research targets · open commitments

The framework's open questions, named explicitly. Each carries a research target: what would refute, what would confirm, where the question lives in the formal apparatus and in the prose. A framework whose open questions are precisely posed is a framework that can be developed. The list below is the document's standing invitation to that work.

Formal mathematical questions

Three questions from §8 of the math doc that the framework names as open targets within its own formal apparatus — positions where the existing proofs do not yet reach.

Q1. The direction of $T_{\mathrm{mix}}$ under Mode C

The question. Does disrupted observation (Mode C: inflating $\mathrm{Var}(\eta_t)$) slow the cohort's convergence toward the policy's stationary distribution, or speed it? The framework's intuition is the former — weaker policy pull means slower mixing — but the proof requires additional structure on the response dynamics $f$, the observation function $g$, and the policy class that the framework does not currently specify.

What would settle it. A formal derivation linking $\mathrm{Var}(\eta_t)$ to the spectral gap $\gamma_\pi$ of the closed-loop transition operator $P_\pi$, or an empirical measurement of cohort mixing time under a Mode C–like intervention at platform scale.

Why it matters. Mode C's strategic role depends on which holds. If disrupted observation slows mixing, the mode buys time during the long mixing-time of Mode A architectural intervention. If it speeds mixing, Mode C may have the opposite of its intended structural effect at the population level.

Where named. Math §8.1.

Q2. The developmental mechanism behind the cohort gradient

The question. The framework asserts as a structural prediction that long-lag interiority $I(\tau)$ is moderated by developmental-stage-at-introduction to the loop (the cohort gradient, §6.1). But the formal apparatus does not derive the developmental moderation from the closed-loop dynamics (1)–(5) alone. The argument is intuitive — pre-closure cohorts have a self-evaluative distribution $\nu_u$ distinct from $\pi_{\hat{R}}$ at first exposure; post-closure cohorts do not — but a formal derivation would specify the developmental learning dynamics of $\nu_u$ formation, and the contractive properties of the user's response dynamics under closed-loop stimulation relative to pre-closure conditions.

What would settle it. A model of $\nu_u$'s formative trajectory with developmental-stage parameters explicit, coupled to the existing Proposition 2 dynamics, that produces the gradient as a derived consequence rather than an asserted prediction.

Why it matters. This is the framework's most refutable claim. If the developmental mechanism is wrong, the cohort gradient may not hold; if it holds, the gradient is a near-direct consequence rather than a free-standing empirical prediction.

Where named. Math §8.2; Citizen After the Pause §2.

Q3. The operating-point characterization of per-register surplus

The question. From equation (5), the per-register surplus $\Sigma^i(T)$ is observable through the marked point process $N_t^i$ and the kernel $\rho_i$. The framework claims that under any policy $\pi$ converged on $\hat{R}$, the joint distribution of $\Sigma^i$ across $i$ approaches a specific operating point: extraction proportional to the marginal yield of each register conditional on the cross-excitation matrix $\alpha_{ij}$. The explicit characterization — a KKT-style constrained-optimization solution over the policy's action space — is not written down.

What would settle it. Either a Lagrangian derivation of the operating point under the platform's revenue maximization (subject to (A1)–(A6)), or empirical identification of the operating point on disclosed platform data.

Why it matters. This would make the framework's prediction about which register the platform over-routes to (and by how much) sharp enough to test against the engagement-event metadata production platforms already collect.

Where named. Math §8.3.

Open structural-empirical questions

Two questions the prose names at the level of structural prediction about the cohort's actually-existing politics, which the formal apparatus does not currently formalize.

Q4. Is cohort-level disruption-mobilization scaled-up Mode C?

The question. Mode C at the individual level bounds the platform's purchase by injecting structural variance into the engagement signal through deliberate inauthenticity, strategic obfuscation, refusal of legibility. The framework's prediction about the cohort's actually- existing politics — the iterative-disruption, rapid-onset, rapid- dispersion pattern — sits in suggestive distance to this. Is the cohort's mobilization-form a population-scale Mode C — does the pattern bound the apparatus's purchase on the cohort the way individual Mode C bounds the platform's purchase on the user?

What would settle it. An information-theoretic measurement of the platform's posterior over a mobilizing cohort during a disruption event compared to its posterior in the quiescent regime. If the posterior is systematically less informative during disruption, the cohort-level pattern bounds platform information in the way Mode C formally specifies.

Why it matters. If the answer is yes, the older analytical cohort's tendency to read disruption-mobilization as failed consolidation is exactly the kind of category error the framework's analytical move (recognition) was meant to correct — the cohort is already doing Mode C politics; the older categories cannot read it.

Where named. Citizen After the Pause §3.

Q5. Does the native cohort's multi-legibilities practice satisfy Mode C?

The question. The native cohort's elaborate construction of online personae (the alt account, the finsta, the parasocial-tested public persona) might look like Mode C from outside but operate, on a pessimistic reading, as another register of platform-articulation rather than the refusal-of-articulation Mode C formally specifies. On an optimistic reading, the cohort has invented Mode C as a native rather than inherited mode, and the multiple-legibilities practice is in fact Mode C's native form for a cohort with no pre-platform self to retreat to.

What would settle it. The same kind of measurement as Q4 but at the user level: is the platform's posterior over a multi-legibility- practicing native-cohort user in fact less informative than its posterior over a single-account user with otherwise-equivalent engagement?

Why it matters. If pessimistic, Mode C joins Mode B as a structurally foreclosed mode for the native cohort, and the only available intervention is Mode A — architectural intervention conducted by political bodies outside the user-platform circuit, on behalf of a cohort whose own intervention-modes have been pre- removed. If optimistic, the cohort is the most fluent practitioner of Mode C the framework predicts — and the older analytic cohort's failure to recognize this is itself the recognition-failure the framework's analytical move was meant to address.

Where named. Adolescence Without Outside §VIII.

Where the empirical record is genuinely thin

Five empirical commitments the framework makes that don't yet have strong empirical confirmation. Named in the reading list's empirical backing section, restated here as research targets.

Q6. The cohort gradient at the developmental level

The empirical commitment. Long-lag autocorrelation in user state is moderated by developmental-stage-at-introduction to the loop. Younger cohorts should retain measurably less interiority at long lags, controlling for total exposure duration.

The literature. Twenge & Haidt make the cohort claim empirically; Orben & Przybylski find small effect sizes in cross-sectional data. The literature is contested.

What refutation looks like. Longitudinal cohort data showing no gradient, or showing the inverse (younger cohorts retain more autocovariance) — see the cohort gradient plate for the predicted shape and the falsification threshold.

Q7. Mutual-information convergence across the four registers

The empirical commitment. Off-diagonal entries $I(u^i; u^j)$ between the Subject/Citizen/Person/Consumer registers fill in under prolonged loop exposure — the dividual condition's measurable surface.

The literature. Kosinski et al. (PNAS 2013) showed single-channel engagement predicts cross-channel attributes; Matz et al. (PNAS 2017) operationalized cross-register inference as persuasion targeting. Both are suggestive but not definitive on the population-scale convergence.

What refutation looks like. Time-series measurement of platform- internal $I(u^i; u^j)$ estimates across cohorts and exposure durations, showing no upward trend.

Q8. Mode C's effectiveness at population scale

The empirical commitment. Practiced at population scale, Mode C strictly bounds platform surplus extraction. (Direct experimental form of Q4 + Q5.)

The literature. None known. The Blackwell-style sufficiency argument is mathematically sound for individual users; the population-scale effect has not been measured to the author's knowledge.

What confirmation looks like. A natural experiment in which a substantial cohort adopts a Mode C–like practice and the platform's disclosed engagement metrics show a measurable bound on extraction.

Q9. Mode A mixing-time prediction

The empirical commitment. Under Mode A architectural intervention (e.g., DSA/DMA regulation), cohort distributions converge to the modified policy's stationary form over months-to-years, not weeks. Evaluations conducted on shorter clock-time will systematically underestimate the structural effect.

The literature. TikTok and YouTube algorithm-change studies show cohort-level distribution changes taking quarters. Markov-chain mixing-time bounds (Levin & Peres) give the formal apparatus. But direct measurement of $T_{\mathrm{mix}}$ under Mode A is still propagating — the DSA/DMA effects are early-stage.

What confirmation looks like. Five-to-ten-year longitudinal data on cohort engagement distributions under EU/UK architectural regulation, showing the predicted mixing-time signature.

Q10. The captive-as-population claim

The empirical commitment. The figure tracked in The Sincere Captives — the post-cynical professional for whom the platform-grammar has been metabolized upstream of articulation — exists at population scale, not just as a phenomenological possibility.

The literature. Phenomenological, with suggestive empirical neighbors (Phillips & Milner; Duffy; C. Thi Nguyen on value-capture) but no rigorous population-level measurement.

What confirmation looks like. Methodologically careful identification of the population's size and demographic distribution. Distinguishing the sincere captive from the cynical professional empirically would require measurement of internal-vs-performed affect-grammar congruence — which is its own methodological project.

On using this list

These ten questions are not the framework's weaknesses to be papered over. They are the targets the framework specifies for whoever takes the framework up — empirically, theoretically, or institutionally. The framework's commitment is to name what it does not yet show, in the same register and at the same level of specificity as what it does show.

The reading list gives the empirical-backing context for each question. The math doc gives the formal position from which Q1–Q3 are posed. The argument gives the prose register in which Q4–Q10 are felt.

If you are taking up one of these questions and would like to talk: the author can be reached through the channels named in /colophon.