Designing an aggregator

This chapter cites law as well as corpus rows, so it carries two closing tables. Sources is the corpus; Legal authorities is everything that is not. Keeping them apart is how a reader checks the project's no-fact-originates-here rule by inspection rather than by trust.

Revision, 2026-09-06 (first pass). The three defendant-canonicalisation defects this chapter names in section V were repaired that morning, and the exhibit was re-run. Not one firing point in the shared-canonical-defendant table moved.

Revision, 2026-09-06 (second pass). The two further defects section V called a floor were then repaired as well, and one firing point did move: at k=5 the all-forums shared-defendant aperture now fires at #47–48 (2025-07-29, C019) rather than #53 (2025-10-17, C059). The table in section IV and the sentence beneath it are updated; section V is rewritten to report the repair instead of the defect. The first pass had missed one of them by aliasing Jeffrey K. Davis without the , Esq. suffix that the data actually carries, so the fix matched nothing — which is why the floor was still there to lift. Every other figure in this chapter is left at the 2026-09-03 values it was written against, including the 188 / 3,463 lattice pair, and is dated rather than wrong.

I. The missing column

This is the last of four demonstrations, and the one that tests the remedy rather than the record. Chapters 4, 5 and 6 showed that the pattern was distributed, narrated and unpriced. This chapter asks the question a reader is entitled to ask next: if an institution had been built to see all of it, would anything have been different?

Chapter 2 ended on a table. Four sovereigns, six restriction regimes, and a last column headed Central register, which read none, none, none, none, none, and one entry that records California state prefiling orders and nothing else. That column is this chapter's subject. The question is not whether the law permits a court to count a filer's actions elsewhere — chapter 2 established that two of the four regimes permit exactly that, in unqualified words — but what an institution would have to build for the permission to mean anything, and what building it would have bought.

The proposal that this project began from named four candidates, in ascending order of institutional cost: cross-sovereign related-case disclosure; a filing-velocity flag on the federal side; extending California's § 391-style listing to federal courts; and a clearinghouse. It set them a test: would it have caught this by action thirty?

That framing contained an assumption, and the exhibit falsifies it. Ordering the remedies by cost implies that yield rises with cost — that the cheap instrument catches little and the expensive one catches much. On this record the two are anti-correlated at the cheap end. The cheapest federal-side fix would have added three actions to the count a judge actually made. A disclosure duty reaching no further than the borders of one state would have added twenty-six. And the answer to the test question turns out to be yes, easily, under every aperture, long before action thirty — which means detection was never the binding constraint, and a chapter that spent itself designing a better telescope would be answering a question nobody was asking.

The correction is recorded here rather than saved for the conclusion, in the manner chapter 2 corrected the thesis statement it inherited.

II. The instrument

The exhibit holds the numeric threshold fixed and varies only the aperture — the set of prior actions a decision-maker at the moment of filing is able to observe. Four of them:

These are a lattice, not a ladder, and saying so is not pedantry. Same court is contained in same state, and both in all forums; all federal is contained in all forums. But all federal is incomparable to same state — wider on districts, narrower on sovereigns. The exhibit measures the incomparability rather than assuming it: across the ordered pairs of actions in this record, 188 are visible to the federal aperture and not the state one, and 3,463 the other way about. Forty-one of the ninety actions this plaintiff commenced never touch a federal court at all, which is why a federal-only aggregator can fire later than the single docket it was built to improve on. It does, twice, below.

The thresholds are stipulated, not doctrinal. Chapter 2's central finding about § 1651 was that it states no predicate whatever, and none of the circuit standards built on it names a number; Safir's factors are qualitative throughout. So k is an analytic device for holding severity fixed while reach varies, not a rule anyone has enacted. The single exception is § 391(b)(1), which does state an arithmetic, and which the exhibit therefore runs as written rather than as stipulated.

A third convention governs what happens when an action moves. An action's court set is every court any of its docket records was filed in or now sits in, and its state set is that court set read through a court-to-sovereign map. So a removal makes an action visible through both the state and the federal aperture, and — the same rule, applied to the same field — a transfer makes it visible through the destination state's aperture as well as the origin's. That is a substantive claim about what a register would hold, not a convenience: an action transferred into a California district is one a Californian register would in fact have carried from the day it landed. Two actions span two sovereigns each on this rule (C046 and C058, both moved out of the Middle District of Pennsylvania into a California district), and the state aperture is defined for them rather than undefined. Deriving the state set from the court set is also what makes same court contained in same state by construction: until 2026-09-10 the state aperture was read from a per-record field recording only where an action began, and C046 — filed in Pennsylvania, sitting in California — came to share a court with the natively-Californian C107 and not a state, which broke the containment this section asserts. The lattice is a claim the exhibit checks, so it failed rather than quietly reporting a lattice that was no longer one.

Two further conventions. The ladder runs over the 90 actions this plaintiff commenced, not the 94 in the corpus: four actions in the record were filed against him (C100, C022, C031, C037), and a rule aimed at a filer does not count the suits other people bring. The snapshot below reports both populations, because the judges were counting the corpus. And a prior action is one filed on a strictly earlier date. Sixteen filing dates in this record carry two or more originating actions — 52 actions in clusters, the largest seven on a single day — so ordering within a day is arbitrary; same-day actions are not allowed to count each other, and where a threshold first fires on such a date the ordinal is printed as a range.

III. What each aperture would have shown, each time a judge looked

Chapter 1's exhibit took the two moments in this record when a court stated a count and recomputed them. This one takes the same two moments and asks a different question: not what existed, but what each candidate instrument would have put in front of the judge who was writing.

Corpus population, 94 actions. Restricted to the 90 he commenced: 22 / 47 / 24 / 63 and 38 / 65 / 40 / 81.

The first and last columns are the validation. The same-court column reproduces each judge's own stated figure exactly — 22, then 38 — and the all-forums column reproduces detection_gap.py's. The exhibit exits non-zero if either stops holding, so aperture (a) is demonstrably the docket-local aperture and not some new construct that happens to resemble it. The middle two columns are the finding, and they are not what the outline expected.

A nationwide federal aggregator would have shown Latella twenty-five where he saw twenty-two. Three actions. Every federal action in the record on that date sat in the Middle District of Pennsylvania but for three in the Southern District of New York — C100, which was filed against him in 2020; C099, filed in 2023; and C098, a New York state action removed there in 2025. Even C046, the case that has since moved to the Central District of California, began in the Middle District and was already on the counting court's own docket. The instrument that a PACER-side rule would use — an index the federal judiciary already operates — was, at the moment it would have mattered most, worth three actions out of the forty-three uncounted, and one of those three was somebody else's suit.

A disclosure duty reaching only across Pennsylvania would have shown him forty-eight. Twenty-six more, and more than half the gap, because the largest uncounted block was Lackawanna County — the adjacent state trial court that had already handed ten removed cases onto the counting court's own docket, as chapter 1 recorded. The instrument that stops at a state line does six times the work of the one that crosses every state line but no sovereign one.

By July the ordering is unchanged and the magnitudes have grown: federal-wide adds three, Pennsylvania-wide adds thirty, all forums adds forty-seven. The shape of this record is that its centre of mass is one county and one district that sit on top of each other, and any instrument that cannot see across that boundary — the state–federal one, at a single geographic point — sees almost nothing that the docket in front of it did not already show.

IV. When each threshold fires

The snapshot asks what an instrument would have shown at two fixed dates. The replay asks the converse: at which action would each instrument first have had enough to act on?

Any prior action. The naive count, and the one that shows the aperture effect most cleanly:

At k=10 the clearinghouse fires sixteen actions and seven and a half months before the single docket does. That is the aperture effect, and it is real. It is also the largest gap in the table: at k=3 the same-court aperture fires at the same action as the clearinghouse, and at k=5 one action behind it, because the record opens with three New York suits filed on one day in 2018 and the early actions cluster in a single court.

Prior actions sharing a canonical defendant — Rule 233.1's actual bar, and the natural trigger for a related-case disclosure duty:

This is chapter 2's organising finding reappearing as arithmetic. A bar that reaches only the same or related defendants cannot catch a filer who rotates them: at k=5 no aperture fires before action forty-seven, and at k=10 every aperture converges on the same action, #72. The widest possible view buys nothing at all once the qualification is narrow enough, because what is scarce is not visibility but repetition against the same party.

Prior actions sharing a dispute group is narrower still, and at k=10 it never fires under any aperture. That is adverse to the chapter and is stated rather than omitted; see section V on why the dispute row is a floor.

Prior actions coded as repeat filings is the most favourable row in the exhibit and the one that should be trusted least. It is labelled an oracle in the exhibit's own output: coding an action RefilingOfPriorCase = Yes under protocols/01-refiling.md already requires knowing that an earlier action existed, against a shared canonical defendant, on the same dispute — which is precisely the cross-forum aggregation being tested for. The row measures what an aggregator that already had this project's answer would have done. It is an upper bound on achievable performance, not a rule an institution could run, and no argument below rests on it alone. For the record: at k=3 it fires at action #8 across all forums and not until #25–26 in the same court, eighteen actions and fourteen months apart.

Velocity is a different instrument, and it behaves nothing like the others. A rate counts the filing in front of it rather than only what preceded it:

Two things fall out. At ≥5 in 180 days every aperture fires at the same action, #11–14 on 2024-07-08 — a velocity flag on this record needs no cross-forum visibility whatever, because the filings that trip it are already in one place. And at ≥3 in 90 days the flag fires on the first day of the corpus, 2018-03-09, on C086, C087 and C088: three suits against three Westchester country clubs, filed together, none of them coded as a repeat of anything, none of them assigned to a dispute group. That is a false positive sitting in this project's own data, and it is named here rather than left for a reader to find.

Section 391(b)(1), run as written, is the only non-stipulated threshold available: five litigations in the immediately preceding seven years, commenced in propria persona and finally determined adversely. Because Outcome is free text and 45 of 104 records carry no OutcomeDate, the exhibit refuses to report a single number and emits a bounded pair — the upper bound counting any adverse outcome text, the lower additionally requiring a dated determination that predates the filing being screened:

Both bounds fire before action twenty, under every aperture including the narrowest. The one threshold in American law that states an arithmetic was satisfied on this record — countable from the New York Supreme Court docket alone — by the beginning of 2023, three years before Latella wrote "at least twenty-two." The five prior actions that clear it there are C086, C087 and C088 (2018-03-09, all discontinued), C089 (2021-04-22, dismissed) and C090 (2022-12-20, discontinued), all in that court, all resolved against him on the face of the record.

And the test question. Action thirty is C006, Tufano v. PPL, filed in Lackawanna County on 2025-04-23. Qualifying prior actions visible on that day:

The federal row is zero because C006 is a state action: a federal-only facility is not merely uninformative at a Lackawanna filing, it is unavailable — the court being asked is not in the system that maintains it. That is the row that most sharply distinguishes an aperture from a rule, and it is why the exhibit models the federal aperture as false at a state filing rather than as federal priors, counted.

Would an aggregator have caught this by action thirty? On volume, yes, under every aperture that was applicable, and by a wide margin. On the Rule 233.1 shape, no, under any aperture, and not for another forty actions.

V. The arithmetic that cuts against this chapter

Stated here, before section VI uses any of it.

The chapter's own exhibit refutes the easy version of its thesis. If the argument were widen the aperture and the threshold fires, the tables above would be a poor advertisement for it: every threshold fires, under every applicable aperture, well before action thirty, and § 391's real arithmetic was cleared inside a single court by action seven. What a wider aperture buys is lead time and magnitude, not the event. Claiming more than that would be an overclaim on page four, and chapter 1 set the precedent for conceding this kind of thing in the text rather than in a footnote.

The counterfactual holds the filer still. "Would it have caught this by action thirty" changes the institution while freezing the behaviour. This filer says out loud that he selects forums — chapter 5 quotes VS697, "a lawsuit in every state I can drive to" — and 41 of his 90 actions are already outside the federal system. An instrument with a federal aperture is exactly the kind of thing a filer routes around. The exhibit measures the detectability of the observed record, and cannot measure deterrence, displacement or adaptation.

N is one, and the subject was selected because he is extreme. The thresholds fire early here because the record is dense. Nothing above supports a claim that a k=5 rule would behave this way on an ordinary docket; it supports only that it would have fired on this one. That is an existence proof plus a mechanism, which is the framing the proposal committed to.

The velocity flag's first firing is a false positive in this corpus — 2018-03-09, C086/C087/C088. Same-day batching is endemic here, 52 of 94 actions sit in a same-day cluster, and a rate calibrated to fire on this record would fire on any self-represented plaintiff who files companion complaints against separately named defendants on one trip to the prothonotary.

Twenty-four of the ninety actions carry no DisputeID, so every dispute-keyed figure is a floor rather than an estimate. Protocol 01's reliability audit found that an unassigned dispute group is uninformative in both directions.

The defendant-sharing row's three known canonicalisation defects were repaired, and the row did not move. When this chapter was drafted, C040's roster omitted the Shopify entities its own claims summary named, Fenix Internet, LLC and Fenix Internet LLC carried different canonical ids with no alias joining them, and C018's roster omitted Taban — so the Rule 233.1 row, the chapter's most doctrinally faithful one, was reported as a floor whose firing points might be late. All three were repaired on 2026-09-06 and the exhibit re-run: every firing point in the table above is unchanged. The defects were real and they were not load-bearing, because each added a shared defendant to an action that had already crossed the threshold on other parties rather than to the action that crossed it.

The row was a floor for two further reasons when this chapter was written, and both have since been repaired — which is how we know it was a floor. Jeffrey K. Davis, Esq., the string C099 actually records, was absent from defendant_aliases, so under the name-matching canonicalisation this exhibit uses his first appearance as a party read 762 days late. That one had survived an earlier repair pass that aliased Jeffrey K. Davis without the , Esq. suffix and so never matched the string in the data — a reminder that a canonicalisation fix is only as good as the exact string it targets. And the C030/C064 removal pair recorded the same two people as four canonical entities, one pair per docket; repairing that turned up five more of the identical kind, all of them a party listed surname-first on the state docket and first-name-first on the federal one.

Repairing them moved this row, and only this row: at k=5 the all-forums aperture now fires at #47–48 (2025-07-29, C019) rather than #53 (2025-10-17, C059), five actions and roughly three months earlier. Every other firing point in the table above is unchanged. The direction is the one the section already argued — the shared-defendant bar is less late than the corpus made it look — and it does not disturb the finding, because the same bar still fires after four dozen actions under the widest aperture available.

Section 391(b)(1) requires the prior litigations to have been commenced in propria persona, and ProSe is blank on 44 of 104 records. The § 391 row above tests the numeric and temporal elements only. It is arithmetic against the statute's counting rule, not a finding that its elements were met, and it could not be a finding anyway: § 391.7 restrains only "in the courts of this state," and two of these ninety-four actions are Californian.

The apertures are not equally real. "All federal" is a facility that exists and can be bought today. "Same state" is not: Pennsylvania has no unified civil docket search, and chapter 3 quantified the resulting asymmetry — 21 coded indicators across 37 M.D. Pa. actions against 7 across the larger block of 38 Lackawanna actions, a fact about what can be read, not about conduct. The twenty-six actions that a Pennsylvania-wide aperture adds are the yield of an instrument that does not exist, and its aperture being narrow does not make it cheap.

VI. Remedy 1 — cross-sovereign related-case disclosure

The cheapest instrument in the list, and on this record the highest-yielding: a duty on the filing party to disclose their own pending and recently concluded actions involving the same parties or subject matter, wherever filed.

There is no such duty in the Federal Rules. Rule 7.1 is the only disclosure statement federal civil procedure imposes on every party in every case, self-executing, filed "with its first appearance, pleading, petition, motion, response, or other request addressed to the court." It requires a nongovernmental corporate party to identify "any parent corporation and any publicly held corporation owning 10% or more of its stock," and, in diversity cases, to name and identify the citizenship of "every individual or entity whose citizenship is attributed to that party." It says nothing about the filer's other litigation. The machinery of mandatory, automatic, first-filing disclosure is built and running in every district in the country; it is pointed at corporate ownership.

That is the whole of the remedy-1 argument, and it is a design argument rather than a doctrinal one. Nothing in chapter 2's four regimes needs amending for a related-case disclosure duty to exist. It would be a rule of procedure, and the vehicle for it already carries a form.

What it would have bought here. A duty reaching every court of the filing state moves Latella's count from 22 to 48 and Munley's from 38 to 68 (section III). It fires at k=5 on any-prior at action #6, and — the row that matters most, because a disclosure duty is naturally keyed to shared parties — at k=3 on shared-defendant at action #7, 2023-02-01, which is C091, Frank Tufano v. American Arbitration Association in New York Supreme Court, filed twenty-one days after C099, the same dispute against the same defendant in the Southern District of New York. Under the same-court aperture that shared-defendant threshold does not fire until action #40–41, on 2025-06-06, two years and four months later.

Where it fails is where Rule 233.1 fails. It runs on the filing party's own initiative and reaches only what that party discloses. Chapter 2 identified that defect in Pennsylvania's rule — it fires on a defendant's motion, so a rule that could have counted a federal prior action never got the chance — and a disclosure duty relocates the initiative rather than removing it. The Shopify sequence in this record shows the shape of the problem plainly: C047 in the Middle District on 2025-02-21, C095 in New York Supreme Court three days later, C050 back in the Middle District on 2025-06-02, C013 in Lackawanna County on 2025-06-06, and C054 in the Middle District again on 2025-07-07 — five actions on one dispute, across three courts in three sovereigns, in under five months. Each of those courts had a form. None of them had a question on it that this sequence would have answered.

VII. Remedy 2 — a filing-velocity flag

The instrument the proposal placed second by cost. It performs worst of the four here, and it fails in two different ways at once.

It is the wrong aperture. The PACER Case Locator is real, and it is better than the argument for building something new usually assumes: "a national case locator index for all federal court records in district, bankruptcy, and appellate courts," updated "every 24 hours, typically nightly," returning "the party name, the court where the case is filed, the case number, date filed, and date closed." A nightly-refreshed nationwide party-name index of every federal case already exists; no restriction standard in chapter 2 is wired into it. But its aperture is the federal one, and section III prices that aperture on this record at three actions. Section IV prices it worse: the federal aperture is the only one that fails to fire the ≥3-in-90-days flag until action #11–14, and at the moment of the test question, action thirty, it sees nothing at all, because the filing is in a county courthouse.

It measures the wrong thing. Velocity is a rate, and the doctrines count repetition. At ≥5 in 180 days the flag fires at action #11–14 under every aperture, which is another way of saying the cross-forum visibility contributed nothing to it. At ≥3 in 90 days it fires on this corpus's first day, on three suits that the coding protocol treats as unrelated to anything. A flag that fires on 2018-03-09 and again on 2024-07-08 has not distinguished the plaintiff who files three companion complaints in one morning from the one who has been relitigating the same dispute for four years, and the distinction is the entire content of the doctrine.

That is not an argument against building it. It is an argument that it is a triage device and should be described as one — the cheap instrument earns its place by telling a clerk where to look, not by telling a judge what to do.

VIII. Remedy 3 — extending § 391-style listing to federal courts

Chapter 2 established the asymmetry this remedy is aimed at, and it remains the sharpest observation available in this body of law. California's § 391(a) defines "litigation" as any civil action or proceeding "commenced, maintained or pending in any state or federal court." Its § 391(b)(1) counts five such litigations in seven years, in propria persona, finally determined adversely. Its § 391.7 then restrains the filer only "in the courts of this state," and § 391.7(f) sends the resulting order to a register that, by construction, can hold nothing but California state prefiling orders. The predicate crosses every boundary in the country; the remedy and the register stop at one state line.

Section IV puts a number on the first half of that. Run as written, § 391(b)(1)'s arithmetic was satisfied on this record between action #6 and action #9 depending on the bound — early 2023 to early 2024 — and satisfied within the same court, without any aggregation at all, by action #7 or #9. Chapter 2's finding was that California can see further than it can reach. The finding here is that on this record it did not need to see far: the counting rule cleared inside one docket, years before any judge stated a count.

Which relocates the remedy. Extending the listing to federal courts is not principally a detection measure; it is a communication measure. Its value is not that it would have let someone count to five sooner, but that it would have created an artifact — a place where a restriction, once imposed anywhere, becomes visible everywhere. This record has no such restriction to publish: zero pre-filing orders, zero monetary sanctions, and two vexatious-litigant requests, neither ever adjudicated (vexatious_motions ids 1 and 2). A register with nothing in it is exactly what a jurisdiction with no restrictions produces, and that is the null result chapter 1 opened on, looked at from the remedy side.

Congress has written a cross-forum counting rule once. Chapter 2 recorded it: 28 U.S.C. § 1915(g), three strikes accrued in any "court of the United States," enacted in 1996 for prisoners proceeding in forma pauperis. The mechanism is not beyond the institution's imagination or its drafting capacity. It has been used once, for one population, and never extended.

IX. Remedy 4 — a clearinghouse

The expensive answer, and the only one that reaches 65 and 85.

Two things are worth saying about it that the outline did not anticipate. The first is that the federal system already operates a cross-district aggregator that works, and it is instructive precisely because nobody thinks of it as one. Under 28 U.S.C. § 1407, when "civil actions involving one or more common questions of fact are pending in different districts," they may be transferred for coordinated pretrial proceedings — and the process may be commenced by "the judicial panel on multidistrict litigation upon its own initiative," not only on a party's motion. The panel is a standing body of seven circuit and district judges, no two from the same circuit. It has a docket, a clerk, and a mandate to notice cross-district commonality without waiting for anyone to ask.

Every defect chapter 2 identified in the restriction regimes is absent from § 1407. It fires on the institution's own initiative rather than a private party's. It has a permanent body whose job is to look across dockets. It aggregates by subject rather than by court. Its predicate is explicitly multi-district. American federal procedure solved this problem in 1968 for a different purpose and has never once pointed the solution at a filer.

Its limit is the same as PACER's: § 1407 reaches districts of the United States. Forty-one of the ninety actions here are outside it.

The second thing is that the courts have already stated the requirement, one filer at a time. The Ninth Circuit's Ringgold-Lockhart v. County of Los Angeles vacated a vexatious-litigant pre-filing order for an inadequate record, holding that "an adequate record for review should include a listing of all the cases and motions" that led the court to conclude a pre-filing order was needed. That is an aggregation requirement. It is imposed on the district court, at the moment of decision, for one litigant, and it is why the remedy is so rarely available: assembling that listing is exactly the work no institution is positioned to do, and a court that cannot assemble it cannot impose the order.

A clearinghouse is therefore not a new idea in this body of law. It is the ex ante, institution-wide version of a thing appellate courts already require ex post, case by case, from judges who have to build it by hand.

Chapter 5's caution applies to the design and is repeated here rather than assumed: an aggregator that counts dockets runs a week behind a filer who announces filings in advance — 32 statements in that corpus announce or threaten a filing before one they name, 21 of them within a week — and an aggregator that counted announcements instead would have counted a lawsuit in every state he could drive to, which was never filed.

X. What detection looks like when it works

Three published decisions, set alongside this record rather than added to it. The sample stays at one.

Molski v. Evergreen Dynasty Corp. (9th Cir. 2007) is the case where the arithmetic worked. A district court counted a serial ADA plaintiff's filings, found the pattern, and entered a pre-filing order that the Ninth Circuit affirmed. Two features of it are absent here. The counting was possible because the filings were in one federal system and mostly one district — the same condition that makes the same-court aperture adequate in section III's first row and inadequate in every later one. And Molski was represented, which puts him outside Pa. R.C.P. 233.1's pro se limit entirely and inside § 1927's counsel-facing liability; the plaintiff in this record is on the other side of both lines.

De Long v. Hennessey (9th Cir. 1990) supplies the ceiling and the reason the ceiling exists. It vacated a pre-filing order for want of notice and an opportunity to be heard, and said that "such pre-filing orders should rarely be filed." Ringgold-Lockhart (9th Cir. 2014) carries that forward and adds the record requirement quoted in section IX.

Read together they say something the design literature tends to elide: the constraint on restriction is not only that courts cannot see, it is that they are told to act reluctantly and on a full record when they can. An aggregator that made counting easy would not make the order easy. It would supply the listing Ringgold-Lockhart demands and leave every other requirement standing — which is the right result, and the reason a clearinghouse is a detection measure and not a restriction measure.

The access-to-courts counterweight belongs to chapter 6, which was drafted on 2026-09-06 and takes it up in its section VI. Its finding there cuts against the reflex this section is guarding: both fee-waiver regimes in this record already carry their own frivolousness screen, and tightening them would have reached almost none of these filings, because where fee status is recorded at all this plaintiff mostly paid. A restriction aimed at people who cannot afford the fee does not reach a filer who is paying it. This chapter states the constraint and defers to it: nothing above argues that more restriction is warranted in this record or any other, and the null result chapter 1 identified is evidence about institutional capacity, not about what any court should have decided.

XI. Detection was never the binding constraint

The exhibit was built to answer whether an aggregator would have caught this by action thirty. It answers yes, comfortably, under every aperture that could have applied — and in doing so it moves the question.

Every threshold in section IV fires early. The only one with statutory authority behind it, § 391(b)(1), fires inside a single court by action seven or nine, three years before the first judicial count. A velocity flag fires on the corpus's first day. What the wider apertures buy is measured and real — sixteen actions and seven months at k=10, twenty-six actions of visibility at the moment Latella wrote — and it is not the difference between seeing and not seeing. It is the difference between seeing a lot and seeing more.

So the missing column in chapter 2's table is missing, but its absence is not what produced the null result. What produced the null result is the thing chapter 2 found and this chapter's arithmetic confirms from the other end: § 1651 states no predicate, so there is no number for a count to meet; § 391 states a number and cannot reach; Rule 233.1 states a predicate and fires only on someone else's motion; and 22 NYCRR 130-1.1 is not a filing rule at all. A count with no threshold to be measured against is not an instrument. Ninety-four originating actions, five trial forums, three states, and every counting rule in the country either satisfied years ago or inapplicable.

The design conclusion follows and is narrower than the chapter set out to reach. An aggregator is necessary and demonstrably insufficient. Necessary, because Ringgold-Lockhart requires the listing and nothing produces it. Insufficient, because the listing was already producible here — a New York clerk could have counted to five in 2023 — and no consequence attached. The load-bearing reform is not the aperture. It is the trigger and the consequence: who is obliged to look, and what follows when they do.