z21 loop · shaped pitch · Lacuna dir 29561

Foresightcheap world-model lookahead for live GUI & web agents

Give a reactive agent a shallow simulator of its own next move — and bet that imagined foresight beats greedy clicking where it has never been proven: the real, non-sandbox web.

Appetite Small-batch spike · one session · prove-or-break
Problem

The agent clicks into a dead-end it never saw coming

A GUI or web agent, mid-task, picks the locally-plausible next control and lands somewhere irreversible — because it never simulated where the click leads.

This is the myopia the Foresighted Planning & World Models direction forms around. Why Reasoning Fails to Plan names step-wise reasoning as step-wise greedy; BALROG documents the "knowing–doing gap." The direction is early and accelerating — 199 papers, a kind-mix of 111 limitations to just 16 findings. An open frontier, which is exactly why it's worth a bet.

Baseline today: teams paper over myopia with retries, guardrail heuristics, and human-in-the-loop babysitting — none of which give the agent actual foresight.

Solution · the mechanism

Observe, imagine a few futures, then commit

A lightweight world model — an LLM prompted (or cheaply tuned) to predict "if I take action X, the state becomes Y" — feeds a shallow lookahead. Roll each candidate out one or two steps, score the imagined outcomes against the goal, and only then take the best real action. No forest, no test-time weight updates.

Observe current state Imagine · depth ≤2 · ≤4 branches Commit best real action
Observe → Imagine (fan) → Commit → Observe — the amber path is the branch the score keeps.

The bet, stated so it can be lost: shallow simulated foresight beats reactive ReAct on a real, non-sandbox GUI benchmark at tolerable token cost. The idea is literature-proven (WebDreamer, Dyna-Mind); its in-the-wild reliability is not. That gap is the spike.

Rabbit holes · patched

Where this bet could quietly eat the cycle

World-model accuracy spiral

patchPredict only a coarse textual state-delta ("a login modal appears"), enough to rank actions — never pixels. If coarse prediction can't separate good from bad, the spike has found its answer and the bet dies cheap.

Cost blowup from branching

patchHard-cap depth ≤2 and branch factor ≤4. Tokens-per-task is a first-class result, measured, not an afterthought.

Benchmark theater

patchRequire a live-ish, non-sandbox GUI/web task set — the exact regime the literature says is unsolved. A clean sandbox would flatter the method and dodge the real risk.

No-gos

Explicitly out of bounds