The Specialists

§ 1 · The fingerprint

A refusal profile that shouldn't exist.

Across the active fleet, when an application draws a substantive refusal, §2(d) likelihood of confusion is among its grounds 64%of the time, and a ground outside the familiar named list — the “other substantive” bucket — 14% of the time. Then there is Robert J. Struck of Law Office L90: §2(d) at 29%, other substantive at 51%, across a 19,109-application docket. No classifier error produces a swing that large on a sample that big.

Reading his office actions explains it. In a 39-document sample spread across 2014–2026, roughly three quarters of his “other” refusals are lawful-use refusals: the mark cannot be registered because the goods cannot lawfully move in commerce under the Controlled Substances Act, with the FDCA appearing alongside for ingestible CBD after the 2018 Farm Bill and drug-paraphernalia grounds close behind. The docket reads the same way — DR. GRINSPOON, LIBERTY HAZE, CHARLOTTE'S WEB HEMP, PERFECT SPLIFF, NYC BUD. His grounds mix was fleet-normal through 2013; at the 2017 peak, 77% of his refusal-bearing applications drew an “other” ground. The timeline is the cannabis policy timeline.

One examiner with a concentrated docket is an anecdote. So we ran the same scan across the whole active fleet.

§ 2 · The desk

Twenty examiners, 40% of the cannabis-named docket.

The scan: every active examiner whose “other substantive” share exceeds 30% on 500or more applications. The fleet mode is 10–20%; the gate is roughly double that, and exactly 20examiners cross it. We then pulled about thirty “other”-bucket office actions for each — 565 documents read in all — and tallied the statutory grounds actually asserted. For fifteen of the twenty, the top ground is the same one Struck writes: lawful use under the CSA. Jeffrey J. Look is the purest case, at 26 of 30 documents.

The concentration receipt is the chart at the top of this piece. Since 2014, these twenty examiners handled 2.0% of the applications examined by the active fleet — and 40.0% of the applications whose mark names contain an unambiguous cannabis term (930 of 2,327). That is twenty times their fair share, on an undercounted subset: a mark name is the least subtle way to sell cannabis, and the proxy only sees names.

The USPTO does not publish assignment mechanics, and nothing in the public record says why these applications land where they land. Deliberate routing and self-reinforcing workload clustering look identical from the outside; what the record shows is the concentration itself. There is an obvious logic to it either way: lawful-use refusals turn on a moving body of law — state legalization, the Farm Bill, delta-8, the FDCA — and a desk that writes them every week has the current law in hand.

§ 3 · The same statistic, a different desk

Two of the twenty see almost no cannabis marks.

Five of the twenty gate-crossers broke the pattern, and two of them had essentially zero cannabis-named filings. Their documents tell a different story. Tara J. Pate refuses on failure to function as a mark in 20 of 30 documents read, with §2(a) refusals in 15 — the informational-matter desk. Judith Michelle Helfman runs failure-to-function in 19 of 30 with ornamental refusals in 8 — the desk where slogans on t-shirts go to be told they are not trademarks. The remaining three run mixed dockets between the two families.

The lesson generalizes. An outlier statistic flags a docket; it cannot name it. The naming took reading the documents.

§ 4 · The corridors

One examiner, one registration, 325 applications.

Ground mix is one axis of specialization. The other lives inside §2(d) itself: examiners who cite the same blocking registration over and over, for years. To measure it, take every active examiner with at least 300 applications carrying an extracted §2(d) citation — 384 examiners — and ask what share of those applications their single most-cited registration appears in. The median answer is 0.8%. The 90th percentile is 1.7%. Repeat citation is rare — which is what makes the exceptions remarkable.

Exactly two examiners clear 10%. Ronald G. McMorrow of Law Office M90 has cited the certification mark TEQUILA (Reg. 5225126) against 325 distinct applications — 22.6% of his entire cited docket, twenty-eight times the fleet median. If you file an agave-spirit mark, there is, in effect, a tequila examiner, and the citation is predictable before you file. Christina Sobral of L90 is the wine-country counterpart: NAPA GREEN (Reg. 3704740) against 114 applications, NAPA VALLEY against 48, 100% NAPA VALLEY against 47.

Below the 300-application floor, two more corridors are forming. John Joseph Schmidt of N10 has cited a registration for the standalone mark AI (Reg. 7629644) against 66 of his 241 cited applications — 27.4%, higher than any examiner above the floor, riding the AI naming boom. Lisa Anne Papale of N30 has cited each of two related 1-800-DOCTORS registrations against 49 of her 209.

A corridor fight is its own genre. The tequila and Napa blockers are certification marks, so the dispute is less about consumer confusion between two brands and more about whether your mark trespasses on a protected designation — a different brief, with its own case law. None of it is visible in a refusal-rate statistic. McMorrow's §2(d) numbers look like a normal tough examiner; the corridor only appears when you count the registrations he cites.

§ 5 · What this means for your prosecution

Know the desk before you brief the refusal.

1. 01

   A cannabis-adjacent mark changes the fight before you file.

   If the goods touch cannabis, hemp, or CBD, the refusal to brief for is lawful use in commerce under Sections 1 and 45 — with the FDCA riding along for ingestible CBD — and the §2(d) landscape matters less than it would anywhere else. The draw is measurably tilted toward examiners who write these refusals every week and know the case law cold.

2. 02

   Check the corridor before you clear.

   For an agave spirit, a Napa wine, or an AI-formative name, the blocking registration is already on the record dozens to hundreds of times. Clearance for these marks should start from the assumption that the citation is coming and answer it in the application as filed.

3. 03

   The statistic can't name the desk.

   Two examiners with the same outlier ground profile turned out to run two entirely different desks — one refuses cannabis marks, the other refuses slogans. The number flags the docket; only the documents say what it is. That read is what an examiner profile is for.

Methodology