Is this working?
The prediction is falsifiable: a post published a year ago should still be receiving a meaningful share of its reads. If it is not, the correct response is to stop and rebuild — not to add features. These are the four criteria, with today's numbers beside them.
Post half-life beats a conventional blog
Target: Meaningful share of reads at 12+ months.
36.7% of reads (22/60)
30% of posts revised within 60 days
Target: ≥ 30%.
0.0% (0/6 posts)
Marginalia loop forming
Target: ≥ 1 annotation per 200 reads.
11.67 per 100 reads (7 notes)
Memory compounding
Target: ≥ 20% 30-day return.
0.0% (0/37 readers)
Reads here mean read receipts — one per reader per post per version — not block impressions, which would flatter every rate by an order of magnitude. If criterion 1 or 2 fails at month 6, the plan says to stop. It is written here so the decision is not made by sunk cost.