Combined Stress Simulation

Several stressors at once: constant friction, slow accuracy drift, occasional bad approvals, and forced scaffold removal every eighth session.

PurposeSee whether a simulated learner’s mastery still rises when several stressors are applied together.

StatusCannot fail — mastery is computed from a fixed formula that the stressors barely touch.

Sample40 simulated sessions, seed 42.

Touches NCFCA codeNo. A self-contained numpy simulation.

AttributionCode AI-generated (tool not recorded). Run in the lab; results unchanged.

ReviewedBy no outside party. The reading in “What it shows” below was done when this page was built, directly from the code and CSV.

What the code does

Each session, mastery rises by 0.06 × (1 − 0.5 × scaffold), where scaffold is 1.0 except every eighth session, when it is 0.0 and the step is multiplied by 1.2. An 8% chance of a “bad approval” multiplies the step by 0.7. Observed accuracy is drawn around 0.85 minus a slow drift plus noise, and feeds only the “effective baseline” column.

What the data show

Mastery climbs by 0.03 a session (0.072 on forced sessions) and reaches 1.0 at session 31. In the CSV it is 0.264 after session 8, 0.537 after session 16 and 1.0 from session 31.
The drift, the noise and the effective baseline never feed back into mastery, so the stressors named in the title cannot change the outcome. Only the bad-approval draw does, by delaying the finish.
The closing line, “System remained functional under heavy multi-vector stress”, is printed unconditionally. No check could make it fail.
Observed accuracy is never below 0.66 and the final effective baseline is 0.817.

What it shows

That the formula behaves as written: a steady climb, a faster step when support is withdrawn, and a small delay from bad approvals.

What it does not show

Robustness of any real component. For that, the simulated stressors would have to act on the quantity being measured, and the pass condition would have to be stated before the run.
Anything about the four-channel architecture.

Download (unchanged from the lab’s run): final_combined_stress_test.py · final_combined_stress_results.csv
The scripts write to a path from the lab’s own machine (/home/workdir/artifacts/); change it to run them elsewhere. Each uses seed 42, so a rerun reproduces the CSV.