Probable run 2 (seed 377,361,921)

One seed reproduces these four posts. On the boot shared with runs 18 to 22 it is the only seed that fits, with the launch 0 to 3 minutes before the first post and on the 15.625 ms tick grid.

The run

Four posts, 1 August 2012, 04:02:36 to 04:35:41 UTC, eleven minutes apart, with the title clock on UTC+8. The batch was opened by the status marker xhllj, one second earlier with the same title. The 2012 build draws Next(8, 64) per post, and the first post is draw 0.

Draw	Post	Posted (UTC)	N
0	xhllk	04:02:36	56
1	xhm7s	04:13:37	8
2	xhmuh	04:24:38	42
3	xhnhp	04:35:41	56

The search

1. No known stream contains the batch. Every count stream already identified was checked for a continuation, and none fits.
2. The boot. Runs 18 to 22 share the boot of 27 July 2012. Runs 21 and 22 have first posts that are not on a round time, so their launch is close to the first post, and they put the boot at 19:13:27 and 19:13:07.5. The boot is therefore no later than 19:13:08. The search allows a few minutes of launch delay, from 19:06 to 19:13:08. This batch lies between run 20 and run 21, on that same boot.
3. The search. Over every seed, a window from that boot up to three posting intervals before the first post gave one seed on the tick grid.

Hypothesis	Candidates	Expected by chance	Seeds found
27 July boot, launch within 7 min	about 54,000	0.006	1 (377,361,921)
27 July boot, launch within 33 min	about 150,000	0.016	1 (the same)
Any boot after run 20 (31 July, 04:19), launch within 7 min	about 5.5 million	0.56	1, at chance level

The alternative, a reboot between 31 July and 1 August that the neighbouring runs do not show, gives one hit at chance level. The 27 July boot is the one the neighbouring runs use.

Checks

• Draws 0 to 3 of new Random(377361921) under Next(8, 64) are 56, 8, 42 and 56: the four counts above.
• The seed is on the 15.625 ms tick grid.
• The launch is 04:00 to 04:03 UTC, 0 to 3 minutes before the first post.
• Draw 0 is a post, as in the 2012 build.

It is counted as a reproduced run, but as a probable one. The chance of a fit this good in the narrowest window is about 0.6 %, and about 2 to 4 % in the wider windows the boot still allows.