7
$\begingroup$

Following up on this question: We have a very large & widening gap between questions and answers. How do we fix it?

Where one of the suggestions was:

We need more answers, and therefore answer-writers

I agree, and I'm wondering whether the datascience stack exchange is causing CV to miss out on some good answer writers and moderators?

Although some overlap between SE sites is unavoidable (CV/Math/Theoretical CS/Signal Processing) - in the case of CV and DS, the overlap is so significant as to be harmfully redundant. Just about any theoretical machine learning or neural network question is completely on topic in either site. Since people can only spend so much time on an SE, it means time devoted to answering and moderating questions by competent people is being split between both sites.

$\endgroup$
5
  • 11
    $\begingroup$ We tend to be pretty mixed about the Data Science site (see, eg: 1, 2). That said, it's beyond our purview to do much of anything about it. $\endgroup$
    – gung - Reinstate Monica
    Aug 3, 2018 at 1:47
  • 4
    $\begingroup$ Long term, questions and people migrate to sites where there are better answers. It can take some years but sites with fewer people or fewer answers or even fewer good questions tend to fade away. And that goes for specific kinds of questions too. So, opportunity for improvement on both sides, not a threat. $\endgroup$
    – Nick Cox
    Aug 3, 2018 at 8:26
  • 9
    $\begingroup$ Some people have a misapprehension that "data science" and "statistics" are distinct, instead of the former being a clever marketing gimmick that renamed the latter. Almost all content on datascience.se is on-topic here, with the exception of questions about how to use software (for which there is a dedicate software forum called stackoverflow). $\endgroup$
    – Sycorax Mod
    Aug 9, 2018 at 16:17
  • $\begingroup$ I'm not sure how answerable this question is without digging into actual answer rates. I don't know how to access that kind of data. But my gut instinct is "yes". @Sycorax this is just my opinion, but I feel like data science has become an umbrella term for statistics and machine learning; if anything, statistics is being left out to dry a bit. Perhaps it could be renamed "the data sciences", much like how library science itself isn't really "science" in the sense of the scientific method, but a collection of techniques for "getting stuff done at libraries". $\endgroup$
    – shadowtalker
    Aug 27, 2018 at 19:49
  • 1
    $\begingroup$ @shadowtalker Could be. Sometimes it's now called "library and information sciences." To my mind, "data science," "statistics,' and "machine learning" are three circles of a Venn diagram with 97% of the area falling in the intersection of all 3. $\endgroup$
    – Sycorax Mod
    Aug 27, 2018 at 19:51

1 Answer 1

Reset to default
4
$\begingroup$

http://data.stackexchange.com/datascience/query/59302/questions-and-answers-per-month

Based on the graphs below you might say that this decline in answers seems to be nicely filled up by the data stackexchange. But I do not think that this is like a Gondwanaland falling apart into an Africa and America that fit so nicely together.

The answer rate curve of 'stats' has a stall/decline like most other (older) stackexchange sites in the science corner (math, mathoverflow, physics, chemistry, biology). So there is already explanation like many other sites also having lower activity.

The answer rate curve of 'data' has a nice stable/rise with an extreme sudden doubling (during only a few months) in answers and questions from 800 per quarter to 1600 per quarter around new-year.

These numbers are only small and would not severely harm the statistics site. That is assuming that there is even a strong causal relation. But the statistics site did not suddenly loose 800 answers per quarter (although for the statistics site at the peak, the first quarter of 2017, there were 5410 answers and now a year ago first quarter in 2018 there were only 5068 answers, this decline does not match the 800 increase at data site which came in a more abrupt way and a slightly different moment, a few months later).

data

Certainly there might be some causal influence but it seems unlikely to me that this entire piece of the data site can be added to the statistics site and I imagine that the effects are only small.

I would say that we can better use the sites for what they are good at individually (I see a difference in the questions, along the lines of the technology science distinction how the sites are currently classified by the stackexchange overview, and this enhances the experience). I suggest to approach them as reinforcing each other and with this as starting point the sites can grow alongside of each-other, together, more strongly rather than having a slightly smaller platform when they would be merged together.

$\endgroup$
3
  • 1
    $\begingroup$ Geological allusion tweaked: Gondwanaland, not Godwana. (Perhaps you're averaging with Godzilla. Let's keep them apart.) $\endgroup$
    – Nick Cox
    Aug 29, 2018 at 13:22
  • 1
    $\begingroup$ +1. But may I make a suggestion? I find it more meaningful to track the sum of answers and closed questions. I hope the reasons for that choice are obvious. Its importance might not be, so let me just add that approximately a quarter of all questions are now closed (and most of those are unanswered, which is why I don't bother to distinguished answered closed questions from unanswered closed questions). Also, because very few answers are deleted, it has been unnecessary to track them. $\endgroup$
    – whuber Mod
    Sep 5, 2018 at 12:25
  • $\begingroup$ That is an interesting suggestion. However, I tried a quick adaptation data.stackexchange.com/stats/query/894703/… and I do not see that a quarter of the questions have been closed. How should I change the query (or what other source does the quarter come from)? Maybe some information is blocked for me? Anyway, I do not think that it will provide much change to the picture of these crude data. There are, of course, clear changes going on, but that is not strange for an 8 year old forum. Yet, the datascience can only have a small side effect. $\endgroup$
    – Sextus Empiricus
    Sep 5, 2018 at 14:27

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .