Monday, June 27, 2011

Reexaminations of Oracle patents: preliminary rejections of claims are relevant but not final

The landmark intellectual property dispute between Oracle and Google has recently drawn more attention than any other litigation of its kind. Unfortunately, that high level of interest also enables inaccurate reports and misleading headlines to spread like wildfire.

In particular, there's one propaganda site out there (with an apparent and long-standing bias in favor of certain companies) that previously wrote: "Oracle Ordered to Reduce Claims Against Google From 132 to 3". That was simply not true. Regrettably a number of reporters amplified that untruth. But that was merely a tentative court order. This is what the judge wrote right at the start of that "order":

Having reviewed the parties' memoranda [...], the Court proposes a three-step process, as below. Each side may file a five-page (double spaced, twelve-point Times New Roman font, no footnotes, and no attachments) critique of the tentative schedule by NOON ON MAY 6, 2011. After taking any critiques into account, the schedule will be finalized. [emphasis mine]

In light of the above, which could hardly be any clearer, one had to be utterly obtuse or, like in this case, a notorious propagandist to claim that Oracle was actually ordered to reduce claims to any particular number. In reality, the final decision on this won't be taken until October. The most recent position by Oracle (also a tentative one) was to assert 21 claims at trial -- seven times as many as that false headline had stated. A filing on May 18 revealed that Oracle even rejected a Google compromise proposal of 10 to 14 claims.

Also, the proposed kind of reduction was just a routine matter of case management (with a view to the court's limited judicial resources), absolutely unrelated to the merits of Oracle's claims. The judge felt that Oracle should be able to concentrate on a few claims. That thinking is important now in connection with reexaminations and preliminary rejections -- I'll get to that further below.

From the same poisoned source that already misstated the envisioned narrowing of claims, some new confusion has emanated. The pattern is the same as before: the audience and reporters were misled by an article creating the perception of a done deal, theoretically mentioning the preliminary nature somewhere (at the end of a very long document) but practically hiding the fact from most people's view.

In a recent article on the ongoing reexaminations of Oracle's patents-in-suit, that blog talked about "rejected" claims and "[O]ffice actions" without mentioning in the article itself that the rejections were just preliminary and the Office actions were so-called first Office actions, by definition non-final. That information could only be gleaned from the text of the relevant document. That text was also published, but few people read such long and highly technical and legalistic texts until the end. The body of the blog post should have stated the preliminary nature of this upfront. It failed to do so (presumably on purpose), and as a result, various reporters ended up writing stories that were either inaccurate or at least missed the point.

By contrast, I strive to point out the scope and effect of every court filing I report on. For example, I reported a lot on Oracle's damages claims but never portrayed Oracle's position as a court order...

First Office actions issued so far

At this point, the U.S. Patent and Trademark Office (also just referred to as "Office" in this context) has issued what is called first Office actions concerning four of the seven patents Oracle asserted against Google. It's true that many of the claims under reexamination have been "rejected" in a temporary sense, meaning that the examiner is inclined to reject unless convinced not to do so in a final Office action. We will soon see Office actions concerning the other three patents (especially for two of them they could be issued any day).

Back in May I tweeted about a post on a very reliable and thoroughly-researched blog named Reexamination Alert. Its author, Scott Daniels, is a seasoned, Washington DC-based patent litigator who takes a particular interest in the impact of reexaminations on lawsuits. His blog is highly recommended.

Oddly, the blog mentioned further above -- the one that misled people as to the first Office actions -- also performed poor statistical analysis. That blog came from the assumption that Oracle asserted all 168 claims of the 7 patents-in-suit. The original complaint just asserted the patents without specifying claims, but actually, the publicly available filings (which that blog routinely publishes) indicate that Oracle's infringement claim charts related to only 132 of them, and in the meantime Oracle had to present a shortlist of 50 claims (which is, however, not in the public record, at least not yet; also, the 50 claims on the shortlist must be a subset of the originally asserted 132 claims). This is a list of the 132 claims asserted according to the claim charts, along with comments concerning their current reexamination status:

So if those first Office actions were final, two of the four patents would be lost entirely for Oracle's purposes -- a fact that others didn't point out in their reports. A third one would be seriously weakened, while the '520 patent is in relatively good shape. And for the other three patents we'll have to wait and see.

I'm sure that Google is quite happy about the current status of the four reexaminations that have already led to first Office actions. I'm also sure that Oracle feels serious pressure. But let's put this into perspective. In the next section I'll provide high-profile examples of similar first Office actions that were reversed. In the final section, I'll also talk about what relevance those numbers have. That's actually a whole set of questions. It concerns the likelihood of a trial in the fall vs. a stay pending reexamination, the probable outcome of a trial (if there is one prior to the finalization of these reexaminations), and the future enforcement of a possible judgment in Oracle's favor.

First Office actions are frequently reversed

After a first Office action is issued, the holder of the patent in question has the opportunity to dissuade the examiner from finally rejecting claims that are "rejected" on a preliminary basis. This includes arguments in favor of upholding but also the possiblity of rewording claims (so long as they are not broader than the original ones). Conversely, those with an interest in invalidation will try to defend the ground they have gained on a preliminary basis and will additionally try to present reasons for rejecting claims the examiner is inclined to uphold.

A high-profile patent dispute that recently ended before the Supreme Court was i4i's XML patent lawsuit against Microsoft. On his Reexamination Alert blog, Scott Daniels wrote:

"The PTO granted the reexamination and issued an Office Action, rejecting all pending claims on June 15, 2009. The '449 patentee traversed the rejection, and late last month [April 2010], the PTO withdrew the rejection."

That reversal was worth hundreds of millions of dollars to i4i...

Sometimes companies defending themselves against a certain patent are prematurely jubilant over non-final Office actions. That's not the case with Google here -- they don't make noise about those reexaminations and can't be held responsible for biased bloggers who try to support them. But it happened to Sun, the very company whose patents are now being asserted by its acquirer, Oracle, against Google.

More than three months ago I reported on Red Hat's license deal with Amphion and its subsidiaries DataTern and FireStar. That patent was also asserted against Oracle and Sun a few years ago (when those were still separate entities). Sun's then-general counsel said this on July 11, 2008:

"These things take time, but last week, we received a response from the PTO in the form of an office action rejecting all of the claims in the patent based on the prior art submitted by Sun. Obviously, we are delighted to get this validation from the PTO. Firestar has two months to overcome this rejection, but given what we presented to the PTO, we believe it will be a challenge for them."

That blog post is a typical example of "He who has the last laugh...". Later, Amphion (FireStar's and DataTern's parent company) bragged about the following outcome:

"[T]he 6,101,502 patent successfully completed a re-examination procedure in the U.S. Patent and Trademark Office in 2009. All 18 claims of this patent were reaffirmed and 26 new claims were added."

That patent is under attack again, and maybe at some point it will be invalidated if someone comes up with the right reasons for that. But the first Office action that Sun celebrated was clearly overturned.

Not only can the examiner change his mind before his final action but it's also possible that an examiner's final action gets appealed to the USPTO's Board of Patent Appeals and Interferences (BPAI). Software patents concerning which the examiners issued final rejections that were recently reversed by the BPAI include (among others):

And even a BPAI decision isn't necessarily final. It can be appealed to the Court of Appeals for the Federal Circuit, and subsequently even to the Supreme Court if the highest court in the United States accepts the case.

The relevance of numbers, put into perspective

In connection with the proposed reduction of claims I already said that the federal judge presiding over Oracle v. Google proposed (on a tentative basis) a narrowing down to only three patent claims (out of 132). Like I said, that was purely based on efficiency considerations. It had nothing to do with how good or bad Oracle's asserted claims are.

That proposal by the judge shows that he didn't feel Oracle would have to successfully assert large numbers of patents to achieve its objectives in this lawsuit. Judge Alsup presumably thought that Oracle should be able to pick its three strongest patent claims, making it reasonably likely that at least one or two claims succeed.

If only one or two or three patent claims are deemed to be infringed, that could be sufficient. What Oracle wants is to assert its ownership of and control over Java. If it had one "Java killer patent", that would do the job -- but it would have to be so broad that Google wouldn't be able to easily work around it. Even seven successful assertions wouldn't be helpful in the future if Google could engineer around all of them.

Oracle obviously wants to assert more patent claims. That's because having more irons in the fire means an increased likelihood of prevailing. But after you have prevailed, the number of successfully asserted claims may be secondary or even irrelevant.

That thinking is also reflected by a damages report submitted to the court by Oracle's damages expert. Let me quote a Google statment on that report. I already quoted that statement in my previous blog post

"Yet Cockburn incorporates billions of dollars into his reasonabl[e] royalty calculation for the time period between 2018 and 2025, when the '720 patent would be the sole remaining patent."

If Google's representation is correct, Oracle's damages expert assumed that even if only one infringed patent is still valid at a given time (the others would already have expired in or after 2018), the commercial relevance of the infringed intellectual property is (potentially) the same as if several patents are infringed and still in force.

Oracle's seven patents cover different aspects of Java. Five of them relate to performance increases, and two of them to security features. If any of the five "performance patents" is successfully enforced and Google has to substantially reduce the performance of Android apps, that could be all the leverage Oracle needs. If any of the two "security" patents is successfully enforced and Google can't engineer around it without creating a security gap, Oracle will also be in a very strong position.

Numbers are important when it comes to the strength of a portfolio in general. In that case, it's about numbers and other factors, such as diversity.

ZDNet's Larry Dignan recently quoted an analyst who said that Oracle owns more than 20,000 patents compared to Google's number of approximately 700. This doesn't mean that Oracle's portfolio is 30 times stronger: the difference could be more or less than that. But the ratio of the sizes of the two portfolios is an important indication because there's at least no obvious reasons for which Oracle's patents should be of a lower average quality than Google's.

It's quite likely that Oracle's vast portfolio contains other patents (other Java-related patents as well as patents in all sorts of fields of technology) that Oracle could assert against Google if the dispute ever widened (such as by Google trying to countersue Oracle).

The bottom line is that numbers are more important prior to a ruling -- and prior to a lawsuit -- than after winning in court.

Judge Alsup has previously indicated that if Oracle wants to assert more patent claims at trial than he considers "triable" (in terms of not being too burdensome on the court and the jury), he may stay the case until all reexaminations of Oracle's patents-in-suit have been completed. A stay doesn't make the case go away -- but Oracle would want to avoid it if possible.

The fact that the USPTO takes a generally rather skeptical perspective on Oracle's patents-in-suit would certainly make the judge lean ever more toward a stay. We'll have to see what the first Office actions concerning the other three patents look like, but the probability of a stay is already higher at this stage than it was a few months ago.

I think it's very likely that Judge Alsup actually wants the case to be resolved within a few months, but it depends on Oracle's cooperation in terms of narrowing its claims. The problem for Oracle is that the worse its patents fare in reexamination, the more claims Oracle will want to assert. It would be possible for the court to rule in Oracle's favor even on patent claims that the examiner rejected on a preliminary basis (the jury would probably not even be informed of that fact). But with each patent claim that gets rejected on a final basis, the value of such a verdict would be reduced.

At some point, even Oracle would prefer a stay over an assertion of too few claims if there's too much of a risk of any or all of those being invalidated later. In that case, Judge Alsup and Oracle might agree to disagree, with the result of a stay. But we're not there yet. In the coming months, we'll see what happens to the other three patents, and maybe there will be additional Office actions withdrawing some of the proposed (non-final) rejections. Maybe Oracle will reword some claims or introduce new ones.

I look forward to future reports by the authoritative Reexamination Alert blog on these reexaminations. This is important stuff. But if Oracle has a few strategically important claims in place that aren't affected by reexamination, and if the number of those elite claims doesn't exceed Judge Alsup's (unknown but likely rather low) limit, then this case will go to trial at the end of October.

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: