# D005 Synthesis — A Stronger Structural Link, Not an Identification ## What is resolved The two phases converge on far more than the framing of their disagreement suggests. Stripped to the facts, they actually agree on every load-bearing element of the record and differ only on what those facts license the wiki to *say*. **1. The Tier-2 registry fact is solid and uncontested.** Both phases treat as established that [[Spark Innovations, LLC]] (filing # 811524872) is a third Delaware foreign LLC managed by [[Michael Montfort]], sharing the Corporation Service Company registered agent and the 2801 Centerville Road, Wilmington DE private-mailbox building with [[Forgelight Ventures, LLC]] and [[Willowbend Capital, LLC]]. The thesis calls this a "registry-anchored shell"; the antithesis calls it "one more shell with the same manager." Neither disputes the fact. **Spark is a Montfort shell, and the Montfort developer-overlap cluster is now three entities, not two.** This is a genuine strengthening of the structural record and it is settled. **2. Spark sits inside the Conway project at Tier-1, and this is real.** Both phases accept that the [[Conway Corporation Project Stratus NDAs]] name "Spark Innovations, LLC, for itself and its parent, subsidiaries and affiliates" as the developer-side "Company," with Conway Corporation's [[Zach Gardner]] as a signatory (2025-10-14). The antithesis explicitly concedes this joint is "real and uncontested." So the Conway project now ties into the Montfort cluster through two distinct documented entities — Forgelight (MOU developer of record) and Spark (Tier-1 NDA counterparty). That this Tier-1 foothold exists is not in dispute. **3. The West-Memphis-to-Spark tie is Tier-3, and the Google confirmation attaches to GROOT.** Both phases agree the [constructionowners.com](../../web%20archive/2026-05-29/www.constructionowners.com/groot-spark-innovations-west-memphis-project-pyramid.md) sentence is a single, vague Tier-3 association ("an affiliate linked to the endeavor"), that it frames Spark as an *earlier-in-time* affiliate "distinct from the current [[GROOT LLC]] proposal," and that the Tier-1 Google confirmation (water-use registration reading "Google Data Center"; Entergy CEO testimony naming [[Altitude Capital, LLC]] as "Google's subsidiary"; the APSC Cypress order) attaches to the **GROOT/West Memphis build, not to Spark's role in it**. The thesis itself lists all of this in its `## Anticipated counterarguments`. There is no factual gap between the phases here — only a dispute over what weight the Tier-3 sentence can bear. **4. The Spark NDA does not name the principal.** Both phases agree the lead Tier-1 exhibit names "a parent, subsidiaries and affiliates that the record does not name" and selects New York governing law that reveals nothing about beneficial ownership. The single document closest to the principal *withholds* it on its face. Uncontested. **Where they decisively part — and where one prevails.** The disagreement is not evidentiary; it is a single inferential question: **does a Tier-2 registry tie between a Conway-side shell (Spark) and a confirmed-Google site's affiliate-history (West Memphis) constitute a "channel" that names the Conway principal?** On this, the antithesis prevails, and decisively, for two reasons the thesis cannot answer on the present record: - **The transitivity is invalid.** "Google at West Memphis (Tier-1)" + "Spark at West Memphis (Tier-3)" + "Spark at Conway (Tier-1)" does not yield "Google at Conway." The inference runs only if Spark is a *Google-exclusive* vehicle, and nothing establishes exclusivity — the antithesis is correct that the same logic would make every CSC client and every Montfort shell a Google entity. A developer or SPV-stack touching a Google project at one site does not convert every project it touches into a Google project. - **The shared-Montfort axis is the dimension D003 already discounted, and its probative value still hinges on an unresolved bracket.** D003 expressly bracketed whether Montfort is an in-house paralegal (strengthening the shared-principal inference) or a Delaware formation-firm organizer (the null hypothesis, under which three same-day-vendor shells prove nothing about a common beneficial owner). The thesis's structural claim rests on a premise the corpus has flagged as unknown; a third instance of a non-probative pattern is still non-probative. The antithesis carries this cleanly. So **the antithesis carries the identification question.** The honest count of independent Tier-1 channels naming Google *at Conway* went from zero (D003) to zero (now), and the gap D003 identified *was* that count. Under the categorical rule of [[Identifying the Unnamed Cooling-Water Data Centers]] ("the circumstantial lead ... is **not** an identification"), a warmer-but-still-circumstantial lead resolves to restraint. **But the thesis carries a narrower point that the antithesis understates, and it should not be lost.** The antithesis is so focused on defeating the identification claim that it slightly flattens what genuinely changed. The thesis is *right* that the Montfort architecture is now better-anchored than the closed two-shell loop D003 weighed: there is now a documented Tier-1 foothold (the Spark NDA) placing a cluster member *inside* the Conway project, and the *same* cluster member has at least one outward-pointing, name-level association with a confirmed-Google build. That is not "another instance of the same thing" in the trivial sense — the cluster has acquired its first outward edge toward a Google-confirmed project, even if that edge is Tier-3, points at GROOT rather than Google, and describes a superseded affiliate. **The pattern is materially better documented than it was in D003, and that strengthening deserves to be recorded** — as a Tier-2 structural finding, not as an identification. ## What is sharper but unresolved The genuine, clarified disagreement is this: **Is the Spark/Forgelight/Willowbend cluster a set of nested SPVs under a single beneficial owner (which would make the developer-overlap pattern strongly probative of a common principal), or is it a serial formation-agent's parallel filings for potentially unrelated clients (which would render the pattern non-probative)?** Both phases agree this is the hinge; neither can resolve it, because the corpus does not disclose Montfort's professional role or any of the three shells' parents. What specific evidence would resolve it: - **Montfort's professional role.** If Montfort is shown to be an in-house corporate officer/paralegal for a single operating company (not a formation-firm partner servicing many clients), the shared-management inference toward a *common* principal strengthens materially. If he is a Delaware formation-firm organizer, the inference collapses. This is the single highest-value missing fact in the entire T003 line. - **Any parent/membership disclosure for Spark, Forgelight, or Willowbend** — a Delaware certificate of formation listing members, a financing UCC-1 naming a secured-party parent, a deed or assignment chain, or any filing that pierces the "parent, subsidiaries and affiliates" placeholder. One such document on *any* cluster member would convert the structural link into (or away from) an identification. - **A Tier-1 channel naming Google (or any principal) at Conway specifically** — the standard the [[Altitude Capital, LLC]] worked example sets and the only thing that would license naming. A water-use registration, sworn testimony, or regulatory order tying Conway/[[Project Stratus]] to a named principal. Nothing in the corpus currently approaches this for Conway. - **Spark's actual relationship to GROOT at West Memphis** — whether Spark's $3B Bollinger Road plan was a Google-directed predecessor to the GROOT proposal, an abandoned unrelated bid, or a separate developer's effort GROOT later superseded. Resolving this would determine whether the one outward edge even points where the thesis needs it to. ## What is bracketed These claims await evidence outside the dialectic's reach and should be carried as open, not resolved either way: - **Michael Montfort's professional role** (in-house officer vs. formation-firm organizer vs. contracted registry technician) — inherited unresolved from D003 and dispositive for the cluster's probative weight, per above. - **The identity and role of organizer "DAVID THOMAS"** named as Spark's Incorporator/Organizer — "a name not otherwise in the corpus." Whether Thomas is a formation-firm employee, a Google/Alphabet agent, or a developer principal is unknown; the antithesis correctly notes the registry points *away from* any identified beneficial owner here. - **Spark's actual operational role at the West Memphis site** — whether it was ever a substantive participant in the Google-confirmed build or merely an early name in the public record, now superseded by GROOT. - **The beneficial owner(s) / parent of Spark, Forgelight, and Willowbend** — the "parent, subsidiaries and affiliates" the NDA withholds. Unknown for all three. - **Whether the Conway/Project Stratus end-user is Google at all.** The thesis's hypothesis (Google) is one candidate; the record does not exclude a different hyperscaler or operator behind the Conway developer ecosystem. The cluster's one Google-adjacent edge is too weak to privilege Google over alternatives at Conway. ## Verdict on tension **Recommended `status:` for [[T003 - Shell-LLC Principal Attribution for Forgelight and Willowbend]]: `resolved-via-D003`.** D005 **declines to revise** D003's restraint verdict. The Spark finding makes the Conway lead warmer but does not make it an identification: it adds zero independent Tier-1 channels naming Google at Conway, its central join (shared Montfort management) is the dimension D003 already discounted and still depends on the unresolved formation-agent-vs-in-house-officer bracket, and its outward edge is a single Tier-3 sentence that points at GROOT (not Google) and describes a *superseded earlier* affiliate. The controlling [[Identifying the Unnamed Cooling-Water Data Centers]] precedent is categorical — a circumstantial lead is not an identification — and adopting the thesis would silently overrule it and import a guilt-by-shared-vendor rule that, applied evenly, would attribute every Montfort/CSC shell in Arkansas to Google. The prior verdict stands: **the Conway principal is unidentified; circumstantial shell-LLC leads are not an identification.** However, `resolved-via-D003` here means *reaffirm the restraint conclusion*, **not** *leave the pages unchanged*. The thesis's narrower point — that the documented structural link is now materially stronger — is correct and must be recorded. The line falls exactly here: **a documented Tier-2 structural link (record it as fact) is not an identification of the principal (do not assert one).** Concretely, the wiki should change as follows: **Do NOT name Google as the Conway principal or "strongest candidate."** Keep restraint. Naming Google — even as "channel-anchored strongest candidate" — would cross the categorical line and is rejected. **Adopt a refined intermediate framing for the *structural* finding** (this is the increment D005 authorizes). On [[The Conway Data Center Project]] and [[Spark Innovations, LLC]], state — and confine the claim to — the following: > The Conway / Project Stratus developer ecosystem is linked, through two distinct documented entities, to the same [[Michael Montfort]]-managed Delaware shell cluster (Corporation Service Company agent; 2801 Centerville Road, Wilmington PMB) that recurs across Arkansas data-center filings: [[Forgelight Ventures, LLC]] as MOU developer of record, and [[Spark Innovations, LLC]] as the Tier-1 counterparty on Conway Corporation's Project Stratus NDAs. Spark is also a third member of this same registry-signature cluster ([[Willowbend Capital, LLC]] being the second). One cluster member, Spark, additionally carries a Tier-3 public association — as an *earlier, since-superseded* affiliate — with the West Memphis site that is independently confirmed as a Google build (the confirmation attaching to [[GROOT LLC]], not to Spark). **This is a strengthened Tier-2 developer-overlap pattern, not an identification of the Conway principal.** The shared-Montfort/shared-vendor registry signature is not, by itself, identification of any ultimate principal; whether the cluster reflects a single beneficial owner or a serial formation agent's parallel filings remains unresolved (Montfort's professional role is bracketed, per D003). The Conway principal therefore remains **unidentified** on this record, and no [[Altitude Capital, LLC]]-style multi-channel Tier-1 confirmation names any principal at Conway. Key drafting constraints for whoever integrates this: - Use "linked to the same shell cluster that also appears at a Google-confirmed site," never "linked to Google." - Flag the West Memphis tie as a *lead* and explicitly as a superseded earlier affiliate; do not let it read as "Spark is at the Google build." - Preserve the existing restraint sentence ("the Conway principal is unidentified; circumstantial shell-LLC leads are not an identification") verbatim or in substance. - The [[Spark Innovations, LLC]] page already reaches this conclusion on its own face; this synthesis ratifies that page's framing and asks the parent synthesis [[The Conway Data Center Project]] to mirror it (upgrading the cluster description to three entities and noting the two-entity Conway foothold), without escalating to identification. Net effect: the wiki gets a sharper, better-anchored *structural* record (three-shell cluster; two documented Conway footholds; one outward Google-adjacent edge) while its *attribution* posture is unchanged — restraint, no Google naming. ## Open questions for future dialectics - **The formation-agent bracket is now the rate-limiting fact for the entire T003 line.** A focused effort to establish Montfort's professional role (e.g., does the same Montfort/CSC/2801-Centerville signature appear on shells tied to *non*-data-center or *non*-Arkansas ventures, which would evidence a formation firm?) could flip the cluster from "non-probative pattern" to "probative of a common principal" — or confirm the null. This deserves its own tension/dialectic if evidence surfaces. - **A cluster-wide enumeration question:** how many entities carry the exact Montfort/CSC/2801-Centerville-PMB signature across Arkansas (and beyond), and do any of them attach to a *named* principal at Tier-1? If even one does, it would either anchor the cluster to that principal or, by attaching to a *different* principal, affirmatively break the single-owner hypothesis. - **Does the Spark→GROOT succession at West Memphis itself disclose anything?** If a future record shows GROOT formally acquired, replaced, or descended from Spark's Bollinger Road plan, that succession chain — not the shared organizer — might become the first genuine outward channel, and would warrant revisiting whether the cluster's West Memphis edge reaches Google after all. - **Methodological:** at what point does an accumulation of Tier-2/Tier-3 circumstantial links become *so* dense that the wiki's categorical restraint rule should yield to a calibrated "probable principal" framing? D005 holds the categorical line, but a future dialectic may need to decide whether the wiki's identification standard admits a tier between "unidentified" and "identified," and if so, what evidentiary threshold governs it. (This synthesis's intermediate framing is deliberately a *structural-link* upgrade, not such a probabilistic attribution tier — that question is left open.)