Tuesday, June 7, 2011

Oracle wants a huge cut of Google's mobile advertising revenue plus compensation for fragmentation of Java

One of the best-kept secrets in the patent dispute between Oracle and Google is what Oracle demands in terms of compensation for past damages and on which terms (monetary and other conditions) Oracle might allow Google to continue to distribute its Dalvik virtual machine as a result of a settlement or, alternatively, if Oracle prevailed in court and obtained an injunction.

The details of those demands are still not known. It was always obvious that Oracle wanted a lot. They wouldn't sue for chicken feed, and if their demands were very limited, this would have been settled a while ago. But for the first time in this entire lawsuit (which began almost ten months ago), a publicly accessible document provides, despite some blackened passages, a pretty good indication as to how demanding Oracle is.

I have analyzed the situation and I can tell you up-front: the word "demanding" is an understatement. The position on damages for past infringement taken by an Oracle expert appears to be such that Oracle would want Google to pay damages for past infringement that would in the worst case far exceed any money Google has made with Android so far -- and would likely expect Google to pay even more going forward.

The two companies are not just miles but light years apart, and it could very well be that a defeat in court would require Google to make fundamental changes to its Dalvik virtual machine -- changes that would likely affect many if not all existing Dalvik-based (.DEX) applications. But even in purely financial terms, there's serious doubt as to whether Google would be able to meet Oracle's requirements while continuing to make Android available without charging a per-copy license fee.

This lawsuit has the potential to bring about a restructuring of Google's Android business in economic as well as technical terms. Interestingly, Google itself admits that it could have done a license deal with Sun (apparently before it was acquired by Oracle) but rejected its terms. That refusal could now prove one of the worst mistakes in Google's 13-year history as a company. In connection with Google's recent hiring of a third law firm to defend it against Oracle's claims, I explained that this litigation is indeed comparable to a make-or-break case.

Let me now show you that filing, and below the document I'll explain what conclusions it allows concerning the damage awards Oracle fights for, and what this implies for the terms and conditions of a possible license agreement. Even a demanding position can still have a legal and commercial logic -- and that appears to be the case.

Public Version of Google Filing Re. Oracle Damages

Oracle's damages expert: Professor Iain Cockburn

The document refers to a "damages report of Oracle's expert Iain Cockburn." That one must have been submitted very recently, but it's not publicly accessible (not even in part -- at least at this stage there's no redacted version for the general public). Fortunately, Google's filing reveals some of what that expert believes Oracle is entitled to.

The expert is Iain Cockburn, a Boston University School of Management professor of economics with a focus on "Strategy and Innovation". His comprehensive CV indicates that he's a sought-after expert on economic issues in connection with intellectual property who, for example, presented at a 2009 hearing of the Federal Trade Commission (a U.S. competition enforcement authority).

Google is so shocked by Professor Cockburn's theories that it claims "allowing him to testify about his conclusions to a jury would prejudice Google." A jury would establish an appropriate amount of damages after hearing the representations of, among others, Oracle's expert(s), and Google seeks to prevent that he even just shares his views (though Google's lawyers would have every opportunity to explain to the jury why they disagree). Google accuses him of "legal errors [that] are fundamental and disqualifying" and doesn't even want the report to be shown to the jury, claiming it's "unreliable, misleading, and inappropriate for presentation to the jury."

Damages vs. license

The first important thing to understand here is that the Cockburn report itself is exclusively about damages for past infringement. Assuming that Google is found to infringe Oracle's intellectual property, the court (with the help of the jury) has to determine how much Google has to pay for that -- from whenever the infringement began until the date of the trial. But whatever Cockburn believes or whatever the court may ultimately decide in that scenario is still a separate issue from the terms on which Oracle may or may not grant Google a license going forward. In its complaint, Oracle asked for an injunction. If Oracle obtained that injunction, Google would have to cease all infringement unless the parties agree on a licensing deal for the future. That deal would not be set by the court: it would be negotiated. There could be a separate court case if Google claimed that Oracle had obligations to grant a license on terms meeting certain criteria, and I'll discuss that further below.

The distinction between past damaages and a license deal for the future is really key. Let us firstly look at what Google's filing reveals about the Cockburn report's take on past damages and subsequently analyze what this suggests for Oracle's possible and likely expectations the terms and conditions of a possible license agreement.

Criteria for damage computation

Google accuses Cockburn of failure to properly "separate out and define the value of the patented technology to both Google and Oracle." He certainly did separate out and define, but not in a way that Google could agree with.

It's correct that the question of damages relates to the commercial benefit Google derived from the alleged infringement as well as the damage this caused to Oracle, which is related to the value those Java-related intellectual property assets represent to its rightful owner. The idea of compensatory damages is "full compensation", or (in other words) to "make [the owner of the infringed rights] whole": Oracle should ultimately be in an economic position comparable to the one it would have had otherwise.

But it's also key to consider that Oracle doesn't ask for merely compensatory damages. In Oracle's original complaint, prayer for relief E demands a "[t]rebling of damages under 35 U.S.C. § 284 in view of the willful and deliberate nature of Google's infringement of the patents at issue in this litigation." I'm sure that Cockburn was not tasked with any analysis of that aspect. As an economist, he was certainly just asked to determine compensatory damages. But whatever the court establishes as compensatory damages could still be tripled if the infringement is deemed to have been willful. So when Cockburn considers a 50% royalty rate on a particular royalty base to be a reasonable rate, the worst-case scenario for Google is not 50%, but 150% (yes, it can be more than 100% -- there's no ceiling in statutory law at 100% or any other percentage, just some case law that argues in favor of certain de-facto ceilings).

Apart from the potential tripling of damages (which is punitive in nature), different damage concepts can't just be added on top of each other if the total exceeds what's needed to make Oracle whole. For example, if one theory is based on a reasonable royalty rate if the parties had entered into a freely-negotiated license agreement and another theory is that Oracle usually receives $X per unit of a licensed Java implementation, those are just two ways to look at the same issue and the resulting numbers can't be added to each other. However, it's not always that simple. I'll get to the issue of "fragmentation of Java" later. It's not clear from Google's filing whether some or all of the damages the Cockburn report attributes to that concept would have to be added to the royalty-based concept. I have a strong feeling that that's actually what the undisclosed report proposed.

A huge issue: the hypothetical royalty base

A percentage of possible royalties is one thing, but in Android's case, the most important question is what the royalty base (the amount against which the percentage is applied) should comprise.

Google complains that the Cockburn report "seems to be arguing that Google's advertising revenue from, e.g., mobile searches on Android devices should be included in the royalty base as a convoyed sale" and instead contends that "the value of the Android software and of Google's ads are entirely separate: the software allows for phones to function, whether or not the user is viewing ads; and Google's ads are viewable on any software and are not uniquely enabled by Android."

On this item, I agree 100% with Oracle and 0% with Google. When it comes to the totality of all the issues relevant to the case, I agree and disagree with both parties on different matters. But this here is an issue concerning which Google's representations defy logic and deny reality. No matter how much I dislike Oracle's software patents, I can't support Google's position on this royalty base issue. I must concur with Oracle because of everything I learned in the IT industry about business models (including "free" business models such as MySQL's open source-based business model), because of my observations of Google's intellectual property licensing practices, and because of my personal experience as a heavy user of Android on a daily basis.

Its business model is Google's -- not Oracle's -- choice. Therefore, it's Google's -- not Oracle's -- problem. Google has chosen to make Android available without charging a royalty but doesn't do this for charity. There's a clear business model, and its most essential part are advertising revenues related to search and other online services.

If one applied Google's proposed logic to other "free" business models, open source companies with a service-based business model could infringe any patents they want and never pay serious damages because the professional services they offer are just optionally available -- but another reason for which Google is wrong is that with Android, the services that generate revenue for Google are anything but optional.

I already have my second Android-based phone -- now the Samsung Galaxy S II, which I think is a fantastic product. I can't see how I, as a user, would even be able to replace Google's search engine with any competing engine, or uninstall some of Google's mandatory closed-source software components (such as its Google Maps client). In fact, I haven't even been able to deactivate that Google Talk client, which I consider a huge nuisance.

After the Skyhook revelations and the antitrust complaints lodged against Google by two South Korean search engines, there's no way Google can seriously claim that it doesn't use Android as a tightly-controlled tool to generate online service usage and the related revenues. By now, everyone knows it. It's even known that Google considers some of its Android-related intellectual property rights a "club" to make Android device makers (such as Samsung and Motorola) do what it wants.

Theoretically, device makers could fork Android. They would lose a lot of the value proposition, which is why I'm not aware of any commercially successful mobile device based on an Android fork. I heard there were plans to do something like that in China. I haven't seen anything like that in the United States or Europe.

Oracle already countered Google's out-of-control excuse months ago in connection with a discovery matter.

Google must consider the rest of the world very stupid to keep telling the fairy tale of an open Android that is sooooo detached from Google's mobile services. Let's get real, please.

Hypothetical royalty rate: 50%

The 50% figure I mentioned before was not a fictitious example. According to Google's filing, the Cockburn report proposes "to apply an unprecedented fifty percent royalty rate to that base" (that emphasis is Google's, not mine).

Since the report itself is inaccessible and since parts of Google's filing are blackened, it's not totally clear what "that base" means. The most aggressive base would be all of Google's Android-related revenues (including any mobile advertising income it owes to Android), and the disclosed parts of Google's filing make that a possibility. Google criticizes Prof. Cockburn for "stating that there is 'no clear economic basis' for apportioning the total value of Android into value attributable to the patents and copyrights in suit and any additional value added by Google." That wording suggests that his position is that the infringed patents (and copyrights, but patents are really far more important here) are so central to Android that there would be no Android without them.

Should that really be the position taken by the Cockburn report, then I do understand why Google argues vehemently against it. One can certainly have an argument over how much of Android we're talking about when it comes to the parts of Android that may make use of Oracle's intellectual property, and I disagree with Oracle's expert (if Google correctly stated what he said) that anything Java-related is all over Android and you can't even tell anymore what Google itself developed. I understand that Oracle takes an Oracle-centric perspective on the world, and I could see some arguments, such as claiming that hundreds of thousands of Android apps run on a virtual machine that allegedly infringes Java-related intellectual property rights. Then there would be the argument that today's mobile platforms are all about ecosystems, and the Android ecosystem stands on allegedly thin ice that largely belongs to Oracle.

But still there's a greater scheme of things and this is just one of 44 (!) Android-related infringement suits that have already been filed (and who knows how many more will be filed). For example, Microsoft is making significant headway now against Motorola, and Apple is suing the top three Android device makers (Samsung, Motorola, HTC). All those other right holders also own a part of Android if their allegations are true, and they can't all be wrong. Saying that all of Android is basically Java-related goes too far in my view (Oracle is not even the only right holder to assert Java-related patents against Android), although it's not as extreme as Google's position on the revenue base, which one can paraphrase as "what's yours is worth zero, and what's worth a lot will always be mine and only mine".

I support Google's position that "damages must be tied to 'the claimed invention's footprint in the market place.'" Should Oracle's expert indeed have denied or not properly applied that concept (for now we only know what Google claims he did), then he should amend his report accordingly.

Damages concept of "fragmentation of Java"

Google also disagrees on how Oracle's "lost profits and opportunities" should be calculated. This related contention is important:

"Most audaciously, Cockburn tries to import into his royalty base the alleged harm Sun and Oracle would have suffered from so-called 'fragmentation' of Java into myriad competing standards, opining that Oracle's damages from the Android software includes theoretical downstream harm to a wholly different Oracle product."

This indicates that Oracle says its damages are not just a question of what the royalty rate would have been but that there's quite some economic damage that's attributable to the mess that Google has created: the inconsistency between Android's denomination of Java and other implementations, in particular, those that adhere to the official Java specification.

Google says that there's no precedent for damages based on that theory, but even in case Google is right (I haven't been able to verify that claim), Oracle would have a lot of business logic on its side, and that business logic includes Google's own positions on how it maintains control over Android and dictates to device makers what they are allowed or not allowed to do if they want to use the green robot logo.

It's really striking how Google uses against Oracle the opposite of the logic it presents in its defense against Skyhook. Let me quote from Google's motion to dismiss Skyhook's competition case:

"Without an assurance of compatibility, adding third party applications [such as Skyhook's location-positioning software] to the operating software would drive end users away from Android, to the detriment of carriers, manufacturers, developers and customers."

There you have it: Google itself explains that not allowing its device maker partners to ship Skyhook's software was just, the way Google describes it, a necessary measure to prevent damage (Google says "detriment", which is the Anglicized version of the Latin word for "damage") from being caused to the whole ecosystem.

But Google does not want to allow Oracle to control Java the way Google controls Android.

I could provide even more examples of how Google's defense against Skyhook contradicts its defense against Oracle. In one case it's about Google's rights to do as it pleases with its own intellectual property (including, but not limited to, the Android trademark). In the other case it's about Oracle's rights to license its IP on its terms. When Oracle wants to assert control and derive revenues, Google claims it attacks open source. But Google's defense against Skyhook also cites the concept that "[a]dvancement of one's economic interest . . . is not an improper motive."

Term of validity

A smaller item (the fourth one in Google's filing) claims that the Cockburn report doesn't take into consideration that the asserted patents and copyrights have different expiration dates. Google's position is essentially that the expiration of any valid and infringed patents would have to lower the royalty rate from that point on. I don't know whether Google overstates this. Maybe the Cockburn report doesn't sufficiently differentiate, but it's also possible that it focuses on past damages and does not go into full detail on possible future damages. After all, what Oracle wants is past damages plus an injunction for the future -- not a determination of what its damage will be five years down the road.

Google's rejection of Sun's licensing offer

Others have previously blogged about what was at that stage just an industry rumor about Google having refused a Java-related license deal offered by Sun (apparently prior to its acquisition by Oracle). Google's filing now admits that fact, accusing Cockburn of ignoring "Google's actual negotiation history with Sun regarding a Java license for the mobile space, which would have included far more than the patents-in-suit and during which Google rejected a proposal to pay Sun [a blackened amount]." (emphasis of "rejected" is Google's, not mine)

Whatever the terms might have been at the time, it's pretty clear that Oracle no longer feels that it has to repeat an offer on those terms at this stage. Google had its chance to do a deal on those terms (whatever they were), turned Sun down, and now has to deal with a far more powerful, aggressive and demanding right owner named Oracle.

Sun's Java-related settlement with Microsoft

Another interesting detail in Google's letter is the claim that in its settlement with Microsoft, "Sun demanded and received $900 million to cover the risk of fragmentation to Java."

The amount itself is not new. The total settlement between Sun and Microsoft was even far greater, but I also recall announcements that suggested that $900 million was the Java-related part. Should Oracle really have attributed the entire $900 million amount to "fragmentation", that would be a bit narrow. The settlement probably included a far-reaching license to various intellectual property rights.

But the fact that Oracle brings up this kind of amount in its litigation with Google indicates to me that Oracle is, as I recently already figured, looking for a settlement with Google that would be similarly big as its recent damage award in the SAP case. We're talking about, potentially, billions.

Conclusions concerning possible license terms in settlement or post-injunction deal

Theoretically, Oracle could take all sorts of damage-inflating positions at this stage of the process but be willing to settle with Google on much more moderate terms. However, we are now less than five months away from the possible trial date, and there's no indication of an impending settlement.

If a patent holder obtains an injunction, he doesn't have an automatic obligation to grant a license to the infringer on any terms. A patent is a monopoly. An enforced patent is an enforced monopoly. There are only two theories based on which a defeated infringer could claim to be entitled to a license on reasonable terms: either because the right holder has a dominant market position according to antitrust law (and it's hard to claim that Java is dominant as a programming language or virtual machine -- Google would have had other viable choices at the beginning and the current dependence of Android on Java could have been avoided) or because a right holder has made contractual commitments to grant licenses. The latter is actually the case here: Oracle made such commitments as part of the Java standardization process, but as long as Android's virtual machine deviates from Java's official specifications, there's no obligation for Oracle to grant a license. Even if Google brought its architecture into compliance with the Java specification, the terms would just have to be "reasonable and non-discriminatory", and if Oracle demands $X per unit and has many customers paying that amount, Google will have a hard time arguing that it should get a license at a lower rate.

Assuming this case goes to trial and Oracle obtains an injunction, I think Google will have to do two things (short of rewriting its entire virtual machine to steer clear of infringement of any Oracle intellectual property):

  1. Pay a very significant per-copy royalty to Oracle -- possibly higher than all of Google's current Android-related revenues, in which case Google would have to change its Android business model and start to charge device makers per-unit license fees (assuming, of course, that Google wants to operate Android in an economically viable way and not just turn it into a "loss leader" in its portfolio).

  2. Formally recognize Oracle's control over Java and make at least some reasonably significant modifications to its code in order to reduce or eliminate the "fragmentation" issue Oracle complains about.

Just for fun, one could add some more items to that list, such as Google's founders having to promise not to buy a yacht larger than Larry Ellison's Rising Sun and not to complain if his MiG fighter jets fly over their houses. But those concessions would actually be much less hurtful -- and not much more humiliating -- than any modification of Android's business model and/or rewrite of Android's virtual machine.

If you'd like to be updated on the smartphone patent disputes and other intellectual property matters I cover, please subscribe to my RSS feed (in the right-hand column) and/or follow me on Twitter @FOSSpatents.

Share with other professionals via LinkedIn: