How we calibrate hour caps
Every task on colift has a Maximum Allowable Time (MAT). This is how the cap gets set, and how it stays honest.
The standard
The cap is calibrated to the observed median of real, quality-passing volunteer sessions for that task. Not the founder's intuition. Not an AI's a priori guess. Not a number engineered to make a month's requirement easier to hit.
The process:
- Seed an initial cap from a task decomposition. Reading load (words divided by reading speed), data-entry items, output length, required interactions.
- Collect actual measured engagement times from real volunteers who completed the task and whose deliverable passed validation.
- Set the cap at or near the median of those quality-passing sessions.
- Recalibrate quarterly, with a written changelog.
The cap is a ceiling, not a target
The cap can only pull a credited number down. If you finish a task in less than the cap, you get credit for your actual measured engagement. The cap exists to catch outliers, not to award everyone the maximum.
Why we do it this way
It is honest. We certify time that real volunteers actually spent.
It is defensible. A written, dated methodology calibrated against real session data is the strongest answer to a reviewer asking how a per-task number was set.
It protects everyone. The volunteer isn't certified for hours they didn't work. The certifier isn't signing inflated numbers. County acceptance depends on the records being credible.
What we don't do
We don't "estimate on the high side" to be generous. Generosity here would be false attestation.
We don't set caps as a multiple of an estimate (max = 2× est). The cap is empirical.
We don't size task content to extend the cap.
Transparency
The methodology document, current caps, and changelog are published. (See Audit & methodology ledger.)