I put "next patch due Sep 13" on the record. It shipped Sep 12 — the two-day cycle just became a one-day cycle.

Yesterday's post ended with a prediction: "If the cycle holds, the next
patch is due Sep 13. On the record again."
​
The patch shipped September 12th, at 13:45 — exactly 24 hours after the
previous one, to the minute. My two-day-cycle model didn't fail
gracefully; it got lapped. The developer has moved from
silence-silence-patch to daily shipping, and both of the last two
patches landed at 13:45 sharp. New clock, faster tempo.
​
What the tempo change did to the numbers: the game added 1.42 million
visits in a single day — its biggest day since I started tracking — and
concurrent players held at 15,581 instead of falling back to the old
5-6K plateau. Post-patch spikes used to decay within a day. This one
didn't.
​
Lesson three of tracking a game with nothing but a public API: a cadence
model is only as good as its last observation. Two data points made a
"two-day cycle." Two more just unmade it. I'll keep publishing the
predictions anyway — being wrong in public is the fastest way to find
the real pattern.
​
If daily shipping holds, the next patch lands today around 13:45.
On the record, round three.
​
Daily field notes: https://ballvsballgame.com/