Splitting Easy MathSE Questions into a New NoviceMathSE Is a Bad Idea

(But there are some positive changes that can be made)

This is the final chapter in my series about the state of internet fora, and Math.SE and StackOverflow in particular. The previous chapters are Challenges Facing Community Cohesion and Ghosts of Forums Past. Unlike the previous entries, this also sits on Meta.Math.SE (and was posted there a week before here). (As I write this as a moderator of Math.SE, I refer to the Math.SE community as “we”, “us”, and “our” community).

A couple of weeks ago, there was a proposal on meta.Math.SE to introduce a third level of math site to the SE network. Many members of the the MathSE community have reacted very positively to this proposal, to the extent that even some of the moderators have considered throwing their weight behind it.

But a NoviceMathSE site would be doomed to fail, and such a separation would not solve the underlying problems facing the site.

To explain my point of view, we need to examine more closely the arguments in favor of NoviceMathSE.

HWMath.SE as stated in the proposal is not appropriate for the SE model

In the proposal itself, the goal is stated to

act as a place where students are solving their homeworks all together.

  • Some students will learn a lot by answering their friends questions. They are usually discouraged to write answers on MSE since usually language and notation is less formal.
  • Discussion between them might be more helpful than a discussion where one stays very formal.
  • They will put more effort on questions, since usually on MSE even a challenging/tricky questions gets a hint immediately.

Encouraging lots of discussion between students solving homework together is a mixture between subjectivity and localization, two things SE tends to avoid.

Maybe someone could create a tool where a school/college/university course would have a SE-like forum/Q&A allowing students to work together on a SE-like framework. This style of tool is used already in some MOOCs to facilitate learning environments (especially since the ratio of students to instructors can be enormous). Some MOOCS reset the forums each term/year to foster additional rounds of student involvement. I don’t know if this sort of tool already exists (if not, then maybe someone should go make one).

This sort of tool belongs there, not on the SE network.

But I think much of the positive reaction to the proposal wasn’t for exactly the same proposal as in the OP, but instead for the thought of adding a lower-level Math Q&A.

For this reason (and because certainly SE would not want to be explicitly viewed as a place where students go to get their homework done for them), I refer to the potential site as NoviceMathSE instead of HWMathSE. (I note that Jyrki has suggested calling it MathTutoringSE, which is also better than HWMathSE).

Levels of math on SE

The proposal asks about “a third level of math site”. Implicitly stated in this proposal is the distinction between Math.StackExchange and MathOverflow as being a difference of the level of the question. But this is not an accurate description of the differences.

MathOverflow is not an ordinary member of the StackExchange network. MathOverflow is run by a non-profit organization which has an agreement with SE to host their site. It did not start through the typical experimental-beta-public StackExchange model, and does not have the same culture (or even all the same rules) as the rest of the StackExchange sites.

It is more appropriate to compare MathOverflow with PhysicsOverflow, which is separate from the StackExchange network.

In essence, MathOverflow has content that is interesting to research mathematicians. This consists largely of research level mathematics, but sometimes it also consists of essentially basic questions that are of interest to mathematicians. This is exactly how MO was founded (it’s older than MathSE).

It is not true that once a question hits a certain level of difficulty, it should be asked on MathOverflow instead of MathSE. Instead it is the audiences that are different.

With this in mind, it is not appropriate to think of creating another math site as something making a three-step trinity of NoviceMathSE, MathSE, MathOverflow.

Can we isolate the signal from the noise?

The goal makes sense. Right now, most of the noise on MathSE comes from low-level questions. The major intent behind this proposal is to raise the ratio of signal to noise on MathSE by removing most of the noise.

But this cannot hope to work, because we cannot achieve consensus on how to distinguish “signal” from “noise”. There are already endless disagreements on what is on-topic or off-topic. It is unreasonable to expect MathSE to be able to draw a clear line on what is on-topic and what is off-topic now.

I cannot begin to imagine the moderating headache that would come from attempting to identify and close these questions amidst the various sources of ensuing community backlash. It would be one thing if MathSE had consensus on the various choices facing it, but this is not the case.

Dumping bad questions

More worrying to me is that this proposal seems to be supported most strongly by users who want to dump bad questions somewhere else. (It is possible that I am misinterpreting this, but I don’t think so.)

Such a site is doomed to fail. It would indeed be full of noise. There would be fewer experts there because there are fewer interesting questions, and novices would often prefer to not post there because there would be fewer experts there. Users want good answers, and depending on novices to help other novices is more appropriate for peer-learning environments than a SE Q&A.

One of the major reasons the SE model has been effective is that each site is created to be a place with very high quality content, where experts want to answer interesting questions, and where people looking for good answers can find good, accurate information.

Yes, migrating lower quality questions to NoviceMathSE from MathSE might improve the condition of MathSE, but the signal/noise ratio of NoviceMathSE would almost certainly spiral out of control towards 0 and the site would fail.

We cannot expect to migrate all the lower-quality content (assuming we could even identify what that means) to another site. If the goal is to remove lower-quality content, then the appropriate course of action is to try to find a way of identifying and removing it. Why bother trying to find somewhere else to dump it?

We want to create a site for them to solve our problems

Many of the comments and posts in favor of a NoviceMath.SE seem to want it to exist in order to solve problems of low quality content on Math.SE. It is unreasonable for a group of us to try to create a site for some other group. That is, it doesn’t make sense for a group of MathSE members to decide on a site that other people should go and populate.

If a group of people want to make NoviceMathSE (or some variant thereof) happen and be a part of that new community, then it would be a good idea for them to step forward and begin establishing what they want and what they’re missing from Math.SE. This is how new communities are established. Too much of the discussion essentially concerns ghettoizing low quality questions. This is against all principles of self determination on the network.

Proposals for other ways to improve site quality

But I think there are some other ways to improve the quality of MathSE that don’t rely on fragmenting the community.

  1. Implement a Triage queue here. StackOverflow has a special review queue called “Triage”. The goal is to quickly sort potentially problematic posts into categories that can be routed elsewhere. In short, questions are sorted into three categories: Looks Ok (where it goes to the front page), Should be Improved (where it has limited visibility on the front page and goes into a help and improvement queue), and Unsalvageable (where it goes to mod review or a close/delete queue).
  2. Consider creating an Ask a Question Template (like the one being experimented with on SO). It is a hard question to determine what someone might put into a question template, but it may just work.
  3. Improve awareness of the ability to favorite and ignore tags, and to hide questions from ignored tags. Did you know that you can not only favorite tags, but you can ignore them? And did you know that you can hide questions from ignored tags? This seems to be little-known, but the fact is that every additional method of filtering towards content that you prefer is better.

But I should note that these come with caveats. The Triage Queue is resource intensive. SE has declined to implement it on other sites in the past because it requires tweaking lots of Machine Learning algorithms (i.e. lots of maybe continuous work) and it requires many people looking at review queues to identify questions quickly. As noted here, triage was tailored to the needs of StackOverflow. This doesn’t preclude its use elsewhere, but that’s a discussion which needs to be had separately. Fortunately, triage makes sense on the largest sites on the network, and Math.SE certainly fits that bill (second largest on the network).

An Ask-a-question template is somewhat complicated, since there are many different questions that can be asked. But in the AB testing on StackOverflow there has been some success. I think it may be beneficial to try to develop a template on Math.SE and proceed with some AB testing as well. (The worst that happens is that it doesn’t work, right?).

This entry was posted in Mathematics, SE and tagged , . Bookmark the permalink.

Leave a Reply