Sunday, September 18, 2011

Not much left of Google's 20 affirmative defenses against Oracle

[NOTE] This is the first of three posts analyzing the prospects and possible terms of a settlement on the eve of court-ordered mediation between Oracle and Google. The final post is an "executive summary". [/NOTE]

The logical starting point for an analysis of the pre-mediation situation in the Oracle v. Google lawsuit is the state of Google's defenses. After the original complaint in August 2010, Oracle had to amend its complaint to be more specific about its copyright infringement allegations. The amended complaint was filed on October 27, 2010, and Google's defenses to the amended complaint followed on November 10, 2010. The amended complaint and the related answer are still the only relevant sets of claims and defenses.

In this post I recite Oracle's prayers for relief (the decisions it asks the court to take) and will then go over Google's affirmative defenses one by one. The bottom line is that Google's defenses are in bad shape. It's not over yet, and only a small part of the overall evidence is in the public record, but the publicly available material does provide an indication as to where things roughly stand. It's not over yet, but Google would need an immense amount of good luck to come out of a trial more or less unscathed.

Oracle's claims and prayers for relief

Oracle asserted -- and still asserts -- the infringement of seven patents and various copyrights (12 files that fall under literal copying, eight of them involving decompilation; and 37 API specifications).

In this series of postings, I will refer to the asserted patents and the asserted patents collectively as "the asserted IP" (intellectual property). Wherever I refer to only one category of IPRs (intellectual property rights), I will refer to them, specifically, as patents or copyrights.

Oracle asked -- and still asks -- the court to

  1. hold Google liable for infringement of the asserted IPRs,

  2. order a permanent injunction against continued infringement of the asserted IPRs,

  3. order the impoundage and destruction of "all copies made or used in violation of [Oracle's] copyrights, and all means by which such copies may be reproduced",

  4. award Oracle damages, with interest,

  5. triple the said damages based on the allegedly "willful and deliberate nature" of the alleged patent infringement",

  6. award Oracle its costs and legal fees, and

  7. anything else the court may determine in Oracle's favor.

Oracle has the burden of proof concerning all of the averments on which it bases its claims. However, it's important not to overestimate the standard to be applied by the court. This is not a murder case where the charges need to be proven beyond reasonable doubt. Google will have to raise much more than reasonable doubt. For example, Oracle's patents must be presumed valid, and only clear and convincing evidence (as opposed to a mere preponderance of the evidence or just reasonable doubt) can lead to an invalidity finding.

One-by-one analysis of Google's 20 affirmative defenses

Google brought 20 affirmative defenses and additionally raised 15 counterclaims. In terms of substance, the 15 counterclaims merely duplicate certain defenses. Therefore, I will not analyze them separately.

1. "No Patent Infringement"

a) Java-Dalvik differences

After the start of this litigation, some people thought that Dalvik (Android's virtual machine) wouldn't infringe Oracle's Java-related patents since it doesn't execute Java .class files directly. Those are previously converted into .dex (Dalvik Executable) files, and Dalvik has a different architecture. However, an infringement analysis in a lawsuit like this isn't based on such broad and vague considerations. It's about interpreting the asserted patents and comparing their scope to what the accused technology does. Patents can be very broad -- broad enough that certain general techniques related to the performance and security of virtual machines may be infringed regardless of any code file conversions that may take place along the way.

Contrary to misinformation spread by unreliable sources, Judge William Alsup, the federal judge presiding over this litigation, is perfectly aware of the differences between the concept of a programming language and the concept of a virtual machine, and he knows that "Java" can mean one or both of those things. Various of his orders reflected his awareness of the two meanings of "Java".

b) claim construction

The first major milestone in any patent infringement analysis is "claim construction": the interpretation of key terminology that appears in the language of the asserted patent claims. The patent claims define the scope of the monopoly right each patent constitutes. Interpretation of the claims means to assess the scope of the asserted monopoly rights.

In claim construction, an alleged infringer can try to narrow the meaning of the terms in order to be in a better position to deny infringement. That's what defendants try most of the time. Sometimes they also push for broad definitions in order to have better chances of proving that a patent should be invalidated based on prior art.

Judge Alsup wanted the parties to agree on the interpretation of as many terms as possible but offered to interpret up to six disputed terms. On May 9, 2011, he entered his claim construction order. He interpreted five of the controversial terms put before him but refused to construe one of them (at least at that stage) because he determined it "is really at least three separate terms" and therefore exceeds the limit he set.

For three of the five terms, Google proposed narrower meanings than Oracle in an attempt to avert an infringement finding. With respect to one of those terms, the judge sided completely with Oracle. On another one, the judge came up with his own definition, which is simply a better-phrased version of what Oracle declared to be the ordinary meaning of the term. Google tried to narrow the meaning of that term, and the judge rejected that attempt. This is also a clear win for Oracle. On a third term, Google didn't succeed in full but managed to convince the judge that a "symbolic reference" is resolved dynamically. Oracle wanted to avoid this because it expects Google to try to capitalize on it in the further infringement dispute. This is an incomplete win -- but still a win -- for Google.

c) contributory infringement of method claims

With a view to method patent claims that Google could only infringe contributorily (meaning that publishing the software does not yet constitute an infringement, but it is an important contribution to the actual infringement of such a claim by the users), Google wanted to bring a summary judgment motion that would have enabled Android and other free-of-charge software to violate such patent claims with impunity. However, the judge denied Google's request to file such a motion. There will be a summary judgment decision on contributory infringement, but it will only relate to Google's liability under U.S. law for the use its software by device makers outside the United States.

d) summary of situation regarding infringement

There will still be a lot of work to do on the infringement analysis (and let's not forget that only the infringement of valid patent claims matter, which is a different defense), but looking at the extent to which the judge sided with Oracle so far, Google will have a hard time convincing the jury (which the judge will instruct to construe those disputed terms according to his order) that it doesn't infringe at least a significant number of Oracle patent claims.

2. "Patent Invalidity"

a) abstractness fallacy

In its answer to Oracle's amended complaint, Google placed disproportionate emphasis on the theory that Oracle's patents are allegedly "invalid under 35 U.S.C. § 101 because one or more claims are directed to abstract ideas or other non-statutory subject matter". Google started its invalidity defense with that assertion and repeated § 101 again in the second paragraph of its invalidity argument. That's the paragraph that was at issue in the Bilski case, in which the U.S. Supreme Court rejected a business method patent application because it was considered too abstract.

There's no indication in the public record that anyone buys Google's § 101-based "abstract ideas" argument. Those Java-related virtual machine patents are software patents, and there's little reference in them to physical components of the computer they run on. But none of the first Office actions issued by the USPTO concerning the seven patents-in-suit argues on the basis of § 101. This was likely just an attempt by Google to raise false hopes among members of the open source community concerning the post-Bilski legal situation. There will -- and still are -- dreamers and manipulators in that space who claim that all or many software patents can be invalidated on the grounds of being excessively abstract, but that's certainly not the case.

b) reexaminations: first Office actions are only preliminary

I mentioned the USPTO's first Office actions. Google requested reexamination of all seven patents-in-suit. After looking at the patents in light of Google's invalidity arguments, the USPTO found a vast majority of the asserted patent claims invalid on a preliminary basis. However, many patent claims that are deemed invalid at that non-final stage are still reaffirmed at the end of the entire reexamination process, which can take years. I gave examples of reversals involving patents at issue in high-profile lawsuits.

Just like the original granting process is lopsided because the examiner only sees the applicant's arguments (other parties are highly unlikely to take action at an early stage since there are so many patent applications under examination at any given point int ime and third parties don't know yet which ones will be asserted against them later), first Office actions are also unbalanced because they are issued before patent holders get the chance to make their argument, which Oracle is working on. At a recent virtual bootcamp organized by the Electronic Frontier Foundation, a speaker referred to a study that looked at how the different examiners processing reexaminations tend to decide at the stage of first Office actions. The study found that there was a considerable number of examiners who had a 100% rejection rate at this stage -- unbelievable, but I don't doubt it.

Since those first Office actions are so unbalanced, courts usually don't admit them as evidence. So if this case goes to trial before there are final Office actions concerning the patents in question, the jury will probably not even be informed of the USPTO's preliminary findings. It will, instead, be instructed by the judge on the basis that patents enjoy a presumption of validity that can only be overcome by clear and convincing evidence.

c) reaffirmed claims enjoy enhanced presumption of validity

On at least one of the challenged patents, Google's reexamination request has already backfired. Oracle is taking the necessary steps to have a reexamination certificate for the '520 patent issued as soon as possible, which will give that patent the benefit of an enhanced presumption of validity. Whether that will happen before a trial is another question.

d) impermissible broadening

Google also failed to convince Judge Alsup that Oracle's RE'104 patent should be deemed invalid due to an impermissible broadening. Google raised that argument in the invalidity section of its answer to Oracle's amended complaint, and recently asked the judge for permission to file a summary judgment motion. However, the judge denied it because he thought Google was unlikely to succeed with it.

e) claim construction

While Google mostly tried to take advantage of claim construction (explained further above) in order to narrow the meaning of some of the asserted patent claims, it also made two construction proposals that were part of a plan to push for invalidity. But to no avail.

The judge sided 100% with Oracle when Google tried to broaden the meaning one of the terms, as well as in another case in which Google claimed that a term was "indefinite", which would have been sufficient all by itself to arrive at an invalidity finding.

f) misconceptions regarding narrowing of claims

There was some confusion out there -- not among readers of my blog, but due to what others tried to make the world believe or propagated after they were, without any fault of their own, misled -- concerning a possible obligation for Oracle to reduce the number of patent claims it asserts in this litigation.

Oracle started with 132 patent claims from seven patents. The judge at some point made a tenative -- again, only tentative -- proposal that Oracle should pick the three strongest ones of those.

That proposal was purely driven by case management motivations: ensuring that the jury wouldn't have to spend many months inc ourt to talk about a huge number of patent claims. It had nothing at all to do with the validity of Oracle's patents.

I addressed the related misconceptions, created by a notorious propagandist, in the first part of a blog post on the aforementioned first Office actions.

g) judge disallowed untimely invalidity contentions

In early August, the judge denied Google's request to outmaneuver Oracle with additional invalidity contentions.

h) "you win some, you lose some"

Neither in a trial nor in the reexamination process would it be likely that Oracle can defend all of the challenged patent claims. There are simply too many invalid patents, and patent claims, out there. All that matters for Oracle is that a sufficient group of patent claims are found both valid and infringed, and that Google can't work around them too easily. In fact, even a single "killer claim" could be enough to give Oracle all the leverage it needs regardless of what happens to all other claims. But more likely, the outcome will reflect the you-win-some-you-lose-some nature of the patent assertion business.

While invalidity is the one defense on which Google is far more likely to score at least some points at trial (as well as in the reexaminations going on in parallel) than on any of its other defense theories, and recognizing that Google still has hundreds of invalidity contentions in play, it would be a long shot for Google to narrow Oracle's case to such an extent that the remaining patent claims will not be found infringed, or that Google can work around the remaining ones to the extent they are found infringed.

To the extent that Google tried to win on invalidity-related issues in the process so far, all of those attempts failed miserably, which doesn't bode well for the rest of those contentions.

3. "Patent Unenforceability (Waiver, Estoppel, Laches)"

In that context, Google makes various arguments as to why Oracle cannot go after Google even if it was found to infringe one or more valid Oracle patent claims.

a) "inexcusable" delay in bringing suit

Oracle itself was obviously very quick with filing this lawsuit only about six-and-a-half months after its acquisition of Sun Microsystems. The question here is whether Sun, prior to being acquired by Oracle, knew about this for so long that August 2010 was too late a point in time for bringing an infringement lawsuit. Fending off a patent infringement suit on this basis is a long shot, and there's no indication in the public record that this argument is likely to succeed.

b) Oracle/Sun "did not intend to enforce its patents" according to its statements and actions

The Mercury News, the leading Silicon Valley newspaper, mentioned in a report on a court hearing that Google's lawyers presented a statement by former Sun CEO Jonathan Schwartz about how he didn't object to Google's use of Java. Google also points to one of Schwartz' numerous blog posts.

I found the text of that post and didn't get the impression that this was sufficiently specific to the Android case to constitute a license grant or to result in any kind of estoppel.

c) alleged waiver

Google's defense filing also claims that "upon information and belief, Oracle knowingly waived any right it may have to enforce the Patents-in-Suit". This could again relate to the Schwartz blog post, but this item as well as the previous one (on the intent to enforce) couuld also be part of Google's "open source Java" story, which I'll address in more detail in connection with Google's 17th, 18th and 19th defense.

d) RE'104 broadening (again)

The allegedly impermissive broadening of the scope of the RE'104 patent again came up in the unenforceability context. As I said further above, the judge didn't even allow a motion based on this theory.

4. "Substantial Non-Infringing Uses (Patent)"

Google argued that all of the accused technologies "have substantial uses that do not infringe and do not induce or contribute to the alleged infringement of the asserted claims of the Patents-in-Suit".

This is not even a long shot: it's downright absurd. The alleged infringements, which Oracle documented with detailed claim charts, are certainly of a Dalvik-specific nature. Certain kinds of patent infringement cannot be pursued with respect to, for example, staple ware -- goods that are used for many purposes. Even staple ware can be infringing on its own, but if the infringement itself takes a lot more than staple ware to infringe, then the staple ware part of it won't have liability issues. For example, a software patent cannot be enforced against public utilities that provide electricity with which computers infringing the patent may run. After all, electricity is neutral and has substantial non-infringing use. But that's not the case with the specific parts of Android's Dalvik virtual machine accused by Oracle.

5. "Limitation On Patent Damages"

Google raises some defenses related to damages with respect to past infringement. Google cites three paragraphs of U.S. patent law. This defense is just about damages -- not about liability and the possibility of an injunction.

The first one relates to the recovery of damages that predate the filing of the lawsuit by more than six years. The lawsuit was filed in August 2010, and six years earlier, in August 2004, Android had not even been published. Consequently, the reference to that statute was totally pointless in the first place.

The second statute cited relates to patent marking. Google asked the court for permission for a summary judgment motion on this one. The judge didn't allow it, arguing that the number of patent claims to be asserted at trial would be much lower than the ones in play right now, so he wanted to evaluate the marking question with respect to a smaller number of patents. This one is unlikely to succeed at trial.

Finally, Google's limitation-on-damages argument also makes reference to a statute concerning the assertion of patents that contain invalid claims. There's no indication so far that Google can benefit from that statute in this case.

6. "Misuse"

A patent or copyright holder who overstates the scope of his rights can lose the right to enforce such intellectual property against the alleged infringer. Google's "misuse" defense firstly refers only to the Patents-in-Suit, but then, more generally, to intellectual property.

A misuse defense only works if there's some rather outrageous conduct of the kind for which there is no indication so far in this case.

7. "Exceptional Case"

Google basically suggests that this lawsuit is a frivolous one that should never have been started in the first place, so if Oracle loses it, it should have to pick up Google's legal defense fees. This only plays a role in a scenario in which Google prevails anyway, and in that case it would make only a gradual difference. But at this stage it's pretty clear already that Oracle had a reasonable basis for suing, even in the (unlikely) event that all of its assertions should be defeated.

8. "Use By The United States"

Google points to a statute that doesn't allow the enforcement of certain rights against technologies "used or manufactured by or for the United States, including but not limited to the Department of Defense". Google won't be able to convince the court that Android is essential to the national security of the United States. It may be used by the Pentagon in some ways, but that doesn't make it a national priority. After all, it could be replaced.

9. "No Copyright Infringement"

Google asked for summary judgment that there is no copyright infringement, but failed almost entirely. It can still use this defense at trial, but the judge wasn't willing to buy Google's arguments.

10. "Elements Not Protected by Copyright"

See item #9 ("No Copyright Infringement"). The judge almost entirely disagreed with Google's non-copyrightability argument.

11. "Copyright Unenforceability (Waiver, Estoppel, Laches)"

The arguments brought under that headline mirror all of the unenforceability arguments concerning the asserted patents with the exception of the impermissible-broadening issue that is specific to one of the patents.

12. "Fair Use"

This is a copyright-related defense and received the same treatment from the judge as most of Google's summary judgment motion concerning copyright infringement (see item #9 above).

13. "De Minimis Copying"

This is a copyright-related defense and received the same treatment from the judge as most of Google's summary judgment motion concerning copyright infringement (see item #9 above).

14. "No Intent to Induce Copyright Infringement"

Google's lawyers claim that there was no "purposeful, culpable expression or conduct designed or intended to result in others infringing Oracle's Asserted Copyrights", but Google's intentional copying and other use of copyrighted Oracle material is pretty clear by now. Here's an example.

15. "Independent Creation"

This is again a copyright-related defense, and unconvincing. It didn't even play any major role in the discussion of Google's summary judgment motion on copyright.

16. "Third Party Liability"

This is still copyright-related. When Google raised this argument about third-party liability, it impressed only the impressionable. In reality, it was always clear that one cannot commercial distribute infringing material without some kind of liability. Also, only Google develops the Android source code. It's not an open project to which anyone can contribute. Google's declared strategy includes the rule of "Do not develop in the open".

This argument didn't play a role in the discussion of Google's summary judgment motion. I doubt that it's going to be of much importance at trial.

17. "License"

Google claimed to be "licensed to use the Patents-in-Suit and the copyrights in the works that are the subject of the Asserted Copyrights". If Google had implemented Java according to its official specifications, or if it had used GPL-licensed Java code from Oracle/Sun such as the OpenJDK, then one could at least talk about this defense. But since Google adopted only parts of the Java standard (which Oracle calls "fragmentation"), it can't claim to be licensed in any way. There's no indication that the judge believes that Google already has a license.

18. "Implied License"

Google again refers to "Oracle's and Sun's statements and actions". All that I have seen in this regard is a blog post by former Sun CEO Jonathan Schwartz, and like I said before, I don't consider that one to constitute a license grant, not even an implicit one.

An implied patent license would be the case if Google used Oracle's GPL-licensed code, but for commercial reasons it chose not to.

19. "Unclean Hands"

The way Google's lawyers worded that section appeared to be duplicative of the "misuse" defense (item#6).

20. "No Injunctive Relief"

This last item is a highly important one. As I've mentioned before and will mention again in connection with the mediation effort, even the substantial amount of damages that could be awarded in this case is not the number one priority. The most important question in this case is Oracle's ability to obtain an injunction. That would give Oracle all the leverage to impose its business terms.

Google's lawyers argue that "Oracle is not entitled to injunctive relief" and imply that it would be sufficient for Oracle to be awarded damages should any valid IPRs be found to be infringed. However, an injunction is pretty much a given in the event that Google's patent infringement is proven to have been willful, and the so-called Lindholm email could play an important role in that regard, provided that it is admitted as evidence against Google's claims of attorney-client privilege.

Apart from documents that may prove willful infringement, Oracle would also have a chance of obtaining an injunction based on the aforementioned "fragmentation" argument. But that one would have to pass the four-factor test.

[NOTE] This was the first of three posts analyzing the prospects and possible terms of a settlement on the eve of court-ordered mediation between Oracle and Google. The final post is an "executive summary". [/NOTE]

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 and Google+.

Share with other professionals via LinkedIn: