Good statistics, Give Cool Hand Luke a Barnstar Options

[Poetlister]

Just to prove that it is possible to apply good statistics to alleged sockpuppetry, consider this work by Cool Hand Luke:



http://en.wikipedia.org/w/index.php?title=...oldid=193665402

http://en.wikipedia.org/w/index.php?title=...oldid=193779572

Basically, this shows that the comparative editing patterns of Runcorn, myself and others are not typical of those of a group of sockpuppets.

* mod note: Fixed image URL - Nathan

[One]

The way this is written, it seems like Cool Hand Luke really wishes that all of the Runcorn accounts were a match in a similar way so that this methodology could retroactively endorse cabal-approved blocks. But they don't, so he has to offer explanations to placate Durova:

So what gives? One possibility is that these accounts edited from different computers at different locations through the day. For example, a work computer could have been used typically for the accounts that peak in the daytime, while a the Runcorn account might have been used from home, explaining why it peaks in the evening. Separate computers would also explain how these account could edit while continually evading checkuser evidence. (It's also possible that some of these accounts weren't sockpuppets, but that's not relevant here.)

It's "not relevant here." Heh. I think this is one of the reasons that ArbCom can't bless any kind of statistics. More rigorous scrutiny could undermine a lot of blocks, especially the casual duck tests that go on every day.

It's like when George W. Bush opposed DNA tests for the executed in Texas because he already knew that they were all guilty.

This post has been edited by One: Sun 9th March 2008, 10:30pm

[Poetlister]

(It's also possible that some of these accounts weren't sockpuppets, but that's not relevant here.)

If this is meant to be a demonstration of how to use editing patterns to prove that people are or aren't sockpuppets, it's pretty breathtaking to suggest that the issue of whether they are indeed sockpuppets is irrelevant!

[Milton Roe]

Just to prove that it is possible to apply good statistics to alleged sockpuppetry, consider this work by Cool Hand Luke:

Basically, this shows that the comparative editing patterns of Runcorn, myself and others are not typical of those of a group of sockpuppets.

And by eyeball, you can see that Runcorn gets up 3 hours earlier in the AM than any of his purported socks. That's just not credible. If a man is going to edit under ONE name just after getting up in the AM, I can't see any reason for him never to use a sock first thing, either. Of course, from here on out after they know this kind of analysis is available, the puppeters are going to get very tricky about this. But this is in the past. Pretty open and shut not-guilty.

Milt

So what gives? One possibility is that these accounts edited from different computers at different locations through the day. For example, a work computer could have been used typically for the accounts that peak in the daytime, while a the Runcorn account might have been used from home, explaining why it peaks in the evening. Separate computers would also explain how these account could edit while continually evading checkuser evidence.

The "computer at different location = work" theory is easy to check by looking differentially at weekend edits. If the guy only edits socks at work, or some non-home location, you'll see dropouts for certain days of the week. The home account is the one used every day of the week, since just about everybody is home on any given day of the week, over time. But that *usually* isn't true of work, and if it is, there's a big difference in frequency (says Milton, while posting this from a non-home work computer on a Sunday afternoon Never say "never.").

-- Milt

[Somey]

If this is meant to be a demonstration of how to use editing patterns to prove that people are or aren't sockpuppets, it's pretty breathtaking to suggest that the issue of whether they are indeed sockpuppets is irrelevant!

I don't mean to excuse Mr. Luke's rather callous "it's irrelevant" remark, but I think what he's trying to say with all that graphing and analysis is that while dissimilar editing times don't disprove sock puppetry, similar editing times are much more indicative of it.

It's not an unreasonable stance, really... It's just that he happened to choose the wrong case as an example, since we essentially know exactly what was "going on" with those particular accounts. They're just unwilling to admit that ours is the most logical and rational explanation, and have focused on one or two supposed "gotcha" edits that are fairly easily explainable in order to ban the lot of them - and really, they're doing that mostly to get payback on Poetlister for being a "WR person." Nothing any of those accounts did on WP was damaging in any way, or even non-constructive.

Cool Hand Luke is a decent admin, for the most part... And I don't mean to single him out, but I just think it's sad when people put so much effort into producing "evidence" and documentation about something that's this obvious and clear-cut. The Poetlister ban is revenge, pure and simple. The Mantanmoreland non-ban is head-in-sand wagon-circling, pure and simple... What's difficult to understand about this?

[Yehudi]

The "computer at different location = work" theory is easy to check by looking differentially at weekend edits. If the guy only edits socks at work, or some non-home location, you'll see dropouts for certain days of the week.

Both Newport and R613vlu edited on Saturday evening and Sunday but not Friday evening, Saturday or festivals. That is not true of the others. This is of course consistent with these two being Jewish.

[KamrynMatika]

Is it me or would this be incredibly easy to get around? If I had two socks, I'd just edit in the morning with sock one and then in the evening with sock two. Instant alibi?

[Poetlister]

Is it me or would this be incredibly easy to get around? If I had two socks, I'd just edit in the morning with sock one and then in the evening with sock two. Instant alibi?

Nobody can ever prove that two different accounts aren't sockpuppets. (Indeed, it seems that in some circumstances nobody can ever prove that two different accounts are sockpuppets.) All we can do is establish a clear balance of probability one way or the other by the accumulation of evidence.

[Brutus]

Poetlister,

Did you enter all that data in manually? ...if so u must have plenty of spare time.

Or is there software around that automatically converts users edit history into a comparision chart?

[Derktar]

Poetlister,

Did you enter all that data in manually? ...if so u must have plenty of spare time.

Or is there software around that automatically converts users edit history into a comparision chart?

Actually that's the work of Cool Hand Luke, he did that during the Mantanmoreland RFAR.