« "How's My Driving?" for Everyone | Main | A Wonder of Modern Communications Post »

May 17, 2006


Feed You can follow this conversation by subscribing to the comment feed for this post.


FYI. You have a typo on the cover.

"Patent Holdouts and the Standard‐Settig Process"

Michael Martin

Your discussion of injunctions in patent cases probably deserves an update.

Doug Lichtman

The typo is being fixed. Somehow SSRN mistranscribed the title from the title page when they were making that cover sheet. Thanks for catching that one; I hadn't.

On the update in light of eBay, meanwhile, I have a few short comments in this draft on that score, and plan to come back and say more in a later draft. For now, though, eBay does not much alter my core puzzle in this paper, so I thought it best not to get too distracted with parsing the new opinion. (Clearly, even after eBay, we still have a patent holdout problem, and will for the foreseeable future.)

Thanks for the comments.

Michael Martin

There might be an analogy between the situation of the plurality of patentees whose claims read onto an industry standard and the plurality of creditors whose claims cover a bankruptcy estate. In a sense, the diversity of the creditors is what makes cram down possible, right? Maybe this is your point? Maybe the analogy is crazy.

Steve R.

Copy of mt postt from the Technology Liberations Front Webpage
Lichtman's paper is highly focused, consequently many fundamental problems that I see with the patent system are unfortunately outside the scope of the paper. Though his paper proposes a reasonable solution, based on its narrow definition of a patent problem, it does not get to what I believe is the fundamental problem - we have a broken patent system.

Lichtman's paper acknowledges that there are a plethora of companies scavanging for patent dollars. But the paper does not really question why there are so many patents. Lichtman notes "To make matters worse, patent language is subject to hopelessly nuanced rules of interpretaion." He goes on to say that "More troubling still, the Patent Office as it stands today is not particularly reliable when it comes to evaluating proposed inventions and weeding out those that do not meet patent law's stringent eligiblity thresholds."

To solve this problem we need to "break out of the box", we need a better definition of what is patentable and maybe even reduce the period (time span) for when a patent holder has a monopoly right.



Very interesting idea. One problem I would have is that the holdouts (i.e. trolls) will not all come forward at the same time. Therefore, no one has a reliable estimate of how many ways the holdout pool will be split among trolls. When the first troll comes forward, you would be willing to pay that troll the entire holdout pool (minus a discount, essentially uncalculable, of expected future trolls). You would again be willing to pay the same amount to the second troll who comes forward after you settle with the first troll.

Your MFN idea would ameliorate this concern by essentially putting a new twist on the reasonable royalty calculation and dilute the power of individual trolls. But I am not sure it would work. The troll litigation posture is usually Troll vs. Group. If it is worth $100 to the Group as a whole to continue using the standard, the MFN clause would require most of that $100 to be split among members of the Group. But the Troll isn't bound by the MFN clause, and can simply threaten to shutdown the entire Group unless Group collectively agrees to hand over the $100 entirely to Troll (i.e. waive the MFN clause collectively). All it requires is for Troll to sue the entire Group.

Steve R.

Not to be off topic, the Electronic Frontier Foundation is reporting (May 18, 2006) "At the request of the Electronic Frontier Foundation (EFF), the U.S. Patent and Trademark Office (PTO) will reexamine a controversial patent for online test-taking from Test.com. The reexamination order is the second granted in just two months after petitions from EFF's Patent Busting Project. EFF filed the reexamination request because the extremely broad patent claims to cover almost all methods of online testing. Test.com has used this patent to demand payments from universities with distance education programs that give tests online."

Again, the issue of "Discouraging Patent Holdouts" requires a clearer definition of what can be patented so that we do not have "Patent Holdouts".

Steve R.

The Electronic Frontier Foundation (May 18, 2006) had this quote from Justice Kennedy's on the Ebay lawsuit.

"An industry has developed in which firms use patents not as a basis for producing and selling goods but, instead, primarily for obtaining licensing fees. . . . For these firms, an injunction, and the potentially serious sanctions arising from its violation, can be employed as a bargaining tool to charge exorbitant fees to companies that seek to buy licenses to practice the patent."


1. There is an information tension in the paper. The problem you address is the threat of patent holdouts to use reliance on their infringing technology to acquire supra-normal royalties. This problem exists because the producing firm cannot identify all potential infringing patent-holders. But, you then argue a firm can plausibly argue to a holdout that there are many other holdouts, so the firm cannot afford to pay that holdout much money. Thus, a firm must know the number of patents they are infringing and the value of those patents, but not know whose patents they are infringing. I find this implausible.

2. As the thousands of holdouts come forward for their slice of the pie, the number of potential holdouts decreases. At some point, there may be only a few left. Then, they will be able to extract much more of the "hostage" value of stopping distribution because the previous payouts are now fixed costs.

Doug Lichtman

Thanks so much for the comments. The ones here, plus a number received by email, have helped me tremendously in terms of sharpening the paper. I really appreciate the pushback.

A few responses:

C&D, you raise two great points. On the information tension, though, I think descriptively it is what happens on the ground. The RFID example, for instance, is real. We know that there are likely 1000s of relevant patents, and yet no one has the exhaustive list. On your sequential point, meanwhile, that is why I like the MFN solution. If I sign a deal with a patent holder but include the MFN clause, then that patent holder still will be sharing in the hostage value and thus the number of hostage-takers remains functionally constant.

TJ, your point about the group dynamic is also really good. I'm not sure I agree with you about how it plays out, however. If the Troll demands that the 100 firms in essence renogiate all of their deals, the firms will balk. Each firm, after all, will think to itself, "I am in almost as good a position as the Troll. I too can demand that the other 99 firms pay me a share of the holdout. Why should I give that position up?" Of course, with 100 firms all thinking that, we are right back to the ironic benefits of massively overlapping patent claimants. (Does that sound right? I need to figure out how to say this more clearly, but I think this is how it would play out.)

Again, sincere thanks. Please do send along additional comments either here or by email. I plan to revise this paper in a few weeks, and hopefully have it ready for law review submission in August.



Your point, if I take it right, is that once the troll raises the prospect of shredding the intra-Group contracts, ever member of the Group is now free to act like the troll and go after all the others.

I think that is unlikely because of the asymmatry between practicing entities and trolls. Practicing entities are members of Groups, are repeat players, and cross-license to each other (or face mutually assured destruction). The only clause that you suggest adding to a standard cross-license is an MFN clause.

Now suppose that all practicing entities have large sunk costs and the cross-licenses are somehow declared invalid. Each of the parties have a choice, holdout or new cross-license? Although the chicken dynamic might suggest that even practicing entities should choose to holdout if the pool is large enough, the combination of sunk costs and repeat playing usually means such entites choose to cross-license.

Trolls are different, because they never cross-license -- they always win the game of chicken. Thus a troll imposing the "waive MFN" term on a Group faces no realistic threat that the Group will disintegrate into mutually assured destruction.

Of course, an assumption is that members of the Group are all practicing entities. The participating of a non-practicing entity like a University in the Group would complicate the dynamic since a University has much the same incentives as a troll except the repeat-play and the ex ante objective.

Doug Lichtman

TJ -

That's interesting. Can you explain a bit more?

Start with your point about repeat play. Usually we talk about repeat play in instances where some firm can do a bad act, but we think the firm will not do it because other firms will hold the original firm accountable in later interations. Here, why would that discourage the behavior I describe? After all, joining the holdout does not hurt the other firms any. No matter what, they are all on the hook for the hostage value of their sunk cost investments. So why would they disapprove of their partner actually living up to his MFN rights? (And then, as you know, living up to the MFN ironically turns out to help the other firms by discouraging the holdout from actually fighting.)

Turn now to your point about sunk costs. I agree that everyone has sunk costs here, but once a patent holdout steps forward, those sunk costs are at risk. Period. Indeed, the sunk costs will be fully extorted if all the patent holders waive their MFNs and give in to the holdout. So, again, why would sunk costs cause patent holders not to use their MFN clause?

Note, too, that even if you are right to think that cooperating firms feel some pressure to renegotiate in the face of an extortionist patent holdout, the number of firms involved might make renegotiation difficult. Each firm might slightly miscalculate how much its contribution should be; it might be hard to get 300 firms on the same page; and so on.

ps. I really like your several points here (the University point is especially good) and want to build them into the analysis. Thanks for taking the time to iterate with me on them here.

J C Cooper

Dr. Lichtman,
Thank you for your very interesting and thought provoking paper. If I may, I would like to offer a few comments for your consideration.
In respect to the class of patent not yet issued, as you recognize there will always be those patent applications which are not yet issued. Many are buried in USPTO examination and some of the better patents may take a decade or more to see the light of day. This offen occurs because of inefficiency by the examiner, or the holder's refusal to accept less coverage than that to which they feel entitled. The latter is common for patents the owner fees have great value.
With respect to the general theme of the paper, you might address the problem of enforcement capabilities of the patent holder. The reality today is that many manufacturers engage in product design and development without any recognition of or attention to others having patent rights affecting that product. A typical product design starts with marketing and sales people who pick and choose features of competitive products to put in their own product. Sales and marketing also adds a few new features in response to customer input. The feature list then becomes the basis for an engineering program, ultimately resulting in a new product for the manufacturer. Meanwhile, the customers are providing the same desired feature input to multiple manufacturers, big and small alike. As you point out, these manufacturing companies turn a blind eye toward any sort of investigation of whether any of the selected features might infringe another's patent. Independent inventors and investors may also identify the customers' desired features as an opportunity to found a startup company.
It is no wonder that after a product is introduced, multiple patent holders are knocking on the manufacturers' doors. Many large manufacturers mitigate their potential liability by reliance on their size. These companies hope that they can either arrange a suitable cross license with other large manufacturers whose patents they infringe, or in the case of the small manufacturer, startup and independent inventor, they hope to simply crush them with their overwhelming resources. After all, even with contingency fee arrangemetns very few small companies or individuals have the resources to take on a big company in court and succeed. I would suggest than in many instances large corporations use these tactics to crash through a patent thicket.

Anthony Topper

I served on committees in five different high-tech industry standards bodies over an 18 year engineering career (now I am a patent agent), so I have first hand experience in this (committees have included large and small practicing entities and non-practicing entities).

Practicing entities have more than sunk costs to protect. Opportunity cost is a primary cohesive force in standards groups. Later round accountability is also a very real part of that cost. Certainly, sunk costs are at risk if someone dares to assert, but so are opportunity costs which are often more significant. This may seem odd, but remember the reason for standards in the first place is to enhance market growth or even to establish markets. Holdout value is a complicated calculation involving many factors such as damage to market size and growth, goodwill costs, and veto-like retaliation which is in proportion to patent thicket size (TJ is correct that when faced with MAD or shared booty, rational self-preservation prevails).

The idea of diluting holdout value through MFN, though, is a secondary effect for large practicing entities, but certainly worthwhile. The key use of MFN is in defending against the non-practicing attacker. I agree with Doug. A partner would not disapprove of another’s MFN. Rather, the self-interest of all the non-holdouts would have them assert MFN and rally in other ways to defend the “shared opportunity value” of the standard. (I was involved in a real standards battle with major tech companies where something close to this did happen and MFN, of course, was even not in play.) I contend that an MFN group is more akin to a cartel with cartel-like group dynamics (Perhaps that is what TJ was alluding to with group dynamics) - an MFN establishes the market price of the “commodity” that the cartel defends. Another model might be a Mexican stand-off where threats to the status quo of self-preservation are repelled. I think it is important to establish a known framework in which to examine the scenario where the MFN devolves.

If a troll attacks an MFN protected group he has to contend with the shared interest of the MFN “defensive cartel”. Once threatened, each firm in the cartel would certainly reconsider whether to negotiate anew its membership in the cartel or attack other members. In doing so they will weigh the initial impetus for joining the MFN cartel, such as the opportunity value of the market, their sunk costs, its defense, etc. So the question is when does an attack on an MFN defensive cartel break it? Members with weak interest may break-off and attack.

Cartels are often broken, not because of attack, but rather the attack reveals inherent weakness in the shared interest - the net present value of the market opportunity is weak againsts the more certain return from attack. Strong cartels have high net present opportunity value in the market for a majority of members. As a case in point: see how major technology companies cooperated and rallied to defend UNIX against the SCO attack, even though no MFN existed and it was a copyright case. Protecting the market opportunity was paramount (not just sunk costs) and no one would dare suffer the loss of goodwill that SCO did.

The end result, taking Doug’s argument to the extreme, is that patents, far from being, merely non-detrimental to a commons when a patent thicket is sufficiently large, can actually protect the commons. Perhaps rephrased: when can a defensive cartel protect a commons? What conditions are necessary to establish, maintain, and defend a defensive cartel so an associated commons is protected by proxy? Is MFN a viable mechanism to do so?

I hope this doesn’t pull the discussion too far off-track.


The analysis in your paper assumes that the only damages a given patent holder could obtain are a fraction of the total value of the product, and thus as the number of patent holders increase, the value to each patent holder of holding out in the hopes of engaging in litigation decreases.

The amount of damages for a given patent holder can be larger than simply a fractional share of the whole, however. For instance, depending on the damages showing, there is the possibility that a patent holder who asserted its patents shortly after the standards had been set could obtain a damages award equal to or greater than the value of the product (in the case of a willfulness determination). Thus, you would end up with the situation where a party that held out until the standards were set, and then asserted their patents, could be in a better position than either a party that participated in the setting of the standards, or waited a significant amount of time after the standards were set.

The problem seems similar to that involved in determining class-action awards, where the award to the present class must be balanced against the possibility of new litigants stepping forward in the future. In the context of patent litigation, an early litigant can potentially gobble up the entire value of the accused product, something that the other patent holders may be powerless to stop.

Under your MFN proposal, if a new patent holder were able to exact a large damages award (or use this threat to obtain a higher payment for its license), not only would the cost of the patent pool be increased by the added cost of the new infringer, the cost of each pre-existing license would also increase. The possibility of such a large jump in costs could have detrimental effects on whether players decided to participate in the licensing scheme in the first place, since they could potentially be subject to large cost increases due to unknown patents.

To put it another way, under the MFN situation, you finish the licensing negotiations with a fixed cost for the entire patent pool of C. In deciding whether to make your product, you balance this cost C versus the potential profitability of your product. If a new patent holder comes along, and gets a high patent value ("N"), the cost of the patent pool is now C + C * (N-C) + N. In essence, your total cost pool has grown (and by more than just N). If it is greater than the total value of the products for all participants, you will be left with no more players- not an optimal result for anyone.


Another comment I had relates again to the fractional value of a new patent holder. You state that "Every infringer would know that hundreds of other patent claimants are waiting in the wings to extract their share of the holdout value." The problem with this is that in order for an infringer to properly value an individual patent holder's fractional share of the whole value pie, the infringer must necessarily know how many players their are. But this number is necessarily unknown. So how would an accused infringer know not to agree to pay more than 1/100th of the total value, as opposed to 1/500th or 1/1000th?

Doug Lichtman

These comments are *so* helpful. Thank you all so much. I will try to build all of this into the next draft.

Jason, I wanted to think more about your two posts:

I couple of people have made the point you make above, and I think that I just must be unclear in this draft of the paper. I am not trying to say that *damages* have to fall below a certain cap. Given the complexities of litigation, damages could indeed by crazily unrelated to value. I am trying to say instead that patent holders will pay only so much to get out of an injunction.

Teasing that out a bit: patent holdouts tend to sue early in the life of the relevant product. They ask for cash for infringements that already took place, and hopefully the courts give only a reasonable number for that, but who knows. I am focusing on the next stage: the holdout gets an injunction with respect to future infringement, and my question is how much the holdout can extort in exchange for waiving the injunction.

Again, the draft is obviously unclear on this point, as you are not the first to read it the other way. I'll try to clarify in the next iteration, and I will also say more about damages and how they might skew incentives if the courts miscalculate them routinely. But the bulk of your point I think goes to the damages version, not the injunction point that is what I mean to be writing about.

On your second point, meanwhile, I agree: no one knows for sure how many patents are relevant. I think the dynamic works so long as we all are sure that the number is large. Whether 500 or 1000 or 10000 might not matter much in terms of incentives, etc.

Obviously, let me know if I'm misunderstanding; and, again, thank you all so much. The feedback here and elsewhere has really helped tremendously in sharpening both what I want to say and how to say it.


Doug- your argument makes significantly more sense when I think of it in terms of what the patentee is willing to pay to avoid a future injunction (i.e.: remain in the market), rather than in terms of a combinaton of the price of the injunction together with past damages. I'll have to think more about it in that light, and I look forward to reading the next version.

I know this has been rehashed in some of the other comments on this blog, but the valuation problem still nags at me. It will be impossible for the participant to accurately value a single threatened injunction when they don't know how many other hold-outs there are still out there. I realize the cost of each individual participant will go towards zero as the number of patents goes up (thus making the difference in price that a company would pay for a 1 out of 500 patent situation versus a 1 out of 10,000 patent situation negligible), a participant can't negotiate an infinite number of licenses- there are transaction costs for each negotiation. So while a participant may be able to deal with the transaction costs of negotiating, say, 100 licenses, they wouldn't necessarily be willing to do so for, say, 10,000 licenses. While the amount they're paying to the pool of licensees would be the same, the total amount (including transaction costs) would be significantly higher for the 10,000-license scenario.

I'm not sure this affects your ultimate conclusion, but it highlights how difficult valuing a single patent holder's rights can be.


Sometimes I really wish for an edit feature. To clarify, I meant to say "I realize the cost of each individual patent will go towards zero," not "participant."

The comments to this entry are closed.