Examiner Wilbert L Starks has allowed 950 of 1,186 decided applications in Computer Architecture, Software, and Information Security.
Lynch LLP represents applicants in patent prosecution before the USPTO.
Talk to the attorney who would handle the matter.
Within 36 months of the first office action, 76.53% of Examiner Wilbert L Starks's applications had been granted and 17.72% had been abandoned. This counts each application from its own first office action, so it covers a different population than the overall allowance rate above, which is measured across decided applications only.
Counts mailed office actions on the merits — non-final and final — over this examiner's 950 allowed applications. A first-action allowance counts as zero, and a notice of allowance is not an office action.
The weighted peer average (1.71) blends the per-art-unit averages, weighted by this examiner's allowed applications in each: 2122 (1.71), 2129 (1.72), 2121 (1.73). Each art-unit figure is the unweighted mean of the individual examiner values in that art unit — the peer-typical figure, not the art unit's own ratio.
Wilbert L Starks maintains an 80% allowance rate across more than a thousand decided applications in Technology Center 2100 (Computer Architecture, Software, and Information Security). His record spans three art units. The allowance rate ranges from 77% to 86% across these art units, reflecting variation in the decided-application pools within each unit. This pooled figure represents the proportion of applications that issued or were abandoned among all decided cases, and does not reflect pending applications.
A pooled allowance rate aggregates results across multiple art units into a single historical figure. This aggregate describes past decisions and is not a prediction of any specific application's outcome. The range across individual art units indicates that allowance rates vary by unit, reflecting differences in application subject matter, complexity, and decided-case composition within each unit. The pooled rate is a summary statistic of the examiner's overall record.
These are aggregate statistics from this examiner's past public record — not predictions about any specific application. The per-art-unit figures below show how the record varies across art units. Our approach to patent prosecution →
Appeal outcomes are counted from USPTO transaction records. Where an examiner has few appeals on record, these figures describe what happened in those cases and should not be read as a rate.
Categories report what happened to the appeal, not the application's final fate — most appeals resolve without a Board decision, and an application whose rejection the Board affirmed can still be granted afterward.
These figures describe outcomes already on the public record. They do not predict the outcome of any application and are not legal advice.
Reassigned applications. The USPTO publishes only the examiner currently assigned to an application. Where an application has changed examiners, its entire history — including office actions issued by a predecessor — is attributed here to the current examiner.
Examiner identity. The USPTO publishes no stable examiner identifier. Identity on this site is derived from the examiner's published name together with the art unit. An examiner's record is pooled across the art units they worked in within a technology center. An examiner who has worked in more than one technology center appears here more than once, with their record split between those pages. Two examiners who share a name within a single art unit may be combined into one record.
Comparing across technology centers. Rejection-ground rates (§101, §102, §103, §112) track the subject matter a technology center examines, not examiner severity. Technology Center 2100's rates are comparable with other art units inside TC 2100 and are not comparable with rates from a different technology center.
Comparing examiners. An examiner's rates are shown here against the average for their art unit, which is the closest available comparison because art units group related subject matter. Comparisons get less reliable the further you move from that — between art units, and especially between technology centers, differences in subject matter, applicant population, and prior art make raw rate comparisons misleading. Where an examiner has few applications on record, small changes in the underlying data move the percentages a great deal.
Applications not counted. An application whose currently assigned examiner has moved to a USPTO business unit outside the examining technology centers is absent from this record, together with every office action on it — even where those actions issued while the application sat in an art unit shown here.
Coverage is incomplete. This site does not include every patent application. Applications that may not appear here include those examined in technology centers not listed on this site; those that moved elsewhere within the Office, such as to the Patent Trial and Appeal Board on appeal, to central reexamination, to the Office of Petitions, or to design or plant examination; those never published, including applications filed with a nonpublication request and those abandoned before publication; provisional applications, which are not examined on the merits; anything filed or acted on after the refresh date above; and individual applications or office actions that the USPTO's public services did not return when we retrieved them. That list is illustrative, not exhaustive.
Counts may be low. The individual outcomes described here are matters of public record. The counts and rates computed from them are ours, and they are incomplete. Because of those gaps a count here may be lower than the true figure. We cannot state the size of the difference, and it is not uniform — one examiner's page may be close to complete while another's is materially short. Rates are computed from these same counts, so the uncertainty sits in both the numerator and the denominator. Do not treat a figure here as a count of record; verify against Patent Center, PatentsView, or the Patent Examination Data System before relying on it.
An absent figure. Where a figure is absent, we did not retrieve the underlying records. It does not mean the value is zero, and it does not mean the USPTO holds no such records.
Rejection-ground retrieval. Office-action rejection records are retrieved from the USPTO's office-action dataset in date-bounded batches sized so that each is returned whole in a single response. Retrieved this way, repeated pulls return the same set of records — the retrieval is reproducible. That is a property of our retrieval, not a guarantee that the USPTO dataset itself is complete; the coverage caveats above apply to these figures as to every other.
Compiled from USPTO records as published on August 27, 2026. Prosecution continues after that date and is not reflected here.
Each section benchmarks this examiner against that art unit's average. Figures are this examiner's own public record within the art unit; the overall rate above pools them.
Primarily examines neural-network / biological-model computing, and machine learning.
Grounds can co-occur, so the four don't sum to 100%. The art-unit figure is the unweighted mean across examiners in the art unit; where an art unit has no computed average for a ground, that row says so rather than showing a comparison.
Allowance rate for applications with an examiner interview versus without one.
A correlation, not proof that interviews cause allowances. Based on 132 decided applications with an interview and 581 without.
Primarily examines neural-network / biological-model computing, and machine learning.
Grounds can co-occur, so the four don't sum to 100%. The art-unit figure is the unweighted mean across examiners in the art unit; where an art unit has no computed average for a ground, that row says so rather than showing a comparison.
Allowance rate for applications with an examiner interview versus without one.
A correlation, not proof that interviews cause allowances. Based on 56 decided applications with an interview and 368 without.
Primarily examines neural-network / biological-model computing, and machine learning.
No rejection records for this examiner were retrieved from the USPTO office-action dataset. This is an absence of retrieved data, not a rate of zero.
Lynch LLP represents applicants in patent prosecution before the USPTO. These are general resources about the firm's services — not advice about this examiner or any specific application.
Methodology. This page pools every art unit in which Examiner Wilbert L Starks has a public record within Technology Center 2100. Statistics are computed from publicly available USPTO records, refreshed on a recurring schedule. This page's data was last updated August 27, 2026. The overall allowance rate is total allowed divided by total decided applications (allowed plus abandoned) across all art units — not an average of the per-art-unit rates; pending applications are excluded. Figures are rounded for display. Pooled sample: 1,252 applications.
Rejection rates. Each §-rate is the share of this examiner's applications that drew at least one office-action rejection in which that statutory ground appears; applications with no rejection on record are excluded, and because grounds can co-occur the four do not sum to 100%. The art-unit figure beside each is the unweighted mean of the per-examiner rates across the art unit, published for §101 and §103 only. Beside the overall allowance rate we show a benchmark: for a single-art-unit examiner it is exactly that art unit's average, labeled “art-unit average”; for an examiner spanning several art units it is the “weighted peer average” — the per-art-unit averages, weighted by this examiner's application count in each — labeled distinctly because it is a blended figure, not any single art unit's average. Both are built from the same per-art-unit averages the panels show.
Office actions per allowance. Mailed office actions on the merits — non-final and final — averaged over this examiner's allowed applications. A first-action allowance counts as zero and is included; a notice of allowance is not an office action. The comparison figure is built the same way as the allowance-rate benchmark — from each art unit's own average for this measure — weighted here by this examiner's allowed applications in each art unit, since that is this measure's own denominator. Each art unit's figure is the unweighted mean of its examiners' individual values, so it is the peer-typical figure rather than the art unit's own ratio.
Lynch LLP is not affiliated with, endorsed by, or sponsored by the United States Patent and Trademark Office. Examiner statistics are derived from publicly available USPTO data.
These statistics describe past examiner behavior and do not predict the outcome of any particular application. Past results do not guarantee future outcomes. Where this page compares an examiner's allowance rate to an art-unit average, that comparison is a factual description of the public record, not a characterization of any individual examiner's conduct or competence.
This page is for general informational purposes and is not legal advice. No attorney-client relationship is formed by viewing it. Consult a registered patent practitioner about your own matter. These figures are reproduced and recombined by Lynch LLP from the USPTO's Patent File Wrapper bulk archive and Open Data Portal APIs; they are not an official USPTO publication, and the USPTO has not reviewed, endorsed, or verified them. If a page here is wrong, tell us at lynchllp.com/contact and we will correct or remove it. Full disclaimers →
ATTORNEY ADVERTISING — Sean Lynch, Partner, Lynch LLP