# The Oversight Simulator Source: https://dinand.com/experiments/oversight-simulator/app/ Author: Dinand Tinholt > Set delegation depth, reviewer capacity and verification coverage, pick an irreversibility tier, and watch the lab's published curves answer, including how long a fault runs at a given reviewer budget and review cadence. A simulation from published coefficients, not a deployed system. Instrument · one set of controls, four published studies # The Oversight Simulator Set how deep your agents delegate, how many reviewers stand behind them and what share of the work gets verified. The lab's measured curves move as you go. What this is · simulation from published coefficients, not a deployed system Every curve on this page is a number the lab has already printed, wired to a control. Panel 01 multiplies your lever changes by the slopes in *Pricing Agent Autonomy* and adds them, which gives a change from wherever you start. The article prints no intercept and no intervals, and the panel shows neither. Panel 02 reads two studies. Its upper half takes a reviewer budget and a review cadence from [the Schouw paper](https://dinand.com/research/when-watching-every-hour-starts-to-pay/). Below it, the Trial Balance audit desk. Panel 03 shows what SIGIL counted about the errors a blocking reviewer can reach. Each panel computes from its own study alone. All of them read the same settings and report to one scoreboard at the foot of the page. 01 · Delegation Cliff coefficients ## What the levers do to the quality margin Modelled change in the quality margin 0.00 *Each bar is your change in that lever times its coefficient. The pale bar behind it is one full increment. A negative total spends margin and a positive total buys it. The other nine levers in the article stay at zero here. Pricing Agent Autonomy, numbers table · linear-probe coefficients, Sobol sample, band-only, n = 4,948 · unit: band-standardised increment* Show the numbers ### Each lever alone, the other two held where you set them *Every strip shares one vertical scale, so a steeper line is a stronger lever. Because the article fits straight slopes, each sweep draws a straight line and leaves out the interactions it mentions. Pricing Agent Autonomy, numbers table · coefficients −1.499, +1.456, +0.308* Show the numbers 02 · Schouw cadence, reviewer budget ## How long a fault runs before someone catches it The Schouw chart loads with the page script. ### The audit desk Published on 1 October. Its budget counts tokens per audit. It has its own axis and its own two controls, and nothing else on the page reads them. Reading budget, tokens per audit 1,600 Diary length, requisitions A Trial Balance for Agent Omissions, figure 3 · threshold from section 4 · 24 audits per cell *Share of injected omissions named correctly, 24 audits per point, one auditor model. The dashed line is the 0.80 threshold the paper uses. Click or press Enter on a budget to pick it. A Trial Balance for Agent Omissions, figure 3 · synthetic seeds 1 and 2* Show the numbers 03 · SIGIL irreversibility tiers ## How much of the worker's error a blocking reviewer can reach A blocker stops approvals. At each tier SIGIL counted how many of the worker's errors were wrongful approvals, the kind a blocker can stop, and how many were wrongful rejections. *Each bar splits the unsupervised worker's errors at one tier into the two kinds. Hover or focus a row for the counts behind its share. Tier 3 produced no worker errors. Click or press Enter on a row to pick its tier. SIGIL paper, figure 2 · convex campaign, pooled over three seeds · net values from the SIGIL executive piece, the one table* Show the numbers Scoreboard ## Your settings, read back --- Site guide for AI agents: https://dinand.com/llms.txt