14
$\begingroup$

So, can someone help me understand what the scope of the site is?

I ask a question about Monte Carlo search, which is one of the core algorithms behind the Go playing AI bot, AlphaGo, and it is closed off as off-topic, citing this reason This question does not appear to be about artificial intelligence.

So, my question is: Why isn't it about AI? Isn't AlphaGo an AI bot? Why does asking about an AI algorithm of an AI bot make it off-topic?

Can someone(maybe one of the close-voters) take the example of AlphaGo and explain what an on-topic question and an off-topic question(<-- You can use mine if you want to.) would look like?

I already asked a question about the scope of this site, citing another example, where I'm yet to get a clear answer.

If all the questions get closed as on-topic in DS and CV, then why do we even have this site? (Sorry if I sound rude, but I really want this site to grow. So, the early we sought out our scope, the better.)

$\endgroup$
  • 5
    $\begingroup$ I respectfully suggest that we should be doing more to get a handle on this. In the short term, we will lose interested posters, in the longer term potentially the entire undertaking. I'm finding it difficult to see where the necessity for such strict demarcations actually arises. If it's possible to take a useful AI-based perspective on any question, what's the big deal? Attempting to restrict this site to the 'top levels' of AGI won't leave anything meaningful to talk about. $\endgroup$ – NietzscheanAI Aug 7 '16 at 15:33
2
$\begingroup$

The initial scope on Area 51 proposal was:

Conceptual questions about life and challenges in a world where "cognitive" functions can be mimicked in purely digital environment.

Of course this isn't a strict rule, because the final scope is defined by community based on the questions being asked, so if you have any great question related to AI, please ask. So after some time this site can find a distinct and unique scope in comparison to other existing network sites.

However please note that the questions about programming, algorithms, implementation and data modelling are already on-topic on the other dedicated sites and are likely to be off-topic here in order to avoid huge overlap.


Basically the scope is still about artificial intelligence, but coming from the technical background, asking the right question could be challenging (because we've already a lot of sites dedicated to different aspects of AI). You can think about it like Programmers SE site, but without asking actual programming questions.

$\endgroup$
  • 4
    $\begingroup$ "questions about programming, algorithms, implementation and data modelling can be asked on other sites". Here's a potential issue with that: even if the questions don't feature such things, the answers almost will certainly have to, otherwise there's really not so much to say. AI is cross-disciplinary, and hence necessarily cross-site - let's embrace that, rather than create too much a stick to beat posters (and ourselves) with. $\endgroup$ – NietzscheanAI Aug 7 '16 at 15:42
  • $\begingroup$ Updated post to be less strict, let me know if that helps. $\endgroup$ – kenorb Aug 7 '16 at 16:48
  • $\begingroup$ that's very accommodating of you, but it still doesn't work for me personally. That said, I'd rather see consensus than "have my own way". To take the specific restriction at the top of your post, it just seems highly likely to me that people who hear that Monte Carlo Tree Search is used in insert latest AI news story will be coming to AI SE first, only to be sent to a less AI-focused site. I should add the rider that I very much respect the effort that people have put in to get the site this far. I hope we can find something that both suits everyone and actually works. $\endgroup$ – NietzscheanAI Aug 7 '16 at 16:54
  • 1
    $\begingroup$ For me this doesn't work either, so we can hope it's going to be changed. $\endgroup$ – kenorb Aug 7 '16 at 16:56
  • $\begingroup$ "...questions about programming, algorithms, implementation and data modelling are already on-topic on the other dedicated sites and are likely to be off-topic here in order to avoid huge overlap...." That might mean that the niche becomes very small if the scope is defined as everything AI - everything AI already included somewhere else. While it increases focus it decreases the total size and means considerable work closing questions when askers miss the fine points. A criticial mass might not be reached. Just want to point out the disadvantages of this approach, next to the advantages. $\endgroup$ – Trilarion Sep 6 '16 at 20:27
  • $\begingroup$ Thats a very stupid scope to be blunt: No AI experts are going to be drawn to this scope, it's essentially only useful for the worldbuilding audience, which already has a popular SE site. If this is going to be called AI SE, it needs to be a place attractive to actual AI experts in the field, not just science fiction enthusiasts speculating about challenges of a world with AI. Questions about fundamental AI design needs to be on-topic. Not programming questions. Just structural, fundamental design questions. $\endgroup$ – Viziionary May 17 '17 at 10:23
4
$\begingroup$

I was one of the close voters.

First up, the close message you see is the generic off-topic message - we only get one reason under the "off-topic" branch of the close dialogs because we currently have no moderators to create and approve off-topic reasons. Therefore, anything deemed off-topic will get that one message. It's not that your question wasn't about AI, it wasn't about AI as defined in the help center (or, again, since we have no moderators yet, as defined on meta).

Your question, in my understanding, is about specific algorithms and how they work. We're not really into the math/statistics/implementation on this site, because those are already well covered by existing places.

I think that the question could be reopened if it was adjusted to ask something like "Why is Alpha Go's approach more appropriate for games than existing technologies?" Then the question wouldn't be about a specific algorithm, but answers could still dive in if they wanted.

As for whether we have a scope, we're still working on that, as evidenced by our abundance of meta posts about topicality! I think we do have at least some sketches of what should by on- and off-topic, though.

$\endgroup$
  • $\begingroup$ Monte Carlo search is a heuristic search algorithm. So, it is clearly an AI algorithm. So, why is asking about an AI algo. off-topic in an AI site? $\endgroup$ – Dawny33 Aug 6 '16 at 14:27
  • 2
    $\begingroup$ @Dawny33 This AI site isn't about algorithms and implementation. (Admittedly, the name is somewhat confusing then.) Those things are covered by existing sites. $\endgroup$ – Ben N Aug 6 '16 at 14:29
  • 2
    $\begingroup$ The thing is: If we close down all the algorithm-related and stats related posts on the site, close to 60% of the current posts need to be closed down. And that would definitely lead to a premature death of the site. $\endgroup$ – Dawny33 Aug 6 '16 at 14:37
  • $\begingroup$ @Dawny33 Unfortunately, duplicating existing sites would also result in the site being closed. To survive, we have to show that we bring something new. $\endgroup$ – Ben N Aug 6 '16 at 14:38
  • $\begingroup$ We've continued this discussion in chat. Pl go through it for further context :) cc @BenN $\endgroup$ – Dawny33 Aug 6 '16 at 15:36
2
$\begingroup$

I don't know, but that shouldn't have been closed. It should be, in almost every case, sufficient to simply ignore any question which falls into a "grey area" regarding scope. We should only close questions which are blatant spam, trolling, or so wildly off-topic that a 2 year old could see it (like a question about the best fuel injector cleaner to use for 1972 Ford Pinto, or something).

If questions are desired by the community, they'l bubble to the top. If they aren't, they'll die from lack of activity. Explicitly closing a question is an aggressive and hostile act and should always be a measure of last resort.

$\endgroup$

You must log in to answer this question.

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