QUOTE(Eva Destruction @ Wed 13th May 2009, 4:36pm)
![*](style_images/brack/post_snapback.gif)
QUOTE(thekohser @ Wed 13th May 2009, 4:46pm)
![*](style_images/brack/post_snapback.gif)
He goes on to clarify, "OK, time of least disruption. That would be around 2 to 5 am EST. I have found that few edits are made during those periods."
Sounds reasonable to me, but Iridescent jumped on him as some kind of American swine. Really, Iridescent? Isn't it possible that Wikipedia does, indeed, have a period of least use?
It has a time of least use, by definition, but I doubt it's statistically significant. I'd seriously question whether "least use in America" equates to "least use globally" even on en-wiki, and would very strongly suspect that the reason he sees fewer changes at 0200-0500 EST is that his watchlist is by definition going to be biased towards subjects that interest him,
which, if his list of interests is anything to go by, are all on Virginia and thus will be most heavily edited when Virginians are active. Someone with their main interest in Indian history, say, will see completely the opposite pattern; someone like me, with interests in Nevada, NY and England, sees a fairly constant "density of changes". (Also remember, those servers cover all the languages – the time of least use on en-wiki will be the time of peak use on Chinese Wikipedia and Russian Wikipedia, too).
What does "statistically significant" mean in this context? Have you looked? I bet it's a relatively stable difference,
as it was in 2002. There are just more English-speakers on the hemisphere defined by the U.S. through England. Most of this clump is asleep/busy between about 430-930 UTC, especially the 145 M or so in the Eastern Time Zone. That's a good chunk of the world's English speakers, and considering that much of the English-speaking population resides in places with relatively low internet connectivity (Nigeria, India), there's a visible difference in most site's traffic. The Aus/NZ people pick up some of the slack, but there's a hell of lot more wired English-speakers on this side than theirs.
I studied 3627 accounts with between 1000-2000 edits in 2007 (Alanyst supplied the data). This group will be skewed toward people with better-than-average connectivity, but it's not a terrible approximation. Here's how editing broke down in half-hour chunks:
00:30 118986
01:00 112178
01:30 109202
02:00 108120
02:30 103555
03:00 104917
03:30 101280
04:00 97680
04:30 93538
05:00 88560
05:30 84079
06:00 79893
06:30 74081
07:00 70112
07:30 67648
08:00 64384
08:30 63604
09:00 61723
09:30 64020
10:00 64468
10:30 66155
11:00 67090
11:30 69164
12:00 73177
12:30 79655
13:00 83857
13:30 88869
14:00 95249
14:30 99334
15:00 107948
15:30 112925
16:00 118651
16:30 120670
17:00 124953
17:30 125176
18:00 128772
18:30 128903
19:00 130397
19:30 132664
20:00 135841
20:30 134375
21:00 137407
21:30 133877
22:00 132343
22:30 128386
23:00 126125
23:30 121800
00:00 119391
Considering other languages, I think the low-period would shift a little more toward continental Europe, but I would bet a lot of money that there's a stable low period, and I would not be surprised if it was between 2-5 AM EST.
Or we could all be American swine. Oink.