I was clicking around and wanted to post a question. Started doing research to be able to ask an interesting one that I have in mind, and, well:

Avoid “easy” questions

It’s tempting to start with easy, superficial questions: surveys, polls, and rudimentary questions like “what are some good books on this topic” or “what are the best blogs on this topic”. Those are not good questions for the private beta – they don’t reflect the actual content that we want this site to contain, and are not representative of it.

I think that there will be a rush to farm reputation on the first hours. Something similar was seen recently on stackoverflow-docs.

I do not complain about the reputation itself, I was rather hoping that the site deals with topics more related with intelligence, deeper than the technical way to train an ANN or how to program a mobile robot in a specific platform. For that, we already have crossvalidated, data science, electronics, robotics, and stackoverflow itself. However, the first bunch of questions will define a path (am I right here?).

I understand that many of us got involved in the stackexchange network because we solve things in a sort of engineering way, but I think the site for AI has much more interesting questions (for instance philosophy of AI). The thing is, those questions cost more time and effort to ask and answer, so I could expect a slower pace than in stackoverflow or english sites.

Which kind of topics are we aiming for?

See also this Related question


1 Answer 1


To answer the title question, easy-to-Google questions are not OK for the private beta. Flooding the site with trivial questions and simple answers is a great way to demolish any chance of attracting big-name experts. Artificial intelligence site proposals have already failed a couple times - once explicitly because of terribad pedestrian questions.

If we want to survive and grow, we have to keep quality high. We can do that by downvoting low-effort questions and closing non-constructive questions (e.g. requests for off-site resources). And of course, we'll need to dig into the literature to see what kinds of good questions we can explore.

  • 1
    $\begingroup$ In order to encourage downvoting of poor questions, it helps to up-vote the remainder, so people have enough rep to spend on down-votes. $\endgroup$ Commented Aug 2, 2016 at 17:43
  • 2
    $\begingroup$ @SeanHoulihane On private beta sites, anyone can downvote, even with 1 rep. Still, upvoting good content is critical too. $\endgroup$
    – Ben N
    Commented Aug 2, 2016 at 17:43
  • 1
    $\begingroup$ AI has become the new baseline for “failed to make it out of private beta” → ouch! Great link, those are indeed things to know! $\endgroup$
    – Luis
    Commented Aug 3, 2016 at 5:20
  • 1
    $\begingroup$ Artificial intelligence site proposals have already failed a couple times - once explicitly because of terribad pedestrian questions. Interesting.. I felt it failed for the exact opposite reason... over emphasis on building an "experts only" (in practice, if not in word) site, which was so hostile to newbies and hobbyists that it ran them all off. And guess what... there just aren't that many "AI experts" out there! If we go down this path again, we'll once again be sacrificing the "good enough" in the unnecessary pursuit off "perfect". $\endgroup$
    – mindcrime Mod
    Commented Aug 4, 2016 at 0:48
  • 1
    $\begingroup$ I'd add "general reference", such as questions that can be answered by the most basic textbooks on the matter at hand. $\endgroup$
    – Raphael
    Commented Aug 12, 2016 at 9:49

You must log in to answer this question.

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