What problems does the two-level split solve?

Question

Kaveh asked what problem the two level split into research and non-research sites would cause. This has now gotten a couple of answers. I asked what problem is solved which I think is more important, and Kaveh thought this would be a different discussion, so here it is.

The only argument I've heard is that you need separate research sites to not waste the time of experts, but that is a strange statement, answering questions is almost by definition time wasting, you do it because you enjoy it, not because you have to or make money from it.

And in any case experts generally do not have the time to sit and pounce on questions when they arrive (although there are some experts like that on Stackoverflow too) instead you will check very narrow special interest tags, or you will look for questions with bounties or questions other people haven't been able to answer. Simple questions will be answered by others.

So the claim that an all-inclusive site would waste the time of experts is clearly incorrect. Each expert chooses how much of his time and one what he wants to waste. The existence of simple questions does not force anyone to answer them.

So: What problems does the two-level split solve?

Update:

We have been discussing this for a while now in several threads, and I think we are starting to run out of steam. My conclusion is this:

The problem a two-level split solves is that some researchers doesn't want to hang on a generic site. The arguments for why are to a large part clearly just rationalizations of their feeling on the subject, but some arguments make sense, like that the research-level questions tend to be more discussions than answers, and hence are a different type of questions. But rational or not, the opinions of the researchers should be acknowledged.

In my opinion it would have been better to start with a generic site, and split out a research-level site if there seems to be a demand for this (ie research level questions), but not enough supply (ie answers), although this is of course too late with cstheory.SE.

Answer 1

My impression is that a high proportion of the highest rep. MO users are willing to participate in maths.sx, but a much lower proportion of mid rep. users. I saw that 9 of the top 10 rep. MO users are 2k+ users on maths.sx, while from the first column (7 users, rep. 3.1k to 3.4k) on page 5, only 4 users have maths.sx accounts at all, and all but one are under 500 rep.

This suggests that the most motivated users want to put their energy into making both sites work, but that there are many experts who are just interested in an experts-only site. Thus I claim a pure experts' site catering to broad academic discipline that has a need for a non-subjective Q&A, like maths and physics, will do better at attracting experts than the general sites.

Incidentally, do note that most academics split time between teaching and research, and to them the sites follow the same distinction: the one is for expert researchers, the other is for expert university-level teachers, who largely, but not wholly, are the same people, but which serve completely different activities. Aaro suggests that the distinction between advanced and beginner-level is soft. I think the distinction between teaching- and research level is pretty sharp: I guess that the number of questions on MO that are "controversial" (i.e., there is serious dispute about whether they are research level) among 1k+ rep. users is a very small fraction of the total.

Answer 2

The research/non-research split is mostly not a split on topic, but it's not really a split on level either. It's more of a split on the target audience: researchers in X form a very distinctive subcommunity amongst the wider community of people interested in X. Charles Stewart cites some statistics on cross-participation between MO and math.se; my interpretation is that there are researchers who spend a lot of time on discussion sites, and they frequent both sites; while other researchers spend little time, and they stick to MO where the content that's most interesting to them is easy to find.

The split is also a split on content, even if the distinction is not the topic per se. Most of the content of MO is not interesting to non-researchers because it is incomprehensible; and most of the content of Math.SE is not very interesting to your typical math researcher because it's not advanced enough. It's easy for non-researchers to ignore the questions they don't understand; it's a lot harder for the researcher in a hurry to find the gem amongst the boring (to him, not in the absolute!) lower-level questions. (No, tags help pick questions based on their topic, but not based on what I would call their flavor.)

There are precedents for sites with overlapping topics on Stack Exchange:

• Server Fault vs Super User: a lot of system administration is on-topic on both sites. SF is specifically for professionals who share common prerequisites and interests; SU is more generalist and caters to amateurs.
• Ask Ubuntu vs Unix SE: all the questions on AU could be asked on Unix.SE. But they would (and do) get different answers, because the sites cater to different communities (AU primarily to people who care specifically about Ubuntu, Unix.SE to more general tinkerers).

TL,DR: the two-level split solves the problem of different audiences expecting different content. And there are precedents that work.

Answer 3

Two levels serve the same purpose as splitting by topic: focus.

High-level questions need careful consideration and answers by experts. Expecially in TCS, the number of peers is generally low. A lifespan of hours as on SO is not sufficient for that; our questions are active for days.

Low-level questions need active users with the drive to look through a lot of questions and (ideally) superior knowledge. Apparently, not all experts see themselved in that role.

Also, I guess that a different kind of answer can be and is expected. On research level, there likely is no short (read: a few sentences) answer if any at all. References to published work and/or a long, iterative answering process has become typical on cstheory. There is usually no need for precision; the exchange of ideas and hunches if often sufficient amongst peers.

On the other hand, questions on lower levels can often be answered very shortly, but might need to be precise wrt to a certain notation and have a didactic finish in order to help the novice.

As you can see, even though the topic does not change between levels, the needs and preferences of users do.

PS: Note that I myself have long struggled with this question. I have come to accept that important users do not like to be part of a larger community. We do not have to like that, but it is fact. Therefore, a merge would not benefit the research community, which is our goal. However, other users would like to help beginners, too, and are likely to frequent such a site had we one.

Answer 4

A part that was not mentioned is that a research level site also helps in learning for research level students. When you are a graduate student, one of the things you need to learn (apart from all the difficult technical knowledge of your field) is the softer skills of what kind of questions and presentations are interesting to your peers. When I visit MO or cstheory I am able to get a flavor of what researchers find interesting. If there is a general site (say math.SE), then although I can ask my questions on it, by browsing the site I don't get a feel for what researchers find interesting.

When I am browsing cstheory, I learn something from reading and answering other's questions. If I go to more general sites then I learn only from my own questions. Thus, the two level model has a chance to increase the amount of effort experts will invest in browsing and answering questions. The reason researchers seem to prefer research-level sites is not snobbery or a disdain for the non-researcher: it is the fact that like any person they have limited time and energy.

Answer 5

From the point of view of the research level site there's basically only one reason: to have a good site you need many researchers (because there are still many subfields so you need to increase the probability that you'll find your peer and ideally more of them to start a discussion). There are two ways to achieve that: a) SO way of being huge and that means even relatively low count of experts translates into big enough absolute number; or b) just try to collect the experts somehow. I am not sure how good 'a' is for doing research (is there any research going on at SO?) but 'b' definitely works.

There is also a secondary reason of lot of experts not wanting to be a part of low level community. Again, 'b' solves this.

Answer 6

I do not know exactly which problems the two-level split solves. I can write my guess, but I am not sure if it is of any use. Not only that. I do not know whether two is the best number or not. Maybe one is better, maybe three is better. Why two? I do not know.

But why should I even care? The split between Math Overflow and Math Stack Exchange seems to be working. I am not interested in sociological study about why the two-level split works. The only important fact for me is that it seems to work. It is only reasonable to imitate what is working now. When I think about Stack Exchange for TCS, it is the most natural to imitate MathOverflow and Math Stack Exchange than, say, Stack Overflow.

Answer 7

The two-level split is an attempt to solve the problem of SE sites not having a "difficulty level" filter. Advanced questions about a particular topic are in many ways like a separate topic from trivial questions about the "same" topic. And while SE sites allow for filtering of topics, there is no way to distinguish between the "advanced" version of a topic and its "beginner" counterpart.

In the past, some proposals have been made on SO meta about adding a "difficulty level" to questions. These proposals have always failed because nobody has ever suggested a good way to implement such a feature. Usually, it's because the proposal was to allow the person asking the question to rate the difficulty of the question -- but I don't think this would ever work well (and apparently most others agreed, hence the proposals always have failed).

The two-level split that we see with some sites is, IMO, simply an implementation of this same "difficulty level" feature, albeit not necessarily a great way to implement this feature. I would propose that, in addition to the two voting buttons, each question on an SE site should have two additional buttons: "Too hard" and "Too easy". Experts who don't want to bother answering a question because it is too basic, simply click "too easy". Others who don't know the answer to a question because it is beyond their level of expertise click on the "too hard" button. This way the community decides what the difficulty level of a particular question is.

There would also need to be a mechanism for filtering by difficulty in addition to filtering by topic. This would solve the same problem solved by the "two-level" system, but IMHO, in a much cleaner way.

In any case, no matter how this problem is solved, I think a solution for it is desperately needed. Certain SE sites (SO, in particular) are becoming so full of very basic questions that they are beginning to fail as a good resource for experts seeking answers to difficult questions.

Answer 8

The point of the split is that people can decide how to divide their time between the sites.

You will find a lot of experts who are not willing to wade through 95% questions that are uninteresting by level, because of the remaining 5% questions still only a small percentage will be pertinent.

On the other hand, if they can get what they need from the research site, they might well decide to pass 10% of the time on the general level site as opposed to 0% on the frustrating general site. And researchers who pass 10% of the time on the general level site will have a higher reputation there than on the other site.

You have to accept that a researcher very rarely learns mathematics from answering undergraduate questions, many people with high rep on math.SE see their few own questions go unanswered.

And no, tags don't do the same thing as a split. On math overflow the questions on the frontpage indicate clearly that a standard undergraduate question is not on the same level, no information like this is conveyed when posters here (mis)use subject tags.

Level tags are not allowed here and posters would not use them correctly. This is not just about blocking basic questions. If I am not interested in research in set theory, say, I cannot block the corresponding questions here without being unable to see undergraduate questions on de Morgan's rule I could and would answer.

I came here to math.SE BECAUSE I first heard about MO and then heard about math.SE there.