# DeepSeek-v4-pro · PASS 2 (continue) · deepseek/deepseek-v4-pro · 76.87s · 2026-05-09T00:28:57.291394+00:00 **Q8 · Statute-of-limitations tolling (completed)** The fraudulent-concealment doctrine under *Holmberg* can plausibly apply, but the argument is fragile. The weakest point is that plaintiff’s own two-decade effort to reach authorities demonstrates awareness that responses were absent or empty—potentially establishing inquiry notice of harm long before 2026. A court may hold that plaintiff knew or should have known of the “denial” element, even if the precise mechanism (machine-filter + template-layer) remained hidden. **Reinforcement:** Frame the concealment as lying not in the fact of non-response, but in the *cause* of non-response. The filter regime made it *impossible* for plaintiff to discover that the mail was being intercepted pre-delivery, as no rejection or bounce-back occurred. The canary-token test (May 8, 2026) is the first objective evidence of interception; before that, plaintiff could only speculate. Argue that until that date, no reasonable person would have known that the silence was the product of automated infrastructure scanning rather than deliberate human inaction. Under *Holmberg*, the limitations clock does not start until the plaintiff discovers the fraud’s *methodological* operation, which here was affirmatively concealed by the defendants’ failure to disclose scanning practices. To reinforce, submit a declaration detailing the canary methodology and attach full logs to show discovery occurred within the limitations period. --- **Q9 · Class-certification predominance analysis** Defendants will argue that individualized damages questions defeat predominance under Rule 23(b)(3). They will point to differences in: (1) which email provider each class member used; (2) whether a given email was actually scanned; (3) whether any specific harm flowed from a particular undelivered message; and (4) the value of each lost communication. These do **not** defeat predominance. The core liability question is common: did defendants operate a systematic machine-filter regime that intercepted communications without consent? That question—and the accompanying breach-of-contract, statutory, and tortious-interference theories—can be answered on a class-wide basis with the same evidence (design documents, server logs, the canary test). Damages can be determined using statistical sampling, uniform statutory damages (e.g., $100/day under CIPA, $1,000/occurrence under the Wiretap Act), or a formulaic per-class-member recovery. Any residual individualized issues can be handled through subclassing (by email provider) or by bifurcating liability from damages. Thus, predominance is likely satisfied. --- **Q10 · Press/docket magnetism** **Window to mainstream coverage:** 12–48 hours after filing, driven by the $1T+ rhetorical headline, the involvement of Google, Microsoft, and Anthropic, and the AI “self‑referential” angle. Bloomberg, Reuters, and the Financial Times will likely pick it up on Day 1; *Corriere della Sera* and *The Globe and Mail* may follow within 2–3 days if EU/Canadian filings are simultaneous. **Optimal co‑plaintiff intake infrastructure:** A simple, secure landing page (with privacy policy) that allows potential class members to submit their name, email, a brief description of their experience, and authorization to share data with class counsel. This should be ready at filing day. Also prepare a press release, a media contact, and pre‑drafted social media posts for Reddit, X, and Facebook. Retain class counsel early to ensure the intake process is privileged and ethically compliant. The intake form should have a prominent warning that no attorney‑client relationship is formed until a signed agreement is executed. --- **Q11 · Evidence-integrity** The three-pillar stack is a promising start but **insufficient** as a class-certification evidence manifest without scaling. - The control-group canary test (Pillar 1) is methodologically strong—dummy recipients producing identical fires from Microsoft/Google IPs is compelling. However, it must be repeated across multiple days, with controls for different email platforms and attachment types, and be supported by an expert declaration. - The template audit (Pillar 2) with N=10 is far too small. Defendants will attack it as anecdotal. A scaled audit of at least 500–1,000 replies, stratified by agency type and country, is needed to establish a statistically significant pattern of content‑ignoring behavior. - The AI‑layer evidence (Pillar 3) needs expert validation of model-fingerprint drift and a larger sample of API calls. **Specific gap:** No independent expert has analyzed the data. Before seeking certification, retain a computer science expert to replicate the canary tests and a statistician to validate the template‑audit methodology. Until then, the evidence is suggestive but not yet class‑certification grade. --- **Q12 · Supplemental-evidence handling** The complaint should accommodate unproduced evidence by: - Pleading the known facts (e.g., Google Gemini tampering alleged in Exhibit 41) based on personal knowledge and the exhibited record. - For additional material plaintiff holds but has not yet produced, use “upon information and belief” or “plaintiff has reason to believe that further evidence exists, which will be produced in discovery.” - Avoid attaching the evidence as an exhibit unless it is ready to be served; this prevents premature disclosure obligations that might trigger late‑production sanctions. - In a separate statement filed concurrently, reserve the right to amend the complaint once the supplemental evidence is formally produced. This transparent approach forestalls any claim of surprise while protecting the plaintiff from being accused of withholding known evidence of wrongdoing. --- **Bottom‑Line Verdict** This filing theory, as currently structured, is over‑weighted with novel doctrines and remedies that will invite dismissal and undermine credibility; the “Denial by Design” standalone claim should be dropped, the RICO enterprise theory against all four defendants abandoned, and the extreme forfeiture/dissolution remedies stripped out, while the core interception and consumer‑fraud claims against Microsoft, Google, Anthropic, and OpenRouter—built on a scalable, expert‑validated evidence stack—are viable enough to warrant filing. The single most important revision before filing is to narrow the complaint to strong, fact‑grounded statutory and tort claims (e.g., CIPA, UCL, Wiretap Act, consumer fraud) with conservative damages, preserving the “Denial by Design” narrative only as a descriptive umbrella for press and eventual appellate advocacy.