tea.mathoverflow.net - Discussion Feed (A reputation table bug)

Gerry Myerson comments on "A reputation table bug" (11206)

Gerry Myerson — Thu, 02 Dec 2010 14:56:39 -0800

I couldn't help noticing on the screen shot that started this discussion Anton's reputation is an exact power of 2, as are his gold and bronze badges. He needs 6 more silver to bring him up to 32 (or did, 2 days ago).

Anton Geraschenko comments on "A reputation table bug" (11200)

Anton Geraschenko — Thu, 02 Dec 2010 07:36:05 -0800

It makes me wonder if the bug extends to applying a single upvote to Andy to both Andy and Franz.

You can verify that this isn't the case by looking at their reputation graphs and picking a very small window of time to see exactly where they've been getting their rep.

It seems that Franz has accepted an answer, thereby getting two points. Think of the sharks, Franz!

I figure there's about a dozen of us in this thread -- so about 10 Anton-downovtes each ought to do the trick.

Much more efficient way to do it is to flag posts as spam. A total of 12 spam votes on two non-CW posts (one flag on each from six users) does 224 points of damage. Then you're only 15 more downvotes away.

Cam McLeman comments on "A reputation table bug" (11199)

Cam McLeman — Thu, 02 Dec 2010 06:05:45 -0800

I think we need to make this a more controlled experiment. What we need to do is downvote 254 points from Anton so that he's at the same rep level. I figure there's about a dozen of us in this thread -- so about 10 Anton-downovtes each ought to do the trick.

Gerry Myerson comments on "A reputation table bug" (11197)

Gerry Myerson — Thu, 02 Dec 2010 00:39:09 -0800

Has anyone ever seen Franz and Andy in the same place at the same time?

Will Jagy comments on "A reputation table bug" (11196)

Will Jagy — Wed, 01 Dec 2010 22:37:33 -0800

That is really interesting, as when I just looked, they matched again. This has happened at 7908, then 7928, 7938, now 7948. It makes me wonder if the bug extends to applying a single upvote to Andy to both Andy and Franz. Or whether Franz and Andy each votes up something of the other when they temporarily fail to match. Also there are sharks with lasers.

Hailong Dao comments on "A reputation table bug" (11194)

Hailong Dao — Wed, 01 Dec 2010 21:52:21 -0800

@Andy: so I have to upvote you every time I want to see Franz's profile (like I just did!) ? That's neat (-:

Andy Putman comments on "A reputation table bug" (11185)

Andy Putman — Wed, 01 Dec 2010 17:18:45 -0800

Sshh Anton! Let's keep the plans a surprise. Everyone else will find out soon enough...

Anton Geraschenko comments on "A reputation table bug" (11184)

Anton Geraschenko — Wed, 01 Dec 2010 17:04:34 -0800

[redacted]

Edit: Andy's right. I should've just kept my mouth shut.

Will Jagy comments on "A reputation table bug" (11183)

Will Jagy — Wed, 01 Dec 2010 16:45:56 -0800

Are there going to be sharks with laser beams attached to their heads?

Andy Putman comments on "A reputation table bug" (11182)

Andy Putman — Wed, 01 Dec 2010 16:38:45 -0800

The sinister "phase 2" of my plans now can begin...

Will Jagy comments on "A reputation table bug" (11176)

Will Jagy — Wed, 01 Dec 2010 11:41:02 -0800

Wednesday, December 1, 11:39 Pacific time, the bug is back, Andy Putman has eaten Franz Lemmermeyer.

Edit: yes, I went back and just typed Franz in the User search field, he exists and has the same number of points as Andy.

Harry Gindi comments on "A reputation table bug" (11161)

Harry Gindi — Tue, 30 Nov 2010 11:28:34 -0800

It was Leonid, although I don't remember his last name. It seemed like he just got PO'd about something and left forever.

HMŠ comments on "A reputation table bug" (11160)

HMŠ — Tue, 30 Nov 2010 10:56:22 -0800

@Willie: and if anyone really wants to know who 002 is, there are always searchable MO dumps (searching his ID number in users.xml does the trick). Although I think that wouldn't be too polite (I did it anyway), since 002 left us and returned his points to the bank...

And after Willie's edit and the link given there this post is obsolete too.

Jonas Meyer comments on "A reputation table bug" (11158)

Jonas Meyer — Tue, 30 Nov 2010 10:50:25 -0800

I believe that 002 deliberately got their score to 2. I don't know the full details of how it worked, but I noticed at the time that the user, after deleting all unaccepted answers, community wiki-ing all posts, and awarding massive bounties on questions, for a short time had 1 reputation. My guess is that a downvote was placed, bounties put the reputation to 1, and the downvote was removed to bring the reputation up to 2, but again I don't know exactly how it worked. All of the user's comments were also deleted, the name was changed to 002 (it had formerly been the user's real name), profile information was removed, and the user stopped contributing.

Edit: Oops, this is somewhat redundant after Willie Wong's post.

Will Jagy comments on "A reputation table bug" (11157)

Will Jagy — Tue, 30 Nov 2010 10:50:01 -0800

Live and learn. There is a user name Snuffleupagus or related spelling who posted here, saying that it had answered under various temporary names and changed email addresses so as to avoid the reputation nonsense. As B Conrad expressed simlar sentiments I guess it is a real thing, protesting the system. And user PowerPuff answered my most difficult question, at least to the extent of saying my very strongest conjecture followed from Schinzel's Hypothesis H. So there is at least one out there, possibly pretty famous, who does answer stuff but not only wants to stay anonymous (fairly easy I imagine) but dislikes the points system. 002 could be similar.

WillieWong comments on "A reputation table bug" (11156)

WillieWong — Tue, 30 Nov 2010 10:44:08 -0800

I think I have a vague memory of that individual. He also deleted all his comments. In two of the questions he asked you can find a clue to his identity from the comments other people left.

I also have a vague memory of a discussion on Meta on the distribution of of Rep score for the low numbers (say, < 50). Ah, here it is.

EDIT: And in fact, user 002 was discussed in that thread.

Cam McLeman comments on "A reputation table bug" (11155)

Cam McLeman — Tue, 30 Nov 2010 10:22:03 -0800

Very peculiar about 002! While in principal it's easy to see how to get 2 points (an accepted answer and some downvotes), it doesn't look like that's the case here. Very many questions and accepted answers, but all community-wiki, earning no points. It looks like 002 amassed about 1300 points, then community-wiki'd all of them. Ooh, in fact, his reputation history chart shows him at -500 reputation, so the reputation of 2 was almost certainly an artifice of having an even number of positive reputation points, and having an even number subtracted when he/she wiki'd, and the software refusing to reduce the current reputation to 0 or below.

Will Jagy comments on "A reputation table bug" (11154)

Will Jagy — Tue, 30 Nov 2010 09:34:10 -0800

I have done Bruce a disservice.

In my last comment last night, initially I suggested that we have somebody upvote something of Andy's and see what happened. Then I thought that wasn't funny and tended to curb the illusion, so I edited to what is there now. Then it hit me that "power" was better than "powers" but I was in bed.

I don't expect anything, but while matches at the high end of reputation are not so common, there should be page after page of usernames with rep 1, so I will check that.

EDIT: 100 points currently at about page 45, 1 point starts page 94, in between there are many scores with several pages worth such as 11. No sign of repeats. I didn't see how to go to a page number I type in, perhaps there is none, or none for my browser. One user with exactly 2 points, not sure how to accomplish that. And the username was 002.

WillieWong comments on "A reputation table bug" (11153)

WillieWong — Tue, 30 Nov 2010 09:16:32 -0800

@Will: I don't think that's the time-line. I first saw this thread after Bruce's comment but before Cam's. And I checked the Users page and, lo-and-behold, Andy appears twice, and I thought Bruce was making a sort of in-joke.

Then some minutes later I saw Cam's post and realize that something screwy is going on, and when I went back to look at the Users list, Franz has been revealed. So I am pretty sure Bruce managed to solve the puzzle before it became trivial.

Will Jagy comments on "A reputation table bug" (11152)

Will Jagy — Tue, 30 Nov 2010 08:41:48 -0800

I just logged on, clicked on Users first and saw how Andy and Franz were now separate by 20 points. My guess is that the first (evil) Andy got two votes overnight and the names corrected when the first one happened. Overnight in the sense of California. So perhaps Bruce saw what I saw, but an hour earlier than I did.

It is also possible that Andy and Franz are in a cabal, part of their mission is to vote up Gerry Myerson.

WillieWong comments on "A reputation table bug" (11151)

WillieWong — Tue, 30 Nov 2010 08:30:00 -0800

The second Andy has now been revealed.

@Bruce: how did you manage to figure that out?

Cam McLeman comments on "A reputation table bug" (11150)

Cam McLeman — Tue, 30 Nov 2010 07:58:47 -0800

Wow, nice catch!

So I guess the software has a bug if two people on the cusp of the page-turn have the same reputation?

@Will: Identity theft, in this case Franz', is a pretty good sign of evil.

Bruce Arnold comments on "A reputation table bug" (11149)

Bruce Arnold — Tue, 30 Nov 2010 07:40:49 -0800

The real name of the second Andy is Franz.

Will Jagy comments on "A reputation table bug" (11139)

Will Jagy — Mon, 29 Nov 2010 22:09:30 -0800

Later Edit: how do we know he is using his power for evil?

Gerry Myerson comments on "A reputation table bug" (11138)

Gerry Myerson — Mon, 29 Nov 2010 21:56:37 -0800

On a mathematics website where I have more reputation than Richard Stanley, Andreas Blass, and other notables, anything is possible.

Ryan Budney comments on "A reputation table bug" (11137)

Ryan Budney — Mon, 29 Nov 2010 21:50:18 -0800

Andy might have two accounts with the same name and exactly the same rep. He keeps one on the 1st page and another on the 2nd, so that it's difficult to spot his evil genius.

Will Jagy comments on "A reputation table bug" (11136)

Will Jagy — Mon, 29 Nov 2010 21:20:21 -0800

His CV has his telephone with a 510 area code. I think that's suspicious.

theojf comments on "A reputation table bug" (11135)

theojf — Mon, 29 Nov 2010 21:12:56 -0800

I'm getting the same thing. So the effect is at least 45 minutes robust. It only seems to be this one case on pages 1-2 --- e.g., Jim (rep=5031, 5 silvers, 28 bronzes) does not appear at the top of page 3.

Ryan Budney comments on "A reputation table bug" (11134)

Ryan Budney — Mon, 29 Nov 2010 21:07:16 -0800

Andy Putman is a quantum topologist.

Will Jagy comments on "A reputation table bug" (11133)

Will Jagy — Mon, 29 Nov 2010 20:43:58 -0800

I don't believe it is a bug in the usual sense. My first guess is that the reputation table is not Hausdorff. There is a similar example in Munkres Topology about how the Hausdorff condition really is needed in the definition of a smooth manifold.

Also, we cannot be sure Andy Putman is Hausdorff.

Andrey Rekalo comments on "A reputation table bug" (11132)

Andrey Rekalo — Mon, 29 Nov 2010 20:23:10 -0800

I wonder if something similar has been observed before. As I'm posting this, Andy Putman features simultaneously on pp. 1 and 2 of the reputation table

http://img502.imageshack.us/img502/2016/page1i.png

http://img213.imageshack.us/img213/4333/page2wh.png