Direction C · Play with the upside
Build a better research day.
Spend six upgrades on an imaginary lab. Discover why improving several steps can beat improving only one.
You have six upgrades. Where will you put them?
Try one path first. Then spread the upgrades across the lab.
6 upgrades left
Each token adds 50% of that factor’s starting value. Equal token costs are a game rule, not real research economics.
What this model assumes
The starting lab runs 10 experiments per batch, succeeds 10% of the time, and runs one batch per day. That gives one expected useful result per day. Each of six fictional upgrade tokens adds 0.5 to one starting factor. Output = (1 + 0.5a) × (1 + 0.5b) × (1 + 0.5c), where a + b + c ≤ 6.
All six tokens in one factor give 4 × 1 × 1 = 4. Two in each give 2 × 2 × 2 = 8. This advantage comes from our chosen equal-cost, independent-factor assumptions. Real budgets, diminishing returns, coordination costs, and dependencies can change the best allocation. The optional validation limit caps usable output at five per day.
This is a positive example of a general mechanism, not a simulation of a particular AI model, biological discovery rate, or real laboratory.
How this fits into a Substack post
Save the image above and add it to the post. Link the image, or a nearby “Try it yourself” link, to this page. The image also works in email. Substack does not support custom HTML or iframe code in its post editor. The compact version is for websites that allow iframes.