w2g3v
short pre-cutover
An app start too short to pin down a seed. Some have candidate seeds, listed on their pages; their counts are plotted by draw with the draws numbered from the first post (an assumption).
- Period (UTC)
- 2012-07-05 07:09:59 → 2012-07-05 10:10:02
- Posts
- 4 (short: 4)
- First / last post
w2g3v/w2kle- Draws
- Not known: no seed was found, so the posts are numbered from the first post, at draw 0, as consecutive draws (an assumption; the 2012 build starts at draw 0, the 2013 build at draw 1 after its startup draw)
- Posting interval
- 60 min
- Title clock
- UTC−7 at the first post
- Started
- Wed 2012-07-04 23:09–Thu 2012-07-05 00:09 Pacific (estimate: within one posting interval before the first post)
Notes
- 4 hourly posts before runs 11/12; undecided: candidate seeds 70,597,546 / 2,764,093 on a new boot after run 10 stopped, 2 hits vs 0.51 expected; run 10 boot excluded
Possible seeds
Two seeds reproduce all four posts. They fit a boot that the run's neighbours do not show: a reboot after run 10 stopped. Against the chance matches expected there, the two-hit result is not strong, so the run is not counted as reproduced.
The run
Four posts, 5 July 2012, 07:09:59 to 10:10:02 UTC, hourly, with the title clock on UTC−7. The 2012 build draws Next(8, 64) per post, and the first post is draw 0.
| Draw | Post | Posted (UTC) | N |
|---|---|---|---|
| 0 | w2g3v | 07:09:59 | 57 |
| 1 | w2hsk | 08:10:00 | 17 |
| 2 | w2j9u | 09:10:01 | 40 |
| 3 | w2kle | 10:10:02 | 45 |
Why it is hard
Each post has 56 possible values, so four posts carry 23 bits. A random seed matches them by chance about once in ten million, which over all seeds is about 200 chance matches. A useful search needs a window of a few million seeds or fewer.
How it was narrowed
The first post is not on a round time, and in 2012 such first posts came 0.5 to 7 minutes after launch, so the launch is between 07:02:59 and 07:09:59 UTC.
| Step | Constraint | Candidates | Expected by chance | Seeds found |
|---|---|---|---|---|
| Run 10's boot | Run 10 (boot 2 July, 09:51 to 10:51) continued | about 260,000 | 0.03 | 0, so excluded |
| New boot | A boot after run 10's last post (4 July, 09:34:11), launch within 7 minutes | about 5.0 million | 0.51 | 2 |
Run 10's last post was at 09:34 on 4 July, and its next hourly slot, at 10:34, never came. Runs 11 and 12 are on a boot of 5 July, between 09:16 and 10:55, and their first post came at 21:03. So this run sits between run 10 stopping and the reboot of 5 July. Its next slot, at 11:10, would have come just after that reboot, the same pattern as the other short pieces.
The title clock lead drops from about +109 seconds in run 10 to about +61 seconds here, which fits a clock resynchronised by a reboot in between. The titles are only to the minute, so this is a hint, not proof.
The candidates
| Seed | Draws 0 to 3 (posted) | Draws 4 and 5 (never posted) | Boot (UTC) | Reading |
|---|---|---|---|---|
| 70,597,546 | 57, 17, 40, 45 | 35, 17 | 4 July, 11:26 to 11:33 | A reboot about two hours after run 10 stopped (04:26 Pacific, 4 July) |
| 2,764,093 | 57, 17, 40, 45 | 51, 28 | 5 July, 06:16 to 06:23 | A boot 46 minutes before the launch (23:16 Pacific, 4 July) |
Both reproduce draws 0 to 3 against the stream of the .NET generator. The launch is between 07:02:59 and 07:09:59 UTC for both.
Reading. 70,597,546 ties two events together: run 10 stopped after 09:34, and the machine came back at about 11:30. 2,764,093 needs a boot during the night of 4 to 5 July, and another before runs 11 and 12. Both also need the reboot of 5 July that comes after. Neither is counted: two hits against 0.51 expected happens by chance about one time in eleven, and the two cannot be told apart from these posts. Chance matches cluster as well, so the empirical probability is about 0.08.
What could decide it
A second piece on the 4 to 5 July boot, or an independent record of run 10's stop or the restart time.
Counts
Posts
| Post | Title | Posted (UTC) | Draw | N | Mask seed | Status | Note |
|---|---|---|---|---|---|---|---|
w2g3v | 201207050011 | 2012-07-05 07:09:59 | 0 | 57 | 265 | explained | 4 hourly posts before runs 11/12; undecided: candidate seeds 70,597,546 / 2,764,093 on a new boot after run 10 stopped, 2 hits vs 0.51 expected; run 10 boot excluded |
w2hsk | 201207050111 | 2012-07-05 08:10:00 | 1 | 17 | 46 | explained | 4 hourly posts before runs 11/12; undecided: candidate seeds 70,597,546 / 2,764,093 on a new boot after run 10 stopped, 2 hits vs 0.51 expected; run 10 boot excluded |
w2j9u | 201207050211 | 2012-07-05 09:10:01 | 2 | 40 | 515 | explained | 4 hourly posts before runs 11/12; undecided: candidate seeds 70,597,546 / 2,764,093 on a new boot after run 10 stopped, 2 hits vs 0.51 expected; run 10 boot excluded |
w2kle | 201207050311 | 2012-07-05 10:10:02 | 3 | 45 | 640 | explained | 4 hourly posts before runs 11/12; undecided: candidate seeds 70,597,546 / 2,764,093 on a new boot after run 10 stopped, 2 hits vs 0.51 expected; run 10 boot excluded |