The Association Bonus

One new feature that's come along with the transition is the "association bonus", which allows members of Math Stackexchange --- and other sites that are much further removed from the mission of MO --- to cast votes here immediately without having asked or answered any questions. (As I understand it, these people receive a "bonus" of 100 points which is enough to enfranchise them.)

This, I think, has the potential for a great deal of intentional or unintentional abuse.

For example: One user received an association bonus for participation on Math Stackexchange, where his/her highest voted question begins

In all applications of the theory of sets, all sets under investigation take place in the context of the universal set $U$.

His/her other questions include "What is the difference between an internal category and a subcategory?", "Does every category have a functor?", a request for an explanation of why the statements $A\cup B=B$ and $A\subset B$ are equivalent, and various questions about evaluating the sorts of integrals that pop up in freshman calculus. He/she has never participated in MO in any substantive way (zero questions, zero answers, zero edits, etc.), but has nevertheless cast 28 votes here in the past week, and 46 in the past month. That's a lot of votes, and I think it's a fair guess that they're not terribly well-informed.

A bigger potential problem is that this creates a gateway for unscrupulous users to create new identities, participate on sites where they can earn reputation points simply by posting calculus problems, and then enter MO empowered to vote themselves up. There have certainly been cases of users creating fake identities for this purpose, but at least in the past they needed to ask or answer a real question or two along the way; now that obstacle is removed.

I realize we are not currently overrun by barbarians at the gates, but as a veteran of Usenet, I also believe that eternal vigilance is the price of sustained quality on the Internet --- and I think the association bonus is a significant problem.

Edited to add: Lest anyone think the above example is unique, there's another user, again with no questions or answers on MO --- or at least none that have survived --- who's recently asked on MSE how it's possible to make sense of an ordered pair containing the empty set as a component, who has also recently asked on MSE how the same integral can evaluate to both $g(t)+C$ and $g(t)-g(t_0)$, and who has, thanks to the association bonus, cast 34 votes on MO this month.

• Is there any way for us to get data on this? What percentage of votes are cast by users with 101 rep? Jul 11 '13 at 7:15
• @Noah: Once MO has public data dumps on the Data Explorer, someone with basic SQL capabilities can try and write a query which pulls out people with zero activity and a lot of votes (or with little activity and a lot of votes).
– Asaf Karagila Mod
Jul 11 '13 at 9:36
• @AsafKaragila Voting is anonymous and the data exposed through the Data Explorer doesn't include user ids for upvotes/downvotes.
Jul 11 '13 at 15:27
• @Anna: I assume it does include the vote count, though. It shouldn't be difficult, methinks, to searc for users that didn't ask, answer or edited but have a positive vote count.
– Asaf Karagila Mod
Jul 11 '13 at 15:49
• @AsafKaragila Ah yes, you're right. My bad.
Jul 11 '13 at 15:59
• A meta-comment. For 'feature-requests' that are either sure to be controversial, or difficult to persuade SE of the value of, could we begin with a [discussion] tagged posted, and then a subsequent [feature-request] post that takes into account the discussion and makes a strong and balanced case for the software change? I think we'll do much better at persuading SE to help us out if we follow a process like this. Jul 11 '13 at 17:02
• I've taken the liberty of retagging here. Jul 11 '13 at 17:04
• Adding a bit more data: The user in question is now up to 49 votes cast on MO this month, and has posted a question to MSE asking why $0$ is not an element of the set $\lbrace 1,1/2,1/3,\ldots\rbrace$ Jul 12 '13 at 1:14
• I think voting up or down questions influences the scope of the site in a similar fashion as votes to close or reopen a question. Thus I think letting new users vote quickly may be problematic for the site. Also, voting is not really necessary for using the site in terms of asking and answering questions and leaving comments. It is rather a means to help maintaining a certain standard, and to provide feedback on what colleagues think about a particular question or answer. Both purposes are served the better the more knowledgable voters are on average.
– Stefan Kohl Mod
Jul 12 '13 at 16:31
• I think I recall that there was a suggestion that the upvotes should have different weight based on the reputation of the user casting the vote. (I cannot find a link now, I think that it might be at meta.math.SE.) This might help with your barbarian problem. But I doubt such a feature can be implemented - it is too radical change compared with the system used so far. Jul 14 '13 at 8:29
• More data: The user cited in the post has just cast another 3 votes on MO. His/her most recent posting to MSE asks whether an empty set can have a non-empty collection of subsets. Jul 15 '13 at 15:24
• Perhaps in the end we need to introduce the guideline that upvotes should be accompanied by comments.
– user9072
Jul 20 '13 at 19:50
• When viewing users, on the voters tab you can see reputation and number of votes by users (for the last week/month/quarter/year/all time). Of course, a user having low rep need not necessarily be a "barbarian at the gates" (to use the terminology from the OP). Jul 23 '13 at 7:43
• I have just received an Association Bonus on MO for no apparent reason, i cannot check why i've got it (maybe because i've enabled email notifications with my email used elsewhere on SE). I do not understand the purpose of it. A reputation is a reputation, association bonus is not a reputation. Jul 26 '13 at 14:56
• Not that I think this is necessarily a realistic proposal, but it sounds like valuing the vote of someone with $p$ reputation points by an amount proportional to $\log(p)$ would allow more experienced individuals to have a bigger influence on the site's valuing of question and answer quality without drowning out the opinions of inexperienced users. May 9 '14 at 13:05

I'm a 101-rep user on MO. I created an account because it's very easy to do so when you go to MO.

I've never voted and never will, as there's no way I could understand enough to do so intelligently.

I suspect that it would be a lot easier to change the privilege levels (which, in fact, already vary from site to site) so that "vote up" is 125 on MO, the same as "vote down".

I can't see that the other privileges granted to us 101s are likely to be harmful.

• Just curious, do you expect to have any active participation in the site except this meta thread?
– Asaf Karagila Mod
Apr 25 '14 at 18:36
• No. I'm a bit of a meta wanderer, and was trying to help. Apr 25 '14 at 18:37
• I see. Thanks...
– Asaf Karagila Mod
Apr 25 '14 at 18:37
• I'm like you. Oh, new SE site? Create an account. It doesn't matter if I have no understanding of the topic whatsoever: I just always log in on SE sites, and if that means creating a new account then so be it.
– TRiG
Apr 25 '14 at 21:30
• I also suspect that this is the simplest approach to deal with this problem (if it is a problem), so I've upvoted it ... which I probably wouldn't be able to do if it were implemented.
– SamB
Feb 2 '15 at 21:29

I expressed my view against the association bonus being granted on MO already in older discussion and upvoted the feature request.

However, some points to make my view more precise:

1. I agree the main issue is voting, so one could also ask for less of a reduction, that is just no voting with association. In fact this request exists on meta.SO presented by Mad Scientist about a month ago. It could also make sense that those that care about it, voice there opinion there too, or at least vote on it. (I did not yet myself, but I think it is as simple as just login in as on MO and then one can vote due to the association bonus, somewhat ironic, though not quite as the meta.SO is a special case for a site.)

2. For some other things I now think it is somewhat reasonable. I did not yet say anything on math.SE but created an account recently, the actual motivation was to test something as a low-rep user somewhere, but also I could imagine to say want to say something on a question after it was migrated there, and then it would feel odd if I could not comment there, so there is some sense to this association thing, too. (Sure I could get the points otherwise there too, but if I want to leave a comment "now" I would be not happy if I could not.)

3. For sock-puppet voting I see less of a problem. Since for somebody to get something out of having sock-puppets here to vote the need at least to be in some sense in touch with the subject of the site. And, for those it is and also was never a real issue to ask some questions also on MO that would get them two upvotes total and not too many downvotes. (This is according to old rep, the new one is rather harder regarding this, so it is not all bad the changes.) And then we had also some that had some "colleague" ask a question that they then answered, so I would say it was always easy enough.

4. However, I still agree that for voting in general it is a bit of a problem, as it likely will still give more votes to the soft-ish things that already get too many votes, IMO. But then this is also not a big problem in my opinion, yet it is still a problem although even if only a minor one.

• I just registered to upvote Mad Scientist and found out that in addition to +100 on meta.SO I got +100 on Mathoverflow, too. Not sure if that should occur, although I won't complain since I lost much more rep than 100 when we changed from MO1.0 to MO2.0. Jul 11 '13 at 15:37
• I want to emphasize that I consider the problem to be pretty much confined to hot questions that receive a lot of external visitors. I don't think 101 rep users voting has any significant effect on the site at large. The post score is anyway a rather noisy measure, it can be skewed in a lot of different ways.
– user35354
Jul 11 '13 at 15:45
• @BenjaminSteinberg: I was surprised too when it happened to me, but it is by design see mathoverflow.net/help/whats-reputation specifically: site association bonus: +100 on each site (awarded a maximum of one time per site) If you are an experienced Stack Exchange network user with 200 or more reputation on at least one site, you will receive a starting +100 reputation bonus to get you past basic new user restrictions. This will happen automatically on all current Stack Exchange sites where you have an account, and on any other Stack Exchange sites at the time you log in.
– user9072
Jul 11 '13 at 15:51
• @MadScientist: thank you for the clarification, I am not sure I agree. MO could be somewhat of a special case, aslo due to the sizeable math.SE existing with the in some sense same but then also quite different scope.
– user9072
Jul 11 '13 at 15:53
• There is also this feature request similar to the one by Mad Scientist: Prevent questions on Hot List from being upvoted by casual visitors (only rep is from association bonus) Jan 19 '17 at 6:11

1.1% of upvotes on MO are from users who have never posted. Hardly seems worth getting worked up over, but I'm not a mathematician.

Forking an SQL query mentioned by ScottMorrison, I found:

1,333 MO users with no posts have given 9,159 upvotes and 2 downvotes. (query)

There are a total of 35,678 MO users, 854,611 upvotes and 53,939 downvotes. (query)

Meaning, non-posting but voting MO users make up 3.7% of the total userbase (which presumably also includes people who have never voted or posted), and that 3.7% of the userbase is responsible for 1.1% of upvotes on MO.

Maybe there have been different trends if you sliced up the data differently, such as just looking at the last year. But then maybe you'd also find non-posting voting users later become posting users. There's much more analysis that could be done. I it doesn't seem like a major issue to me.

I'm not sure if these sorts of queries were already done? I just stumbled onto this by accident. I know the discussion is a bit old.

• I think the issue was more about the concentration of those votes on specific questions/answers. I think there were follow up discussions (which you may want to check of you are interested in the issue). Aug 8 '14 at 16:48
• Sure, if you point me to them. Aug 8 '14 at 16:55
• meta.mathoverflow.net/… Aug 8 '14 at 16:57
• Data on voting by 'association bonus only' users Aug 8 '14 at 16:58
• The two queries linked are identical.
– j.c.
Oct 2 '15 at 17:09

One possibility, that might be less controversial, would be to have a cap on the number of upvotes you can use without having a single upvoted question or answer (maybe 20 votes?).

• That might be better than the status quo, but it would have the disadvantage of signaling that MO is indeed open to voting by non-participants, just not too often (and it wouldn't help at all in the case that question goes viral and attracts lots of random people, like the batman question on MSE). Jul 11 '13 at 17:43
• The "hot question" issue is separate in my mind. It's a known problem for all sites that I'm pretty confident is on SE's "to fix" list (though I doubt it's very high on that list). Jul 11 '13 at 17:55
• I think doing it like this could have a quite unwanted side effect. In my mind an advantage of this association thing is that people do not need to try to "get some points" to somewhat use the site; this might prevent some not-so-good questions (or still worse answers) that otherwise might be made to this end. Now, perhaps this would not be that frequent. But if you give somebody 20 votes so they get somewhat in the habit of voting here, and they used them up, then I would assume not few will just ask something. (Also they wil suceed, 15 points is not much.)
– user9072
Jul 11 '13 at 18:30
• FWIW in Math.SE we have a zero participation user who has voted 57000 times. All upvotes, I dare not speculate what kind of posts are the benficiaries :-/ Jul 18 '20 at 22:20

A bigger potential problem is that this creates a gateway for unscrupulous users to create new identities, participate on sites where they can earn reputation points simply by posting calculus problems, and then enter MO empowered to vote themselves up. There have certainly been cases of users creating fake identities for this purpose, but at least in the past they needed to ask or answer a real question or two along the way; now that obstacle is removed.

Moderators have many awesome tools to detect this1. While the association bonus has been abused in the past for sockpuppetry, it's caught quite easily, especially if socks are being used for reputation inflation. There are system scripts that reverse serial voting too.

He/she has never participated in MO in any substantive way (zero questions, zero answers, zero edits, etc.), but has nevertheless cast 28 votes here in the past week, and 46 in the past month. hat's a lot of votes, and I think it's a fair guess that they're not terribly well-informed.

I personally have seen that there are people who upvote bad posts (intentionally, unknowingly, or by accident). However, the number of good votes usually outnumber these, enough to render the "bad upvote" irrelevant. Unless the site is being overrun by voters from other sites (doubt it), this should not be a major problem.

1. Unfortunately I cannot share the details of these tools as we're supposed to keep them secret -- understandably so, as the more one understands something, the easier it is to circumvent it. Rest assured, they're quite thorough, and are able to identify and verify suspicious cases quite efficiently.

• But even if the downside of the association bonus is small (which I'm still not convinced of), the upside seems to be virtually nonexistent. Anyone who shows up here with the knowledge and the willingness to participate usefully is able to vote very quickly, with or without the association bonus. Pretty much the only thing the association bonus adds, as far as I can tell, is voting power for the sort of user who needs to ask for an explanation of why $A\subset B$ is equivalent to $A\cup B=B$. Jul 11 '13 at 2:30
• @steven The voting power isn't the upside. The upside is commenting and the removal of new user restrictions (which are in place to prevent spammers) Jul 11 '13 at 2:48
• FWIW, it's easier to get thr 15 rep required to vote via suggested edits than it is to gear u sockpuppets on other sites. Jul 11 '13 at 2:49
• Manishearth: Good point! I'm totally on board with eliminating the points for suggested edits. :) Jul 11 '13 at 3:09
• @steven not quite what I meant -- IMO there is no problem in the first place to fix (so I'm against that change personally) but yeah. Jul 11 '13 at 4:02
• I don't think that @Steven is talking about sockpuppets. I'm thinking that he points out that people who know absolutely nothing about research level mathematics get a say in what is a good question or answer on a research mathematics site. This is an actual problem, even if at the moment there are only a few users which do that, it doesn't mean that September is very far. It is perfectly reasonable that it's coming right for us, or don't. But it might.
– Asaf Karagila Mod
Jul 11 '13 at 9:35
• @AsafKaragila I know. However, it's been my experience that the only effect people from other sites have is when a question goes "hot". If someone is able to understand the posts here to an appreciable degree, then there's no reason not to let them vote. If someone is trying to troll by voting randomly -- well, fortunately people usually aren't that bored in life. If someone doesn't understand the posts here, I don't see any reason other than trolling for them to come and vote for stuff. Why would you waste time on a site where you're not learning anything or sharing any knowledge? Jul 11 '13 at 10:39
• Of course, I'm talking about persistent voters. I myself don't understand (hope to someday) most of the posts here, but I've voted for a couple that are within my league. Jul 11 '13 at 10:40
• @Manishearth: Understanding most of the posts on MO is a ridiculously ambitious goal --- most mathematicians wouldn't. Jul 11 '13 at 10:55
• @MarkMeckes s/most/greaterthanfive :) Jul 11 '13 at 10:59
• @Manishearth: There are certainly reasons not to let someone vote just because they think they understand the posts here to an appreciable degree. One is that they may be deluded: anyone who has to ask why $A \cup B = B$ and $A \subset B$ are equivalent is incapable of judging whether they understand MO questions. Note that this question was asked one week ago, so if it was a honest question at all, then the person who asked it has no business voting on MO a week later. Unfortunately, the world is full of people who like math and have delusional beliefs about their own understanding. Jul 11 '13 at 14:15
• A second issue is that even if people did vote only on things they genuinely understood, adding a voting contingent of people with no research-level background just distorts the voting system and makes the site less research-focused. I don't think this is as problematic as deluded voting, but it would itself be a good reason to eliminate the association bonus on MO. Jul 11 '13 at 14:19
• @HenryCohn Re:the second issue: why would someone waste time reading a random site that they don't understand? And the cases where someone is deluded/etc are low, I don't see that happening often. My point is that people only come here and vote if they have interesting stuff to read (which will only happen in the case of "hot" questions). People won't read without motivation. With that in mind, I don't think that there will ever be a flood of bad votes. A small trickle, maybe. Jul 11 '13 at 15:21
• @Manishearth, I think you might be underestimating the number of people who think they understand some mathematics. This is a field particularly prone to incompetence so severe as to make recognizing one's own incompetence impossible. Jul 11 '13 at 16:56
• @Manishearth: I believe the number of people on the internet who think they understand research-level mathematics is much higher than the number who actually do. They often have an impressionistic, vague understanding and enjoy reading about mathematics research. There are enough such people to overwhelm MO if they all really wanted to participate, but you're right that that's not going to happen. On the other hand, I think it's much safer to set up reasonable rules now, rather than waiting to see whether a serious problem develops (at which point it will be much more difficult to fix it). Jul 11 '13 at 17:05

The minimum reputation for setting bounties is 75, which is less than the association bonus of 100. So any person with an association can immediately post a bounty.

I suggest that the association bonus should be smaller than the reputation needed to set bounties.

• Why? Why shouldn't someone new to MO be allowed to set a bounty? May 5 '14 at 5:55
• I imagine that you are thinking of users using bounties to keep bad questions from being closed. If this is the case, note that a bounty cannot be put on a question until that question is 48 hours old, which is more than enough time for the regular MO users to close such questions. (See the MSE bounty faq.) May 5 '14 at 7:28
• @GerryMyerson, setting a bounty is a privilege, which calls attention to a question, and I find it distracting rather than stimulating to give new users that privilege. In particular, it distracts me when I look at "featured questions" and find questions that that do not warrant featuring. Why should there be a reputation requirement for setting bounties at all? Your answer there may answer your own question too. May 7 '14 at 21:12
• The bounty here is a good example: mathoverflow.net/questions/369520/… Aug 25 '20 at 13:50