Exception handling

Exception handling describes what a system does when the evidence is incomplete, contradictory or outside the expected range: carry on as if nothing is wrong, or explicitly flag the case for a human. A system without exception handling treats every case the same, even the ones where it does not actually have a good answer.

What it means for the calculation

In an estimate this is the difference between a system that gives a number anyway when in doubt, and a system that says: I am not sure about this, look at it yourself. The second is slower per case, but prevents an uncertain estimate from looking like a certain one.

Where this comes up

In any moment where evidence is missing or conflicting, and in the choice to make that visible instead of hiding it behind a seemingly certain answer.

Common mistakes

Giving an unclear case an ordinary answer anyway because handling an exception takes more work, so the uncertainty becomes invisible instead of resolved.

Is exception handling the same as an error message?

Not necessarily; it can also be an explicit "unknown" or "check this" instead of an error that stops the process.

Why does this matter for an estimate?

Because a silent wrong answer is more expensive than a visibly uncertain one.

How does Blake handle an incomplete or conflicting case?

Blake is designed to surface incomplete or conflicting evidence for the estimator to review, rather than silently filling the gap with an assumption.