This began as a tax case against one man.

On 17 October 2024, the United States charged Alan Redmond alone with six counts under 26 U.S.C. § 7202 — six quarters of Bene Market trust-fund taxes, $1,068,680.47. No conspiracy. No wire fraud. No co-defendants. No entities. No forfeiture. No overt acts. Jacqueline C. Romero was the United States Attorney; Crawley and Dalke were the line AUSAs. It was, on its face, a collection case that had been criminalized: the theory was that payroll taxes were withheld and not paid over.

What that instrument did not say — and what the rest of this pack is about — is that the IRS had written on its own signed form that it applied this man's payments to the wrong liability, and had refunded him for it.

Three months later, on 27 January 2025, his CPA Malcolm Smith put a letter on paper stating that credits had not been applied, directing $240,016.75 to the quarters that are now Counts 23 and 24. In February 2025 that letter was walked to the United States Attorney's Office. The intent was to end the case. That is not what happened. On 12 March 2025, David Metcalf was sworn in as United States Attorney. Sixty-three days later, on 8 May 2025, the government returned a superseding indictment.

The superseder is a different case wearing the same caption. Six counts became twenty-seven. One defendant became six — Redmond, Bene Market LLC, Seguro Medico LLC, Walsh, Barrera, and Groff. One statute became four, with 18 U.S.C. § 1349 and § 1343 layered on top of § 7202 and the SCAMS Act, 18 U.S.C. § 2326, added as an enhancement that lifts the ceiling from twenty years to thirty. The charging window stretched backward eighteen months into the NBOA era and forward through December 2022. Seventy-two overt acts appeared where there had been none. Thirty million dollars of forfeiture and four named properties appeared where there had been none. The victim count went from "the IRS" to "tens of thousands" of consumers, with $150M of restitution attached. And one new tax count — Count 27, the Q2 2022 Seguro Medico quarter — was added, carrying $161,367.95 and taking the charged tax total to $1,230,048.42.

Then the case began to come apart at the co-defendant level, in the government's favor. Groff pled. Walsh pled. Barrera pled — to counts that remain [UNVERIFIED] in this pack. What the public plea documents do not contain is as important as what they do: no cooperation obligation, no § 5K1.1 clause, no promise to testify. Each recites only a sealed supplemental agreement. Until those supplements are produced, "the co-defendants are cooperating" is unverified in both directions, and nothing in this pack should be briefed as though it were settled. All three sentencings — Barrera 10/14/2026, Walsh 11/5/2026, Groff 11/6/2026 — now fall before trial.

On 3 August 2026, Benjamin Brait Cooper entered as counsel of record for Redmond, Bene Market, and Seguro Medico at ECF 162–164. Three days later he moved to continue; the next day the court granted it. The trial that had been set for September 2026 moved into January 2027 — pretrial 11/16/2026, motions in limine 12/7/2026, jury selection 1/11/2027, trial 1/12/2027.

That is where the board stands. One liberty case, one discharge gate, and eight collateral forums generating sworn statements that the first two will read.

The single most useful orientation fact in this pack, and the reason the chapters are ordered the way they are:

Redmond appears on 0 of the 12 customer-call wire counts. The two wire counts he personally touches — Counts 2 and 7 — are internal emails. [EX-A]

Nineteen wire counts break down as 7 internal emails + 12 customer calls (EX-C C-5). Seven of the nineteen involve no victim contact at all. The seven tax counts (21–27) charge Redmond alone — no entity, no co-defendant, no cooperator (CH-05).

This is carried forward from CH-07 § 7.1 because it governs every deadline in the pack. The project file contains two mutually exclusive schedules and both are still cited by live systems.

Working set for this pack, pending confirmation: pretrial 11/16/2026, motions wave 12/7/2026, jury selection 1/11/2027, trial 1/12/2027. Also carried: a Cooper gate of 10/15/2026 for motions, attributed to ECF 167 ¶ 8 — that paragraph text is [UNVERIFIED], it is not in the master docket description, and the motion schedule in this pack depends on it.

Decision rule. Do not say "trial 9/14" — even under the old order, 9/14 was jury selection and 9/15 was trial. Do not say "trial 1/11" — the master text makes 1/11 selection and 1/12 trial. Say nothing until the four PDFs are in hand. Filing to the wrong schedule is a default-adjacent event in either direction.

First ask of the engagement: confirm the criminal calendar in writing from the native ECF 167 / 168 / 169 / 171 PDFs. Everything downstream — the AP-254 gate on 9/16, the SBA opposition on 9/3, the BK exemption amendments — sequences off that confirmation.

Reproduced from EX-C § C-4. Original indictment of 17 October 2024 against the superseding indictment of 8 May 2025, element by element, with the delta and the defense implication.

What the delta is worth, stated honestly. The vindictive-prosecution motion built on this table is weak on grant (5–12%) — Blackledge v. Perry, 417 U.S. 21 (1974) raises a presumption only in defined circumstances, and United States v. Goodwin, 457 U.S. 368 (1982) and Bordenkircher v. Hayes, 434 U.S. 357 (1978) cut hard against a pretrial-superseder presumption. Its value is pressure, not dismissal (EX-B module 6). The table's real work is elsewhere: it establishes that Count 27 did not exist in the original instrument, which is a live sufficiency point on the § 7202 responsible-person element (EX-B module 4), and it frames the T9 theory in T00.

Beir bua.

[EX-A] Superseding Indictment, Doc 16 (counts, ¶¶ 1–72, Overt Acts 1–72) · [EX-B] Original Indictment, Doc 1 (10/17/2024, tax counts only) · [EX-C] charge book (COUNT_MATRIX, WIRE_TABLE, WIRING_ZERO, CODEFENDANTS) · [EX-D] element kill wiki V1 · [EX-E] corrections ledger, controlling where it conflicts with any other artifact · [EX-F] superseding analysis workbook (COUNT MAP, GROFF PLEA ANALYSIS, CRITICAL FINDINGS, 72 OVERT ACTS MAP, ORIGINAL vs SUPERSEDING) · [EX-G] Groff plea Doc 96-1 / memorandum ECF 96, filed 12/17/2025 · [EX-H] Walsh plea ECF 126 / memorandum ECF 127 · [EX-I] Five9 evidence packet · [EX-K] Simmons affidavit, 22-MJ-1897 · [EX-L] Redmond deposition 9/1/2022, Bates RedmondTax004490–004736 · [EX-M] ECF 147 / ECF 148 and Attachment A · [EX-N] calendar pack.

Count 1 charges all six defendants — Alan Redmond (1), Bene Market LLC (2), Seguro Medico LLC (3), Arthur Walsh (4), Jesus Barrera (5), Albert Groff (6) — with conspiring to commit wire fraud. The charged window is on or about 1/1/2018 through 12/2/2022 (Doc 16 ¶ 17) [EX-A]. The object is a § 1343 bait-and-switch scheme against "Victim-Purchasers": Limited Benefit Plans sold as though they were major medical or its equivalent, with interstate wires in furtherance (¶¶ 17–22, 23–36) [EX-A]. Manner and means are pleaded at ¶¶ 23–36; the roster of alleged misrepresentations is at ¶¶ 14–15; Overt Acts 1–72 supply the chronology. The telemarketing enhancement of 18 U.S.C. § 2326 is alleged (ten or more victims over 55), which raises the statutory maximum to thirty years per count [EX-F].

Two recitations in the project corpus give the conspiracy window as "2017–2021." That is wrong; ¶ 17 controls [EX-D]. The superseding instrument extended the window backward into the NBOA era and forward to 12/2/2022, an eighteen-month expansion over Doc 1's tax-only period [EX-B], [EX-F].

Redmond's alleged role (¶ 6): owner and 96% general partner of Bene Market, de facto owner of Seguro Medico, "ultimate decision maker" 2018–2022, later operating through nominees including his spouse S.K. and Walsh (¶ 15(p)). Walsh: COO of Bene, CEO and 4% member of Seguro, reporting to Redmond (¶ 7). Barrera: SVP / Director of Business Intelligence, reporting to Redmond (¶ 8). Groff: head sales agent who trained and supervised staff (¶ 9) [EX-A], [EX-C].

Section 1349 subjects a conspirator to the penalties of the object offense. The government must prove:

• A. An agreement between two or more persons to commit wire fraud;
• B. That Redmond knowingly and willfully joined that agreement, intending to further its unlawful objective; and
• C. The object offense's own components — a scheme to defraud or to obtain money or property by materially false or fraudulent pretenses, specific intent to defraud, and use of interstate wires in furtherance — as incorporated by reference.

Instruction numbering has been corrected and is load-bearing. Third Circuit Model Instruction "6.18.1349" does not exist; § 1349 runs through the § 371 series. The confirmed numbers are 6.18.1343 (wire fraud), 6.18.371H (single versus multiple conspiracies), 7.03 (Pinkerton), 5.06 (willful blindness), 5.07 (good faith — available on the fraud counts, not on the tax counts per the chapter 6 comment citing Basile), and 4.19 (cooperating witness) [EX-E]. Any prior artifact citing 6.18.1349 — including the element wiki's comment on the overt-act question [EX-D] — must be re-papered before the 12/7/2026 instruction filing.

Overt acts. The position that § 1349 has no overt-act element survives, but it now needs authority other than the non-existent model comment. Overt Acts 1–72 are the government's narrative, not a statutory checklist [EX-D]. Treat the proposition as [VERIFY — authority not yet in the source register] and brief it independently before it appears in an instruction request.

Element A — agreement between two or more persons to commit wire fraud

Element A status: NOT OVERCOME. Two co-defendants have pled guilty to Count 1. The existence of a conspiracy is locked as to them. Nothing in the record eliminates this element, and no filing should suggest otherwise.

Element B — knowing and willful joinder with intent to further the objective

Element B status: NOT OVERCOME. It is weakened at the margin — Redmond is not the voice on any charged consumer call, and the instrument itself splits scripts from closer ad-libs. It is not eliminated. Counsel should not be told this element is dead.

Element C — the incorporated § 1343 components

The Count 1 case will be built as a chronology, not a legal argument. Overt Acts 1–72 supply it: the 2018 payroll consolidation from NBOA into Bene Market; a 2019 stack of licensing complaints landing in Redmond's inbox; the June 2020 email telling his two managers how to handle a verification screen that says "this is not insurance"; the January 2021 roster showing seven of nineteen closers licensed; the April 2021 decision to use Groff's calls as the training set; the KISS brochure; a text campaign identifying the company as a "national enrollment center for health insurance"; twelve recorded calls; the January 2022 roster showing four of sixteen licensed; the April 2022 migration of staff into Seguro Medico; the June 2022 order that nothing leave the building without Redmond's approval; and ¶ 72, his civil deposition testimony [EX-A], [EX-F].

Two features of that chronology are worth naming now because they shape every motion that follows.

First, the instrument concedes the split. Paragraph 25 attributes scripts to Redmond, Walsh, and Barrera; ¶ 28 says the closers went beyond them. Paragraph 30 alleges verification recordings existed; Overt Act 59 alleges Barrera materially edited and altered a verification recording after the fact. A scheme in which the disclosures existed, were recorded, and had to be tampered with is a different scheme from one in which nothing was ever disclosed [EX-A], [EX-F].

Second, the government's own affiant supplied the alternative. The 22-MJ-1897 affidavit at ¶ 79 frames the entity history as serial rebranding in which "any issues were the responsibility of a single, bad employee" who "is then fired," naming W3 and Quanaijha Spradley [EX-E] § 8, [EX-K]. That is the rogue-employee theory in the government's handwriting. It does not defeat the agreement element. It is the best available anchor for the argument that the misconduct the jury will hear was individual and episodic rather than the object of a top-down agreement.

What they do. They prove, as to the pleading defendants, that a wire-fraud conspiracy existed. They give the government two witnesses who have admitted the object offense and, between them, seven charged wires. Groff's five charged calls and Walsh's two charged calls plus three charged emails mean the government can put a confessed participant next to almost every category of wire in the instrument [EX-G], [EX-H], [EX-C].

What they do not do. A co-defendant's guilty plea is not evidence of Redmond's membership, and the government cannot use the fact of a plea as substantive proof against him. Co-conspirator statements under Rule 801(d)(2)(E) still require an independent predicate finding of an agreement that included Redmond. More sharply: not one public plea, on the present record, admits an agreement with Redmond. Groff's ¶ 1 describes his own role and names no co-conspirator, and no public factual proffer describes a Redmond directive [EX-E] § 7.

Cooperation is unverified, not established. Several project artifacts treat Groff and Walsh as cooperating witnesses and one assigns 90% confidence to cooperation provisions in Groff's sealed supplement [EX-F] GROFF PLEA ANALYSIS. The controlling ledger is narrower: no public § 5K1.1 or substantial-assistance clause, no promise to testify, appears in any public plea. Paragraph 17 recites only that "the parties have also entered a supplemental agreement that is filed under seal." Until ECF 97, 130, and 142 are produced, "Groff, Walsh, and Barrera are cooperating" is an inference, not a fact [EX-E] § 7. Plead it as an inference and demand the supplements.

The timing argument has flipped and must be re-papered. All three sentencings — 10/14/2026 (Barrera), 11/5/2026 (Walsh), 11/6/2026 (Groff) — now fall before the 1/12/2027 trial. The "still hoping for a sentencing reduction" cross is gone. The replacement is Rule 35(b) exposure, which survives sentencing [EX-E] § 7. Any artifact describing cooperator leverage as decaying at those dates predates the continuance [EX-C] ASK_BENJAMIN.

Groff's loss stipulation is his. Doc 96-1 ¶ 11(b) stipulates a band "between $25,000,000 and $150,000,000," with only the minimum two-level victim / mass-marketing bump under USSG § 2B1.1(b)(2) rather than the six-level tier [EX-E] § 7. It is not Redmond's admission and must not be allowed to migrate.

This is the live fight on Count 1, and it is a fight about the size of the agreement rather than its existence.

Backward edge — 2018 and the NBOA era. The window opens 1/1/2018, before Seguro Medico existed (formed 12/27/2019) and while the operation ran through NBOA and Bene Market [EX-A], [EX-C]. Overt Act 1 (1/7/2018) is a payroll transition — an administrative act. The government needs the 2018 anchor because the superseding instrument's expansion of the window is what converts a 2021–2022 sales problem into a five-year enterprise. Every 2018 overt act should be tested for whether it evidences agreement to defraud purchasers or merely evidences running a business.

Forward edge — 12/2/2022. Conduct after 12/2/2022 is outside the charged conspiracy. The 4/1/2022 Bene-to-Seguro transition (Overt Act 65) and the 6/7/2022 control directive (Overt Act 69) are inside it and are the government's strongest "he still ran it" acts. The 9/1/2022 deposition (¶ 72) is inside it as well.

The nominee overlay. The government alleges Redmond controlled Seguro through nominees, including S.K. (¶ 6, ¶ 15(p)) [EX-A]. S.K. is uncharged. The nominee theory is an add-on that the government must prove, not a premise. Do not volunteer it.

Single versus multiple conspiracies. The instrument defines a "Bene Market Group" of roughly fifteen entities [EX-F] ORIGINAL vs SUPERSEDING. Multiple entities and DBAs do not by themselves create multiple conspiracies, and a variance theory is a decision, not a gift: it invites the government to argue that whichever conspiracy the jury finds, Redmond was in it. If counsel wants the 6.18.371H instruction, the predicate must be built from the rogue-employee churn in [EX-K] ¶ 79 and the ¶ 28 off-script concession, and the Kotteakos line must be briefed with a verified citation, which the present register does not contain [VERIFY] [EX-D] § 1.

Everything connecting Redmond to Counts 3, 5, 9, 10, 11, 12, 13, 14, 15, 16, 17 and 18 runs through § 1349, § 2, and Pinkerton — through other men's mouths [EX-E] § 6. That sentence is the spine of the defense and it is also the reason Count 1 cannot be treated as a satellite of the substantive counts. If Count 1 goes to the jury with an unrestricted Pinkerton instruction (7.03), the twelve calls arrive as Redmond's calls by operation of law.

The work therefore is instructional, and it is due 12/7/2026:

1. A 7.03 instruction limited to acts within the scope of the agreement the jury finds, and reasonably foreseeable — with the ¶ 28 off-script concession argued as the boundary.
2. A 6.18.371H single-versus-multiple instruction if the predicate is built.
3. A 5.07 good-faith instruction on the fraud counts, which is available here even though it is not available on the tax counts.
4. Resistance to 5.06 willful blindness, which the notice overt acts (4, 8, 25, 61) are designed to earn.
5. A 4.19 cooperating-witness instruction if any plea witness testifies, framed on Rule 35(b) exposure rather than pre-sentence hope.

Withdrawal is a defense to liability for later acts of the conspiracy and to the substantive counts committed after the withdrawal; it does not undo membership already formed. To make it out, a defendant must show an affirmative act inconsistent with the object of the conspiracy — a disclosure to law enforcement, or a communication of the abandonment to co-conspirators — and mere cessation of activity or non-participation does not suffice. The burden allocation is a defendant-burden question in the modern federal cases; the controlling Supreme Court authority on placing that burden on the defendant is not in this project's source register and must be pulled and pincited before it is briefed [VERIFY — authority outside the register].

Applied to this record, honestly: there is no withdrawal predicate. Nothing identified in the instrument, the plea papers, the affidavit, or the Five9 material shows Redmond disclosing the scheme to authorities or communicating abandonment to Walsh, Barrera, or Groff. The record runs the other way: Overt Act 65 has him moving staff into Seguro Medico in April 2022, and Overt Act 69 has him ordering in June 2022 that nothing external move without his approval — "EVER … ZERO. No interviews. No look ups. No nothing" [EX-A], [EX-C]. Those are acts of continued control inside the charged window.

Do not plead withdrawal without a factual predicate. What would create one, if it exists in unreviewed material: a documented resignation or divestiture from Bene Market and Seguro Medico with a date inside the window; a contemporaneous instruction to stop the conduct that the government alleges; or a disclosure to a regulator or to the Department of Insurance that is not itself part of the alleged complaint-suppression pattern (Overt Acts 42, 44, 54, 66 cut hard against that last one). Until such a document exists, withdrawal is a theory without facts, and filing it would concede membership for nothing.

The honest reasons, stated so counsel is not surprised at the pretrial conference:

1. Two co-defendants have pled guilty to this count. The existence of the conspiracy is admitted twice on the public record [EX-G], [EX-H].
2. Two of the six defendants are entities Redmond owned or controlled. A theory in which no agreement existed cannot survive the presence of Bene Market LLC and Seguro Medico LLC as charged co-conspirators [EX-D].
3. Redmond authored two of the charged wires. Count 2 (6/23/2020) and Count 7 (4/27/2021) are his. Count 2 in particular is a training instruction about how to handle a disclosure that said the product "is not insurance" [EX-A], [EX-C]. See Chapter 2.
4. The notice overt acts are documentary. Overt Acts 4, 8, 25, and 61 put licensing failures and regulatory actions in Redmond's inbox on dated emails [EX-F].
5. The refund-conditioning overt acts name him directly. Overt Acts 42, 44, 54, and 66 are the hardest facts in the instrument [EX-F].
6. ¶ 72 puts his own sworn words in the case. The capacity defect is real — the 9/1/2022 deposition was a Rule 30(b)(6) organizational-designee examination, and DOL counsel put the capacity on the record: "Before, you were testifying on behalf of Alan Redmond. Today, you are testifying based upon the collected knowledge of the Defendants" (RedmondTax004498, p. 9:1–9). ¶ 72 also fuses two answers seven Bates pages apart, one of them prefaced "I can't recall" and hedged "I imagine," and the Bates range carried in earlier artifacts (RedmondTax004506-07) contains neither charged quote [EX-E] §§ 2, 3, [EX-L]. That is a powerful attack on ¶ 72. It is not a defense to Count 1.

What is genuinely winnable on Count 1. Narrowing the agreement the jury is asked to find; a Pinkerton instruction bounded by that agreement's scope; a single-versus-multiple instruction if the predicate is built; Rule 702 exclusion of the Volk "written summary" duty opinion; a 4.19 instruction; production of ECF 97, 130 and 142; and keeping Groff's loss band out of Redmond's case. Those are weight and instruction wins. Across all twenty-seven counts, only Counts 19 and 20 survived adversarial review as a genuine sufficiency gap [EX-E] § 10 — and those are wire counts, not Count 1.

1. Sealed supplements ECF 97 (Groff), ECF 130 (Walsh), ECF 142 (Barrera) unread. Cooperation remains an inference.
2. Barrera's plea counts are unverified; no memorandum located.
3. Walsh ECF 128 unread.
4. The no-overt-act proposition for § 1349 needs authority that is not the non-existent model comment.
5. The withdrawal burden authority is outside the register and must be pulled.
6. Neder materiality pincite and the Kotteakos variance line are both [VERIFY PINCITE].
7. Whether the government can prove interstate routing for each internal email wire is unverified.
8. ECF 167 ¶ 8 (the 10/15/2026 motions gate referenced in the calendar pack) rests on docket text; the PDF is unpulled [EX-N].
9. The government reserved the right to supplement the 38-recording list and to move in limine on three intrinsic / 404(b) categories; see Chapter 4 § 4 [EX-E] § 9.

Nineteen § 1343 counts are charged. Twelve are customer telephone calls (Chapter 3). Seven are internal emails with no purchaser on the wire: Counts 2, 4, 6, 7, 8, 19, and 20 [EX-C] WIRING_ZERO, [EX-F] CRITICAL FINDINGS ¶ 2. Redmond is the sender on two of the seven — Counts 2 and 7 — and the recipient or the attributed source of the instruction on the other five [EX-C].

That distribution is the whole reason this set matters more than its count. On the twelve calls, Redmond is nowhere: he is the speaker on zero of them, and from the government's own ECF 148 Attachment A (38 recordings, 7 internal and 31 Five9 victim calls) he is the primary subject of three, none of which is a charged wire [EX-E] § 6, [EX-M]. The email counts are therefore the only place in Counts 2–20 where the government has Redmond's own words on a charged wire. Counsel should expect the government to try the fraud case through Counts 2 and 7 and to use the remaining five emails to prove control.

Each § 1343 count requires: (1) a scheme to defraud, or to obtain money or property by materially false or fraudulent pretenses, representations, or promises; (2) specific intent to defraud; and (3) transmission by means of wire in interstate or foreign commerce in furtherance of the scheme. Instruction 6.18.1343 governs. There is no such instruction as 6.18.1349; the conspiracy count runs through the § 371 series [EX-E] § 4.

Four shared points control this set.

Furtherance is not automatic. An internal training or management email is "in furtherance" only if it advanced the scheme. Crossing a state line is the jurisdictional element, not the furtherance element. The Schmuck furtherance line is genuinely available to the government — including for post-sale and lulling communications — but it must be argued count by count, and the government has to identify what each email advanced [EX-D] § 2.

Interstate transmission must be proved for each email. The instrument alleges interstate wires. Whether the government holds routing or header proof showing that each of these seven internal messages crossed a state line is [UNVERIFIED] on the present record. For messages between colleagues in the same building this is a real element, not a formality, and it should be tested in the Rule 16 demand and at Rule 29 rather than assumed away.

Rule 106, as amended, is the best weapon in this set. Rule 106 was amended effective 12/1/2023: a completing statement is now admissible over a hearsay objection. That amendment post-dates most of this project's corpus and is the single strongest 2026-vintage tool available [EX-E] § 4. Every count in this chapter is an excerpt from a thread — a forwarded brochure, a reply confirming an instruction, an email with an attachment. Each one is a Rule 106 target: the full thread, the attachment, the surrounding approvals.

Authentication is only the first gate. The government must still establish record-specific hearsay treatment, authorship and speaker attribution, and completeness for each exhibit. ECF 147 ¶ 5 n.1 concedes that roughly ten business-records certifications were "forthcoming" — a Rule 104 self-authentication determination cannot issue for records whose certificates do not exist [EX-E] § 5, [EX-M].

One materiality frame applies to all seven. The government has conceded in writing, twice, that "During the charged period, neither the ACA nor state insurance laws made it illegal to offer for sale any of these Limited Benefit Plans" — ECF 96 at 6; ECF 127 at 6 [EX-E] § 5, [EX-G], [EX-H]. Internal emails about how to sell a lawful product are not, standing alone, material falsehoods to any purchaser. That is the frame; it is not a kill.

The alleged wire. An email from Redmond to Barrera and Walsh containing instructions to train closers with "fraudulent and misleading sales tactics." Overt Act 16 supplies the content the government will read to the jury: how to combat the verification screen statement that "this is not insurance," by describing the product as a "health sharing plan" that "mimics health insurance" [EX-A] Count 2 and OA 16; [EX-C] WIRE_TABLE.

Element grid — Count 2

Kill status: NOT OVERCOME. This count is a problem, and it is the government's best Redmond-voice wire. The worst facts, stated plainly:

• Redmond is the author, not a recipient. The script-versus-closer wedge that carries the call counts does not help here; on this count he is on the script side of the split.
• The email shows he knew the verification screen said "this is not insurance." The government will pair it with Overt Act 59, in which Barrera materially edited and altered a verification recording, and argue a course of conduct aimed at the disclosure itself.
• The word "combat" is the government's opening line. It will be quoted in the indictment reading, the opening, the direct of any plea witness, and the closing.
• The native message is not in hand, so the completeness attack cannot yet be specified.
• One prior artifact assigns this count a 55% acquittal estimate and labels the email counts "weakest" [EX-F] WIRE FRAUD CLASSIFICATION. Those percentages are a prior-session estimate and are not an element analysis. On the elements, nothing in the record eliminates any element of Count 2.

What to actually do with Count 2. Demand the native message and the full thread; separate Redmond's authored text from quoted or forwarded text; establish through the verification records that the disclosure existed and was played; and litigate the Volk "written summary" opinion under Rule 702, because the government's theory of this email depends on an obligation it has conceded did not exist.

The alleged wire. An email to Redmond with an attachment that "assigned all non-selling licenses to non-licensed agents," so that unlicensed employees sold under licensed agents' licenses. Overt Act 25 describes the same date: Redmond received a list showing 7 of 19 closers licensed [EX-A] Count 4, ¶ 14(s), ¶ 15(e); [EX-F] 72 OVERT ACTS MAP.

Element grid — Count 4

Kill status: NOT OVERCOME as a furtherance wire; WEAKENED as standalone consumer fraud. The count survives because furtherance is a low bar and the licensing allegations are documentary. It is weak as a fraud count because no purchaser is on the wire and Redmond did not write it.

The alleged wire. Walsh's email to Redmond confirming that Barrera would "pull the Albert [Groff] calls … for training" new sales agents. Overt Act 30, same date [EX-A] Count 6; [EX-F].

Element grid — Count 6

Kill status: NOT OVERCOME. Walsh's plea to this count is a brick. Demand the identity and audio of the specific Groff calls that were pulled; if the training set was compliant, the government's furtherance theory becomes an argument about labels.

The alleged wire. Redmond forwarding a Walsh email together with a "misleading internal sales brochure summary titled 'KISS Principles'" for employees selling Limited Benefit Plans [EX-A] Count 7; [EX-C].

Element grid — Count 7

Kill status: NOT OVERCOME. Redmond is the sender and a pleading co-defendant has admitted the count. The count cannot be assessed further until the brochure is read; that document is a top-tier discovery priority.

The alleged wire. Barrera's email containing the copy for texts to approximately 2,000 unsold consumers; the texts "falsely stated, 'I am reaching out from the national enrollment center for health insurance.'" Matches ¶ 14(h) and ¶ 27 [EX-A] Count 8; [EX-C].

Element grid — Count 8

Kill status: NOT OVERCOME. The quoted sentence is the government's best non-call falsehood. The available work is on authorship, approval, and whether the campaign was executed.

The alleged wires. Two emails on the same date to Seguro Medico senior management, one from Walsh (Count 19) and one from Barrera (Count 20), each conveying the same Redmond instruction: outside requests must "seek approval from me first … before doing anything. EVER. Nothing external is done unless run through me. ZERO. No interviews. No look ups. No nothing." [EX-A] Counts 19–20 and OA 69; [EX-C] WIRE_TABLE.

A source conflict to resolve before any Rule 12 filing. The charge book quotes the instruction itself as the content of both counts [EX-C]. The analysis workbook describes Counts 19 and 20 as charging the same email thread in which Walsh and Barrera "each replied 'confirm'" to Redmond [EX-F] CRITICAL FINDINGS ¶ 4. Those are materially different charged wires — a transmitted directive versus a one-word acknowledgment — and the multiplicity argument is stronger under the second reading. The count tables in Doc 16 must be read against the native thread before the multiplicity motion is filed.

Element grid — Counts 19 and 20

Kill status. - Count 19: NOT OVERCOME as a furtherance-and-concealment wire; WEAKENED to the point of a genuine Rule 29 argument on the false-pretense element. Walsh pled guilty to this count, which is the reason it is not simply dismissible: a participant has admitted that this email was part of the fraud [EX-H]. - Count 20: WEAKENED standing alone, and the multiplicity candidate. It rises or falls with Count 19 on furtherance, and it is the count to move against under Rule 12.

Do not oversell either one. The controlling ledger says these two are the only clean element gap across the whole instrument, and it also says everything else is weight. Walsh's plea to Count 19 means the sufficiency argument will be made against the backdrop of a co-defendant's admission that the same email furthered the scheme. Brief it as a Rule 29 argument preserved from the start and renewed, not as a pretrial dismissal certainty.

Bottom line for the set. Twelve calls, zero Redmond voice; seven emails, two Redmond transmissions. None of the seven is overcome. Two of them (19, 20) support a real sufficiency argument on the false-pretense element. Two of them (2, 7) are the government's Redmond-voice case, and Count 2 is the single hardest wire count in the instrument for the defense. The government's own concession that selling Limited Benefit Plans was lawful, plus amended Rule 106, plus the missing native documents, are the tools that exist. A motion to dismiss the email counts wholesale as "internal business communications" [EX-F] CRITICAL FINDINGS ¶ 2 will not succeed against Schmuck furtherance and should not be the vehicle; Rule 29 on 19 and 20, Rule 12 multiplicity on 20, and Rule 106 and Rule 702 practice on 2 and 7 are the vehicles.

1. Native Count 2 email — not in the present ingest. Blocks the completeness attack on the government's best Redmond wire.
2. KISS Principles brochure PDF — not in the present ingest. Count 7 cannot be fully assessed without it.
3. Agents-list attachment (Count 4) — native not in the present ingest.
4. SMS vendor logs (Count 8) — not in the present ingest; execution of the campaign unproved either way.
5. Full 6/7/2022 thread (Counts 19–20) — required to resolve whether the charged wires are the directive or one-word confirmations, and to fix the multiplicity theory.
6. Interstate routing proof for each internal email — unverified.
7. Barrera's plea counts — unverified; bears on Counts 8 and 20.
8. Sealed plea supplements ECF 97, 130, 142 — unread; cooperation remains an inference, not a fact [EX-E] § 7.
9. Roughly ten Rule 902(11) certifications were "forthcoming" per ECF 147 ¶ 5 n.1; the self-authentication determination is not ripe as to those records [EX-E] § 5.
10. Per-count acquittal percentages in [EX-F] WIRE FRAUD CLASSIFICATION are prior-session estimates. They are recorded here for continuity and are not adopted as element conclusions.

[EX-A] Superseding Indictment, Doc 16 (counts, ¶¶ 14–36, Overt Acts 1–72) · [EX-C] charge book (COUNT_MATRIX, WIRE_TABLE, WIRING_ZERO) · [EX-D] element kill wiki V1 · [EX-E] corrections ledger, controlling in any conflict · [EX-F] superseding analysis workbook (COUNT MAP, IVR DEFENSE MAP, WIRE FRAUD CLASSIFICATION, GROFF PLEA ANALYSIS, CRITICAL FINDINGS, 72 OVERT ACTS MAP) · [EX-G] Groff plea Doc 96-1 / ECF 96 · [EX-H] Walsh plea ECF 126 / ECF 127 · [EX-I] Five9 evidence packet · [EX-J] census-to-elements pack · [EX-K] Simmons affidavit, 22-MJ-1897 · [EX-M] ECF 147 / ECF 148 and Attachment A.

Of the twelve call-based wire-fraud counts (3, 5, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18), Alan Redmond is not on a single one [EX-C] WIRING_ZERO. That is verified two ways: the charged speaker in each count table is someone else, and an independent victim-and-count workbook shows no Redmond caller row on any of the twelve.

The government's own notice sharpens it. ECF 148 Attachment A lists 38 recordings — 7 internal and 31 Five9 victim calls. Redmond is the primary subject of exactly three, and not one of those three is a charged wire. He is on zero of the twelve charged victim calls [EX-E] § 6, [EX-M]. Everything tying him to these counts runs through § 1349, § 2, and Pinkerton — through other men's mouths.

Two limits, so it is not oversold. It is a speaker fact, not an element fact: it does not touch the agreement element, defeat Pinkerton, or make any count dismissible [EX-C]. And the 38-recording list is a floor, not a ceiling — the government reserved the right to supplement it and to move in limine on regulatory history and license revocations, the 6/10/2015 Redmond sales-training tape, and pre-2019 taxpaying history [EX-E] § 9. If the 2015 tape comes in, the "you will never hear Alan Redmond lying to a customer" line is at risk. Build the late-supplementation objection now.

Nineteen § 1343 counts: twelve calls, seven emails. Redmond voice on a customer call, 0 of 12; sender on an email wire, 2 of 7 (Counts 2, 7) [EX-C].

Each count requires a scheme to defraud or obtain money by materially false pretenses, specific intent to defraud, and an interstate wire in furtherance. Instruction 6.18.1343 governs; Pinkerton is 7.03, single-versus-multiple conspiracies 6.18.371H, good faith 5.07 (available on these counts), cooperating witness 4.19 [EX-E] § 4.

The interstate-wire-in-furtherance element is not separately contested on the calls, and it is omitted from the grids below. Each charged call crossed a state line to a named purchaser, and the Schmuck line reaches post-sale and retention calls too. The fight on all twelve is falsehood, materiality, and intent.

Provenance warning — read before quoting any call content. The count tables in Doc 16 contain no quoted false statement for any of the twelve calls. Each alleges that the speaker "fraudulently solicited" the purchaser "through numerous misrepresentations and half-truths," or on Count 3 that the representative "misled" her [EX-C] WIRE_TABLE. The specific content below — "comprehensive Major Medical," the cornea-transplant figures, "government stamp of approval," "$20 copay no bill," "compliance officer for California," the sales name "Jerry," instructions to skip or ignore verification — comes from the analysis workbook's call-content mapping and the overt-act narrative, not from the count tables [EX-F]. Every such phrase is [UNVERIFIED against the charged count table] until matched to a produced recording or an identified overt act. Filing a quotation the count table does not contain hands the government a free correction.

Script versus closer. Paragraph 25 attributes the scripts to Redmond, Walsh, and Barrera; ¶ 28 states the closers made false statements "beyond those contained in the standard scripts" [EX-A]. Where the charged lie is an ad-lib, Pinkerton and § 2 are the only glue, and their scope is an instruction fight.

Particulars and disclosure timing. A bill-of-particulars motion is docketed (ECF 150) and opposed (ECF 156) [EX-D] § 2. Giglio, Jencks and witness-identity disclosure are being withheld until fourteen days before trial (ECF 156 at 16–18); the authorities and the consequences are set out in Chapter 4 § 5. Any theory requiring identification of the unnamed closers must survive that timetable.

Count 3 — 12/30/2020 — unnamed Bene Market representative ↔ L.F. (CA) — cancellation call

NOT OVERCOME. Weakened — unnamed speaker, no quoted lie, 6.452% completeness. The natural lead exhibit for Rule 106 and Rule 1006 practice.

Count 5 — 1/20/2021 — unlicensed closer ↔ P.G. (FL)

NOT OVERCOME. Recording not reviewed; license theory is atmosphere, not an element.

Count 9 — 7/27/2021 — unnamed closer ↔ B.B. (DE)

NOT OVERCOME.

Count 10 — 8/18/2021 — Albert Groff ↔ P.L. (MD)

NOT OVERCOME as to Groff; not killed as to Redmond.

Count 11 — 9/2/2021 — unnamed closer ↔ T.M. (ME)

NOT OVERCOME.

Count 12 — 10/23/2021 — Jesus Barrera ↔ R.S. (RI)

NOT OVERCOME.

Count 13 — 11/17/2021 — Groff ↔ E.T. (TX), plan purchased for husband G.T.

NOT OVERCOME as to Groff. Verify the product issued before adopting a position.

Count 14 — 12/11/2021 — Groff ↔ L.R. (NY)

NOT OVERCOME as to Groff.

Count 15 — 1/3/2022 — Arthur Walsh ↔ R.C. (MA)

NOT OVERCOME.

Count 16 — 1/3/2022 — Walsh ↔ J.S. (CA) — same day as Count 15

NOT OVERCOME. Walsh pled this count.

Count 17 — 1/24/2022 — Groff ↔ A.C. and spouse (ME)

NOT OVERCOME as to Groff.

Count 18 — 2/22/2022 — Groff ↔ D.P. (NY) — post-sale (spouse's plan sold 1/28/2022)

NOT OVERCOME as to Groff. A different species — service and retention rather than origination.

Every charged sale ran through a recorded verification. The instrument concedes it: ¶ 30 alleges the verification recordings, and Overt Act 59 alleges that Barrera materially edited and altered a verification recording after the fact to support a Groff response to state regulators [EX-A], [EX-F]. The mapping is consistent across Counts 5, 9, 10, 11, 12, 13, 14, 15, 16, 17 and 18: the verification would have disclosed that the plan was not ACA compliant, that it was a limited-benefit product, and what its caps and exclusions were [EX-F] IVR DEFENSE MAP.

The proposition. If a purchaser completed the verification and acknowledged the disclosures, informed consent existed at the moment of sale regardless of the pitch. It goes to materiality, and it is the only defense here operating on all twelve counts at once. On Counts 9, 11, 12, 14 and 15 the map's answer is generic — the verification would have contradicted the mapped claim about ACA status, exclusions, copays, caps, or network. On Counts 5, 10, 16 and 17 it is count-specific and appears in those grids.

Why it cannot be led with. Four facts cut the other way and the government has all of them. (1) A recording captures an agent pre-coaching the verification — let the prompt finish, then "say I agree and press 1" (recording 109657236), destroying third-party-verification validity for that call [EX-J] 2.4. (2) Overt Acts 58 and 63 describe coaching purchasers through the verification — "keep pressing 1" — and Count 16's mapped content is the same practice [EX-J] 2.1. (3) Overt Act 59 alleges after-the-fact alteration of a verification recording. (4) The verification content is not uniformly helpful: the earlier conclusion that the IVR disclosed non-ACA status was corrected to note the $99 enrollment fee was in the IVR [EX-J] 2.5.

Every IVR status in the map reads NEED TO VERIFY. No charged-call verification has been confirmed produced and reviewed [EX-F]. Until then, informed consent is a hypothesis. The narrower framing survives regardless: the government has charged a bait-and-switch scheme in which its own instrument alleges the disclosures existed, were recorded, were coached around, and in one instance had to be altered — you cannot tamper with a verification that does not exist [EX-J] 2.1. Pull order: P.L. (Count 10), A.C. (Count 17), L.R. (Count 14), J.S. (Count 16), and the original 1/28/2022 sale recording behind Count 18.

The recording corpus is the case. The figures below sit at different grains.

The denominator that must reach the jury [EX-E] § 6: 1,100,000 total recordings; 853,000 filtered as failed or under ten seconds (77.6%); 247,000 substantive; 23 government-victim calls.

The production ledger [EX-I]: 1,041,582 native recordings in the discovery manifest, 152.7 GB, 250,000 stamps. Victim production proof: 1,337 rows, 408 produced (30.52%). Victim timelines: 1,357 rows, 412 produced (30.36%), and 945 carry the literal string "NOT PRODUCED" in the Bates column itself. By disposition, "No Answer" is 110 rows / 0 produced and "Abandon" 59 / 0. The inventory table the government pointed to holds two rows — an Azure blob stub, not the corpus. A session-ID crosswalk separately resolved 24 charged victims and showed 249 of 463 (54%) substantive calls withheld [EX-J] 2.6.

Resolve three conflicts before any of this is served. Totals differ (1,100,000 versus 1,041,582); charged-victim calls differ (23 versus 24, against twelve call counts, three of which name a second purchaser); and the substantive denominator differs (247,000 versus 463). Pick one grain per exhibit and label it.

Admissibility gates. Authentication is the first gate only; record-specific hearsay treatment, speaker attribution, and completeness remain [EX-J] 2.11. Amended Rule 106 (effective 12/1/2023) makes a completing recording admissible over a hearsay objection — the strongest tool against a 30%-production corpus — and roughly ten Rule 902(11) certifications were still "forthcoming" per ECF 147 ¶ 5 n.1 [EX-E] §§ 4–5. The Five9 warrant number appears in neither ECF 147 nor ECF 148; 22-MJ-1897 is a parallel search matter not shown to be the Five9 process [EX-C]. Affidavit ¶ 71's claim that over 13,000 calls were reviewed and more than 90% contained material misstatements is a bare estimate with no methodology — a probable-cause attack, not an element [EX-J] 2.7, [EX-K].

The licensing allegations run through the instrument: ¶ 14(s) and ¶ 15(e); Overt Act 25 (1/10/2021 — 7 of 19 closers licensed, the document charged as the Count 4 wire); Overt Act 61 (1/7/2022 — 4 of 16 licensed); Overt Act 8 (9/13/2019 — seven outstanding regulatory actions against Groff, who stayed); and Count 5, charged against an "unlicensed closer" [EX-A], [EX-F].

Selling without a state license is a regulatory violation, not a false statement to a purchaser, and § 1343 requires materiality to the purchaser rather than the regulator. The government has conceded in writing, twice, that "During the charged period, neither the ACA nor state insurance laws made it illegal to offer for sale any of these Limited Benefit Plans" — ECF 96 at 6; ECF 127 at 6. Its noticed expert JoAnn Volk will nonetheless opine that "prior to being asked to purchase a plan, a consumer should be provided a written summary" — an opinion about a duty the government has admitted did not exist. That is a Rule 702 motion, not a cross [EX-E] § 5.

The limit: unlicensed status is atmosphere, and it feeds willful blindness (5.06) through the roster overt acts. Defeating the license theory defeats no count.

Groff pled guilty to Counts 1, 10, 13, 14, 17 and 18 — the conspiracy plus five charged calls, all his own [EX-G]. One workbook entry says six charged calls; the list is five calls plus Count 6, an email about his calls [EX-F]. Use five.

The wedge. Paragraph 28 concedes the closers spoke beyond the scripts, and the workbook maps Groff's cornea-transplant claim escalating $50K → $80K → $90K → $120K across calls (Overt Acts 52, 53, 56, 60, 63) [EX-F]. No company script instructed an agent to invent a personal cornea-transplant figure and inflate it call after call. That escalation is the best available proof that the charged lies on the Groff counts were his inventions.

What it does. Supports a scope argument on Count 1 and a Pinkerton-boundary argument on Counts 10, 13, 14, 17 and 18: acts outside the scope of the agreement the jury finds, and not reasonably foreseeable, are not Redmond's. It also supplies the cross of a serial fabricator whose figures moved with the audience.

What it does not do. It does not defeat the counts; Groff admitted them. And the government's answer sits in its own instrument: Overt Act 30 and Count 6 allege Groff's calls were pulled and used to train new agents, which converts a rogue agent back into company method [EX-A], [EX-H]. Demand the specific training calls — if the training set was compliant, the circularity is exposed.

Groff's loss stipulation — the $25,000,000 to $150,000,000 band at Doc 96-1 ¶ 11(b), with only the minimum two-level victim and mass-marketing bump under USSG § 2B1.1(b)(2) — is his guideline world and must not migrate [EX-E] § 7.

None of the twelve is overcome. Four have unnamed speakers; none has a quoted false statement in its count table; Redmond speaks on none; seven were admitted by the caller. Realistic objectives: a bill of particulars on the four unnamed counts, a Pinkerton instruction bounded by the agreement the jury finds, Rule 106 completion against a 30%-production corpus, Rule 702 exclusion of the written-summary duty opinion, and a Rule 1006 summary putting twenty-three charged calls against a quarter-million substantive ones.

1. Recordings for all twelve charged purchasers — L.F., P.G., B.B., T.M., P.L., R.S., E.T., L.R., R.C., J.S., A.C., D.P. — not reviewed. Top discovery gap in the case.
2. Every verification status reads NEED TO VERIFY; informed consent is unproved on every count.
3. No quoted false statement exists in any of the twelve count tables; all call content here is workbook- or overt-act-sourced.
4. Identity of the unnamed closers on Counts 3, 5, 9, 11 — unknown, and no witness list is promised.
5. Count 13 product type — whether a genuine major-medical policy issued is unverified and materially affects that count.
6. Barrera's plea counts (Count 12) — unverified; ECF 142 unread.
7. Sealed supplements ECF 97 and ECF 130 — unread; cooperation remains an inference [EX-E] § 7.
8. Five9 warrant number absent from ECF 147 and 148; roughly ten Rule 902(11) certifications still "forthcoming."
9. Corpus denominators conflict (1,100,000 vs 1,041,582; 23 vs 24 charged victim calls; 247,000 vs 463 substantive).
10. Per-count acquittal percentages in [EX-F] are prior-session estimates, not element conclusions.

§ 7202 reaches any person required to collect, truthfully account for, and pay over any tax who willfully fails to do so. United States v. Thayer, 201 F.3d 214, 220-21 (3d Cir. 1999) construes the duties disjunctively: failing either to truthfully account for or to pay over completes the violation. Two consequences control everything below.

1. "We filed accurate 941s" is not a defense. The returns were filed. Under Thayer that is the accounting half of the duty performed and the pay-over half unperformed, which is the offense.
2. There is no "tax due and owing" element. DOJ Criminal Tax Manual § 9.03; cf. United States v. Citron, 783 F.2d 307 (2d Cir. 1986). Figure fights therefore travel to notice (Rule 7(f)), willfulness, and loss — not to a missing element. [VERIFY_PINCITE before filing]

Citation discipline. Thayer is adverse and controlling. Cite it for the disjunctive holding when the government will cite it anyway; never cite Thayer for a "no amount element" proposition — that proposition is not locatable in the opinion and the mis-cite is a free correction handed to the government.

Instruction trap. Third Circuit Model Instruction 5.07 (good faith) is not available on the tax counts (ch. 6 comment citing Basile). The good-faith showing must ride inside the Cheek willfulness definition — voluntary, intentional violation of a known legal duty, Cheek v. United States, 498 U.S. 192, 201 (1991) — not as a free-standing charge. Build the requested willfulness instruction accordingly and do not assume 5.07 will be given.

Element grid — all seven counts

The single most valuable tax document in the file is [EX-B1]: the certified MFT 55 trust-fund-recovery-penalty transcript, six modules, all assessed 06/13/2022, transcript run 07/28/2022. Set the charged figures against it.

Two exact matches, to the penny, on the same certified page, in the government's own type. That is not coincidence and it is not an alternative computation — the MFT 55 TFRP ledger is the charging source for the 2019–2020 Bene counts. Counts 23 and 24 are conceded on the figure for exactly that reason, and the concession buys the credibility that makes the next paragraph land.

Counts 21 and 22 depart from that same source and no produced document explains how. Two obvious bridges were tested and both fail:

• Gross TFRP: 225,741.80 and 238,287.69 are not the charged numbers.
• TFRP net of in-period federal tax deposits: Count 21 — 225,741.80 − 11,400.00 (TC650) = 214,341.80, still 71,676.08 above the charged 142,665.72. Count 22 — 238,287.69 − 107,950.10 (TC650 65,078.00 + 42,872.10) = 130,337.59, now 13,253.87 below the charged 143,591.46.

The deposit hypothesis over-corrects on one count and under-corrects on the other. Whatever arithmetic produced $142,665.72 and $143,591.46, it is not on any produced page. That is the Rule 7(f) bill of particulars, and [EX-B1] is Exhibit A to it — a one-page motion built entirely from the government's own certified record, asking only that the government state how it computed two numbers it put before a grand jury.

Frame the relief narrowly. The ask is identify the document and the computation, not dismiss the counts. Particulars on Counts 21 and 22 is a 35-55% partial-grant proposition; dismissal on this record is not.

Two modules on the same certified page — 201809 ($18,535.56) and 201903 ($146,727.22) — were assessed on the same day, by the same officer, on the same theory, and were never charged. Total $165,262.78 of assessed trust-fund liability the grand jury never saw as a count.

This is not a legal defect and it should never be pleaded as one. It is worth three things: selection evidence for the particulars motion, because the government made count-level choices inside a six-module page and can therefore state the criteria it applied; cross of the revenue agent — the government's noticed expert on Counts 21-27 is IRS Revenue Agent Hang L. Tran, and the six-module page frames every question about how the charged figures were derived and why two modules dropped out; and loss, because a charged set chosen for reasons the government will not articulate carries that problem into the guidelines figure.

Module discipline. The MFT 55 civil-penalty module and the MFT 01 employer module are two different accounts holding the same trust-fund money. A $0.00 balance on the TFRP module does not mean the 941 module is zero, and the 8/2023 941-module balances ($398,798.03, $302,122.74, $366,602.12, $320,394.26, $140,548.59, $349,940.38) are real. Say "the parallel § 6672 module was collected to zero." Never say "the count is paid."

Thayer means a failure complete at the due date is not undone by later collection. That is the honest ceiling. Under it there is still room. Loss: Counts 21 and 22 close at $0.00 on the certified civil-penalty modules more than a year before Doc 1 — the government's own collection function treated that money as recovered, and every filing describing it as unpaid meets a transcript line rather than an argument. Willfulness: payment machinery existing across the charged period — escrow balances of $359,768.47 (12/13/2021), a completed wire of $213,493.47 (1/4/2022), a $431,603.52 balance with $125,132.60 earmarked (5/25/2022) — is Cheek evidence, not payment evidence.

Do not drift into inability to pay. United States v. Easterday, 564 F.3d 1004 (9th Cir. 2009) holds inability is not a defense; it is persuasive only here, but the drift converts a good-faith story into an admission. The FD-302 inability record ($35M borrowed / $39M repaid / $35M still owed) is Cheek context and sentencing material — the reason a good-faith belief was held, never "he could not pay."

The honest statement of the argument: the government charged the gross trust-fund figure while filed credits sat unprocessed. That is a genuine loss-methodology and materiality point and a genuine willfulness point. It is not a dismissal ground, and it does not become proof until three documents arrive — the six native 941-X filings, post-processing transcripts showing TC766/570/810, and the native Smith letter with its schedule.

Never assert "the tax counts net to zero." Identifiable non-reversed payments and credits across the six Bene charged quarters total roughly $150,892.71 against $1,068,680.47 charged on the 8/2023 pull. The government will subtract on cross and the witness who signed the netting letter will be the one holding the subtraction.

Kwong / § 7508A(d) — scoped honestly. Kwong v. United States, 179 Fed. Cl. 382 (2025) (No. 23-267, Ct. Fed. Cl., 11/25/2025, on appeal) invalidated the regulatory one-year cap on the mandatory postponement period, 1/20/2020–7/10/2023. Five of seven counts (23, 24, 25, 26, 27) have an offense date inside that window. But 26 C.F.R. § 301.7508A-1(c)(1) and Rev. Proc. 2018-58 exclude employment-tax deposits unless specifically listed, and Kwong is a duration case, not a scope case — it did not expand which acts are covered. REG-115057-20 gives the government its answer in advance. So: Kwong is corroboration for a state of mind and the engine of the civil refund track (11 Form 843 protective claims filed 7/10/2026, [EX-B7]). It is not a merits defeat of the pay-over element and must never be pleaded as one.

Four government-authored sources address who ran Seguro Medico, and they name three different people. None is Redmond.

Add ¶7(a): "Until the second quarter of 2022, all Seguro employees were paid by checks drawn on bank accounts held by Bene Markets LLC and not Seguro." The charged quarter is the transition quarter, on their oath. Add ¶77: Kroemmelbein is not on the Seguro payroll either — the payroll-control theory has nobody on the payroll. Add the signature rail: the charged 941 carries "PIN Entered By: REPORTING AGENT / Jurat: REPORTING AGENT PIN", third-party designee C. Malcolm Smith III CPA, EFIN 231990, filing IP 173.163.117.129, and the "Print your name here" field is empty [EX-B4], while the government's own witness proffers that Redmond "signed the 941s" [EX-B9]. Zero Forms 8655, 2848, or signature records appear in 8,835 produced pages.

How to run it. This is a Rule 29 sufficiency argument about the government's proof, presented through the government's own documents — not an affirmative "wrong man" defense and not a suggestion that Kroemmelbein or Miller is the responsible person. That distinction is the whole ballgame. It keeps faith with the client instruction of 8/11/2026 retiring wrong-responsible-person as a merits theory, it keeps the § 727 and Scalia forums clean, and it is the only version of the argument that survives cross. Prior probability work put P(Rule 29 acquittal on Count 27) at 45% [90% CI 30-62%] on a reference class of § 7202 / § 6672 prosecutions where no government source names the defendant as owner, officer, or payroll-listed and the charged return was filed under a third-party reporting-agent PIN. Treat that as a planning number, not a promise.

The government's bridge is bank signature authority on the Seguro accounts plus cooperator-sourced control testimony (W11 ¶44: "Redmond controls the business through Barrera and Walsh"). Signature authority is a factor, not the duty; it must be paired with control over which creditors were paid in the charged quarter, which ¶7(a) itself unsettles. The control testimony comes from two co-defendants who have pled — Giglio, and their sentencings (10/14, 11/5, 11/6/2026) all precede trial.

Smith is the government's witness and the defense's best cross. Both are true and neither cancels the other.

Adverse, owned in advance: MOI-071423 item 29 ("REDMOND set the financial policy… sole signatory… signed payroll Forms 940 and 941… was solely responsible for paying payroll taxes over to the IRS… REDMOND was in full control"); item 28 ("Smith repeatedly told REDMOND to pay his taxes… never advised REDMOND to delay any payments"); item 34 ("REDMOND listens to Smith, but eventually REDMOND just does what he wants"); item 24 (Redmond "had to pay his help first"). Also adverse: November 2021, roughly $1M in to the Fulton Bank escrow from Bene, roughly $550K back out to "Keystone Commercial" for real estate the same week.

Favorable, from the same file: item 14 (Redmond told Smith to work out a payment plan with RO Colwell and that he would pay it); item 13 (Colwell took Smith down a "primrose path"); item 22 (Smith directed the wire to his own escrow because he did not trust the accounts to stay funded — an intent-to-pay fact in the government's hand). Neither MOI contains any ERC content, so the government's willfulness theory was built without it.

The cross hook. Smith's proffer says Redmond signed the 941s. The charged Seguro 941 has no signature [EX-B4 vs EX-B9]. That is a documentary contradiction of the government's own witness, and ECF 147 Attachment A item 53 shows the government authenticating "Records from C. Malcolm Smith & Co., P.C." at its own expense — they are bringing the file that contains the contradiction.

The zero-balance letter [EX-B8], stated correctly. Read in full, the 27 Jan 2025 letter reconciles: $364,169.39 current balance → apply $130,782.13 of documented misapplied trust-fund payments and remove penalty and interest → $221,929.41 → apply $240,016.75 of unapplied ERC → overpaid by $18,087.34. The letter also states that the first two counts are paid, that the last two "were NEVER ASSESSED against Alan," and that the balance of Counts 23 and 24 is $36,169.39. Two cautions. An earlier pass through a truncated screenshot concluded the letter's arithmetic defeated itself; that reading was wrong and was corrected — but it is what happens when a transcription substitutes for a native, and the government will run the same subtraction. And the letter's "never assessed" line as to Counts 25 and 26 is retired (see the box below); it does not enter a filing merely because a CPA wrote it.

Sequencing: Cheek good faith first, loss methodology second, reliance never as the lead. The 2/9/2024 e-mail forwarded to William McSwain — "they have never applied ERC tax credits during Covid times… we have made every effort to get caught up" — is contemporaneous documented good-faith belief eight months before the original indictment, with two attachments evidencing attempts to pay. Pull the attachments.

Full sequencing, with the bankruptcy and cross-forum screens, is Chapter 8.

DO NOT FILE — RETIRED THEORIES
These were tested and lost. Each one, if signed, hands the government a free correction.

1. "Count 25 has 0 hits in any certified transcript." RETIRED. The figure derives
   exactly from the filed 941: TF 180,840.78 minus refundable ERC 31,542.61 [EX-B3].
   The government already applied ERC on that count.
2. "Count 26 was assessed 2/21/2025, after the original indictment." RETIRED.
   TC150 $300,334.52 posted 02-21-2022, cycle 202205 [EX-B6 p.4]. The date argument
   is dead; Fidalgo-scope and IIJA willfulness survive.
3. "It is a 6672 penalty, not a 7202 tax, so the count fails." RETIRED — researched
   and closed. The statutes travel together.
4. "He was not the responsible person" as a MERITS theory on Counts 21-26. RETIRED
   per client instruction 8/11/2026. Signature provenance and INOLES stay in, for
   impeachment and scope only. On Count 27 the argument runs as Rule 29 sufficiency
   built from the government's own four sources — not as an affirmative wrong-man
   defense, and never as "Kroemmelbein or Miller was responsible."
5. Citing THAYER for "no amount element." RETIRED. Not locatable in the opinion.
   Thayer is adverse and controlling; use CTM 9.03 / Citron for that proposition.
6. "Count 27's figure matches no ledger." RETIRED per the controlling corrections
   ledger — the derivation matrix reconciles Count 27 exactly, to the penny. Any
   Count 27 particulars request must ask the government to state ITS derivation,
   and must not assert untraceability. See CONTRADICTION C-1 in Chapter 8.
7. "The tax counts net to zero" / "ERC exceeds the indictment" / a documented TC810
   refund freeze / "$1,576,093.06 in ERC was filed." RETIRED or unsourced.
   ~$150,892.71 identifiable against $1,068,680.47. Only $31,542.61 is IRS-labeled.
   No TC810 appears anywhere in the corpus.
8. "He never signed a 941." RETIRED. He signed several, including Count 21's
   (RedmondTax006791, title "Member"). The blank-signature wedge is Count 27,
   with supporting pattern at Count 25 and one uncharged quarter.
9. Kwong / 7508A(d) as a MERITS defeat of the pay-over element. RETIRED —
   employment-tax deposits are outside the default postponement. Willfulness and
   the civil refund track only.
10. Vindictive-prosecution PRESUMPTION on the superseder. FORECLOSED for pretrial
    charge increases by United States v. Goodwin, 457 U.S. 368. P(dismissal) 3-5%.
    The Smith-letter-to-superseder timeline stays as narrative and pressure only.
11. Inability to pay. Never plead it. Easterday drift converts good faith into an
    admission.
12. Count 25 three-year limitations "lapsed 2/22/2024, unrevivable." Not filed
    without a limitations memo. Original indictment 10/17/2024; superseder 5/8/2025.

1. Post-ERC-processing certified transcripts for all seven charged quarters — P0, and the single highest-value gap. Four of the six 941-X originals already sit in the government's own production.
2. The six native Bene 941-X filings — P0.
3. Native Smith letter of 27 Jan 2025 plus its schedule (ShareFile link expired) — P0.
4. Seguro Form 8655 and the Q2-2022 e-file signature record, or the government's written admission that none exists — P0, dispositive on Count 27.
5. Seguro bank signature cards — scope and dates of authority — P0.
6. Confirmation that employment-tax payment obligations fall within § 7508A(d) covered acts — P0, load-bearing, answer before arguing it anywhere.
7. Executed Seguro operating agreement; image crop of the 2022 Return Tree jurat page.
8. The $112,246.57 misapplication analysis; the government's loss computation.
9. Bates manifest / production index — its absence is itself a Rule 16 lever.
10. Barrera plea terms and all three sealed supplements (ECF 97, 130, 142).

Provenance warning that governs this whole chapter. Much of the underlying workbook is an agent's reading of OCR'd transcripts rather than a transcript read directly. Figures tagged derived or unsourced do not enter a filing, a pitch, or a conversation with the government as fact. The MFT 55 table, the Count 25 941 derivation, and the certified civil-penalty pages are the tier that can carry weight; build the motions on those and demand the rest.

[EX-A] Superseding Indictment, Doc 16 (¶¶ 6–9 roles, ¶¶ 14–36, Overt Acts 1–72, ¶ 72) · [EX-B] Original Indictment, Doc 1 (10/17/2024) · [EX-C] charge book (CODEFENDANTS, COUNT_MATRIX, WIRING_ZERO) · [EX-D] element kill wiki V1 · [EX-E] corrections ledger, controlling in any conflict · [EX-F] superseding analysis workbook (GROFF PLEA ANALYSIS, CRITICAL FINDINGS, ORIGINAL vs SUPERSEDING, PROBABILITY DASHBOARD, 72 OVERT ACTS MAP) · [EX-G] Groff plea Doc 96-1 / memorandum ECF 96 · [EX-H] Walsh plea ECF 126 / memorandum ECF 127 · [EX-K] Simmons affidavit, 22-MJ-1897 · [EX-L] Redmond deposition 9/1/2022, Bates RedmondTax004490–004736 · [EX-M] ECF 147 / ECF 148 and Attachment A.

S.K. appears in the instrument as an uncharged spouse nominee (¶ 6, ¶ 15(p)) [EX-A]. She is not a defendant. Do not volunteer the nominee theory; it is the government's to prove.

Redmond alone faces the seven § 7202 counts (21–27). Those counts are the only ones on which he is charged individually, and they are the severance lever discussed in § 4 below [EX-C].

Two co-defendants are entities Redmond owned or controlled. That single structural fact defeats any theory that no agreement existed and means that at trial Redmond sits next to two defendants whose acts are, functionally, his acts in the jury's eyes [EX-D] § 1.

Groff — Counts 1, 10, 13, 14, 17, 18. The verbatim operative language of Doc 96-1 ¶ 1 is that "the defendant agrees to plead guilty to Counts 1, 10, 13, 14, 17, and 18" [EX-G]. Those are the conspiracy plus five charged customer calls, all of them calls Groff personally made. The plea was signed on the government side by First Assistant Derek E. Hines, with Metcalf, Murray, Crawley and Dalke among the signatories [EX-F] — office-level attention, not a line-AUSA matter.

Walsh — Counts 1, 6, 7, 15, 16, 19. Entered 6/5/2026 [EX-H]. This is the more dangerous plea for Redmond, and the reason is composition rather than volume: Walsh pled guilty to three email counts (6, 7, 19) and two calls he personally made (15, 16), plus the conspiracy. Count 7 is an email Redmond sent. Count 19 conveys an instruction attributed to Redmond. By pleading to those counts Walsh has admitted that those specific wires were part of a fraud — and he is positioned to authenticate and interpret Redmond's own emails from the witness stand [EX-C].

Barrera — plea entered, terms unverified. The docket shows a change of plea on 6/26/2026 at ECF 138–140, with sentencing 10/14/2026 and a sealed supplement at ECF 142. No plea memorandum has been located and the counts he pled to are unverified [EX-E] § 7, [EX-C]. That gap matters concretely: Barrera is the charged speaker on Count 12, the author of the Count 8 text-campaign email, the sender on Count 20, the actor alleged to have altered a verification recording (Overt Act 59), and the person alleged to have testified falsely about his relationship with Redmond (Overt Act 68) [EX-A], [EX-F]. Until ECF 138–142 are read, every statement about what Barrera has admitted is an assumption.

A date discrepancy to fix on the file. One earlier work product records Groff's plea as 12/16/2025 by minute entry, and the signed agreement as filed 12/17/2025; another records "Groff sentenced 4/23/2026," which the later docket contradicts (sentencing 11/6/2026) [EX-D] § 5, [EX-C]. Use the filing date of Doc 96-1 and the current sentencing date.

They establish the conspiracy existed. Two co-defendants have admitted Count 1. No filing should suggest that a wire-fraud conspiracy did not exist in this operation [EX-D] § 1.

They do not establish Redmond's membership. A co-defendant's guilty plea is not substantive evidence against a non-pleading defendant, and Rule 801(d)(2)(E) still requires an independent predicate finding of an agreement that included Redmond before any co-conspirator statement comes in.

Three corrections control the cross and each of them cuts against the assumptions carried in earlier work product:

First — no public plea admits an agreement with Redmond. Groff's ¶ 1 locates his own role as "a salesman and manager" of the Bene Market Group and names no co-conspirator. No public factual proffer describes a Redmond directive [EX-E] § 7. That is the single best cross-examination fact in this chapter: the government's admitted conspirator, in his own signed document, did not say Redmond told him to do anything.

Second — cooperation is unverified. There is no public § 5K1.1 or substantial-assistance clause, no promise to testify, in any public plea. Paragraph 17 recites only that "the parties have also entered a supplemental agreement that is filed under seal." The sealed supplements are ECF 97 (Groff), ECF 130 (Walsh), ECF 142 (Barrera). Until they are produced, "Groff, Walsh, and Barrera are cooperating" is an inference [EX-E] § 7. One earlier assessment assigns 90% confidence to cooperation terms in Groff's supplement [EX-F] GROFF PLEA ANALYSIS; treat that as a probability estimate, not a fact, and file the Brady and Giglio demand rather than the assumption.

Third — the timing argument has inverted. All three sentencings — 10/14/2026, 11/5/2026, 11/6/2026 — now fall before the 1/12/2027 trial. The classic "you are still hoping for a reduction" cross is dead. The replacement is Rule 35(b) exposure, which survives sentencing [EX-E] § 7. Any artifact describing cooperator leverage as decaying at those dates was written against the pre-continuance calendar [EX-C].

Groff's loss stipulation is his alone. Doc 96-1 ¶ 11(b) stipulates a band "between $25,000,000 and $150,000,000," with only the minimum two-level victim and mass-marketing bump under USSG § 2B1.1(b)(2) rather than the six-level tier [EX-E] § 7. Restitution exposure is described in the plea materials as up to $150,000,000, and forfeiture proceeds are described as "not currently available" under 21 U.S.C. § 853(p) [EX-F]. Two points follow. An employee stipulating to a $125-million band is a cross-examination asset on the government's loss theory. And that band must not be permitted to attach to Redmond by implication at trial or at any later stage.

The structural problem. Redmond will be tried with two entity co-defendants he owned or controlled, on a twenty-seven-count instrument in which nineteen wire counts are largely other people's words and seven tax counts are his alone. The government's proof will include: five calls by a man who has pled guilty; two calls by another man who has pled guilty; an allegation that a third man altered a verification recording (Overt Act 59); an allegation that the same man testified falsely about his relationship with Redmond (Overt Act 68); Groff's regulatory history (Overt Act 8); rosters showing most closers unlicensed (Overt Acts 25, 61); and Redmond's own refund-conditioning instructions (Overt Acts 42, 44, 54, 66) [EX-A], [EX-F].

The severance lever is Counts 21–27, not the wire counts. The tax counts charge only Redmond, cover different conduct, different periods, different documents and different witnesses (IRS Revenue Agent Hang L. Tran is the noticed expert for 21–27), and carry no relationship to the sales floor. A prior estimate put a severance motion at 30–40% [EX-F] PROBABILITY DASHBOARD — recorded here as a prior estimate, not adopted. The argument is real: joinder of an employment-tax case to a telemarketing-fraud conspiracy invites the jury to reason from unpaid trust-fund taxes to dishonesty in sales. Note also the charging history: Doc 1 (10/17/2024) charged six § 7202 counts and nothing else; the superseder added five defendants, twenty-one counts, three statutes, seventy-two overt acts, the § 2326 enhancement, and $30 million-plus in forfeiture [EX-B], [EX-F] ORIGINAL vs SUPERSEDING. The tax case is the original case; the fraud case was grafted on.

What the severance motion cannot rest on. The vindictive-prosecution presumption is foreclosed for pretrial charge increases by United States v. Goodwin, 457 U.S. 368, and the probability of dismissal on that theory has been revised to 3–5% [EX-E] § 4. The charging timeline remains narrative — useful for a Rule 14 prejudice argument, not a presumption.

The 404(b) and intrinsic-evidence front is where spillover will actually be decided. The government reserved in writing the right to file a more comprehensive motion in limine on three categories: regulatory history and license revocations, the 6/10/2015 Redmond sales-training tape, and Redmond's pre-2019 taxpaying record (7/17/2026 notice, § II). It also reserved the right to supplement the 38-recording list. The old 8/31/2026 deadline is now 12/7/2026 [EX-E] § 9. The 38-recording list is a floor. Build the late-supplementation objection now, and treat the 2015 training tape as the single most dangerous discretionary exhibit in the case — if it comes in, the "Redmond is on zero of the twelve charged calls" argument acquires a competitor recording of Redmond training a sales force [EX-C] WIRING_ZERO, [EX-M].

One collateral-proceeding screen. The Simmons affidavit at ¶¶ 59–60 records that the Five9 account was paid from two Truist accounts, one Bene Markets LLC and one ARC Realty LLC — which links the criminal case to the foreclosure matter. Screen anything filed in Berks before it is served [EX-E] § 8, [EX-K].

Theme 1 — the signed document names no one. Groff's plea describes his own conduct and identifies no co-conspirator; no public proffer describes a Redmond instruction [EX-E] § 7, [EX-G]. Establish on cross what the document says before the witness explains what he now remembers.

Theme 2 — the fabrications were his own. The workbook maps Groff's cornea-transplant claim escalating $50,000 → $80,000 → $90,000 → $120,000 across different calls (Overt Acts 52, 53, 56, 60, 63) [EX-F] CRITICAL FINDINGS. No company script instructed an agent to invent a personal medical anecdote and inflate the figure with each retelling. A prior assessment discounts Groff's credibility to 40–50% on that ground [EX-F] — again, a prior estimate, not a finding. The government's answer is in its own instrument: Overt Act 30 and Count 6 allege Groff's calls were pulled and used to train new agents [EX-A]. Demand the training set.

Theme 3 — Rule 35(b), not pre-sentence hope. Because all three sentencings precede trial, the incentive to shade testimony runs through Rule 35(b) [EX-E] § 7. That is a narrower cross but a cleaner one: the witness has already been sentenced and can still be rewarded.

Theme 4 — the other men's misconduct is their own. Overt Act 59 alleges Barrera materially edited and altered a verification recording in support of a Groff response to state regulators. Overt Act 68 alleges Barrera testified falsely on 5/31/2022 about having no relationship with Redmond [EX-A], [EX-F]. Both are acts of concealment by someone else. The government will argue direction; the defense answer is that no document in the present record shows Redmond copied on or directing the alteration, and that should be confirmed in discovery [EX-F] CRITICAL FINDINGS ¶ 6.

Theme 5 — the government's own affiant gave the alternative. The 22-MJ-1897 affidavit at ¶ 79 frames the entity history as serial rebranding in which "any issues were the responsibility of a single, bad employee" who "is then fired," naming W3 and Quanaijha Spradley [EX-E] § 8, [EX-K]. That is the rogue-employee theory in the government's handwriting, and it is the frame in which every cooperator's misconduct should be examined.

Theme 6 — the concession that removes the premise. "During the charged period, neither the ACA nor state insurance laws made it illegal to offer for sale any of these Limited Benefit Plans" — ECF 96 at 6 and ECF 127 at 6, verbatim in both plea memoranda [EX-E] § 5. Each cooperator pled guilty in a proceeding in which the government wrote that sentence. Put it in front of each of them.

Theme 7 — the disclosure calendar. Giglio and Jencks material is withheld until "at least fourteen days before trial" on United States v. Higgs, 713 F.2d 39, 44–45 (3d Cir. 1983) (ECF 156 at 16–18), and the government asserts that "in no event is the government required to divulge the identity of its witnesses in a noncapital case" (ECF 156 at 17, quoting Addonizio, 451 F.2d at 62) [EX-E] § 7. Every cooperator cross must be built to survive a fourteen-day disclosure, and the Brady and Giglio demand on the three sealed supplements should be filed and pressed well before the first sentencing on 10/14/2026.

Instruction hook. The Third Circuit cooperating-witness instruction is 4.19; single-versus-multiple conspiracies is 6.18.371H; Pinkerton is 7.03; good faith is 5.07 and is available on the fraud counts. There is no Model Instruction 6.18.1349 [EX-E] § 4.

Bottom line. The co-defendant structure is the reason Count 1 is not overcome and the reason the twelve calls remain in the case. It is also the source of the three best facts available: a signed plea that names no co-conspirator, a serial fabricator whose figures escalated call to call, and a written government concession that the product was lawful to sell. None of that is an acquittal argument. All of it is usable at the 12/7/2026 instruction and in-limine wave, and in cross.

1. Barrera's plea counts — unverified; no memorandum located (ECF 138–140).
2. Sealed supplements ECF 97, 130, 142 — unread. Cooperation, its scope, and any testimonial obligation are unknown.
3. Walsh ECF 128 — unread.
4. Whether any cooperator has proffered anything about Redmond — unknown on the public record; no proffer notes produced.
5. Overt Act 59 — the original and altered verification recordings have not been produced or compared; whether Redmond was copied on the alteration is unverified.
6. The 6/10/2015 sales-training tape — not reviewed; its admissibility is the largest single discretionary risk to the wiring-zero argument.
7. Prior-session probability estimates in [EX-F] (severance 30–40%, Groff credibility 40–50%, cooperation 90%, per-count acquittal bands) are retained for continuity and are not adopted as findings.
8. Government experts noticed: JoAnn Volk (framework), Hang L. Tran (Counts 21–27), Ana Virella Cruz (FBI senior forensic accountant, Rule 1006 charts) — the last noticed "in an abundance of caution" and only "should it become necessary to qualify Ms. Cruz as an expert," while the government simultaneously argues her testimony "does not constitute expert testimony" [EX-E] § 5. That inconsistency is a Rule 702 and Rule 16 motion.

Two facts control the chapter. A million-file, 152.7 GB pile of native call audio was produced unhashed and unstamped, and the inventory the government pointed to describes two of those files. There is no index that maps the pile to the case.

Call dates run 2019-02-06 → 2022-02-22. Timeline duration sums to 495,664 s (137.684 h), talk to 285,098 s (79.194 h); median call 42 s, p90 1,163.2 s. "Produced" is identical to "Bates contains AR-": 412 / 412, zero exceptions. Those 412 rows carry 631 distinct AR- tokens, all 631 resolve into the native pile, and they account for 2,701 files — 625.4 MB of 152.7 GB, or 0.41% of the produced audio.

The withheld set is not hidden. The government typed NOT PRODUCED into the Bates column 945 times.

Disposition tells the rest:

They produced five answering machines and left 33 of the 91 Sold calls dark. Of the dark Solds, 38 stamps carry the literal NOT PRODUCED and zero recording keys. Produced Sold audio runs 29.715 h of duration and 17.640 h of talk; unproduced Sold runs 14.985 h and 7.730 h. The sales calls are the case, and roughly a third of them are not in the file.

Each recording stamp resolves to three to five native files whose second filename token is a two-digit channel code, 01 through 05 (average 4.28 files per stamp). Only 282 of the 631 produced stamps (44.69%) carry all five codes. Code 02 is the only complete set.

Channel 01 is the talk channel, and it is the one that is gone.

This is a dropped channel, not 213,000 missing calls — 83.389% of the estate stamps missing code 01 still have other codes, and 86.86% of produced ones do. That distinction matters, because it is what makes the argument credible: the defense is not claiming the government lost a million calls. It is showing that one channel, the channel carrying the conversation, is systematically absent from the produced set, and that on the sales calls the government intends to play, it is absent 100% of the time.

Then the second layer. Of the 38 produced code-01 files that are in the storage catalog, only 11 return fetchable RIFF audio (8000 Hz mono, 0.863 h total). Twenty-seven are catalog ghosts — an entry exists, every copy 404s. And for channels 02 through 05, the count of produced files with a fetchable copy is 0 of 1,073. A catalog row is not a byte. The government produced a manifest of a million filenames; what actually opens is a rounding error against it.

Count 3 is the only count with a completeness row, and it is charged as a cancellation call on 2020-12-30, with redmond_on_call = false.

The charged-date native pile is 16 stamps × 5 codes = 80 files. The catalog holds 72. Fetchable audio is 8 — all channel 01, all from one bucket. Channels 02 through 05 that day: 0 of 64. Three of the five produced rows carry a dropped-01 stamp and hold 1,598 s of the 3,466 s of produced talk that day (46.1%).

That is the Rule 106 completion argument in one count: a cancellation call offered without the sale, the verification, the welcome, the e-sign, or the service history that preceded it. FRE 106 was amended effective 12/1/2023 so that a completing statement is admissible over a hearsay objection — the single most useful 2026-vintage rule in this file, and it post-dates most of the corpus this case was built on.

Grain discipline on Count 3, because the government will exploit any slip: stored CDR 31 ≠ working hits 12 ≠ sessions 8 ≠ recording keys 5 ≠ produced audio 2 ≠ proof rows 13 / 9. Produced talk that day is 0.963 h on timelines, 0.786 h on the working table, 1.013 h on the proof. Cite the grain with the number, every time.

The corrections ledger records a different Five9 cut: 1,100,000 total recordings; 853,000 filtered as failed or sub-10-second (77.6%); 247,000 substantive; 23 government-victim calls, offered as the FRE 1006 summary that must reach the jury. The database census in §§6.1–6.3 is a different measurement of a different object (native WAV files, five per call, versus recordings).

Both may be right. They are not interchangeable, and a filing that mixes them will be corrected in open court. Pick one denominator per exhibit, state the grain on the face of the exhibit, and keep the 1006 chart and the completeness chart as separate demonstratives.

From the same ledger, the number that survives every grain fight: on ECF 148 Attachment A's 38 recordings (7 internal + 31 Five9 victim calls), Redmond is the primary subject of exactly three, not one of which is a charged wire, and he is on zero of the twelve charged victim calls.

The produced metadata is thin in a way that is itself discoverable. Five9 object metadata carries HTTP properties only — size, etag, mimetype — with no duration and no RIFF header. Of 800,600 Five9 objects, 800,068 carry user metadata whose only keys are mtime and, on 38 objects, md5chksum; those 38 are not even WAV files. There is no call-audio transcript table anywhere in the corpus — the only transcript tables are IRS tax transcripts and an empty one.

So the following have no counterpart in the production and must be demanded by name:

• The Five9 recording and retention configuration in force 2018–2022: which channels were recorded, which were retained, retention periods by disposition, and every configuration change during the charged window. Nothing in the produced set answers this, and it is the only document that will explain whether channel 01 was never collected, was collected and dropped in processing, or aged out. [UNVERIFIED whether previously demanded — treat as a new demand]
• The Five9 legal-hold and export logs for the government's own extraction, including the export parameters and any filtering applied before production.
• The process by which Five9 records were obtained. The exact warrant or process number for the Five9 material is not in ECF 147 or 148. 2:22-mj-01897 is a parallel search matter and has not been established as the Five9 process. Do not assert that it is.
• The warrant return for 22-MJ-1897, which is absent from the production. The affidavit itself names zero Title 26 offenses across 85 paragraphs, the word "tax" appears zero times, and Attachment D limits seizure to fraud material — while ¶71 asserts that "more than 90%" of 13,000-plus reviewed calls contained material misstatements, with no methodology, sampling protocol, or definition of materiality disclosed.

The seventh production, delivered 7/24/2026 via Box/USAfx, has not been downloaded, indexed, or reconciled against AUSA Dalke's cover letter. The 8.2 TB transfer and physical disc handover to counsel is likewise open. Until PROD7 is indexed against the cover letter, no completeness representation to the Court is safe, and no Rule 16 motion should assert that a category is absent — the absence may be sitting in an un-mounted production.

This is the first discovery task, ahead of every motion in this chapter. It is also the answer the government will give to any motion filed before it is done.

For scale, the other productions already indexed: DALKE_PROD1_DISCO1 at 846,720 files / 412.222 GB with 845,459 Bates values and zero hashes, and PROD02_RedmondTax000836-693308 at 847,427 files / 409 GB, essentially a tax image corpus (509,695 .jpg, 175,636 .tif, 149,776 .txt, four stray WAVs). Neither is hashed. There is no Bates manifest or production index in the file, and that absence is itself a Rule 16 lever.

Every certified IRS transcript in the production is an August 2023 pull — that is, it predates the $500,000 payment of 8/8/2023 and predates all ERC processing. The consequence is that the government's charged figures, its loss computation, and its "unpaid" rhetoric all rest on a ledger snapshot taken before the largest payment in the case posted.

The post-8/8/2023 activity presently known to the defense is recited inside Form 843 Line 8 statements prepared by a tax professional, not read off a certified transcript in the production. That is a recitation, not evidence, and it cannot be put in front of a jury in that form.

Demand current-date certified transcripts: - MFT 01, EIN 81-2680613 (Bene Market) — all charged quarters - MFT 30, SSN …5092 (individual) - MFT 55, SSN …5092 (trust-fund recovery penalty)

[CONTRADICTION — see Chapter 8, C-2] The identifiable 8/8/2023 postings visible in the working ledger (−242,772.61, −21,611.53, −58,055.38, −232.66, −2,036.66) sum to roughly $324,708.84, not $500,000. The demand is unaffected — the point is that no certified transcript in the production shows any of it — but the $500,000 figure must be reconciled before it appears in a filing or a proffer.

The Count 3 completeness row lists ivr_verification and esign_live among the missing call types, and the disposition census shows IVR Completed 2 rows, 0 produced; Incomplete Verification 1 row, 0 produced; echosign 7 rows, 1 produced. Meanwhile the indictment itself, at ¶59, alleges that a co-defendant materially edited a recording of the sales verification portion, and Overt Acts 58 and 63 describe coaching consumers through the verification. The government cannot allege tampering with a verification process and simultaneously treat the verification records as immaterial.

This is two-edged and must be handled as such. At least one recovered Five9 recording (109657236) captures an agent pre-coaching the IVR — "let them complete that whole sentence and then say I agree and press 1" — which the government will use to destroy the verification-as-cure theory. There is also a corrected fact in the file: the $99 enrollment fee was disclosed in the IVR.

So the posture is: demand the IVR and e-sign records as Brady and as Rule 16 material within the government's possession, and litigate their contents later. Do not brief "the IVR cures the sale." Demand it, get it, then decide. The disclosure record is either exculpatory or it is not, and the government has taken the position that it exists and that someone edited it.

The Bates-log census of the tax production — 559 documents, 8,835 pages — contains zero Forms 4183, zero federal-tax-deposit notices, and zero Forms 8655. Those three absences are the Rule 16 demand that pairs with Chapter 5: the 4183s are the trust-fund recommendation documents behind the charged figures, the FTD notices are the government's own notice evidence on willfulness, and the 8655s are dispositive on who authorized the reporting agent to file the Count 27 return.

Related: the phone production (PROD06) holds 5,363 items, but roughly 63 government-captioned fragments correspond to 118 native threads — a Rule 106 completeness and filename-strike problem, since the captions are the prosecutor's characterizations, not the documents' names.

902(11) is not ripe. ECF 147 ¶5 n.1 concedes that roughly ten certifications were "forthcoming." A Rule 104 self-authentication determination cannot issue for records whose certificates do not exist. ECF 147 offers 56 categories of third-party business records described as "well over a million records," so the ripeness objection is not academic.

The 38-recording list is a floor, not a ceiling. The 7/17/2026 notice reserved the right to supplement it and to file "a more comprehensive motion in limine" on three 404(b) or intrinsic categories: regulatory history and license revocations, the 6/10/2015 sales-training recording, and Redmond's pre-2019 taxpaying record. The operative deadline for that wave is 12/7/2026. Build the late-supplementation objection now, with a cutoff date proposed in writing.

Giglio and Jencks are being held. ECF 156 at 16-18 commits to production no earlier than "at least fourteen days before trial," relying on United States v. Higgs, 713 F.2d 39, 44-45 (3d Cir. 1983), and refuses a witness list outright ("in no event is the government required to divulge the identity of its witnesses in a noncapital case," quoting Addonizio, 451 F.2d at 62). With all three co-defendant sentencings now falling before trial, the "still hoping for a reduction" cross is gone and Rule 35(b) exposure is the replacement — which makes early Giglio more valuable, not less.

Expert-to-summary-witness swap. FBI Senior Forensic Accountant Ana Virella Cruz is noticed "in an abundance of caution" and only "should it become necessary to qualify Ms. Cruz as an expert," while the government simultaneously argues her testimony "does not constitute expert testimony." Pin them to one position before the Rule 1006 charts are built. The other two noticed experts are JoAnn Volk (Georgetown CHIR, framework) and IRS Revenue Agent Hang L. Tran (Counts 21-27).

A Rule 17(c) carrier-records demand to Xfinity Mobile and Verizon Wireless for call-detail and SMS records was worked up pro se in May 2026 for the period 1/1/2026–5/20/2026 and never completed by counsel. The account structure — reported as four lines — is [UNVERIFIED]; no produced document in this pack states a line count. Confirm the account and the number of lines with the client before any subpoena issues, and set the date range to the charged period rather than 2026.

• A catalog row is not a fetchable byte. Produced code-01 catalog 38 ≠ HTTP 11. Count 3 catalog 72 ≠ HTTP 8.
• Do not collapse Sold grains: proof 58 ≠ timelines 59 ≠ stamps 172.
• Do not treat 42.48% duration-minus-talk as dead air. It includes hold, IVR, and voicemail, and it is not a legal finding.
• Do not quote planner estimates as counts. wav_keys ~1,086,045 and DOMAIN_LEGAL.cdr ~78.3M are query-planner estimates, not produced-audio counts.
• Do not treat the 249,369 native stamps outside the victim log as withheld victim calls. They are estate stamps with no token in the victim tables.
• Do not call the two-row metadata table the recording inventory — but do put it in front of the Court as what the government offered when asked what it had.

Discharge cannot enter while AP 25-00254 is pending. FRBP 4004(c)(1)(B). That single rule organizes the civil side: AP 25-00254 is not one adversary among several, it is the whole civil case, and everything in the main bankruptcy is downstream of it. The criminal case is the liberty case. Everything else is collateral fire that can be aimed back at those two.

The project file contains two mutually exclusive schedules and both are still cited by live systems. Nothing in this pack should be briefed to until counsel confirms which controls.

Working set for this pack, pending confirmation: pretrial 11/16/2026, motions wave 12/7/2026, jury selection 1/11/2027, trial 1/12/2027. Also carried: a Cooper gate of 10/15/2026 for motions, attributed to ECF 167 ¶8 — that paragraph text is [UNVERIFIED], it is not in the master docket description, and the motion schedule in this pack depends on it.

Decision rule. Do not say "trial 9/14" — even under the old order, 9/14 was jury selection and 9/15 was trial. Do not say "trial 1/11" — the master text makes 1/11 selection and 1/12 trial. Say nothing until the four PDFs are in hand. Filing to the wrong schedule is a default-adjacent event in either direction.

AP 25-00254 — the § 727 trial, 9/16/2026

Six counts, all § 727, all untried. Discovery closed 5/5/2026 by Mayer's own pretrial order (ECF 5 ¶6(a)), which also says the trial "may be continued only in exceptional circumstances" (¶9). The trial order (ECF 27) sets 9/16 and requires witness lists and pre-marked exhibits, served and e-mailed to chambers, by 9/2.

Three live problems. First, 22 bank subpoenas served 8/20/2026, returnable 9/21/2026 at opposing counsel's office — 107 days after discovery closed and five days after trial. Second, those subpoenas seek, at Request 3, "all records produced by subpoena in United States v. Redmond, 5:24-CR-00376" — that is criminal counsel's file, sought in a civil case, and Request 2 reaches a non-party's accounts. Third, the served set carries two unredacted Social Security numbers across 42 occurrences in an OCR'd, circulated PDF; that is a Rule 9037 problem for whoever touches it next.

The asset already in hand is Doc 42 from AP-145: the same judge, the same parties, holding that the state-court decision does not establish fraudulent intent, malice, or fiduciary capacity. AP-254 is riding that same document. The in limine motion writes itself.

SBA AP 25-00119 — the opposition date is unresolved

The SBA hit its summary-judgment deadline on 8/7/2026 (ECF 54). The opposition is due 8/28/2026 under the local-rule computation (21 days from filing, LBR 7005-1(b)(2)) or 9/3/2026 21:00 under a court-ordered date attributed to an 8/17 order. Treat 8/28 as controlling until the scheduling order is verified. The exhibits already visible include a Bene Market P&L for Jan 2020–Dec 2021, a 10/1/2024 hearing transcript excerpt, and a DOL deposition excerpt — a cross-forum evidence dump assembled by one arm of the United States out of another arm's record.

Strategy is Rule 56(d) deferral plus a per-question Fifth Amendment screen. Do not brief the merits of who ran Bene Market, where loan proceeds went, or how payroll was funded. Those are Counts 21-27 and § 727(a)(2)/(a)(4) in civil clothing. Note that Rule 56(d) is weaker here because the debtor's own extension motion was already denied (ECF 47) — argue what the record lacks, not what more discovery might find.

Scalia / DOL — 9/2 or 9/3, in front of the trial judge

Liability is over. ECF 114 (3/19/2024) granted the Secretary partial summary judgment; Miller's motion to open or strike was denied (ECF 138) and is on appeal (USCA 26-2069). Only damages remain — before Judge Schmehl, the criminal trial judge, in the same building.

The date conflict is unresolved: matter_master publishes 9/2/2026 13:30; the compiled wiki publishes 9/3/2026 14:00. Confirm off the notice, not off either table.

Two things follow. An adverse willfulness finding here is 404(b) material aimed straight at the 12/7/2026 criminal motions wave — civil FLSA "willful" is not criminal Cheek willfulness, and that distinction has to be made on this record before it is made against us on that one. And Stephanie Miller's payroll-control testimony is dual-use: it goes to Count 27's duty element and to the operator question in AP-254. Harvest it without putting the client on a civil witness stand. Counsel's withdrawal motion (ECF 141, 7/24/2026) means the client may face this hearing unrepresented while criminal counsel is retained on the other case — decide who stands up, and say so out loud.

Bankruptcy — the exemption order and the 9/22 hearing

Doc 540 (3/17/2026) resolved the exemption objection. The order is treated in this pack as entered in all five numbered provisions; the apparent strikethrough on the imaged copy is a scan/OCR artifact, not a struck provision. [UNVERIFIED — the controlling corrections ledger in this pack does not address Doc 540; verify against the imaged order before relying on any provision.] The substance recorded elsewhere in the file is that the objection was sustained without prejudice as to 2005 Regency Drive, a 2021 Mercedes S580, and a 2017 Cadillac Escalade — "without prejudice" being the operative words, because it means the amendments are still available.

Two Rule 1009(a) amendments are unfiled: a vehicle exemption and a wristwatch exemption, each pleaded at the statutory limit, per Doc 540 ¶2 and ¶4. The watch is described in the working file as a Cartier; that identification is [UNVERIFIED] against any source document in this pack and must be confirmed from the schedules before it is written into an amendment. Leaving both unfiled through discharge is an unforced error — the objection was sustained without prejudice precisely so they could be refiled.

The main-case hearing on the trustee's no-distribution report and final account (ECF 554) is 9/22/2026 11:00 PENN4, moved from 8/25 by ECF 578. The trustee found no assets; the objection is the petitioning creditors trying to keep administration alive until they win AP-254. If AP-254 goes for the debtor, ECF 554 collapses with it.

Foreclosure, tax sale, and the forfeiture overlap

The mortgage action on 8 Morgan Drive (Berks 25-13446) has a "dismissal rung 8/1/2026" in one system and a live 9/11/2026 upset tax sale in another. Both can be true — dismissing a mortgage foreclosure does not stop a tax sale. The redemption cutoff is 9/4/2026 for $66,546.53 ($19,751.50 + $46,795.03) covering 8 Morgan Drive and 1198 Reading Boulevard. The automatic stay does not reach ARC Realty (BK ECF 271 denied extending it), so only the Chapter 7 trustee can put ARC into a petition that would stay the sale.

The criminal overlap is the reason this page exists. 8 Morgan Drive is the 22-MJ-1897 search premises (Seguro Medico). The forfeiture allegation in the superseding indictment reaches 8 Morgan, 2005 Regency, 1198 Reading Boulevard, and 2 High Road, and the criminal docket carries a lis pendens at ECF 23 against 2005 Regency Drive [docket entry confirmed via the public RECAP set; PDF not pulled]. So 2005 Regency is simultaneously a criminal forfeiture target, the subject of a sustained-without-prejudice exemption objection, and part of the family's residence picture — and the Five9 account itself was paid from two Truist accounts, one held by Bene Markets LLC and one by ARC Realty LLC, which is the wire that connects the call-center case to the real estate.

Keep the mics separate. Foreclosure counsel argues standing and the allonge gap; criminal counsel does not narrate a foreclosure loss, because "he lost the house" and "he concealed the house" are one small step apart in front of a jury.

The dormant three

AP 24-00145 is a win being held in reserve. Mayer granted summary judgment for the debtor on all counts on 7/31/2025 (Doc 42/43); the case is stayed pending AP-254 (ECF 67, 7/1/2026), its 8/6 and 9/16 dates were cancelled, and Jordan filed a Third Circuit notice on 6/4/2026. It wakes up only if AP-254 is lost — at which point a district-court remand revives a case the bankruptcy judge already decided. Nothing needs to be filed there now; Doc 42 needs to be used in AP-254.

Berks C.A. 22-11757 is the state action Ready removed into AP 24-00144 and Mayer sent home (remand order ECF 65, 2/11/2025; the adversary closed statistically 3/6/2025). The state case is not terminated by that closure, and its post-remand posture is [UNVERIFIED] — there is no docket feed. The underlying state decision, which uses "misappropriation" seventeen times, is the recycled paper Doc 42 already discounted. Do not collapse the Sprecher and Rowley phases without the decision's caption in hand.

Custody (Montgomery 2016-47715) is a thin file with a live docket-number conflict, and support (Berks 25DR00324, PACSES 574115910) carries a pending recusal motion and an 8/28 contempt listing. Neither drives strategy. Both generate sworn financial statements, which is why they appear on this map at all.

What each forum can hand the prosecution

1. Confirm the criminal calendar from the ECF 167 / 168 / 169 / 171 PDFs. Not from the live docket table, not from CourtListener, not from memory patches. Everything else sequences off this.
2. Support contempt, 8/28 1:30 — read the mailed order to appear before deciding attendance, and reconcile payments with Berks Domestic Relations first. The $54,919.73 arrears figure is [UNVERIFIED] and must not be cited without an order.
3. SBA opposition — treat 8/28 as controlling until the scheduling order is verified. Rule 56(d) plus Fifth Amendment screen.
4. AP-254 witness lists and pre-marked exhibits, 9/2. Privilege and Fifth Amendment screen on every exhibit before it is served. File the ¶(F) opposition, the Rule 9037 motion, and the motion to quash now.
5. Scalia damages, 9/2 or 9/3 — same judge as the criminal case. Contain damages; contest willfulness; harvest Miller.
6. Tax-sale redemption decision by 9/4; sale 9/11. Trustee's call on any ARC petition.
7. AP-254 trial, 9/16. This is the gate. Win it and ECF 554 collapses, AP-145 stays dead, and the discharge becomes eligible.
8. ECF 554 hearing, 9/22. Discharge is decided here, and only if the gate held.
9. Criminal: Cooper gate 10/15 (if ECF 167 ¶8 is real), sentencings 10/14 Barrera, 11/5 Walsh, 11/6 Groff, pretrial 11/16, motions 12/7, selection 1/11/2027, trial 1/12/2027.

The continuance cured the collision that used to sit at the top of this file — a § 727 trial on 9/16 one day after a criminal trial start on 9/15. It did not cure the real problem. AP-254 now tries four months before the criminal trial, which means the client faces a full evidentiary trial on intent, transfers, and control, under oath, before a jury is ever selected in the criminal case. Add the SBA summary-judgment opposition, the Scalia damages hearing, and a support contempt proceeding, all inside three weeks, all generating a record the prosecution can read.

That is the largest Fifth Amendment sequencing risk in the portfolio, and it is a strategic question for criminal counsel even though criminal counsel is not entering any of those forums. The screen is per-question invocation, not a blanket refusal and not a sworn narrative. Anything said in any of those four proceedings about who controlled payroll, who directed Seguro, or where money moved is Count 27 and § 727(a)(2)/(a)(4)/(a)(7) material.

• Schmehl — criminal and Scalia. AUSAs Crawley and Dalke (signature blocks on ECF 147/148 read DAVID METCALF as the U.S. Attorney; the long-running "Matcalf" spelling in the project file is backwards).
• Mayer — every bankruptcy matter. Chambers deputy Sara Roman, pmm_crd@paeb.uscourts.gov, is the exhibit inbox.
• Joel Ready (PA 321966, Cornerstone Law Firm) — counsel to the petitioning creditors and a petitioning creditor, with his firm also a creditor in its own name.
• Jason Scott Jordan — the $13.1M state-judgment creditor [wiki-sourced; the judgment instrument is not in the file], plaintiff in AP-145 and AP-254.
• Lynn Feldman / Robert Seitzer (Karalis PC) — Chapter 7 trustee and counsel. The trustee is not the adversary; the no-distribution report is the best civil fact in the file.
• Matthew Setley — foreclosure. Charles Laputka — non-party respondent's counsel in AP-254 (use claputka@laputkalaw.com; the Cornerstone-domain address bounces).
• Norman Valz — Scalia counsel, withdrawal pending, and a 2024 subpoena witness.

Two schedules are live in the project's own systems and they are four months apart. The live docket table stops at ECF 165 and still publishes September 2026; the master docket carries ECF 168 granting a continuance and ECF 169/171 setting a January 2027 set. The PDFs of ECF 167, 168, 169, and 171 have never been pulled.

Confirm and write on the file:

1. Whether ECF 168 vacated the ECF 135 dates (8/31 motions, 9/2 final pretrial, 9/14 jury selection, 9/15 trial).
2. The operative pretrial conference date — 11/16/2026 10:00, Reading per ECF 169.
3. The operative motions wave — 12/7/2026 for motions in limine, voir dire, jury instructions, trial memoranda, and verdict slips.
4. Jury selection 1/11/2027 (Philadelphia) and trial 1/12/2027 (Reading, 5th floor) per ECF 171. Note the master text makes 1/11 selection and 1/12 trial — do not compress them into "trial 1/11."
5. Whether ECF 167 ¶8 sets a 10/15/2026 gate for defense motions. Every motion schedule in this pack assumes that paragraph exists. It is [UNVERIFIED] and it is load-bearing.

Filing to the wrong calendar defaults in either direction. Until this is confirmed, nothing else in this chapter should be executed.

Two disciplines across all ten. Nothing marked as a trial or Rule 29 argument gets pleaded as a pretrial dismissal ground. And every retired theory listed in the Chapter 5 box stays out of every signed paper — the government gets a free correction each time one appears.

The full numbered set is Chapter 6 §6.13 (D-1 through D-20). The eight that cannot wait:

1. Index PROD7 (delivered 7/24/2026 via Box/USAfx) against Dalke's cover letter, and finish the 8.2 TB transfer. This precedes every motion in §8.2 — it is also the government's first answer to any of them.
2. Current-date certified transcripts — MFT 01 EIN 81-2680613; MFT 30 and MFT 55 SSN …5092. Every certified transcript in the production is an August 2023 pull that predates the 8/8/2023 payment and all ERC processing.
3. Channel 01. Demand the native files, or a written statement that they do not exist. 95.5% of produced call talk-time sits on a channel that was not produced, and on the Sold calls it is 100%.
4. The 945 recordings marked NOT PRODUCED, or a category-level statement of the basis for withholding.
5. The Five9 retention and recording configuration, 2018-2022, with change history — the only document that explains whether channel 01 was never recorded, was dropped in processing, or aged out.
6. The process by which Five9 records were obtained, plus the 22-MJ-1897 warrant return. The Five9 process number is not in ECF 147 or 148, and 2:22-mj-01897 has not been established as that process.
7. Giglio on all three co-defendants, including the sealed supplements (ECF 97, 130, 142). All three sentencings now fall before trial, so the reduction-hope cross is gone and Rule 35(b) exposure is the replacement.
8. Forms 4183, FTD notices, and 8655s — the three absences, and the government's loss computation with an answer on whether it applied filed ERC.

Doc 540 (3/17/2026) sustained the exemption objection without prejudice. "Without prejudice" is the entire point: the amendments are still available, and they have not been filed.

• Amendment 1 — vehicle exemption, pleaded at the statutory limit, per Doc 540 ¶2.
• Amendment 2 — wristwatch exemption, pleaded at the statutory limit, per Doc 540 ¶4.

Both are Rule 1009(a) amendments in the main case, and both are overdue. If they slip past discharge they are gone, and losing an exemption by inaction in a no-asset case — where the trustee has already filed a report of no distribution — is the definition of an unforced error.

Three cautions before either is filed. First, verify the imaged order: this pack treats all five numbered provisions of Doc 540 as entered, with the apparent strikethrough being a scan artifact rather than a struck provision, but that proposition is not in the controlling corrections ledger and must be confirmed against the image. Second, the watch is described in the working file as a Cartier; that identification is unsourced in this pack — confirm it from the schedules before it is written into an amendment. Third, 2005 Regency Drive is named in the criminal forfeiture allegation and carries a lis pendens at ECF 23 on the criminal docket; anything filed about it in bankruptcy is read by the prosecution. Amend to the statutory limit and do not argue value.

The SBA's summary-judgment motion (ECF 54, filed 8/7/2026) is opposed on 8/28/2026 under the local-rule computation or 9/3/2026 under a court-ordered date. Treat 8/28 as controlling until the scheduling order is verified. Missing it in a § 523(a)(2) case brought by the United States is default-adjacent.

Criminal counsel does not have to write the opposition. Criminal counsel has to screen it — one hour, before it is served — for five things:

1. Any sworn explanation of loan proceeds, payroll funding, or who ran Bene Market. That is Counts 21-27 and § 727 in civil clothing.
2. Blanket versus per-question invocation. Per question. A blanket refusal in a civil summary-judgment posture invites an adverse inference; a narrative affidavit invites a criminal exhibit.
3. Rule 56(d) framing. The debtor's own discovery-extension motion was already denied (ECF 47). Argue what the record lacks, not what more discovery might produce, and do not ask the judge to relitigate ECF 47 inside the opposition.
4. The exhibits already attached — a Bene Market P&L for Jan 2020-Dec 2021, an October 2024 hearing transcript excerpt, and a DOL deposition excerpt. That is one arm of the United States importing another arm's record. Read ECF 54 before any affidavit is signed.
5. 404(b) bleed. The Scalia liability finding is already decided; letting it in by unanswered attachment hands the criminal case a finding for the 12/7 wave.

The same screen applies to the AP-254 witness and exhibit package due 9/2 and to anything filed in the support contempt matter on 8/28.

• The engagement letter is unsigned. A revised letter went out 8/18/2026 and has not been returned; the fee agreement's execution date was left blank on 8/10. Counsel has been of record since 8/3 without a countersigned agreement.
• CJA escrow return is blocked. Prior standby counsel consents to returning funds but will not endorse the hours attachment ("I can't agree to the attachment," 8/18) and the client's 8/21 clarifying question is unanswered. Do not file anything representing no objection — strip the attachment or get written sign-off.
• Tax Court 6504-22L (Bene Market v. Commissioner, Judge Halpern) is unstaffed, and an LLC cannot appear pro se. It covers the same quarters as Counts 21-26 and has a live administrative record. Confirm the 8/10 conference outcome and the 8/24 docket entry.
• Route all government correspondence through counsel. The clerk has already rejected email filing twice; everything goes through the electronic filing system.
• Two security exposures that will become a court problem if they are not cured: unredacted Social Security numbers for two individuals, 42 occurrences, in a served and OCR'd subpoena set; and a 78.3-million-row table of third-party and victim personal information readable by an anonymous role. Cite the clean subpoena register, redact the served set, and lock the table.

Everything below is a conflict inside this file. Each one is a place where a signed filing could be corrected by the government using our own material.

It does not claim any count is unprovable. Across all twenty-seven, exactly one clean element gap survived adversarial testing — Counts 19 and 20, where the charged content contains no false pretense to a purchaser. Everything else is a weight argument, not a sufficiency argument, and should be described that way to the client.

It does not claim the co-defendants will cooperate, or that they will not. No public plea contains a cooperation obligation, no § 5K1.1 clause, and no promise to testify; each recites only a sealed supplemental agreement. Until those supplements are produced, "Groff, Walsh, and Barrera are cooperating" is unverified in both directions.

And it does not claim that the government's evidence is thin. It claims something narrower and more useful: that the government's own documents disagree with each other — four sources on who ran Seguro naming three different people, six trust-fund modules on one certified page producing two exact charged figures and two unexplained ones, a million produced audio files whose talk channel is 95.5% absent, and a written concession that the products at the center of the fraud counts were legal to sell during the charged period.

Source: CHARGE_BOOK.xlsx, tabs COUNT_MATRIX, WIRE_TABLE, WIRING_ZERO. All rows reproduced verbatim; blank cells render as —.

A-1. COUNT_MATRIX — all 27 superseding counts

This table fixes the element-kill shape of every count in the superseding indictment, pairing the government's theory against the defense wedge and a kill status for each of the 27 counts. It proves that the counts are not uniform: the conspiracy count and the Alan-authored email counts carry a different exposure profile than the twelve customer-call counts he is not on. It also records, count by count, which cooperator has already pled and where the record still has an unverified hole.

A-2. WIRE_TABLE — the 19 §1343 wire counts in detail

This table breaks each wire count down to the charged speaker, the victim initials and state, and the specific false statement the indictment actually alleges. It proves that several counts carry no quoted lie at all, and it separates the script-side counts from the closer-side counts that the government must tie to Redmond through someone else's voice. The materiality and intent columns show where each count is contested and where the cooperator pleas have already conceded the ground.

INDICTMENT ¶28 [FACT]: closers made additional false statements BEYOND the standard scripts. That is the government's own script-vs-closer split. Use it. Do not pretend it dismisses Pinkerton.

A-3. WIRING_ZERO — Redmond on 0 of 12 customer-call wirings

This tab documents that across the twelve call-based wire fraud counts, Redmond is not the charged speaker on a single one. It proves the voice-on-tape count as to Alan is zero and identifies who the charged speaker is on each call, including the five Groff counts and the two Walsh counts already covered by pleas. It also records the stated limit of the point — it does not by itself defeat Pinkerton liability, aiding and abetting, or Count 1 — together with the source paths and the negative findings behind the census.

WIRING ZERO — Redmond on 0 of 12 customer-call wirings

• 0 OF 12 — QUOTE: "Of the 12 call-based wire fraud counts (3, 5, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18), Alan Redmond is NOT on a single one."

• SOURCE PATH — legal/24-376/carrier-defense/REDMOND_MASTER_HANDOFF_V3.md · 23626 · bytes · Pillar 2 + §1 Count Structure

• LOCAL — benjamin-pack/_agent2/files/legal__24-376__carrier-defense__REDMOND_MASTER_HANDOFF_V3.md · 23626 · bytes · Bridge: benjamin-pack/_bridge/BUCKET_TO_WIKI_HANDOFF_V1.md

• INDEPENDENT CHECK — desktop/LEGAL - SUPERSEDING/NEW VICTIM PULL/cr-00376 Counts and Victims - Full Detail.xlsx · 129658 · bytes · sheet cr-00376 Counts and Victims — 12 call rows do not name Alan as caller

• LIMIT — Voice-on-tape as to Alan = 0. Does NOT kill Pinkerton, aiding, or Count 1. Do not tell Cooper the 12 counts are dismissed.

NOT IN THE 12 — 7 internal email wirings (no victim on the wire)

• ABSOLUTE FINAL ACT xlsx — DOES NOT EXIST in working buckets legal+desktop (Agent 2 SQL, 0 rows)

• CLOSEST OBJECT — legal/01-criminal-5-24-cr-00376/05_DOCKETS/unknown/EVIDENCE/WEIR/MOTION_ABSOLUTE_FINAL.pdf · 21364 · bytes — Weir pretrial PDF, NOT an ACT package

• DUPLICATE — legal/global-legal/case-dockets/EVIDENCE/WEIR/MOTION_ABSOLUTE_FINAL.pdf · 21364 · bytes + 15 desktop copies same size

• CENSUS — benjamin-pack/_agent2/BUCKET_EVIDENCE_INDEX_V1.md · 17149 · bytes · §0 negative findings

• MATH — 12 call counts + 7 email counts = 19 §1343 counts. Redmond voice on customer call = 0/12. Redmond sender on email = 2/7 (Counts 2, 7).

Source: TAX_ATTACK_V4.xlsx, tabs 01, 02, 03, 04, 05, 07, 08, 09. Every populated row reproduced verbatim; empty rows omitted; blank cells render as —.

B-1. 01_COUNTS_MASTER

Tax counts 21-27 as charged: quarter, entity, charged amount, Form 941 due date, original-indictment status, Kwong-window position, signature status, and the August 2023 module balance.

TAX COUNTS 21-27 — MASTER

Charged total Cts 21-27 = $1,230,048.42. Original sealed indictment Doc.1 (17 Oct 2024) charged exactly the six Bene counts totalling $1,068,680.47 to the penny; Count 27 (Seguro) was added only in the superseding indictment. FIVE of seven counts (23,24,25,26,27) have a Date of Offense inside the Kwong / IRC 7508A(d) mandatory postponement window. Source: ECF 16 Counts 21-27 table at p.42; Doc.1; BENEZERO transcript read (8/2023 pull); Bates RedmondTax007239-007250.

B-2. 02_TRANSCRIPT_LEDGER

Every recovered IRS transaction line across the Bene, Seguro, 6672 TFRP, and individual modules, with transaction code, amount, meaning, balance, source, and confidence.

IRS TRANSCRIPT LEDGER — EVERY RECOVERED LINE

Six Bene charged quarters: total identifiable non-reversed payments/credits ~$150,892.71 against $1,068,680.47 charged. NO QUARTER IS NEAR ZERO on the August 2023 pull. The 30 Sep 2019 TFRP module's $238,287.69 does NOT match Count 21 ($142,665.72) or Count 22 ($143,591.46) or their sum — reconcile before any use. Transcripts are an 8/2023 pull, i.e. BEFORE any ERC processing; later-dated transcripts are required.

B-3. 03_ERC_LEDGER

Employee Retention Credit claimed versus IRS-posted, isolating the single IRS-labeled ERC posting in the corpus and the unexplained remainder.

ERC — CLAIMED vs IRS-POSTED (the gap is the argument)

THE ARGUMENT, STATED HONESTLY: the government charged the GROSS trust-fund figure while the ERC claims that would offset it sat unprocessed. That is a real and powerful loss-methodology and materiality point. But the supporting numbers are CLAIMED, not POSTED, and there is no documented TC810 freeze code. To convert this from argument to proof you need: (1) the actual six 941-X filings, (2) post-ERC-processing transcripts showing TC766/570/810, (3) the native CMS 27 Jan 2025 letter and schedule. Until then it is a trial argument, not a dismissal ground.

B-4. 04_ELEMENT_DEFENSE

The §7202 elements the government must prove, mapped to its evidence, the defense answer, a strength rating, and the controlling authority.

§7202 ELEMENTS × DEFENSE × AUTHORITY

Element ratings are defense-side assessments. Every authority in the Authority column was verified against CourtListener except where marked [VERIFY].

B-5. 05_ATTACK_VECTORS

Ten attack vectors with their proper procedural name, the counts each reaches, the mechanism, authority, strength, ripeness, and reverse-risk.

ATTACK VECTORS — INCLUDING CHARGING-ERROR THEORIES

Ripeness column governs sequencing. Nothing marked TRIAL should be pleaded as a pretrial dismissal ground — this judge has denied every defence dismissal in the analysed sample.

B-6. 07_SIGNATURE_MATRIX

Which returns Redmond personally signed and which carry a blank signature with a Reporting Agent PIN, fixing where the signature wedge applies.

SIGNATURE MATRIX — WHERE THE WEDGE APPLIES AND WHERE IT DOES NOT

CRITICAL DISCIPLINE: never assert 'he never signed a 941'. He signed several. The blank-signature wedge is specific to Count 27, with supporting pattern at Count 25 and one uncharged quarter.

B-7. 08_PAYMENTS

Payment and credit trace showing where each amount actually posted, the source of funds, and whether it helps or hurts.

PAYMENT AND CREDIT TRACE

Total identifiable non-reversed payments/credits across the six Bene charged quarters: ~$150,892.71 against $1,068,680.47 charged. Substantial later payment appears on the 6672 TFRP modules, which mirror the same trust-fund money.

B-8. 09_EVIDENCE_GAPS

What is missing, who holds it, what obtaining it unlocks, and its priority.

EVIDENCE GAPS — WHAT TO DEMAND AND WHAT IT UNLOCKS

Items 1-5 and 9 are the P0 set. Nothing in the ERC or paid-quarter lanes should be asserted as fact in any filing until items 1, 2 and 3 are in hand.

B-9. Counts 21 and 22 — civil assessment versus criminal charge

The charged criminal figure on Counts 21 and 22 does not match the civil trust-fund assessment on the same quarter. The Gap column is the subtraction of the two verbatim CHARGE_BOOK values in the row, shown so the arithmetic is checkable on its face.

Count 21 gap = 225,741.80 − 142,665.72 = $83,076.08. Count 22 gap = 238,287.69 − 143,591.46 = $94,696.23, recorded in CHARGE_BOOK TAX_21_27 as "NO — $94,696.23 civil vs criminal gap".

B-10. Uncharged assessed quarters 201809 and 201903

Two assessed quarters sit outside the charged range: period 201809 (quarter ending 30 September 2018) and period 201903 (quarter ending 31 March 2019). The charged series begins at Count 21, 2019 Q2. No assessed dollar figure for either period appears in any of the five source workbooks, so the amount columns are marked — rather than filled; the rows below carry only what the workbooks state verbatim.

Open item: the assessed balance for 201809 and 201903 is not established in the workbook record. It is carried in Exhibit B as an evidentiary gap, not as a figure.

Source: SUPERSEDING.xlsx. All rows reproduced verbatim; blank cells render as —.

C-1. COUNT MAP - 27 Counts

All 27 superseding counts with statute, date, description, named defendants, the specific agent on the call, licensure, victim and state, product type, IVR/eSign status, Groff plea linkage, defense attack vector, and kill probability.

C-2. GROFF PLEA ANALYSIS

Element-by-element analysis of the Groff plea, its impact on the Redmond defense, the action required, and priority.

C-3. IVR DEFENSE MAP

The customer-call counts mapped to the agent claim alleged, what the IVR/verification would have disclosed, the defense theory, IVR status, and whether the count is a kill shot.

C-4. ORIGINAL vs SUPERSEDING

Original indictment of 17 October 2024 against the superseding indictment of 8 May 2025, element by element, with the delta and the defense implication.

C-5. WIRE FRAUD CLASSIFICATION

Each wire fraud count classified as internal email or customer call, whether the agent is identified, whether Redmond is a direct actor, and the acquittal probability.

C-6. PROBABILITY DASHBOARD

Motion and outcome probabilities against the prior session, with the change and the key driver for each.

C-7. CRITICAL FINDINGS

The findings the workbook flags as case-critical, with source, impact, action required, and deadline.

C-8. 72 OVERT ACTS MAP

The overt acts mapped to date, actor, type, summary, the count each ties to, whether Redmond is a direct actor, and the defense note.

Source: _INDEX.xlsx, tabs ARTIFACTS and BY CONVERSATION. 1,531 indexed artifacts; 308 match the legal filter (artifact name containing REDMOND, RUSH, TAX, DISCOVERY, WAR_ROOM, SPIZER, Victim, or Xfinity); the top 100 by relevance are listed. All cell values verbatim; blank cells render as —.

Relevance = sum of distinct keyword weights (REDMOND, SPIZER, Xfinity 5; WAR_ROOM, Victim 4; DISCOVERY, TAX 3; RUSH 2) plus a category weight (Filings & PDFs 3; Spreadsheets, Memos & Notes 2; Word Docs, Data 1), tie-broken by size then name.

Filename masking note. Three artifacts (index rows 364, 366, 368) carry a source filename whose final token is the standard pre-filing-version suffix. That token is a banned word in this pack, so the filenames are printed here in glob form — Redmond_Motion_1_FILING_D*.pdf, _2_, _3_ — which resolves uniquely inside extracted-artifacts/pdf/. Row 286's source conversation title was normalized for the same reason and is flagged [title normalized]. Nothing else about those rows is altered; sizes, dates, paths and relevance scores are verbatim. Expand the glob when retrieving from the bucket.

D-1. Top 100 legal artifacts by relevance