Treat bad lists like game bugs

Common game bugs:

1. Empty lists — no rounds to score; accuracy would divide by zero.
2. Mismatched lengths — zip silently stops at the shorter list, so leftover truths or scores vanish.
3. Threshold outside 0..1 — a cutoff of 1.5 means almost nothing ever counts as a yes.
4. Reporting accuracy alone — hides whether you got false alarms or misses.

These are input and rules bugs. Fix them before you accuse the guessing logic of being “random.”

Always keep one known-good run so you can tell “still broken” from “fixed”:

truth = [1, 0, 1, 1, 0]
scores = [0.8, 0.3, 0.6, 0.9, 0.2]
threshold = 0.5
preds = [int(score >= threshold) for score in scores]

pairs = list(zip(truth, preds))
tp = pairs.count((1, 1)); fp = pairs.count((0, 1))
tn = pairs.count((0, 0)); fn = pairs.count((1, 0))
accuracy = (tp + tn) / len(pairs)
print(preds, f"accuracy={accuracy:.2f}", tp, fp, tn, fn)

Expected evidence:

[1, 0, 1, 1, 0] accuracy=1.00 3 0 2 0

If your “bug fix” changes this healthy output, you broke the working path while chasing a failure.

In your notes, store a tiny mutant:

• truth = [] or scores = [] → refuse to compute accuracy.
• truth length 5, scores length 4 → refuse or warn before zip hides a round.
• threshold = 1.5 → reject as out of range.

Paste the failing message into failure-before.txt. After you add a guard, paste failure-after.txt. Retries do not help a bad list — retrying empty input just fails again.

A good fix is small and readable: check lengths, check ranges, then run the scoreboard. After repair, rerun the healthy fixture and confirm the expected evidence line still appears.