The same people, org tree and luck, simulated twice: once with your institutions (left) and once with institutions repaired — the benchmark. Everything below compares the two. How to read this →
Over time
Operating yield by year
How ICs spend their time real maintenance impression compliance
Engagement
Rules in force
Tech debt
CEO over-belief
Incidents per month
Leavers per month
your company repaired benchmark (dashed)
Where operations lose value
The org map
Ideas
Every idea that arose succeeded failed vetoed
The funnel
Signals, resources and people
Why: which institution causes how much of the loss
Runs all 128 combinations of the seven institutions broken or repaired, on the same world.
What would help
Eight realistic, partial fixes with running costs, one at a time and all together.
Decisions, term by term
The model
Nine assumptions
People are never changed between runs: the same ability, political skill, alignment and tribe in your company and in the benchmark. Only institutions differ. Everything else follows from these assumptions.
How value is produced
Each IC, each month: ability × engagement × real-work time × steering × debt multiplier × market × budget.
Ideas add heavy-tailed payoffs minus their cost; incidents subtract. Output adds up linearly across people; there are no prices or revenue.
Each year
Each month
Not in the model
orgsim v1.1 runs entirely in your browser (Web Workers, no server). It is a JavaScript port of the Python reference, with a different random-number generator, so numbers differ slightly from the Python output but not in distribution. Designed by merging independent prototypes by two AI models, Astra and Fable, then revised after their reviews.