Skilled Searcher Test Allows Estoppel for Unknown References

by Dennis Crouch

Ironburg Inventions Ltd. v. Valve Corp., — F.4th —, 21-2296 (Fed. Cir. Apr. 3, 2023)

The recent decision in Ironburg Inventions Ltd. v. Valve Corp. has significant implications for post-IPR estoppel under 35 U.S.C. § 315(e)(2). In this case, the Federal Circuit adopted a “skilled searcher” standard to determine whether grounds not raised in the original IPR “reasonably” could have been raised at the time. The court also addressed the burden of proof in estoppel cases, with the Federal Circuit holding that it lies with the party seeking the estoppel — a ruling that aligns with traditional practice and the best reading of the statute.  These standards and procedures are important because they provide clarity on the scope of estoppel and help stabilize the law, particularly given diverging district court rulings on these issues. One interesting aspect of Ironburg is the split 2-1 decision on indefiniteness, which I’ll address in a separate post.

Before delving into the case, it is also worth noting that the issue of estoppel and its nuances are currently pending before the U.S. Supreme Court in Apple v. CalTech. In Apple, the Federal Circuit broadly applied the estoppel doctrine, and the Supreme Court is now considering whether to grant certiorari.

Lack of Knowledge and a Skilled Searcher Test: After being sued for infringement, Valve challenged Ironburg’s game controller patent via inter partes review (IPR).  That administrative action ended with a final written decision favoring the patentee — i.e., Valve lost.  Back in district court, Valve attempted to challenge the validity again – this time based upon non-petitioned grounds that it gleaned from a competitor’s IPR petition against the same Ironburg patent. The district court barred Valve’s defenses as estopped under § 315(e)(2)  after concluding Valve “reasonably could have raised” those grounds in its original IPR. Valve argued that it had no knowledge of those references at the time of its IPR petition and therefore could not have included them as challenge grounds. However, both the district court and the Federal Circuit held that lack of knowledge is not a complete excuse.  Rather, in cases where the IPR challenger lacks knowledge of the prior art the appropriate test for estoppel is what “a skilled searcher conducting a diligent search reasonably could have been expected to discover.”

Burden Shifting: One question on appeal is which party has the burden of proving that the ground reasonably could have been raised in the IPR.  The district court (seemingly) placed the burden on the patent challenger to prove that it could not have reasonably raised the issue. In some ways, that burden makes sense because the IPR challenger is in the best position to understand and prove its level of knowledge  (or lack thereof) at the time.  On appeal, however, the Federal Circuit flipped that ruling — holding instead that the burden lies with the party seeking the estoppel to prove estoppel.  In my view, this result better aligns with both tradition and the reading of the statute.

Digging a bit into the weeds: the district court’s holding basically concludes that Valve reasonably should have been able to find the references since the other IPR challenger found them.  On appeal, the appellate court found some faulty logic — an inherent and unproven assumption made by the district court was that the other IPR challenger used only reasonable diligence to find the art. If, on the other hand, it took “extraordinary measures” to find the references and construct the challenge ground, then no estoppel should apply.

The patentee had won a $4 million judgment at the original trial, but on remand, the district court will need to reconsider its estoppel decision. If the result is no-estoppel, Valve will get a third chance at invalidating the patent. This case underscores the importance of conducting a diligent search to identify all potential grounds for invalidity and then putting the best grounds forward in the IPR petition.

40 thoughts on “Skilled Searcher Test Allows Estoppel for Unknown References

  1. 11

    Finding evidence to support patent law is a learned skilled. Like any skill, people who are motivated will learn the skill better than others. Expert ability in this function exists in the private sector. It takes a full year to train a private sector searcher who works full time, 2k hours, learning the skill followed by more learning over time. Contrast that with a patent examiner who is not trained. New examiners receive two days of training in the patent academy and then it is up to their SPE. They might work 200 hours searching in that first year. Factor in that the Office considers searching i.e. the evidence function, to be a basic skill, meaning anyone can do it, leaves applicants having to rely on an untrained part time worker to perform this function. What would happen if evidence was treated this nonchalantly in criminal.

    A diligent search has a name: patentability search. A patentability search looks for evidence to apply to features of the invention. Any invention has a limited number of features limiting the articles of evidence necessary to establish that a feature is known. A strong article of evidence will disclose multiple features. At the end of the process you will have 8-12 references allowing the lawyer to judge the inventive space. Patentability searchers are usually based on technical disclosures not a finished application. You cannot determine patentability unless the application has been subjected to a patentability search. Patentability is an ex ante activity.

    A validity search looks for evidence to apply to claims. Claims provide a fixed target. If a document contains one or more features of the invention but cannot be applied to the claim it will not be forwarded as evidence by the skilled searcher. You cannot conduct a validity search before a patent issues. Validity is an ex post activity.

    Patent examiners are plainly tasked with the goal of validity when all this is possible, at this stage of the process is patentability. It is impossible to do patentability and validity at the same time.

    Patent examiners are not skilled searchers. Skilled searching is the province of private enterprise.

    Within the profession (I use this word loosely) there is no such thing as an ordinary search vs an extraordinary search. Other than the objective of the search searching is searching.

    This decision is a significant step forward in terms of the court recognizing that search skill is a separate and independent function in patent law.

    1. 11.1


      Thank you for your perspective.

      Unfortunately, as much “skill” as you seem to want to import into searching, the Courts are just not likely to take your viewpoint.

      Rather, the fact of the matter being that the Federal Agency charged with the task has placed a MUCH lower skill level to it.

      Conversely (and rather non-intuitively), this lower skill level determination creates the inference that searching must be easy, and ANY diligent search should find any any all prior art.

      Let’s remember — this entire AIA alternative scheme was explicitly formulated and constrained to the “easy” prior art aspects as a Short and Complete alternative to Article III litigation.

      Trying to turn this “searching” aspect into some complicated or difficult task then (as much as that may comport with the viewpoint of professional searchers), simply does NOT comport with either the legislative record or intent.

      1. 11.1.1

        Thank you as well Anon. I absolutely need to hear what opponents to my perspective have to say.

        The lower skill level imposed on the task by the PTO is perfectly acceptable for 95% of all applicants (based on the common assumption that 5% or patents are valuable). Most large entities do not patent to either assert or license so an industry standard patentability search is not necessary and the system works fine for them, no complaints. Among the 5% are innocent patent holders who believed the promise of the constitution and got screwed. Only the 5% require a concerted effort to look for evidence.

        I believe you have conflated my words “searching is a learned skill” to something I did not say. It is a huge leap from learned skill to “complicated or difficult.” It is no more difficult than any other learned skill than say HVAC repair, electrician, or certified auto mechanic. The work is only easy for the person who has been trained. Are you suggesting that anyone can do it?

        When you say “…should find any an[d] all prior art.” Do you mean cumulative? Any and all evidence would be a “collection” or “state-of-the-art” search neither of which addresses patentability.


          Thank you again for further explaining why pre-IPR issued patent prior art searching in the face of threatened or actual law suits is not comparable to the much cheaper [and thus shorter] application prior art searching that the vast majority of applicants are willing to pay for.
          Another good reason is that examiner searches are primarily on the claims originally filed, which all too often are much broader in scope than the issued claims written to distinguish those very search results. Especially with the many applications filed these days without any pre-filing prior art search by the applicants.


          First, I appreciate your willingness to engage – that willingness is of short supply, as most pundits run when they see a viewpoint that they do not like.

          Let’s take this point:

          The work is only easy for the person who has been trained. Are you suggesting that anyone can do it?

          It is less that “I” am conflating anything and more that your perception (and I mean this in the professional sense) has conflated the meaning.

          I very much do ‘get’ the pride you have in being highly skilled and recognizing the value of being highly skilled. But that pride is your downfall, Even in your response here, you seek to insert a value stratification.

          The lower skill level imposed on the task by the PTO is perfectly acceptable for 95% of all applicants (based on the common assumption that 5% or patents are valuable).

          That’s a logical fallacy – in context.

          Whether or not ‘it is perfectly acceptable’ on your value insertion is error.

          Further, it is NOT based on ANY assumption of any level of patents ‘being valuable,’ as ALL granted patents have the same legal status, and it is the separate market drivers that may determine market value. YOU have conflated two very different things.

          Perfectly rigorously pre-searched (prior to USPTO searching) patents often still have zero market value.

          Yes, I do get that those that can MORE afford to play the patent game (and make it a Sport of Kings) conversely can afford to care less about any single particular patent (IBM’s long reign of patent count king testifies to that).

          Yes, I do get that it is a far better strategy to do as rigorous a search as possible prior to writing a patent (for a number of reasons).

          So yes, I do “get” the reality of value stratification in real life.

          But the context here is NOT about that value stratification. The context here IS about ‘the Court AND the USPTO doing MORE than merely suggesting anyone can do it, but outright stating that BECAUSE anyone can do it, the reasonable standard is simply far lower than your professional standard.’

          So while you do not directly state my “complicated” factor, your words do in fact imply that as you are in fact showing the pride of your more highly skilled training than the (base) training than what ‘qualifies’ as learned skill – and the more important legal context at point as to what that lower level means.

          As I stated, that lower level thinking or what it takes (the thinking by the USPTO that anyone CAN do it) necessitates a conclusion that “skilled searcher” as the court uses the term is NOT THE SAME as how you view your own craft.

          It is this very real difference in views that drives the point that I have presented.

          The courts simply are not seeing “certified” as you would see it (for example in your statement of, “no more difficult than any other learned skill than say HVAC repair, electrician, or certified auto mechanic.” you use ‘certified,’ wherein the USPTO and the courts would NOT bother with that word, and would indeed set the standard at an UNcertified level.**

          As I stated, it is this LOWER level that means that in the legal sense, a “diligent searcher” WILL have results that differ from your own professional perspective – and the elevated results [in your professional context] of say a cumulative or state-of-the-art search IS the same as what “just about anybody” would come up with.

          And the phrasing ‘cumulative’ IS what the USPTO uses for its search mode – as the examiner’s “consideration” is tracked to their bringing ‘the best’ art for rejections – cumulative of all art.

          Again – I ‘get’ the professional viewpoint of real world value stratification, but that is just NOT the context of the legal point here.

          (I would ask that you explicate your use of ‘patentability’ in your phase of, “Any and all evidence would be a “collection” or “state-of-the-art” search neither of which addresses patentability.” as this does not appear to be a correct legal use of the word – but you may be using that word differently.)

          ** By the way, I have been perhaps the leading member of those regular commentators here that PUSH for higher quality examination as a true worthwhile goal of legislative change. The problem is NOT the current set of law, but instead how the OFFICE enacts those laws. It is NOT – As Paul would deign to phrase it – the fault of applicants (or even lack of search after a first action) that far more rigorous searches are done in litigation mode. It is not – as the typical examiners here would have it that their own internal metrics are sufficient.

          The law is clear that examination – which includes search – is to be done to render an examined patent valid under the law. The set price for a search does NOT mean that an inferior search is ‘perfectly fine.’ My clients want patents – and they want valid patents. The Office AND Examiners taking shortcuts hurts my clients. It has been made abundantly clear – and across many different actions, that the pricing offered by the Office is NOT reflective of the level of effort, and it dissembling for those that know this to continue to misrepresent the situation.

          Let me add one [complicating] facet here – I can easily see AI rendering your valued and more highly skilled trade craft one of buggy-whip manufacturing, and bring the actual more highly skilled level DOWN to a level of just anyone pushing a button without truly understanding the craft work. AI may actually resolve the (mis)perceptions of paid price and level of search.

  2. 10

    Seems from the cheap seats that in many instances, an IPR is a poisoned chalice barely worth the putative cost savings, esp. because the WMD of subject matter eligibility is excluded from the process. I’d prolly take my chances in an Article III court for a few dollars more, if even, these days.

    1. 10.2

      ? Filing pre-trial IPR petitions and filing Alice type 101 subject matter ineligibility motions are neither conflicting or incompatible. Both are normally decided faster, vastly cheaper, and more likely to succeed than trying to win a jury trial on 103 unobviousness with patent or publication prior art. Since it is normally only law suit defendants using either, their client interests [not such adverse plaintiff client or plaintiff attorney views] will be controlling.

      1. 10.2.1

        The slow parts of a district court proceeding are mostly pre-discovery/experts. IOW, the cheap parts.

        Filing pre-trial IPR petitions and filing Alice type 101 subject matter ineligibility motions mean two entirely separate proceedings. That’s costly and not speedy, and the IPR may well poison the Article III effort.

        The IPR isn’t as travel intensive as a court case, but it’s not local either for 90% of petitioners.


          There are still a lot of instances out there where the 101 problems are so laughably obvious that it does not take a whole lot of time or skill to tank the claim.

  3. 9

    This is not about the burden-shifting issue, rather to note some confusion below about the IPR statute, including § 315(e)(2). IPR estoppel [as for reexams] is expressly for asserted patents or publications prior art OR any other such art that the petitioner “reasonably could have raised.”
    This Ironburg Inventions Ltd. v. Valve Corp. decision hold that: “..the “skilled searcher” standard is consistent with the statutory requirement that a petitioner be estopped from asserting “any ground that the petitioner . . .reasonably could have raised . . . Both parties before us agree we should apply this standard here, . . . and we do so. . . . Accordingly, we hold that, provided the other conditions of the statute are satisfied, § 315(e)(2) estops a petitioner as to invalidity grounds a skilled searcher conducting a diligent search reasonably could have been expected to discover, as these are grounds that the petitioner “reasonably could have raised” in its petition.”
    Note especially that a “diligent search” is part of the “skilled searcher” requirement. As it should be, since obviously the amount of searching time purchased from the diligent searcher is related to the likelyhood of finding more relevant and material prior art.
    This, however, still leaves judicially undecided the question of how much should be spent on such a patent & publication prior art search sufficient to avoid estoppel? Years ago more than $20,000. was considered a reasonable search by some. But that is a miniscule % of a typical patent suit defendant’s pre-trial expenditures, much less their infringement exposure, and it is only 80 hours at $250/hr. Thus, not sufficient in my view. I would be interested in what amount others think might be sufficient?

    1. 9.1

      P.S. Since on remand here the district court below is to reconsider its estoppel decision, the latter question may not be academic?

    2. 9.2

      “ Years ago more than $20,000. was considered a reasonable search by some. But that is a miniscule % of a typical patent suit defendant’s pre-trial expenditures, much less their infringement exposure, and it is only 80 hours at $250/hr.”

      I would think that for most of the grown-up arts, increases in digitization (database size) and computer speed would mitigate against needing more time than that.

      In the logic arts, of course, it’s a different story (and will remain so absent any effort to force claim drafters to comply with a systematic format to facilitate searching and comprehension of the allegedly new logic and/or algorithm).

      1. 9.2.1

        ^^^ the cognitive dissonance is blaring.

        “Grown-up arts” are simply not what you feel them to be. Your animus against the computing arts blinds you. Refusing to understand Kondratiev waves of innovation is the epitome of NOT being grown-up.

        This is of course no surprise.


          I have nothing against the logic arts, also known as the art of writing instructions for instructable machines that use logic to process data. But the idea of shunting that art into a utility patent system designed for non-abstract arts was always idiotic and will always remain so.

          Pretending otherwise in 2023 is the hobby of deluded f o o l s or worse.


            Feel free to abstain from any of that utility that is there.

            Note: this is not a new invitation for you, and it is guaranteed that you will still not take up this invitation and will still avail yourself of the utility of these innovations.

    3. 9.3

      Payment to the person looking for the evidence in the subject matter of this case would be 2-4k, probably on the lower end. If the question is “how much does it cost to pay the person who finds evidence to invalidate patent claims,” you need to ask people who do that job. I question the cost (especially from years ago). $250.00 an hour may be what a law firm charged however the business model is that the task is contracted out.

      To my knowledge there is only one study that looked at the question of prior art and it reveals a lot. The data is for IPRs in 2017

      link to

      66% of all wins were based on patent information and 22% in a separate “art of record “category. In both categories the skilled searcher found the invaliding art using standard search procedures.

      Years ago the invalidating standard was “every element of the claim had to be disclosed in a single document” i.e. a 102. Back then patent information was not well organized and access especially to foreign patents was very limited. Depending on how far back you go (15 years) we relied on a system built for paper that was transitioning to meeting modern demand. The task has become much more efficient in recent years coincident with access to foreign documents.

      Today a document that discloses multiple features of the invention (using very similar language), and was not utilized during prosecution is a sufficient primary reference combined with 1 or 2 secondary references that provide the missing features (using similar language), i.e. 103. This is where the work of the searcher ends.

      Because of advancements in modern tools finding pertinent information has become easier and practically impossible to repress. As recently as Y2K bad actors could still hide references. It was once common for a firm to hire multiple searchers (in the US and in foreign countries) but the aforementioned advancements has practically eliminated that necessity.

      Public sector searchers also sell infringement searches but this has a different legal objective than validity and results are not likely to overlap. For a skilled searcher the cost are in the same range as validity.

      1. 9.3.1

        Thanks for that cited article, which you indicate is the only one on this subject of What prior art was being cited in IPRs: Stephen Yelderman, Prior Art in Inter Partes Review, 104 Iowa L. Rev. 2705 (2018), available at:
        link to

  4. 8

    Seems like the CAFC got it right. Estoppel should be the default position under these sorts of circumstances. The question about how to handle the situation where prior art “suddenly appears” that couldn’t “reasonably” have been ascertained by the patent challenger in an earlier-filed IPR is interesting. I’m trying to imagine the most common examples of a “hidden” “publication” that would overcome the default estoppel but the coffee hasn’t kicked in. Foreign language publications without any “online”presence, I suppose?

    1. 8.1

      Yes, a foreign language publication not available “online” [or in a major technical library] might be one example of a publication that might be excused for not being found by a “skilled searcher” conducting a “diligent search.” Others may include non-secret technical service manuals, descriptive literature or technical papers distributed to those in that art at trade fairs or technical conferences, indexed thesis’s, etc. [Fed. Cir. decisions on what can be a “publication” are quite broad.] Nor would being “behind a paywall” be likely to be an excuse for not finding a publication.
      [It would be harder to find excuses for not finding relevant material in issued patents or published patent applications these days since most, or their equivalents in other countries, are available on-line, unless perhaps the relevant disclosure was only in a drawing?]
      Failure to search for similar subject matter among the prior-filed patents and pending applications of the patentee – suit plaintiff itself, and prior publications of the named inventors of the patents in suit, might also not be excusable?


          “Mere Earthlings, I can find any prior art for any issued patent or application (and if not, I’ll make something up that will fool you).”

          “Now bow down to your new searcher!”

          — Chat GPT


          “ what is an AI-assisted skilled searcher enabled to find”

          The virtual nose in front of its virtual face and little else, at the moment.

  5. 7

    Is a “skilled searcher” performing an ordinary search a reasonable test standard?

    In my experience, at least with regard patentability searches, when you give two different searchers the same search, they each return two different sets of references, often (possibly usually) with no overlapping references. Perhaps validity searches are different in this regard, but I still find it hard to believe that anything below an extraordinary search is likely to find all references has any sort of precision to it. There also is a difference between finding a reference and recognizing its potential significance. The most relevant part of the reference can sometimes be a phrase berried in the reference, in a tangential remark, which is unlikely to be found without a thorough reading of the entire reference.

    Of course, in hindsight it is easy to say, a searcher of ordinary skill should have found the reference.

    1. 7.1

      “ In my experience, at least with regard patentability searches, when you give two different searchers the same search, they each return two different sets of references, often (possibly usually) with no overlapping references.”

      This is opposite of my experience. The only times this happens is when someone provides the searchers with a claim and zero context and/or the searchers are searching through two completely different sets of data (i.e., one is searching for text strings in claims only and the other is searching for text strings in abstracts only).

    2. 7.2

      Note that you can only really give two different searchers the “same patentability search” if you give them the same claims and spec to search with adequate searching scope and time, as for IPRs, but usually not for inventor disclosures for proposed patent applications.
      Which of the uncovered prior art documents is “the best” can of course also be somewhat subjective, differing between different searchers, petitioners or IPR APJs.

  6. 5

    >In my view, this result better aligns with both tradition and the reading of the statute.

    I’d say the exact opposite. The original concept of IPRs was that the defendant could pick an Article 3 court OR the PTAB. See also the promise of “gold plated patents.” The idea that a defendant can get two-bites-at-the-apple merely by first performing an “ordinary” search for the PTAB bite, then an “extraordinary” search for Article 3 bite, is completely contrary that concept.

    Put differently, the defendant’s 2nd bite should be an extraordinary remedy, something the *defendant* has to justify. Particularly so given the inherent subjectiveness (i.e., semi-randomness) of the modern obviousness inquiry.

    1. 5.2

      Fake edit: I guess I’d be OK with a rule where the patentee had to formally plead estoppel (i.e., just that “there a previous PTAB case”). Then, after that very-basic fact is established, then the burden shifts to the defendant to establish why the estoppel should not apply i.e., an exceptional circumstance exists.

      Substantively, the key is that the patentee wouldn’t have to prove that a skilled searcher would/should have found the new reference and/or evidence.

  7. 4

    The Federal Circuit continues to misconstrue the term “ground” as used in the Patent Act. I think that term refers to legal theories grounded in 102 and 103–e.g. public use bar, on-sale bar, prior knowledge, obviousness–not particular prior art combinations. All should normally be barred once you lose an IPR.

    The term “ground” appears to come from the 1952 Act. 282(b) sets out defenses to an infringement claim, including (b)(2): “Invalidity of the patent or any claim in suit on any ground specified in part II as a condition for patentability,” referring to Sections 102 and 103. The key is to remember that Section 102 was long the center of action, and the various 102 defenses (“grounds”) each had their own jurisprudential lineage. It’s relatively recent that they were all condensed into “anticipation” and relegated to the backbenches of patent litigation.

    The IPR statute seems to borrow “ground” from 282(b) and continually draws a distinction between “grounds” and “evidence”. 35 U.S.C. 311(b) lets you petition for IPR on “a *ground* that could be raised under section 102 or 103” but “only on the *basis* of prior art consisting of patents or printed publications.” 312(a)(3) makes you “identif[y]” “each claim challenged, the grounds on which the challenge to each claim is based, and the evidence that supports the grounds for the challenge to each claim, including [(A) copies of prior art patents/pubs, (B) expert decls]. It’s telling that the prior art patents/pubs are listed as one bullet, and expert decls the other.

    In sum, an obviousness ground is an obviousness ground, regardless whether it relies on the teachings of Smith and Lee, or Jones and Conner, and it should be estopped en toto in district court under 315 upon an adverse FWD.

    1. 4.1

      100% correct, that is exactly what i thought when the AIA was first created. you should be estoppel from bringing any 102 or 103 argument of any kind if you already brought a 102 or 103 ground in the IPR irrespective any particular combination of references, rational, secondary considerations, etc. Otherwise the estoppel is fairly meaningless and would be limited to the exact same arguments and only those.

  8. 3

    Rather, in cases where the IPR challenger lacks knowledge of the prior art the appropriate test for estoppel is what “a skilled searcher conducting a diligent search reasonably could have been expected to discover.”

    There’s some cognitive dissonance involved here. Patents are accorded a presumption of validity, which is essentially a presumption that the examiner is a skilled searcher who conducted a diligent search and found everything that the examiner could reasonably have been expected to discover.

    Meanwhile, Ironburg demonstrates that the Federal Circuit also believes either that an examiner is not a skilled searcher or that the examiner of Ironburg’s application didn’t conduct a diligent search. If they didn’t believe this, then the fact that the examiner didn’t find Valve’s newly cited references during examination would be evidence that the references were unlikely to be found by a skilled searcher, and thus Valve would be permitted to continue with that invalidity defense in court. But the lack of a diligent search by a skilled searcher during examination undermines the justification for the presumption of validity – or, conversely, the statutory presumption of validity undermines the “skilled searcher” test.

    So, Federal Circuit, which is it?

    1. 3.1

      “the Federal Circuit also believes either that an examiner is not a skilled searcher or that the examiner of Ironburg’s application didn’t conduct a diligent search.”

      Everyone knows that the PTO does not provide diligent searches, as that term would be understood in the world of IPR search.

      But I don’t believe there’s any cognitive dissonance because the presumption of validity is not actually rooted in examination, since it dates back into the registration era of US patents.

      1. 3.1.1

        I think that it is less “the search,” and more “the consideration” afforded any search results.

        I’ve seen some immensely large “nets” thrown out there for which “consideration” could ONLY be a specific key-word match.

  9. 2

    The Federal Circuit seems to brush aside concerns about the patent challenger asserting privilege when the patentee seeks discovery to meet its burden. How will district courts efficiently resolve this issue?

  10. 1

    Would this not be a Pyrrhic victory, given that the art in question is still challenging the Patent, albeit by someone else’s IPR?

Comments are closed.