Probable run 1 (seed 293,378,646)

One seed reproduces this five-post burst in the only window the timing allows. Even if the launch is allowed up to an hour before the burst, it is still the only one.

The run

Five posts within 20 seconds, on 1 July 2012, 02:49:29 to 02:49:48 UTC, with the title clock on UTC−4. It is a burst test, not a schedule. The 2012 build draws Next(8, 64) per post, and the first post is draw 0.

Draw	Post	Posted (UTC)	N
0	vv2q4	02:49:29	10
1	vv2qe	02:49:34	39
2	vv2qq	02:49:39	63
3	vv2qx	02:49:43	37
4	vv2r5	02:49:48	20

The search

1. No known stream contains the burst. Every count stream already identified was checked for a continuation, and none fits.
2. A full search. Every seed from 0 to 2³¹−1 was tested against the first three draws. Five seeds reproduce them. The first search used a boot window too narrow to include the right one. A wider window, which allows for a machine whose clock had folded (see step 3), leaves exactly one: 293,378,646.
3. The boot is pinned. Two other runs on the same machine, vn92j and vsb20, were posted at launch. Their titles put that machine's boot at 19:12:40 on 15 May 2012, 83 ms apart over 2.5 days. The clock had folded: the machine had been up for 46 days, and the seed counts the uptime from 2³² ms minus the seed. Against that boot, the first posts of the other runs come 0.5 to 7 minutes after launch, which is the delay used here. The seed puts the launch at 02:45:48 UTC, 3 min 41 s before the burst.

Hypothesis	Candidates	Expected by chance	Seeds found
Folded 15 May boot, launch within 7 min	about 420,000	0.001	1 (293,378,646)
Folded 15 May boot, launch within an hour	about 3.6 million	0.007	1 (the same)
14 June boot, launch within an hour	about 4.0 million	0.007	0
A new boot after run 09 (29 June, 23:31), launch within an hour	about 98 million	0.18	0

Checks

• Draws 0 to 4 of new Random(293378646) under Next(8, 64) are 10, 39, 63, 37 and 20: the five counts above, exactly.
• Draw 0 is a post, as in 39 of the 40 runs of 2012.
• The seed's count stream does not contain the later vvjmg run.
• A chance fit in this window is about one in 500. Allowing for the several boots and launch windows that were tried, the chance of a coincidence is about one in 100.

It is counted as a reproduced run, but as a probable one: the seed rests on the boot argument above, not on a match across many posts.