Studio — Rebuilding a Draining Product into a Returning One · Design as Displacement

Module 8 — Studio — Rebuilding a Draining Product into a Returning One

Learning objectives

Diagnose any product as a wrong-attractor engine: name its ground state $S^0$, the stable wrong configuration $W$ it traps users in, and the specific framework condition (DC5 or DC7) its design violates.
Convert each extractive mechanism into its returning counterpart by re-targeting design away from displacement $\xi$ and onto displacement cost $\Phi$.
Distinguish profit-as-glitch ($G = \Phi - k|\delta|$) from fair profit-by-volume ($\Pi = m\,n\,\bar\Phi$, DC8), and judge a redesign by which one it generates.

Exposition

This is a synthesis studio. You have a method; here you run the whole loop on one artifact. The corpus hands us a fully worked extractive product — the engagement-optimized social feed — so we use it as the bench and then generalize. The method has five moves.

1. Locate $S^0$. Every product acts on a human system that has a minimum-cost configuration. For attention the corpus names it precisely: $S^0_{\text{attention}}$ is intentional, cognitively autonomous attention — voluntary focus, metacognitive awareness, compulsion-free disengagement, and boredom tolerance (the most diagnostic feature). Displacement is $\xi(t) = \rho(s, S^0)$, the distance a user sits from that state at time $t$; instantaneous cost is $D(\xi)\ge 0$.

2. Name the wrong attractor $W$. An extractive product does not merely displace; it engineers a stable configuration that is not $S^0$ and makes it self-reinforcing. The social-media paper identifies four mechanisms that build the basin: variable-ratio reinforcement (the "slot-machine" pull-to-refresh), infinite scroll (removal of stopping-cue landscape features), notification systems that externally re-induce $D(\xi)$ across distance, and algorithmic amplification of negative affect. Doomscrolling is the sharpest instance: a loop where elevated $D(\xi_{\text{threat}})$ drives more scrolling as pseudo-information-gathering, which raises $D(\xi_{\text{threat}})$ further — a basin whose depth grows the longer you stay.

3. Find the violated condition. Two recur. DC7 (Fairness): a fair product prices on the work it performs, $C(T)=k\Phi_T$; an extractive one prices on the severity of displacement, $\xi(t_0)$. The feed optimizes engagement, and engagement correlates with $D(\xi_{\text{threat}})$ and $D(\xi_{\text{social}})$ — so the product's metric is the user's displacement. DC5 (Irreversibility): the return path is broken. Infinite scroll deletes the natural stopping cue, so $\Phi_{\text{return}} > \Phi_{\text{departure}}$ — exit costs more volitional work than entry. The glitch is then the accumulated $\Phi = \int_0^T D(\xi)\,dt$, summed across users and sold as ad inventory, while net displacement $\delta$ stays near zero (no goal advanced). The ledger shows profit; the field shows debt.

4. Invert each mechanism. The diagnosis makes redesign mechanical: re-target every feature away from maximizing $\xi$ and toward lowering $\Phi$ and $\Phi_{\text{return}}$. Take each basin-deepening feature and build its return-supporting inverse. Variable-ratio reward becomes predictable, completion-shaped delivery — a bounded set you can finish, killing the dopaminergic anticipation loop. Infinite scroll becomes restored stopping cues (a "last page," a session end), repairing DC5 so $\Phi_{\text{return}}$ falls. Notification induction becomes default-off, batched, user-scheduled signals, so $D(\xi)$ decays off-platform. Negative-affect amplification becomes ranking on anxiety-to-action conversion, not arousal.

5. Re-base the business model. This is the load-bearing move, because individual willpower is insufficient against an engineered landscape — a slot machine is not resisted by motivation. The corpus is explicit that the wrong attractor "is not a side effect of engagement design; it is the product." So a redesign that leaves the advertising incentive intact will silently re-grow the basin. The fix is DC8: decouple revenue from engagement and tie it to completed work or user-reported wellbeing, so the firm earns $\Pi = m\,n\,\bar\Phi$ — margin on the volume of real work done for users — instead of harvesting $G$ from their depth of displacement.

A returning product, formally, is one whose own success metric decreases $\xi$ and whose architecture lowers $\Phi_{\text{return}}$. Test: does the design profit when the user reaches $S^0$ and leaves, or only when they cannot?

Worked example

Take the catastrophe feed and redesign it into a calibrated-attention news client — the open "design problem" the doomscrolling paper poses.

$S^0$: calibrated attention — proportional monitoring with an achievable termination condition (the functional criterion, P2). Above the information ceiling $I^*$, more monitoring yields no preparedness, only anxiety.
$W_{DS}$: the threat-monitoring loop where more information raises anxiety rather than lowering it — the relation $\partial\xi/\partial I > 0$ — sustained by P5 algorithmic amplification. Violations: DC7 (ranks on outrage, a proxy for $\xi$) and DC5 (no stopping cue).
Redesign: (a) a bounded daily briefing that ends, restoring the termination condition and DC5; (b) every item tagged with whether it converts to an action the user can take, ranking on conversion not arousal, repairing DC7; (c) an explicit "$I^*$ reached" signal when further reading adds no preparedness; (d) revenue from a subscription tied to retained-but-calmer usage, not session length, satisfying DC8.
Result: the product earns by reliably returning users to $S^0$ and letting them disengage. $\delta \to$ informed-and-calm; $\Phi$ is small and paid for as work, so $G \to 0$.

Exercises

Pick a product feature whose explicit purpose is to prevent disengagement. Identify the natural stopping cue it removed and state, in DC5 terms, why $\Phi_{\text{return}} > \Phi_{\text{departure}}$ results.

For one engagement metric (e.g. daily session minutes), argue whether it tracks $\xi$ or $\Phi_T$. If it tracks $\xi$, propose a replacement that satisfies DC7 and show the substitution would change a real ranking decision.

(Open-ended) Choose any extractive product outside attention — a pricing model, a subscription, a game economy — and run all five moves: name $S^0$, $W$, the violated condition, the inverted mechanisms, and the DC8 re-basing. Then state the honest cost: what revenue does the firm forgo by no longer extracting $G$, and is the returning design viable without structural (landscape-level) intervention?

Sources

Social Media Addiction as Engineered Wrong Attractor (`social_media_addiction.tex`, Phronesis Research) — the four engineering mechanisms, "the wrong attractor is the product," and the individual-vs-structural return-path taxonomy.
Doomscrolling as Threat-Monitoring Wrong Attractor (`doomscrolling.tex`, Phronesis Issues #527) — $W_{DS}$, the functional criterion (P2), the Safety Paradox, the information ceiling $I^*$, return paths, and the calibrated-attention "design problem."
The Displacement Framework (`displacement-framework.tex`) — DC5, DC7, DC8; $G = \Phi - k|\delta|$; $\Pi = m\,n\,\bar\Phi$.

All three are archived live on Zenodo.