliance monitoring

3efforts of the Microsoft decree, and it was essentially

4copied by the European Commission in the work that

5they're doing in Microsoft as well.

6And it had not been done before. There were

7lots of times where in complicated markets people had

8used monitoring trustees, I shouldn't say there were

9lots of times, but there were examples of monitoring

10trustees being used, usually they were in things like

11prison condition litigation, where there was some pretty

12complicated oversight that was needed, but hiring

13technical experts to help out was an innovative thing to

14do and I think has proven to be a pretty successful

15component of the Microsoft decree.

16Now, you also may need technical assistance when

17you're trying to figure out whether or not somebody has

18violated the decree and you actually want to go after

19them for contempt. I think the Microsoft model doesn't

20quite fit so well in that context, because it's a little

21hard to see how you can justify the party who you're

22going to be pursuing in contempt actually paying for the

23expert that you're going to be using, to go after them

24in contempt, but it's something that people -- you want

25to think about, and at least have the resources and


1capability to get that kind of help on board.

2So, I have probably 30 seconds at this point

3left. The last thing I would say is that licensing

4remedies are incredibly common in technology markets.

5They can be useful, and I think can work well, but I

6think they work particularly well in the context where

7you know or have a very good idea of what the

8intellectual property is or what the asset is that needs

9to be licensed, are there particular patents who needs

10them, and again, if you go back at the very beginning,

11to those are things that you can think about early on

12and figure out and they'll help you determine whether or

13not a licensing remedy is likely to be successful.

14And of course when you're doing that, you need

15to think about the policy issues that are associated

16with compulsory licensing of intellectual property,

17which is a hot topic these days.


19MR. HILLEBOE: Thank you so much, Renata, for

20those comments.

21Michael Cunningham is general counsel at Red

22Hat, Inc. Prior to joining Red Hat, he served as

23associate general counsel at IBM, where he had legal

24advisory responsibilities for the Business Consulting

25Services Division for Europe, the Middle East and


1Africa. He was also a partner and associate general

2counsel at PricewaterhouseCoopers.

3 Michael?

4MR. CUNNINGHAM: Thank you, and good morning.

5I'm pleased to have the opportunity to participate in

6this important consideration of Section 2 remedies, to

7do so before distinguished representatives of the

8government, as well as with this particularly

9knowledgeable panel.

10I'm the general counsel of Red Hat. I'm going

11to make a little disclaimer, I'm a technology lawyer,

12I'm not principally an antitrust lawyer. I hope that I

13can offer some comments, however, as an executive of a

14technology company that are relevant to these inquiries.

15With your indulgence, I would like to describe a

16bit about our business that I think is relevant

17innovation, given the debate about antitrust remedies

18stifling innovation, I think it's particularly

19appropriate this morning.

20The software solutions that Red Hat offers, and

21for which we provide services, are developed by very

22broad horizontal communities that are without

23geographic, organizational or political boundaries. The

24community of innovators that unleash the value of open

25source are not contained within Red Hat. Some of its


1contributors are, but it's not.

2The contributors include the customers and

3vendors of hardware and software. It includes

4academics, it includes many, many motivated individuals

5that we call hackers, it includes persons from every

6continent and from multiple political subdivisions.

7The development environment is also not

8controlled by any single individual company or political

9entity, it is instead a free, meritocratic marketplace

10of ideas. Individuals take these ideas and they place

11these ideas with their individual name and reputation

12into the marketplace in a particular software

13development project to which their idea is relevant.

14There are literally thousands of these projects

15out there. In one of our offerings, Red Hat Enterprise

16Linux, hundreds of projects are represented. These

17ideas are then reviewed by that development community,

18for that project, and only those ideas that can handle

19the open scrutiny of this open source community are then


21In this way, the best ideas and the bets bits of

22ideas bubble up. Moreover, if there happen to be a

23serendipitous discovery that is made in one of those

24projects that's relevant to another project or might be

25an entirely new approach, the contributor or any other


1person is free to contribute it to that project or

2indeed to go out and start a new project to take the

3technology in a new direction.

4This model has produced and continues to produce

5copious innovation. It also accelerates and multiplies

6innovation, I would argue, by providing tools of

7innovation, such as information ideas to a broader and

8more diverse community than development within any one

9firm is possible could provide.

10The open exchange of information and ideas is an

11innovation force multiplier. For example, sophisticated

12business and other users of software frequently take the

13modular pieces of well crafted software that's developed

14in the open source community, cobble bits and pieces of

15it together, modify it, append to it and create

16solutions for problems that heretofore were not solved,

17or new problems that arise in their business.

18Similarly, the creative juices of the lone

19teenager in North Dakota in some remote location can

20contribute to that process, so can a Cal Tech physicist

21who is wondering why there hasn't been a software

22development that would help in his or her research. And

23so are many, many others unleashed in the creative

24process through this open development and collaboration



1The modular and open nature of open source

2software has fueled much innovation, but it is by no

3means limited to software. It is not a software-only

4phenomena. No, I would submit to you that the relative

5ubiquity and low cost of the Internet, and collaboration

6tools like email and dedicated web sites portends for

7joint collaboration that is unleashing all sorts of

8innovation across the world.

9If you've read the best selling book by Tom

10Friedman, The World is Flat, you will get a very good

11sense of some of these trends, I think. I would also be

12happy to comment on some other areas where that

13innovation is being unleashed in the questioning, if

14that's helpful.

15With that bit of an introduction, maybe I should

16turn my attention now more directly to remedies. First,

17I believe that in the software space at least, the

18relevance of the antitrust law hangs on the issue of

19remedies. I can think of no way as a practitioner and

20an executive in a company in the industry to more

21starkly illustrate that point than to disclose my actual

22advice to my client in pursuing whether to participate

23in or pursue any monopoly-related case, whether that be

24in a government-related case or in private litigation.

25I would tell my client, it's too expensive for


1you to fully embrace and do that. You cannot do it.

2You don't have enough money to pursue it, it's certainly

3over $10 million, it will be a long time, and it is

4likely, I would submit to you, at least this would be my

5advice, it is likely and substantially likely that the

6remedy that will result will be of limited utility. So,

7therefore, those sorts of expenditures would not be


9And guess what? Those that the government

10representatives seek to regulate know this, and they

11know it well. By way of illustration, a high-ranking

12representative, indeed a very high-ranking

13representative of a party found to have market power by

14multiple international competitive authorities has

15aggressively and indeed smugly advised Red Hat that

16there is no competition authority in the world that this

17firm will not outspend, outlast, and seek to thwart.

18In short, the system seems broken in terms of

19speed, cost, and effectiveness of remedies, at least

20from my little corner of the world. You know, why is

21this the case? Well, as others have said, technological

22change is very rapid and litigation is not. The rate of

23change at least in information technology is in very

24short cycles, three to five years, maybe six to eight

25years, certainly not longer than that in many, many


1areas of information technology.

2Remedies that only address a particular market

3complained of, and established at great expense, will

4often be too late to provide meaningful relief. A

5remedy focused on future conduct would address some of

6those limitations and in many instances I think is


8 I also am intrigued by the idea of smaller

9simpler cases with speedier trial times that would focus

10on future contact to make the law more relevant.

11Clearly cost and delay undermine the perceived and

12actual effectiveness of the antitrust laws in our

13competitive zone.

14In that way, some of Professor's Lao's writing

15on the role of the intent in finding liability seem a

16fruitful avenue for further inquiry to me.

17Second, technology can be manipulated. The

18speed with which information technology moves and can be

19molded provides real opportunity for conscious

20manipulation by the monopolist away from the market

21complained of. The government enforcement actions

22against Microsoft are an example of the timing

23challenges, I'm thinking now about the European Union,

24even the most aggressive threats by the EC are mired in

25delay, seemingly extended without limit.


1According to the most recent statistics we've

2seen, Microsoft continues to gain in the operating

3system worker group server market, meanwhile the market

4continues its very rapid evolution, probably reducing

5the relevance of any remedy that may eventually be

6enforced and/or issued.

7I guess I should also point out that private

8enforcement actions have not solved the problem either,

9this won't be a surprise from my earlier comment. The

10antitrust law, like the Ritz Carlton, is open to the

11rich and poor alike. The most entrepreneurial and the

12most innovative firms, the small fledgling ones are

13without means to mount private antitrust cases.

14Let me turn my attention for a few moments to

15innovation. Protecting competition does not mean

16stifling innovation, I don't believe. While there is an

17inevitable tension between the intellectual property law

18and the antitrust law, competition law cannot achieve

19its purpose if regulators and courts are preoccupied

20with a concern that remedies affecting some intellectual

21property rights will necessarily stifle innovation.

22That focus on IP, that is intellectual property,

23a legal concept, is misguided. The focus should be on

24true innovation, not patents and copyrights, public

25grants of a monopoly.


1Why is that the case? Well, first I think

2equating innovation to the accumulation of intellectual

3property is suspect, at least in the software world.

4The software patent approach in the United States is

5being broadly questioned, and that's the case for at

6least two or three different reasons.

7First of all, the software industry in

8particular survived for almost 20 years with very

9limited forms of software patents, not the broad range

10that we now see following State Street and other court


12Second, I would submit to you the relationship

13of software patents to innovation is suspect. I

14regularly review the academic literature in this area

15and I am aware of no convincing argument that software

16patents have unleashed -- and no empirical study --

17that they have unleashed and spurred additional


19Third, the news is regularly filled with stories

20of highly suspect software patents, patents that are not

21new and innovative, ones that are anticipated by prior

22art and ones that common sense tell us lack sufficient

23novelty to warrant 20 years of protection.

24Of course that shouldn't be surprising, there

25are well publicized challenges in the Patent & Trademark


1Office, there's no effective and searchable database on

2prior art for software. There's also serious challenges

3in retracting and retaining the kinds of experts that

4Renata talked about to actually evaluate what is seeking

5to be patented.

6I say that just to suggest that the innovation

7reflected in software patents is questionable at times.

8Therefore, giving, you know, complete deference to

9intellectual property in that context seems misguided.

10Even more important to this debate, as my

11opening remarks sought to illustrate, there are broad

12communities of collaboration that are massively

13innovative. Please note that their style of

14collaboration is not readily or naturally susceptible to

15patent protection, given the open and collaborative

16nature of their exchanges.

17Thus, innovation of the firm is not the only or

18even the most effective form of innovation to be

19considered or protected when facing the market

20disruptive effects of monopolists. Powerful new

21innovation paradigms are upon us now and they're growing

22and they need to be considered and measured in balance.

23But even if we were to assume that the firm is

24the epicenter of innovation, the smallest and perhaps

25most innovative are without the means to challenge the


1innovation of the monopolist that is purported to be

2reflected in intellectual property. The combination of

3suspect software patent quality and the disparity of the

4cost to acquire a patent versus the cost to defend

5against it skew IP protection in favor of larger

6enterprises with market power.

7Cost of acquiring a patent, let's say, is

8$25,000 to $35,000. It absolutely pales in contrast to

9the cost of a proper infringement defense. That is

10variously $3 to $5 to $7 million, and by all accounts is

11growing at present.

12Moreover, the monopolist can disrupt the

13business of smaller competitors merely by suggesting to

14consumers that its IP is infringed, without any proof

15whatsoever. If you consider Steven Bommer's recent

16statements that the users of Linux have an undisclosed

17off balance sheet liability to Microsoft, which were

18offered without any substantiation whatsoever. And the

19SCO litigation that is ongoing I think offers some

20interesting and vicarious variance on the same theme,

21which I would also be happy to comment on in the

22question and answer period.

23Keeping on the intellectual property theme, an

24effective remedy needs to prevent the extension of

25market power. A company who has acquired market power


1through anticompetitive conduct shall not be permitted

2to be able to hide behind intellectual property

3protection to reinforce and extend its market power. I

4think there is an interesting lesson in history on this

5that deals with data formats.

6In particular, I would like to contrast how

7Microsoft came to compete in word processing, versus how

8it now competes. The background is as follows:

9Software products manipulate and ultimately store

10customer data after that manipulation. To the extent

11this data is then placed into storage formats, that are

12claimed as either proprietary or protected by

13intellectual property of the software vendor, then the

14ability of a competing product to make effective use of

15the stored customer data and break into and compete in

16that market, which is likely reinforced by very strong

17network effects, can be precluded.

18Take, for example, Microsoft's word processor

19competition against the then-important market position

20of the WordPerfect product in the 1980s. Because the

21data format's inability to represent the data with

22substantial fidelity was possible, Microsoft could

23compete at the enterprise level by saying, give me a try

24in parallel with WordPerfect. If I do better, then

25incur the cost of switching out your old technology and


1taking on our technology.

2In contrast today, I would submit to you the

3formats of Microsoft alphus data have been and are

4increasingly being obscured by Microsoft and cannot be

5presented, that is the data cannot be presented with

6true fidelity by any competitor, like OpenOffice, which

7thereby extends the time of their dominant position and

8permits extension of power into adjacent markets.

9It is the case that Red Hat cannot effectively

10compete with open source personal productivity

11applications, like word processors and other things, at

12the enterprise level against Microsoft, it can't get its

13foot in the door. If a client wants to give someone a

14try and you can't render their existing data in a

15meaningful fashion, that prevents anyone from entering

16into that market, I would submit to you, or doing so

17easily, anyway.

18Microsoft controls, I would submit to you, a

19facility of competition through the extension of IP and

20proprietary formats that is needed to meaningfully

21render and manipulate customer data. I have no doubt

22that's why you're seeing states like Massachusetts

23aggressively consider the open document format, a truly

24open standard in format in its procurement processes.

25The mono type litigation of Red Hat is another


1example that illustrates that that I would be happy to

2comment on later.

3In summary, I guess I would say that innovation

4does not equate to intellectual property, and therefore

5greater focus on preserving and promoting true

6innovation in the marketplace is warranted. Further,

7there are numerous ways in which the use and assertion

8of intellectual property rights can be a pretext that

9chills competition and extends monopoly power.

10Thank you.


12MR. HILLEBOE: Thank you very much, Michael, for

13that, and I think we will take about a ten-minute break


15(Whereupon, there was a recess in the


17MR. HILLEBOE: Thank you, everyone. William

18Page is a Marshall M. Criser eminent scholar at the

19University of Florida Levin College of Law and he is

20also an alumnus of the Antitrust Division, where he

21served as a trial attorney in the 1970s.


23MR. PAGE: Thank you. Rather than speak in

24generalities about Section 2 remedies in high-tech

25markets, I want to zero in on one highly technical and


1seemingly obscure provision in the final judgments in

2the government's Microsoft case that has turned out to

3be the most difficult and the most problematic in its


5The provision requires Microsoft to license to

6software developers communications protocols that

7Microsoft uses in its Windows Client operating systems

8to interoperate with Microsoft server operating systems,

9either in corporate networks or over the Internet.

10Communications protocols are the rules for transmitting

11information between different devices.

12So, in a computer network, the protocols allow a

13user of a client computer, for example, to store

14information on a network drive or send an email or

15display a web page, among many other things.

16This sort of interoperation is relatively easy

17when the client computer's operating system and the

18server operating system share a common base in code.

19It's like they speak the same language, so they can

20interoperate easily.

21Where the client computer, usually a Windows

22client, has to interoperate with servers from other

23vendors, then the problem with interoperability becomes

24much more difficult, but there are ways of solving them.

25There are recognized ways of solving them. Some involve


1installing a client on Windows that would allow

2interoperation with the non-Windows server and

3applications running on it.

4There are also standard protocols that are

5available and supported in Windows. This provision

6requires another way of assuring interoperation, that is

7requires Microsoft to disclose its proprietary

8protocols, to license them to software developers so

9that they can interoperate. The near-term goal would be

10for them to be able to write programs that will

11interoperate as well with Windows clients as

12applications running on Microsoft servers.

13The long-term goal is to allow -- is to preserve

14in this network context the so-called middleware threat

15that was the focus of the government case. The

16middleware applications running on servers, the concern

17is, may eventually evolve into platforms that could

18rival the Windows desktop and thereby erode the

19application's barrier to entry. Essentially the theory

20of the government case.

21In spite of its apparent obscurity, this

22provision has been given an unusual amount of importance

23by the District Court enforcing the Microsoft judgment.

24She's referred to it as the most forward looking

25provision in the final judgments and as necessary to


1assure that the other provisions don't become

2prematurely obsolete. It's now being implemented by the

3two sets of plaintiffs in the Microsoft litigation, the

4Antitrust Division and the nine settling states, and

5also by the group of non-settling plaintiffs who were

6awarded essentially the same relief, but there are

7different enforcement mechanisms.

8There's the technical committee that Renata

9referred to in the Antitrust Divisions's consent decree

10and there's a technical consultant to the non-settling

11states under their decree, but they're coordinating

12their enforcement efforts. Both of these judgments went

13into effect in 2002.

14And the plaintiffs in both cases and Microsoft

15has been filing status reports every two months about

16the enforcement of both of the judgments, and I have

17studied these reports with the help of a research

18assistant, who was also a software developer and a

19management consultant, and so he has been sort of my

20technical consultant. He provided all of the technical

21expertise in this study, because I certainly claim none.

22 The enforcement of this provision, this one

23provision in these judgments has dominated these

24reports, particularly in recent years. It by far

25occupies most of the reports and certainly most of the


1time of the technical committee. And I'll argue that

2this provision has not accomplished its purpose, and

3that we can draw some lessons from that experience.

4So, I want to first describe what I take to be

5the principles of Section 2 remedies, I'll then suggest

6that most of the provisions in the Microsoft judgments

7adhere to these principles, but that this provision, the

8protocol licensing provision, departs from the

9principles and that is part of the reason why it has not

10been successful.

11I'll describe briefly how it has been

12implemented and then in the end I'll try to draw some

13lessons. And incidentally, this is a very brief summary

14of a much longer article which I hope to post on SSRN


16The goals of Section 2 remedies should be to

17restore competitive conditions that would have existed

18but for the illegal conduct. They should not be to try

19to restore or to create some sort of ideal competitive

20condition or to supervise market outcomes. I take the

21primary antitrust remedy to be deterrence, through fines

22and covered damages. If deterrence can be effective, if

23an optimal penalty can be imposed, that's always going

24to be preferable to having an administrative structure

25imposing remedies. It's simply the direct costs of


1imposing those remedies will be -- will impose a greater

2cost than effective deterrence.

3Assuming that some sort of injunctive relief is

4required, I would suggest that injunctions should be

5limited to preventing reoccurrence of proven

6anticompetitive behavior. The Sherman Act, unlike

7sector-specific regulation, I believe reflects the

8assumption that if specific impediments to competition

9are removed, then private contracting within the market

10will lead to the efficient outcome. And if that would

11not be the case, then that would argue that the market

12should be regulated.

13Beyond that, I would suggest that injunctions

14are problematic. First, divestiture, at least in the

15case of a unitary company, should be a last resort,

16primarily appropriate to dissolve recent combinations.

17Regulatory decrees also, as many have observed, should

18be avoided. As the Supreme Court said in Trinko, they

19require antitrust courts to act as central planners,

20identify improper price policy and other terms of

21dealing in roles for which they are well suited.

22Most of the Microsoft final judgment provisions

23reflect these principles. They do not require any form

24of divestiture, and most provisions respond more or less

25directly to the liability holdings in the case that were


1affirmed by the D.C. Circuit in 2001, prohibiting

2retaliation against computer manufacturers for promoting

3rival software, requiring uniform licensing terms,

4giving computer manufacturers the flexibility to remove

5the visible means of access to Microsoft middleware

6products and so forth.

7The protocol licensing provision does not

8respond directly to any illegal conduct. Server-based

9applications were mentioned in the findings of fact,

10only to exclude them from the market.

11Interoperability in networks was not an issue in

12the case, and in fact developing and refusing to license

13incompatible proprietary software was not held illegal,

14in fact, it was specifically held to be legal, if

15nothing more than that were shown.

16So, where did this come from? The idea for this

17provision actually arose, according to Ken Alletta's

18book on the Microsoft litigation, after the findings of

19fact had been issued. In other words, after the record

20was closed in the case. The feeling was that Microsoft

21essentially was not going to continue the conduct that

22was actually the subject of the litigation, the browser

23wars were over, Microsoft had already stopped the

24discriminatory pricing, it had gotten rid of the

25exclusive terms in its contracts, so we needed to be


1more forward looking and what was forward was this

2network environment.

3The fear was that in this -- you've got to, you

4know, as the computer market moved toward networks, both

5local corporate networks and the Internet, it was

6necessary to assure that Microsoft would not

7discriminate in allowing rivals to interoperate with the

8dominant Windows client.

9And, so, various proposals for various

10interfaces by Microsoft were made. After the original

11judgment was reversed, of course the Antitrust Division

12reached an agreement with Microsoft on the consent

13decree and it included a version of this. The protocol

14licensing provision, which essentially we now have, in

15both that consent decree and in this -- the states'


17Judge Kollar-Kotelly approved this provision,

18even though she recognized that the government was not

19strictly entitled to it, because it was not responsive

20to proven illegality, and she also recognized that there

21were these other ways in networks of achieving

22interoperability besides requiring Microsoft to license

23its proprietary protocols.

24Nevertheless, she found that -- and here's the

25key language, it's closely connected to the theory of


1liability in this case, and furthers efforts to prevent

2future monopolization.

3So, under this program, Microsoft has developed

4the Microsoft communications protocol program, which is

5an extension of its Microsoft developers network, and

6under this program, it offers a license to these

7protocols, and technical documentation. In the initial

8response in August 2002, actually before the consent

9decree was approved, but nine months after it was

10originally agreed to by the parties, Microsoft produced

115,000 pages of technical information, documentation, on

12the protocols, which it reported with a product of the

13work of five technical writers working essentially

14full-time for nine months.

15By July 2003, however, eight months after the

16entry of the final judgments, only four developers had

17licensed these protocols. And Judge Kollar-Kotelly told

18the parties in a status conference, this is reported in

19the report, that she was very, very concerned that

20nobody was taking these licenses. And both Microsoft

21and the government responded to this by various efforts

22to promote them. Microsoft took out ads, they

23evangelized these protocols, but with very little

24success. And finally the government conducted a survey

25of developers asking them why aren't you licensing this


1material, and they gave a list of reasons, some of which

2focused on the license itself, said it was way too

3complicated, it was pages of technical terms, and they

4were too expensive, the technical documentation was

5insufficient, the royalty was too high, whatever. But

6some said, we just don't need them for our development


8All of these, except that last one, were

9addressed over the next three years. The license term

10has been extended, the limitations in it have been

11relaxed, and simplified, the royalties have been

12reduced, many of the open standard protocols that

13Microsoft supports have been made available under the

14royalty free license. Microsoft has made its source

15code available to licensees.

16Now, to become a licensee, you need to show you

17have a legitimate purpose. So, you can't go and ask to

18see the source code, but if you are a licensee and you

19can show that you have need for it, under the license,

20then they'll show it to you and they'll actually provide

21support to show you how to use it. It's also provided

22500 hours of free premier technical support, it's

23provided a dedicated account manager, it's provided

24three-day, what they call plug fests, where you can

25bring your product and test it and Microsoft engineers


1will work with you to try to make sure it interoperates

2well with Windows. It's created an interoperability

3lab, and I should mention, when we had the first plug

4fest, only two licensees signed up for it, no one has so

5far signed up for the interoperability lab.

6So, over the years, what's most dramatic about

7these status reports is the accounts of how Microsoft

8and the technical committee have tried to improve the

9technical documentation of the protocols.

10In July 2004, the technical committee and

11Microsoft agreed on a 40-page specification that the

12documentation was supposed to meet. And the technical

13committee undertook to develop what it calls prototype

14implementations of each protocol. There are about 100

15and 120 protocols, and in order to assure that the

16documentation of them was sufficient, the technical

17committee has undertaken to try to actually write a

18little application using the protocol.

19And, so, if they could do that, then that would

20show that the documentation, it could actually be put

21into effect by the developer. Where they run into

22problems, if they ran into problems, they treated that

23as an issue, and they reported that to Microsoft as a

24bug to be addressed, and depending on its importance,

25they gave them seven days or, you know, longer time


1limits to respond to it.

2And this was the approach for about a year, but

3by early 2006, the technical committee had reported to

4Microsoft about a thousand of these issues, and only

5about 300 of them -- 300 of them had been resolved, and

6in May, this is about a year ago, the plaintiffs

7reported to the judge that the project had reached what

8it called a watershed, and at that point, someone who I

9take to be a strong personality, Robert Muglia, who is

10the senior vice president of Microsoft and formerly was

11the head of server division, reviewed this program and

12said that this process of trying to respond to bugs one

13by one, as they're reported by the technical committee,

14was just not working, and that we would need to start

15from scratch and rewrite all of the technical


17And, so, last summer, incidentally, it was at

18this point that the technical committee made contact

19with the European Commission's monitoring trustee, which

20is also administering an order to Microsoft to disclose

21protocols, and in connection with those communications

22had with Microsoft, agreed on a new overarching

23specification. This is now the third standard that will

24be used to judge the documentation.

25And Microsoft was given a new set of milestones,


1time tables, to complete the project. At this point, it

2was clear that the decrees were due to expire in the

3fall, and it was pretty clear that that was not going to

4be enough time to do all of this, and so that's when the

5parties agreed to extend the term of the judgment for up

6to five years.

7Meanwhile, Microsoft has suspended royalty

8payments entirely for its licensees, until the

9documentation is deemed to be sufficient, and the

10technical committee has continued to develop these

11protocol implementations, and interestingly, Microsoft

12has also undertaken to do something similar, developing

13what they call test suites, which it's one of the

14practices of software developers when they're working on

15an application, they come up with suites of testing

16applications to see if they work, and Microsoft has

17undertaken sort of a parallel or duplicate testing


19And in this most recent status report, which was

20issued earlier this month, the plaintiffs reported that

21although they've had some questions about Microsoft --

22apparently Microsoft discovered some new protocols that

23they hadn't identified before, they said that this new

24documentation is looking better, although significant

25additional work needed to be done.


1So, Microsoft now has been -- remember the first

2project, it had a few technical writers working for a

3certain number of months to produce these 5,000 pages.

4They now have 313 employees working on this project.

5And the technical committee also has increased its staff

6to 40 engineers, and they now have offices both in

7Redmond, Washington and in Silicon Valley.

8The bottom line, as of this month's status

9report, of the thousands of developers writing

10applications for servers, for server operators, to run

11on server operating systems, only 27 firms have taken

12the royalty-based license, and all but four of these,

13but for very specific purposes, like media streaming or

14data storage or security, the proxy firewall segment.

15So, and of those 27, only 14 are producing any products.

16And none of these products seems likely to have any

17potential as a platform.

18So, what are the lessons from this experience?

19The original rationale for this project was to preserve

20the middleware threat to the Microsoft monopoly in the

21network environment. If so, at least so far, the

22project has not succeeded, because it's attracted very

23few licensees, despite these enormous efforts, and I

24think quite admirable, and impressive efforts on both



1What this suggests to me is that the primary

2reason why we're not seeing more licensees is that

3licensing Microsoft's proprietary protocols is generally

4not necessary for these firms to develop software

5applications to run on non-Microsoft servers. They can

6use the standard protocols that Microsoft supports in

7Windows, or they can develop their own windows client

8which then could run on the Windows client and

9communicate directly through Microsoft's application

10programming interfaces.

11So, to boil it down, what I would say is that

12what this remedy does is to treat the Microsoft

13protocols as if they were an essential facility, except

14that they're not essential. There are other ways of

15accomplishing the same thing.

16So, what I would take to be the two primary

17lessons are first, injunctive relief, particularly in

18high technology markets, should be limited to responding

19to a proven need, and the most important proven need is

20to -- is to interdict and remove anticompetitive

21practices, proven anticompetitive practices.

22 So, if Microsoft is proven to have engaged in

23practices that violate the antitrust laws, those should

24be enjoined. But as we've seen, the protocol licensing

25provision did not respond to a proven violation, and did


1not even address technology -- and it addressed

2technologies that were not even the focus of the

3liability phase.

4During the remedial proceedings, there was a

5record developed on network computing and there was

6evidence introduced of various so-called bad acts, as

7Judge Kollar-Kotelly characterized them, but she treated

8them as being essentially irrelevant, because they had

9not been shown to be anticompetitive, or at least if

10they were anticompetitive, they may have had

11pro-competitive justifications that had not been


13The second, under this heading of only

14responding to a proven need, I don't want to rule out

15the possibility that forward-looking or fencing in kinds

16of provisions may be necessary, but if they are, then I

17think there should be -- there should be a record built

18to support the need for them. And I think in this case,

19for example, we know that the government at one point

20actually surveyed software developers to see what their

21needs were in this area.

22I'm not sure what was done during the

23negotiation of the consent decree, but perhaps more in

24that direction could have been done to find out

25precisely what was needed to ensure adequate



2And also I would just add that the Court of

3Appeals in the 2001 decision cautioned that remedies

4should be proportional to the strength of the proof that

5Microsoft's illegal actions actually reduced

6competition, and that was why the Court of Appeals said

7that divestiture was probably not going to be an

8appropriate remedy, because as they put it, the harm to

9competition for Microsoft's actions, in other words,

10whether they had actually prevented Netscape's browser

11or Java from evolving into a rival platform, that was

12established by only -- as they put it -- by inference,

13in other words, there was no evidence that that actually

14would have happened. And where you have that relatively

15weak evidence of likely anticompetitive effect, then you

16need more evidence to support more Draconian remedies.

17And divestiture is certainly that, but I also

18think regulatory relief is also a Draconian remedy, and

19that brings me to my second lesson, and that is to avoid

20regulatory decrees, especially in high technology

21markets. And this was recognized, Judge Kolar-Kotelly

22rejected one principle during the remedial proceedings,

23on the grounds that it would result in too regulatory of

24a decree.

25Well, the protocol licensing has become highly


1regulatory and direct government supervision of price

2and other terms of dealing and especially quality.

3Direct government supervision of quality that's being

4produced. And the device of the technical committee

5certainly has provided a high level of expertise, but in

6effect, what its created is a regulatory body, and I'm

7not sure that the structure of the technical committee

8and its relationship to the plaintiffs and the court

9establishes an effective regulatory agency.

10So, just to conclude, if in the future cases

11have these characteristics, those should be treated as

12warning signs, and addressed in the -- in the relief.

13And with that I'll sit down.


15MR. HILLEBOE: Thank you, Bill. Marina Lao is a

16professor of law at Seton Hall Law School. She

17currently serves on the executive board of the section

18on antitrust law of the American Association of Law

19Schools, and she's an alumna of the Antitrust Division,

20where she was a trial attorney. She has published

21numerous articles on antitrust law and trade regulation,

22and somewhat surprisingly on this high-tech panel, she

23is the only speaker with slides.


25MS. LAO: I guess it's even more surprising


1given that I am usually the least high-tech person on

2the panel. Thank you very much for inviting me and I'm

3happy to have the opportunity to participate in this


5I agree with a number of the speakers who have

6gone before me who have said that remedies are often

7treated as an after thought. Unfortunately, that's not

8a very good idea, because success in proving liability

9often does not translate into success in remedying the

10anticompetitive situation, and so it's often best to

11work your vision of remedy into the case development

12much earlier on.

13What I'm going to do, since I'm bringing up the

14rear, is to try not to overlap too much with what has

15been said; I'm going to focus on three main points in my

16comments and I will be skipping over some of the slides.

17First, where network effects are substantial in

18the industry that's affected by Section 2 violation, I

19probably differ from Bill, in that I think that there's

20a need for broader rather than narrower remedies for

21some of the reasons that I'll talk about later.

22Second, again, I guess on this issue I differ a

23bit from Bill as well. I'm going to talk about the

24importance of forward-looking remedies. I would call

25them affirmative remedies that reduce rivals' costs and


1some of the problems in crafting them. I do agree that

2tayloring these remedies to the problem is a bit

3 difficult.

4And lastly, I'm going to discuss whether there's

5any value in bringing Section 2 enforcement action if

6there is no effective judicial remedy. My conclusion is

7that there is deterrent value to bringing an enforcement

8action, even if it is irremediable, so to speak.

9Let me start with a few words about the ongoing

10debate among antitrust commentators on the application

11of antitrust in the dynamic high technology markets.

12The question that is often raised is: Do we need more

13rigorous antitrust enforcement or do we need a more

14hands-off approach? Those who say that less

15intervention is necessary generally argue that because

16there is rapid innovation, product cycles are short, and

17so dominance is fleeting. And there are continuous

18opportunities for fringe firms to overtake the

19incumbent. The Microsofts of the world will have to

20constantly innovate or they're going to be left in the


22And so for that reason, there's really not that

23much of a need for antitrust intervention in order for

24markets to remain robust. In fact, too much antitrust

25intervention could stifle innovation and competition.


1While there's obviously some truth to that

2argument, I think the Microsoft case itself tells us

3that rapid technological change can cut the other way,

4especially when you have substantial network effects

5which tend to operate as significant barriers to entry.

6If these are substantial network barriers to entry, a

7clearly dominant firm can much more easily exclude even

8superior technologies, up to only a certain point, of

9course, if it can ensure that the rival technologies

10remain incompatible.

11And, the dominant firm can also control research

12avenues, up to a certain point. What's more, even

13without any antitrust violations, there are natural

14benefits, that flow from network effects of those

15natural benefits, I think dominant firms can more easily

16use tying and other exclusionary strategies to preserve

17their dominance and to exclude competitors


19So, my conclusion is that antitrust intervention

20is not only not redundant, but there is perhaps an even

21stronger need for it when you have markets with strong

22network effects.

23With respect to remedies, there's a similar

24ongoing debate among commentators. There are those who

25say that with fast moving technologies, you need milder


1remedies, remedies that are less severe, because of

2several reasons. First, there is the self correcting

3market rationale, which postulates that the market is

4going to correct itself much faster than antitrust

5intervention can correct it. Second, advocates of mild

6remedies warn of the possibility of unintended

7consequences, that is where market conditions in the

8future are uncertain, one may not know what to prohibit

9and what not to prohibit, and so the remedies adopted

10today may not be sensible a few years hence.

11And, so, they argue it is probably safer to

12adopt milder forms of remedy in order to lessen the risk

13of chilling innovation and competition from the dominant


15First of all, I happen to think that high-tech

16markets do not that easily, at least self correct, not

17if network externalities exist, because by definition, a

18self correcting market, requires innovation and new

19entry, but network effects raise entry barriers and

20reduce access to the network.

21Obviously easy entry markets are not going to

22easily self correct.

23As to the argument that uncertainty about future

24market conditions means that we should perhaps take a

25more hands-off approach and apply the mildest remedy


1possible, I also do not completely agree with that. I

2think that if market conditions are uncertain, we have

3to exercise more care in defining the future boundaries

4of the relevant market, and in identifying the

5participants in this future market, and in crafting the


7But we should not overlook the danger of doing

8too little too late, which carries its own risk as well.

9Another possible solution to the uncertain market

10condition problem is to have a continuing jurisdiction

11clause in the remedial order, which I know is not a

12common practice. With a continuing jurisdiction clause

13either party can go back to the court for modification

14if it turns out that the remedies agreed upon do not

15work because of changing market conditions.

16As to the "potential chilling effects" argument,

17it's often said by advocates of milder remedies that

18compulsory licenses of IP rights and other affirmative

19remedies tend to chill innovation on the part of the

20dominant firm, that's basically one of the points

21Justice Scalia made in Trinko.

22What is often lost in this discussion, though,

23is that competition and innovation from fringe firms are

24also very important, and if remedies for an antitrust

25violation are insufficient, innovation and competition


1from fringe firms could be chilled. The AT&T

2divestiture experience is very instructive. Few would

3disagree that the structural remedy in the AT&T case

4unleashed innovation from smaller telecommunications

5firms on an unprecedented scale, which enhanced consumer


7Another point that we should not lose sight of

8is that with high technology markets, it's extremely

9difficult to resuscitate a competitor, after the

10competitor has been crushed. The convergence of factors

11that produced a competitive challenge before it was

12anticompetitively excluded, may never re-appear, not in

13the same fashion, anyway.

14The factors together call for a solution that is

15less hands-off.

16They also lead me to conclude that narrowly

17focusing the remedy on the specific conduct found to be

18unlawful, will not return competition to the status quo;

19thus drafting or crafting forward-looking remedies is

20quite important.

21Of course I do realize that forward-looking

22remedies have to be carefully tailored.

23The problem one faces in crafting

24forward-looking remedies is that you have to understand

25the market. You've got to analyze the likely evolution


1of the market, predict which way the market is headed,

2the innovations will likely emerge, what will be the

3next generation of innovations, and how these

4innovations might change the path of the market.

5Unless you have a pretty good grip on these

6issues, it's very difficult to predict what remedial

7actions would work to break down entry barriers and

8facilitate competition, and what would not.

9If we do not know what is going to work, then we

10risk adopting an injunction that constrains conduct that

11no longer needs to be constrained, but does not

12constrain conduct that needs to be constrained. Perhaps

13the prime example of this is the first Microsoft consent

14decree, which prohibited Microsoft from "per processor"

15licensing which it had engaged in. But by the time of

16the decree, Microsoft no longer needed to engage in that

17strategy, because its competitors in the operating

18systems market were already defunct and the prohibition

19accomplished nothing.

20Another problem, I think, that is rather

21peculiar to high-tech markets is having to anticipate

22how dominant firms might circumvent the judicial

23constraints imposed and still achieve their

24anticompetitive ends, and then block these alternative

25paths in the in the decree as well. Fast-changing


1markets tend to be pretty malleable, thus giving the

2dominant firm myriad ways to achieve its anticompetitive


4To understand how Microsoft or any dominant firm

5might sidestep an injunction and still achieve its end,

6we need to know what the possible alternative strategies

7are. But dominant firms generally have an information

8asymmetries advantage over the government that's quite


10That is, the government knows much less than the

11dominant firm about what the potential new innovations

12and the possible alternative strategies to achieving the

13anticompetitive objective are. So how can the

14government overcome the information asymmetries problem?

15I think the simplest solution is to just enlist the

16assistance of the dominant firm's competitors or

17potential competitors, who probably are in a much better

18 position than any outsider, including government

19enforcers, to know about the industry, to know what

20remedies might work and what might not work, and what is

21the innovation trend, et cetera.

22Oftentimes, when this is mentioned as a possible

23solution, you hear the argument that, well, then, the

24department or agency might be subject to capture. I

25think that simply relying on competitors to educate


1government enforcers on the market is not equivalent to

2capture, and is also entirely consistent with the

3principle that we should protect competition and not


5Let me turn, briefly, to the importance of

6implementing creative affirmative obligations. The

7problem with conduct remedies and I'm not discussing

8structural remedies at all, because it's been discussed

9in detail already is that generally speaking, if the

10dominant firm has already successfully excluded its

11competitor and potential competitors, simply stopping

12the conduct and preventing its recurrence is not going

13to be enough to restore competition. That is because

14stopping the exclusionary conduct will not unravel the

15dominant firm's accumulated market power.

16Instead, what would be helpful would be to

17impose affirmative duties on the dominant firm. I call

18it lowering rivals' cost as opposed to raising rivals'

19cost. The Post-Chicago school has said that dominant

20firms can exclude competition anticompetitively by

21engaging in strategies that raise rivals' costs. For

22 remedy purposes, we need to go a little bit beyond

23prohibiting acts that raise rivals' costs; we need to

24impose some obligation on the part of the dominant firm

25to reduce rivals' costs.


1Some affirmative duties are pretty well

2established in antitrust jurisprudence, and are not very


4One is compulsory licensing of IP rights, with

5or without royalty fees. The case that springs to mind

6involving forced licensing is the Xerox case brought by

7the FTC in 1975. The FTC in that case imposed a

8compulsory licensing obligation on Xerox. In Microsoft,

9as Bill just mentioned, there was also a compulsory

10disclosure of information component in the decree as

11well Microsoft was required to disclose its APIs and

12also its communications protocol.

13Another typical affirmative duty is the

14obligation to sell to all customers on a

15non-discriminatory basis, and that was part of the order

16in the Ninth Circuit Kodak case.

17The third example that I have listed on the

18slide is also not terribly controversial, and that is

19unbundling. For example, in United Shoe, the defendant

20was required to unbundle its machinery and its repair


22The fourth category is probably the most

23controversial, and that is requiring the defendant to

24create products to comply with industry standards and

25not just with its own proprietary standard. This is the


1remedy that the State of Massachusetts asked the court

2to impose in Microsoft, in the case that Massachusetts

3continued to pursue after Microsoft settled with the

4DOJ. Incidentally, the District Court did not grant

5that request.

6I was going to talk about the Korean Microsoft

7case, which I found very interesting, but I don't think

8I will have time for that, so let me just end with two

9points. I have alluded to the first point earlier, and

10that is the usefulness of a continuing jurisdiction

11clause in a remedial order. Perhaps those of you who

12are still in government can enlighten me as to why the

13government does not seem to want to include these

14jurisdiction clauses in their remedies anymore, back in

15the 1950s and 1960s.

16Having a continuing jurisdiction clause is

17helpful in a dynamic high technology market because it

18allows the court to assess the success of the remedy,

19and to assess future development. The purpose of

20assessment is not so much to ensure that strict

21compliance with the decree itself is occurring, although

22that is very important too, but to ensure that there's

23movement toward the ultimate objective set by the court.

24I think Professor Hovenkamp in one of his articles

25suggested that perhaps a continuing jurisdiction clause


1would be very, very helpful, because it would allow the

2court to look at whether the decree has been successful

3or not. I think of success as not simply whether the

4defendant has complied with the specific terms of the

5decree, although that is obviously a part of it, but

6whether the decree is doing anything at all to make the

7market more competitive.

8One final note, and that is I think there is

9value to Section 2 enforcement even if no effective

10judicially-imposed remedy is available, on two

11conditions: if there is really an egregious violation

12of the antitrust laws, and if there is substantial harm

13to consumer welfare. The reason enforcement is

14important even if the violation is judicially

15irremediable is that I think the defendants would

16moderate their behavior somewhat, simply because

17litigation has been brought. And they may even

18voluntarily discontinue some of the challenged


20I think it is commonly acknowledged and commonly

21known that Microsoft relaxed enforcement of its

22exclusive dealing contracts with the OEMs during the

23process of the litigation. And, as far as I can tell,

24Microsoft does not seem to be using against the type of

25tactics that it had engaged in against Netscape and



2I am not a very tech savvy person, but it would

3seem to me that there must be strategies similar to the

4kinds that Microsoft had employed against Netscape and

5Java, and yet they have not engaged in them against

6Google. Of course we will never know how much of their

7reticence is the result of the deterrent effect of the

8government's enforcement action.

9Finally, for public policy reasons the

10government should not just step back and say, well,

11there is no effective remedy, so what's the point of

12bringing a lawsuit? If consumer harm is substantial,

13and if the act is egregious, I think it is bad policy to

14take no action because it sends a wrong signal. Taking

15enforcement action can deter the Microsofts of the

16world. Who knows, it might deter Google at some point.

17With that, I hope I haven't repeated too much of

18what has been said.


20MR. HILLEBOE: Thanks, Marina. This is the

21portion of the hearing where we allow each of the

22speakers to comment with what they've heard before, and

23I'll start with Howard, please.

24MR. SHELANSKI: Well, I thought a number of the

25presentations raised provocative, extremely provocative



2Let me start with Michael Cunningham's comments

3about the problems that companies like Red Hat still

4face, even in the wake of the decree.

5I found his comments extremely interesting,

6because they suggested both at the same time a need to

7be very aggressive against anticompetitive behavior,

8because it has lasting effects, but also to raise real

9questions about what can be done about those effects,

10and if one were to translate that into a recommendation

11about remedies, it would be hard to know -- it would be

12hard to know exactly what the result is.

13On one hand, it might be taken to suggest that

14we need very aggressive kinds of remedies of the kinds

15that Professor Lao just suggested, with continuing

16supervision, and more creative solutions to lowering

17rivals' costs.

18On the other hand, I think that Bill Page raised

19very good reservations that I share about pursuing that

20kind of aggressive oversight.

21So, where I come out from Michael's comments is

22to say that we do need to pursue these cases. We need

23to pursue these cases to understand what kind of conduct

24is likely to lead down the road to problems that are

25very hard to uproot. And in concert, I think, with what


1Professor Lao just suggested, even if we're not sure

2that the remedy will work, pursue the case so that next

3time around, we can uproot the conduct earlier and have

4a remedy that will be effective, but I think, Michael,

5you pointed to some really very difficult challenges.

6With regard to Renata Hesse's comments, I think

7I shared very, very much your point of view. I think

8you were a little bit more cautious about the likelihood

9of success of injunctive remedies, I thought you raised

10some very good points there, but I continue to think

11that particularly in the high-tech sector, injunctive

12remedies will take the form of a negative prohibition of

13thou shalt not are likely to be the most fruitful

14remedial avenue overall.

15Professor Page, I found that story fascinating,

16but I think the detail was extremely instructive, and

17very helpful. And I guess on one hand, I might be

18inclined to say, well, does that mean we shouldn't go

19deep into these kinds of continuing remedies; on the

20other hand, I might say, well, maybe this is very costly

21to Microsoft, with little benefit to competitors, but

22maybe costly to Microsoft in and of itself, isn't so


24But maybe costly to Microsoft in and of itself

25isn't so bad. Maybe it's a very back-handed form of


1disgorgement remedy through the front door.

2I say that partly tongue in cheek, because I

3don't know that they really notice that kind of spare

4change over there.


6MR. SHELANSKI: No, but it does raise some very

7serious questions about how even the most carefully

8wrought and technologically sophisticated attempt at an

9affirmative remedy can be very difficult, and that's a

10lesson that I take very much to heart. So, I've learned

11a lot from all of you. Thanks, very interesting.

12MR. HILLEBOE: Thank you very much, Howard.


14MS. HESSE: Sure. I think -- I don't think the

15mic' is on. I think the thing that I took away from

16everyone's comments was very similar to what Howard just

17said, was that there seems to be a sort of inherent

18conflict between these two views of both the difficulty

19and in some cases I think impossibility of imposing

20remedies in technology markets, and yet at the same time

21the view that we really need to keep trying, even though

22we're not likely to be successful.

23And I haven't come up with a good way of

24bringing those two points of view together, other than

25to say that I think, you know, courts, and not in the


1antitrust context, but in lots of other contexts, have

2over the years dealt with a lot of very difficult

3issues, which people, I think, over time, have thought,

4well, you know, how could a court ever figure out how

5to -- I'll use, you know, prison conditions litigation,

6which I think I talked about before, you know, school

7desegregation is another one.

8Difficult problems that are not within the core

9competency of either courts or lawyers, and everybody, I

10think, has thought that a social benefit derives from

11intervention in those areas, and at least an attempt to

12try to solve them in some way.

13And I don't really see technology markets as

14being different in any -- I mean, they're obviously

15different in terms of the substance that they deal with,

16but not different in terms of the importance of the

17issues that you're dealing with, in terms of the

18importance of markets to both not just America's

19economy, but the world economy, and to the every day

20consumer. I mean, these products and services are

21things that we all use on a daily basis, and spending

22time thinking about, A, whether or not the law is being

23violated in those areas, and B, if it is being violated,

24how can you do the very best job you can to try and

25solve the problem seems to me to be a worth while


1expenditure of not only government time, but also in

2some cases in private litigation time, too.

3Keep at it, I guess, is my final conclusion.

4MR. HILLEBOE: Michael, and also I would ask you

5to address your points of the speed and cost of

6antitrust litigation are duly noted. If you have any

7profound suggestions with respect to those or practical

8suggestions or any other type of suggestions.

9MR. ELIASBERG: Or those quick and speedy cases,

10I was very interested in that.

11MR. CUNNINGHAM: Right, profound thoughts

12probably won't be forthcoming, but I will try and offer

13a couple. I take a pretty simple approach as a business

14person. I have a difficult problem, I keep working on

15it and keep attacking it until I come up with a


17I think, you know, serious examination of the

18effects of the Microsoft remedies is worth while, but

19there is assuredly deterrent value. One part of the

20advice that I tell my client, which I didn't mention

21before, is that I believe it assuredly moderates

22behavior for us to have any participation and then for

23the case to be brought at all.

24Indeed, in the area of some of the protocols

25that have been licensed that Bill referred to, I deeply


1wonder whether Microsoft would have reached out to Red

2Hat and requested our assistance and consultation in

3producing a very, very simple protocol license that's

4one page, we'll never know the cause/effect of both the

5EU action and the U.S. action, but there's reason to

6think that some of that may moderate behavior.

7I think in the case of Bill's examination, also,

8I would just comment that continuing to look at those

9facts are important. For example, Bill pointed out that

10there are other ways to interoperate. Other ways to

11interoperate that are fundamentally disadvantaged is not

12interoperation. It doesn't work.

13The IT community, you know, competes on the

14speed, efficiency, and look and feel of interoperation.

15So, simply concluding that there may be other protocols

16out there that may have issued since the decree, at

17least some of them, may not be complete examination. I

18should point out, Bill was kind enough to provide me a

19draft of his entire paper, which I didn't have a chance

20to look at before, so if it's addressed in the paper, my


22I think that, you know, these are terribly hard

23problems to work on, and I just don't see where, without

24learning and gaining experience in how to better address

25conduct remedies, we're able to make effective inroads


1into some of these fast-moving markets.


3MR. PAGE: I just have a few kind of stray

4 comments. I was struck by Renata's point about focusing

5on a remedy early, and I agree that that is really

6critical, and I would suggest that particularly in a

7case that ends in a consent decree, before litigation,

8it's absolutely essential.

9What I -- part of the problem I saw in the

10Microsoft remedial issue was that the case lasted so

11long that it was a moving target to think about the

12remedy, you know, that at -- that by the time the case

13was over, the remedy that people wanted was different

14from the one they would have predicted early in the


16So, you know, particularly for cases that last

17longer than just a couple of years, it's particularly

18difficult to be sure the remedy from the outset and be

19building a factual basis for it.

20I think the point about avoiding mandatory kinds

21of remedies as opposed to prohibitory remedies is a

22valid one. I would just caution, though, that in the

23 Microsoft case, there was another mandatory remedy to

24reveal the APIs that Microsoft uses to interact with its

25middleware, between the Windows operating system and its


1middleware, and that one seems not to have caused that

2many problems. And I suspect that the reason for that

3is that Microsoft's whole business is marketing APIs,

4and documenting APIs. If they couldn't do that, they

5wouldn't be in business.

6So, that was a much more straightforward problem

7than marketing protocols, their own proprietary

8protocols, and I think that's, you know, perhaps that

9explains some of the difficulties that have been found

10in documenting that.

11So, not all mandatory types of relief will

12necessarily be as problematic as this one. On the issue

13of the technical committee, I want to combine this with

14the idea that the courts should retain jurisdiction, and

15periodically review the experience in enforcement. The

16technical committee I think is one institutional concern

17that I have about the technical committee, certainly

18they are quite expert. I know nothing about them

19individually, but certainly no one would challenge their

20technical capacity, but they were given a single task,

21and that was to assure that the documentation is first

22rate, flawless. And, you know, as Howard pointed out,

23who cares how much Microsoft pays, to do that, and so

24it's a very expensive process to meet that kind of



1On the other hand, I think at some point, the

2court should come back and ask the question, is this

3accomplishing as much as we could accomplish in other

4ways? In other words, the economic question is always

5compared to what? And particularly if we can

6preemptively think about these issues before they come

7up, but also, if we can think about them down the road,

8perhaps as an opportunity for mid-course corrections

9that could reduce costs and perhaps benefit the market


11Just finally, on the issue of whether high

12technology markets require or it's more appropriate to

13use remedies in them because of network effects, I would

14only caution that the literature on network effects

15doesn't exactly say that competition doesn't work in

16these markets. It doesn't necessarily say that network

17effects are bad, I mean, when you think about it,

18network effects are simply economies of scale on the

19demand side. In other words, they benefit consumers,

20and so the concern that they are simply a barrier to

21entry I think somewhat overstates the case.

22Markets converge on a single standard for

23reasons that are actually beneficial to consumers. It

24doesn't necessarily follow, then, that government

25intervention is necessary, and I would add to that the


1issue of compatibility is also not so simple, because

2markets characterized by network effects can sometimes

3compete very effectively with totally incompatible

4systems, as we observed in the video game console market

5where, you know, it's a constant leapfrog competition of

6totally incompatible systems of hardware and software.

7 And that is a very effective model for competition.

8So, it doesn't necessarily follow that we should

9be promoting interoperability in all circumstances.

10MR. HILLEBOE: Marina?

11MS. LAO: I actually only have a few comments.

12I think the presentations today highlight the

13difficulties involved. For instance, Bill's

14presentation focused on the problems that I had tried to

15shy away from, and that is there are major difficulties

16in using and implementing forward-looking remedies.

17And Michael's points, I think, drive home the

18need, for more active government intervention, because I

19think private Section 2 cases are extremely difficult to

20prove, especially since proving anticompetitive effects

21now often requires economic proof. When the violation

22involves technology that hasn't fully emerged yet, it's

23very difficult to show that there is actual

24anticompetitive effect. I pretty much agree with most

25of what Renata and Howard said.


1MR. HILLEBOE: Okay, thank you.

2Bill, just as a point of clarification, I think

3you had indicated that Microsoft was licensing its

4source code. Just to clarify that, I think you probably

5mean it's licensing portions of its source code that are

6associated with interoperability issues. Is that


8MR. PAGE: It's allowing licensees of the

9protocols access to the source code in order to help

10them use the protocols.

11MR. HILLEBOE: Right, but not the crown jewels,

12so to speak?

13MR. PAGE: No, they're not saying here's our

14source code, you can use it, you know, for whatever

15purpose, it's purely to assure -- there were some of the

16licensees, or prospective licensee who said that they

17really needed access to the source code, more than they

18needed the specification of the protocols. And I'm not

19enough of a geek to know why that would be, but this is

20in response to that.

21And interestingly, that is an important

22concession, I would say, on Microsoft's part, because

23that was one of the proposed remedial provisions that

24the non-settling states wanted to have added to the

25final judgment was to require Microsoft to disclose its


1source code for these purposes, and the court refused to

2order that.

3And, so, in this limited sort of disclosure, I

4think is an important concession.

5MR. HILLEBOE: And several folks have talked

6about technical committees, and I wanted to direct a

7question to Renata about that, since she's had a lot of

8experience with that. I was wondering, Renata, if you

9can offer us some insights with respect to setting up

10the technical committees, given that in a conduct

11remedy, when you're talking about high-tech markets, and

12given the lack of expertise of lawyers and the fact that

13we're not engineers, and it seems almost inevitable that

14you're going to have a technical committee, were there

15things that you may have changed from the way you did

16it? Also, are there any differences in the European

17monitoring trustee? Is that a different situation? And

18also your thoughts about having all the parties involved

19in terms of determining who the trustee or the committee

20should be, including the defendant?

21MS. HESSE: I'm looking back at Patty Brink, who

22spent a lot of time with me trying to figure out how to

23construct the technical committee, and truthfully, it

24was in terms of the formation of the company, it was

25like starting a new business. So, we had to work


1through all sorts of issues that you wouldn't ever

2anticipate, and we certainly didn't anticipate when we

3thought about the provision, including how do you set up

4a company so that it doesn't have tax liability, how do

5you hire employees, how are they paid, all of these

6things that none of us really knew how to do, and we

7spent a lot of time consulting with various people to

8figure that out.

9The more important pieces of it, though, I think

10really had to do with the selection of the technical

11committee members, and if you look at the comments and

12the response to the comments to the consent decree,

13there were a number of people who said, whoa, you know,

14Microsoft gets to pick and gets a role in picking at

15least one, so the DOJ and the states picked one,

16Microsoft picked one, and those two people picked the

17third, and, you know, that's just, you know, they're

18going to put one of their own people on there, and what

19good is that really going to do.

20And I think the interesting thing that happened

21was that we really did find three people who were not

22just technical experts, but also had been business

23people, so people who had started technical companies,

24and who really knew how to -- not only run the business

25that they had to run, but also what the business reality


1of the various technical issues that they were advising


3And as it turned out, they really formed a

4whole, and they worked a lot with Craig Hunt, who is the

5nonsettling states group, who is sitting out in the

6audience, also. And they have, you know, coalesced as

7an entity unto themselves and the Microsoft appointee

8plays no different role in -- the Microsoft selected

9person plays no different role than any of the other

10members. And I think that has been really a tremendous


12I think the things that one would go back and

13look at again are the provisions in section 4 of the

14final judgment, which is the technical committee one,

15that relate to what the technical committee can say

16publicly and do publicly. And this is always -- and

17that's a big difference between the monitor trustee in

18Europe, and the technical committee in the U.S.

19In the U.S., the technical committee is not

20allowed to make public statements without prior approval

21of anybody, and their work product can't go directly to

22the court. In terms of a compliance or enforcement

23effort. And I think there were good, reasonable reasons

24to do that, and I think in the end that's probably the

25right way to do it, but in Europe, that's not how


1they've done it. And so their monitoring trustee

2actually will testify at hearings about whether or not

3Microsoft is in compliance with the final judgment.

4And those are two very different roles, and I

5think it's important to think about when you're

6constructing something like this, which of those two

7roles you want the person to play. I think having them

8play both roles is pretty dicy.

9MR. HILLEBOE: And I know Bill from his comments

10expressed some skepticism about having a technical

11committee and having another regulatory body. I was

12wondering what the other speakers thought about having a

13technical committee, and if they don't like that idea,

14if they have some suggested alternatives to that.

15Howard, do you have any thoughts about that?

16MR. SHELANSKI: I mean, I think technical

17committees for the reasons that Bill outlined are likely

18to be extremely tricky, and so the only thing I have to

19add is probably what others have said.

20I think a technical committee should be reserved

21for circumstances in which we have a pretty clear idea

22of what needs to be accomplished, a pretty clear idea of

23the market demand for that outcome.

24MR. HILLEBOE: Michael, do you have some

25thoughts about that?


1MR. CUNNINGHAM: Yeah, I personally think that

2at least if there's going to be a conduct remedy, not

3having a technical committee would be a fatal flaw. The

4technology is simply too complex, too subtle and too

5fast moving to not have, you know, that advice.

6But turning back to some of Bill's observations,

7the fact that the technical committee had a thousand

8comments when they sought to implement the protocols,

9might suggest a massive failure to comply. And, you

10know, the fact that the technical committee ran into

11difficulties, maybe because it's difficult, which is

12partly true, may be difficult because people were not

13trying to comply in good faith. I don't know.

14MR. HILLEBOE: And Bill, did you have some

15alternatives to having this regulatory body?

16MR. PAGE: Just on this one last point, before I

17answer that, most of the status reports do indicate that

18the technical committee, or the plaintiffs, were not

19really questioning Microsoft's effort. I mean, there

20are occasionally comments where they're disturbed by

21this or they're disturbed by that, but in general, the

22tone is one of this is a huge job, and we're having

23problems accomplishing it and we're both trying in good

24faith to do it. That's in general what I thought from

25these reports.


1And I should just say that the reports are

2pitched at a certain level so that there's only so much

3understanding you can get from them. And maybe if they

4were any more technical, I wouldn't understand them at

5all, but I'm a little bit like a denizen of Plato's

6caves seeing the reflections of reality on the wall and

7the reality is really outside of the cave and I can't

8really tell for sure everything that's going on.

9But to some degree, that is the position of the

10court, and as Renata said, the technical committee is

11sealed off from the court, which means that its

12observations need to be mediated by the lawyers, who I

13suspect probably don't understand the technical issues

14much better than I do, and I think that's a problem.

15I mean, we have this technical body that does

16understand the issues from a technical point of view,

17but their antitrust significance has to be mediated by

18people who essentially don't. And I think that's a --

19that's a difficulty that perhaps wouldn't be the case if

20we had a more conventional administrative agency where

21expertise were, you know, the problems of addressing

22expertise and using it in decision-making were more

23formally, you know, implemented.

24MR. HILLEBOE: Marina, do you have any thoughts

25on this?


1(No response.)

2MR. HILLEBOE: Okay. You know, one of the

3outstanding features of these types of markets that we

4look for are the presence of network effects, and some

5people have discussed this, but I think it's important

6to cover this. Is there a consensus with respect to in

7markets where you have network effects, are those

8markets that tend toward monopoly or toward a

9winner-take-all or winner-take-most equilibrium, or some

10people have suggested that, or is that overly simplistic

11or is that a capricious argument. What are your

12thoughts on that, Howard?

13MR. SHELANSKI: Well, first let me say that I

14think that the markets that are truly likely to tip to

15monopoly are few. I think it's a fairly circumstance

16where a network market will precipitously tip to

17monopoly, but it can happen.

18Not all cases where network market tips to

19monopoly yield bad outcomes. First of all, those

20monopolies can be unstable. There's a fair amount of

21research that actually shows that network markets

22flip-flop more frequently under some conditions than is

23good for consumers. Because they're stuck with legacy

24technologies that don't migrate forward to the product

25of new innovator.


1So, I think that just because something is a

2network market doesn't mean that we need to worry about

3some kind of tragedy of tipping. But it -- it can

4happen. And then where it does happen, I think that the

5remedial problem is really a challenging one. The

6structural remedy can break up network effects,

7interoperability remedies can lead to the need for

8behavioral oversight, but also, we want to be careful, I

9think one of the commentators, it might have been Bill,

10pointed out, we don't necessarily want to mandate

11interoperability, even when recommending a network

12market, because new standards come into the market that

13could improve things for people and you don't want to

14eliminate the incentive to try to create the new network


16So, I think network monopolies can arise, one

17should not presume that they are too easily going to tip

18to monopoly, even though their demand side of positive

19externalities. We've seen cases where multiple systems

20exist, and where they do exist, I think the remedy needs

21to be thought about very carefully. Structural remedies

22can be risky, interoperability is not always worth


24So, in those markets, it would seem the simplest

25and baseline remedy would be if there is some kind of


1conduct that is clearly putting impediments in the paths

2of an innovator, enjoin that conduct, whether you go

3farther and engage in structural relief or mandate to

4interoperability should be undertaken with extreme


6MR. HILLEBOE: Renata, did you want to comment?

7MS. HESSE: I guess I think that the presence of

8network effects in a market does at least open up the

9door for the suggestion that the market may be more

10susceptible to a monopoly -- to monopoly power being

11exercised, or existing. I also think that network

12effects can benefit consumers in many ways. So, there's

13a hard balance there, because you don't -- you honestly

14don't want to do something that will then take away the

15benefit of the network effect that the consumer derives.

16But I think they tend to raise barriers to entry,

17whether or not those are long-standing and durable

18barriers is I think the really big question, and if they

19are, how you fix them.

20MR. HILLEBOE: And Michael is somebody who is

21out in those markets every day. What's your view?

22MR. CUNNINGHAM: I'm not sure I can provide a

23broad across the industry, certainly the network effects

24in the markets we participate in is a very, very

25profound -- has very profound effects on competition.


1So, I also can recognize that there are consumer

2benefits to it and I agree with Howard's comments that

3it probably presents some special challenges in

4structuring a remedy and that certainly structural

5remedies could present some real issues.

6MR. HILLEBOE: And Michael, precisely how do you

7think they affect competition if they present a barrier

8to entry? Is that essentially what you said?

9MR. CUNNINGHAM: Yeah, they present a barrier to

10entry. I think they also, because they present a

11barrier to entry, they permit, you know, migration into

12adjacent markets.

13MR. HILLEBOE: And Bill?

14MR. PAGE: One of the observations that was made

15fairly early in the effort to integrate antitrust and

16network effects, and I think it was Mark Rome who stated

17it, one of the observations that had been made was that

18when you're in this period of standards competition, in

19between two incompatible standards and it's not entirely

20clear which is going to become the dominant standard,

21there's a huge incentive for firms to engage in

22practices that don't look rationale. Penetration

23pricing, giving stuff away for free, and so forth, and

24part of the difficulty is that if you look down that

25list of things that they have the incentive to do, a lot


1of them look like antitrust violations. You know, it's

2just rational to engage in practices that can look like

3antitrust violations, and what they are is standards

4competition, they're exactly what the literature would

5predict as standards competition.

6So, that is a serious dilemma for applying the

7antitrust laws in these markets. On the other hand, you

8know, one of the -- one of the supposed paradoxes in the

9Microsoft case was, you know, who cares who the

10Microsoft or Java, for example, wins, or Netscape/Java,

11or Netscape alone, because all you'll have is just the

12new monster. And who cares? You know, you'll just wind

13up with one firm dominating the market and you'll have a

14monopoly and so what.

15And I think there's a very good answer to that,

16that actually came up in the oral argument in the

17Microsoft case, and that I take that the Court of

18Appeals accepted, because they didn't even discuss it in

19their opinion, and that is that you don't want a biased

20choice. In other words, it does matter who wins.

21You're going to have a monopolist, it does matter which

22is the monopolist, and the network effects, the

23literature would suggest, that in some circumstances,

24network effects can exclude even a product that's better

25setting aside the network advantage.


1So, you know, I'm not sure exactly where to come

2down on it. Mark had a few suggestions, in his article

3that was in Connecticut, and I don't remember the name

4of it, but he had a few suggestions on how to, for

5example, distinguish conventional with the sort of the

6predicted penetration pricing from genuine predatory

7pricing and how that might be adapted to network


9MR. HILLEBOE: Marina, do you have any thoughts

10on that?

11MS. LAO: I think it's true that network effects

12can be very efficient, and the example that I'm thinking

13of is not a high-tech one, but is real estate

14multi-listing. No one would say that the network

15 effects there are not efficient, and agree that in

16remedies where network effects are efficient, we have to

17be very sure -- we have to be very careful not to take

18away the efficiencies.

19So, for instance, in the real estate

20multi-listing situation, perhaps you could force the

21network to open itself up to competitors, but not try to

22introduce a competing network.

23MR. HILLEBOE: And moving on to sort of --

24MR. CUNNINGHAM: Just one final thought.



1MR. CUNNINGHAM: Just on the idea of preserving

2innovation through standards competition, perhaps

3apropos my principal comments, innovation also occurs

4through open collaboration about open standards and

5there's ample evidence about that. So, I think it's a

6factor, but I don't think it's the only factor that

7needs to be considered in that circumstance.

8MR. HILLEBOE: Moving on to kind of a nuts and

9bolts issue, Renata suggested that given the speed of

10change in these markets, that perhaps a shorter consent

11decree might be appropriate. Is that something that as

12an antitrust enforcement agency we should be thinking



15MR. SHELANSKI: Maybe I'm too optimistic about

16the ability to advise consent decrees, I should know

17better, I think I litigated waiver number 917 on the NIT

18decree, but I'm not sure that I would shorten the decree

19for the following reason, and I mean, I defer to you who

20implement these daily to know better, but it would seem

21to me that if it was easier to repeal and modify a

22decree than to re-authorize one or to negotiate a new

23one, I might put one in place for a longer period of

24time and back off if it becomes moot and then go in the

25other direction. That's an enforcement question I'm not


1qualified to answer.

2MR. ELIASBERG: If I could follow up on that one

3with Howard. Howard, there were allusions to some sort

4of a review process, in which the court or somehow or

5another would open up the decree, not to see to

6necessarily compliance with the decree, but with the

7effectiveness of the decree. How would you factor that

8into this whole question of term of decree?

9MR. SHELANSKI: Well, I think it's a great idea,

10and I would favor a review provisions, or, you know,

11eventual sunset provisions in the absence of review.

12But review, you know, review is very difficult. You

13know, I'm not sure the second and third triennial

14reviews under the AT&T decree ever occurred, and so --

15and then the question of, well, what gives cause, what

16gives cause to open them up, but having them there in a

17decree so that someone can go get a mandamus and seek


19MR. HILLEBOE: Do any other speakers have any

20thoughts about that?

21 Yes, Bill?

22MR. PAGE: I think in principle, I like short

23decrees. On the other hand, it's a bit of a catch-22

24when you're talking about the compulsory licensing

25provisions, because how do you market to firms the idea


1of building on, say, Microsoft's proprietary base, if

2the license is going to expire in a few years? I mean,

3how -- that seems to be like a contradictory -- I mean,

4not that firms would ever necessarily want to be

5building on Microsoft's proprietary protocols, in many

6instances, they might choose not to do that even if they

7were thought to be perpetual licenses, but I would be

8concerned that at some point, the government is going to

9leave the picture and Microsoft is going to yank my

10protocols under the basis of my whole business.

11So, you know, I guess it depends -- to my way of

12thinking, it would depend on the nature of the remedy.

13If it's a prohibitory remedy to remove specific

14impediments, that would make sense for that to just be a

15short-term one. But if there is a legitimate need for a

16forward-looking remedy, then I think, you know, five

17years is probably not enough, and certainly it hasn't

18been enough in the protocol licensing provision.

19MR. SHELANSKI: Can I just follow up really

20quickly on that?

21MR. HILLEBOE: Of course.

22MR. SHELANSKI: I think Bill makes a good point,

23I think the nature of the conduct really in some sense

24has to derive what the length of the decree is. For

25example, suppose somebody gets a network monopoly by


1penetration pricing, and now they get zero, and then

2they undertake some type of conduct later once they have

3their monopoly that prevents subsequent innovators by

4doing the same thing, by exclusive dealing or something

5else like that. I'm not sure that you want a short

6decree there, because it's quite clear that the conduct

7will always be harmful, and so I think tying it to the

8conduct, there might not be a systematic answer.

9MR. ELIASBERG: Actually, if I can follow up

10with Renata, I think Renata you initially raised this

11point. What are your thoughts on how to determine if a

12shorter decree is appropriate, and also just how long

13that shorter decree ought to be.

14MS. HESSE: That's asking me impossible

15questions. I actually agree with both Bill and Howard

16that what kind of conduct it is that you're talking

17about is going to be an important input into that

18determination. It's clear that the five years was not

19enough, for the section of the consent decree, or that

20at least both Microsoft and all the plaintiffs came to

21the conclusion that they needed more time.

22So, and then there was a lot of work done, which

23I think if you, you know, scour the status reports,

24you'll see they're done to make sure that this problem

25that Bill talked about, which was why would I invest in


1this to begin with if it's going to get yanked out from

2under me in the end, to see that the terms of the

3licenses were flexible enough so that hopefully people

4felt comfortable with that.

5I think that the kinds of things to think about

6when you're trying to decide whether or not a shorter or

7longer decree makes sense have to do with both the way

8in which the market changes, how quickly you think the

9market is going to change, whether or not that matters

10for the ultimate success of the remedy, whether or not

11you think that there's a sort of simple one-shot

12solution to the problem, and that if somebody can -- if

13the particular conduct, if stopped for a period of time

14will result in new entry, or in a lowering of a barrier

15to entry that will be sufficient in a short period of

16time to overcome the prospect of the network effect.

17I think in most technology markets, despite the

18fact that they move fast, this issue that Bill raised

19about there being an underpinning in the monopolist's

20technology that may be an important part of alleviating

21the anticompetitive or the harm from the anticompetitive

22conduct, would tend to suggest that shorter decrees

23actually are not warranted in most cases.

24On the other hand, you know, I think both of the

25agencies have gone away from the idea of doing perpetual


1decrees, ten years is generally the standard. So,

2you're talking about the difference between five and ten

3years, and it's hard to know precisely in what cases it

4makes sense to do one or the other I guess.

5MR. HILLEBOE: I thought Howard made an

6interesting point, and it's something that we touched on

7yesterday, but we kind of had a truncated discussion on

8it, and that is I think there's a recognition frequently

9in a case you see perceived liability, but you recognize

10that it's going to be very difficult to come up with a

11remedy. And the question what is the value of

12proceeding and prosecuting that type of a case, and the

13possible goals might be for deterrence, as Howard

14suggested, or for establishing a precedent, or for

15making it easier to bring a subsequent case.

16I know Howard's view on that, but what do the

17other speakers think about that? Renata, do you have

18any thoughts about that? Or do you want to punt that


20MS. HESSE: How about this, why don't we start

21down there, so Marina can go first.

22MR. HILLEBOE: Marina?

23MS. LAO: I believe that we should proceed if

24the violation is egregious and if the consumer harm is

25substantial, but where it is not substantial, and where


1the act is borderline, then if we don't have a clear

2remedy that is workable, then perhaps we should back


4MR. HILLEBOE: So, sort of a sliding scale in

5your analysis?

6MS. LAO: Sliding scale.


8MR. PAGE: I would suggest that one remedy is

9collateral estoppel, and that, you know, there are

10plaintiffs who will not bring a case for the reasons

11that we've just heard, that because it's simply

12impossible to go up against the monopolist in

13litigation, for practical terms. Just because an

14injunctive remedy is not issued, does not necessarily

15mean that there is not a remedial benefit, because there

16can be follow-on litigation. I mean, the most recent

17estimate I saw of the damages or the settlement amounts

18in the Microsoft litigation was approaching nine billion

19dollars. Even for Microsoft, nine billion, that will

20get your attention.

21So, I suspect that even establish -- and if the

22case were brought with an eye for collateral estoppel, I

23think there's every reason to bring a case.

24MR. HILLEBOE: Michael?

25MR. CUNNINGHAM: It's certainly consistent with


1my visceral reaction and my advice to clients, to my

2client, that it has a deterrent effect for typically

3even more egregious behavior. I do think there are some

4potential evidences that the deterrent effect is real.

5I think in addition to the complaints that Howard laid

6out when dealing with complicated problems the

7experience of competition authorities in learning how to

8deal with them and getting more sophisticated in dealing

9with them is not a value that should be discarded value.

10MS. HESSE: Actually, I think I said this

11earlier, I actually agree with the notion of the

12deterrent effect of taking action, even if you're not

13100 percent sure that you can figure out a way to solve

14the problem perfectly, or even reasonably well, and I

15think there are a lot of people who would say, even

16people who will say both, that the Microsoft decree has

17been a failure, and has done nothing, and at the same

18time say that it was a case that was worth bringing.

19So, and I tend to -- I'm not taking a position

20on whether it was a failure or not, but I agree that

21even if you assume it was a failure, that the case

22itself, both demonstrated that these were markets that

23the government was capable of dealing with, that they

24were capable of litigating against a huge company and

25winning, and that, you know, nobody was, you know, above


1the law. And that's an important point to make.

2MR. HILLEBOE: Bill, I just have a question for

3you. We talked yesterday about various goals in terms

4of antitrust remedies, and you spent a great deal of

5time talking about Microsoft. How would you

6characterize, what's your opinion of what the goal was

7for the government at the time they entered into that

8remedy based upon reading from Charles James articles or

9whatever, and do you think the goal was achieved?

10MR. PAGE: You mean the consent decree?

11MR. HILLEBOE: The 2002 consent.

12MR. PAGE: Well, they're in a position where the

13Court of Appeals had really given them not too much

14choice. The thought of pursuing any type of structural

15relief was impossible at that stage. So, at that point,

16some sort of -- some sort of conduct was all that you

17were going to get, and I suspect that -- well, perhaps

18I'm not the best one to -- I'm certainly not going to

19sort of assume what the goals were, but as I said

20earlier, I think that by and large, the terms of the

21consent decree and the parallel relief in the states'

22remedy are closely tied to the theory of liability in

23the government case.

24Now, certainly the grandest standard by which we

25would judge that would be does it restore the platform


1threat? You know, does it create some sort of rival

2platform that would threaten Microsoft, and by that

3standard, you would have to say that it hasn't done

4that. On the other hand, I think there are other ways

5of evaluating the decree. I mean, one of the provisions

6of the decree is to make sure -- there's an internal --

7there are two, actual, internal Microsoft compliance

8officers, and, you know, if you go back and listen to --

9if you go back and read Judge Jackson's comments about

10Microsoft, it's almost he said they were like, you know,

11young punks or organized crime or, you know, defiant

12organization, criminal enterprise, whatever, and I don't

13think anyone -- well, I'm not sure that anyone would

14necessarily say that that's the case now.

15I think at least, you know, there is a huge --

16in fact, there is one of the status reports describes

17the Microsoft compliance program, I think they said

18something like -- well, they've conducted these

19antitrust compliance seminars worldwide, 15,000

20employees have taken them, you know, all the executives

21are schooled in the requirements of the consent decree

22and the antitrust laws, it may all be window dressing,

23but I suspect that there is a difference in attitude at

24Microsoft because of this case.

25MR. HILLEBOE: Any of the other speakers want to


1comment on that?

2MR. ELIASBERG: Yeah, a question I wanted to

3touch base, actually, and start with you, Renata, you

4indicated or suggested that there could be some

5disruption to structural relief, indeed, sometimes it

6can be cleaner and so forth. But we seem to have some

7language from the Court of Appeals suggesting that we

8should be extremely reluctant about thinking about

9structural relief and indeed it should be the last


11What thoughts do you have about just how

12advisable is it for us to be thinking about structural

13relief right out of the box with respect to such a


15MS. HESSE: I think I read the Court of Appeals'

16decision to be -- and this actually was something Bill

17was talking about, also, to be focusing on the question

18of causation and the importance of establishing

19causation if you're then going to go and impose a

20structural remedy. And that -- I think that is a very

21important question.

22I think the Court of Appeals' attitude toward

23 structural relief probably supports some of the things

24that I said, which is that imposing it occasionally in a

25Section 2 case or demonstrating that you're capable of


1doing that may have a greater deterrent effect, and that

2people perceive that remedy, rightly or wrongly, to be a

3more Draconian one than a behavioral remedy.

4But the question of causation, I think, is

5really an interesting one, because it does get to this

6question of how do you know what the competitive

7conditions of the marketplace would look like without

8the bad exclusionary conduct? And nobody knows, really.

9Nobody knows whether another platform effect would have

10emerged. And so I think it's hard to say looking at at

11least in the Microsoft context, looking at the

12marketplace today, whether or not the decree has been a

13booming success or, you know, an abject failure, if --

14because you really don't know what would have happened.

15And I think the record was -- had some information about

16it, but I don't think anybody really knew whether

17Netscape, in fact, was really a viable platform threat.

18We knew that Microsoft was worried about it and thought

19that it was.

20So, I think I certainly wouldn't out of the box

21say, it's not worth even spending your time thinking

22about, because I think these cases are -- they're not

23only hard to put together and then try, but they're very

24difficult, and you should leave open all of your options

25in terms of thinking about how to resolve, how to remedy


1a problem that you've seen and I think that, you know, a

2structural remedy would certainly be appropriate in the

3right cases.

4MR. ELIASBERG: Howard, did you have something

5 you wanted to add?

6MR. SHELANSKI: Well, my tongue-in-cheek remark

7earlier about the cost to Microsoft aside, I don't

8believe any of us believe that the government should be

9in the business of just creating costs for firms. So,

10we need to be darn sure of the curative potential for --

11I think for any remedy, and I think with a structural

12remedy, I read the Court of Appeals, too, of being as

13insisting on a tight causal link, and I would rephrase

14that slightly as a strong curative likelihood of success

15for the competitive harms.

16And I think you want to be darn sure of that in

17a structural setting, because especially in a high-tech

18industry, I think the unintended consequences of

19structural relief could be many.

20MR. ELIASBERG: Something I also wanted to just

21cover with the panelists, just to be sure we canvassed

22all the views, Marina floated the notion of I'll

23describe it as lowering rivals' costs as a strategy with

24respect to shaping -- creating -- formulating relief. I

25was curious if any other panelists had a reaction one


1way or the other about the advisability or not of such

2imposition. You can either volunteer or I'll just go

3ahead and call on you.

4MR. PAGE: Well, I would say that it's

5appropriate if it's in response to actions that

6anticompetitively raised rivals' costs. I don't know

7that because a violation has been found that all

8methods, and I don't want to characterize you saying

9this, but all methods of lowering rivals' costs have

10been appropriate.

11So, again, lowering rivals' costs is certainly a

12legitimate goal, if the causal link to the

13anticompetitive conduct is established.

14MS. LAO: I really see that as a conduit to

15promoting consumer welfare, and not to benefit

16competitors for the sake of benefitting the competitors.

17MR. SHELANSKI: As a veteran of the unbundling

18wars in Telecom, I twitch a little bit when I hear

19lowering rivals' costs, and I think the one thing that

20would give me pause is I would say maybe, if the cost

21you're lowering is one that the defendant is being asked

22to lower through the remedy is a cost that the defendant

23created, and I think that that would be a tie that even

24before thinking about it I would want to see there,

25because otherwise, I think there's really great danger


1for the agency to become an ongoing regulatory authority

2as opposed to someone recommending particular

3anticompetitive conduct.

4MR. ELIASBERG: One more question.


6MR. ELIASBERG: Actually, this one, Michael, is

7to you. In your presentation, you made a comment about

8situations where steps may be taken by an incumbent to

9change structure of its product so that it could not be

10transferability or used by a subsequent -- front by a

11rival or something of that nature. In a case like that,

12assuming for the moment that there was liability found,

13found for that alteration or change in the product

14design, what would be the type of relief you would think

15would be -- what would be the remedy that you would

16think would be the appropriate remedy in a situation

17like that?

18MR. CUNNINGHAM: In our industry, I guess with a

19strong network effects, some interoperability remedy

20would seem to be the one that you would need. Yeah.

21MR. ELIASBERG: Nothing else comes to mind?


23MR. ELIASBERG: Anyone else have a rationale for


25(No response.)


1MR. HILLEBOE: Well, I note that it's close to

212:30. So, I just want to say on behalf of the FTC, and

3my colleagues at DOJ, I wanted to say thank you very

4much to these speakers, an excellent presentation, and I

5want to remind and thank everyone for coming and remind

6everyone that we have a final wrap-up in the coming

7weeks. Thank you.


9(Whereupon, at 12:28 p.m., the hearing was


















1C E R T I F I C A T I O N O F R E P O R T E R




5DATE: March 29, 2007


7I HEREBY CERTIFY that the transcript contained

8herein is a full and accurate transcript of the notes

9taken by me at the hearing on the above cause before the

10FEDERAL TRADE COMMISSION to the best of my knowledge and



13               DATED: 4/3/07


15               SALLY JO BOWLING


17C E R T I F I C A T I O N O F P R O O F R E A D E R


19I HEREBY CERTIFY that I proofread the transcript

20for accuracy in spelling, hyphenation, punctuation and




24               SARA J. VANCE

Updated June 25, 2015