6
$\begingroup$

Presently (and for the foreseeable future) I have a particular interest in finding especially good fairly recent answers.

Besides the reasonably obvious ones of

(i) following the active queue (seeing very recent answers) and identifying the good ones as they are posted or updated; or

(ii) using the moderation tools to locate the Highest voted among the Answers with extreme votes (by day, week, fortnight, etc), and identifying the good ones by upvotes,

are there some neat ways of finding good answers, whether common to SE or considering the specifics of CV* that I might have missed?

*(e.g. is there some user whose upvotes are especially good indicators of generally good answers, for example, or some other CV-specific indicator?) (not possible as whuber points out)

| |
$\endgroup$
  • $\begingroup$ @AndreSilva Yes, I saw those before - but thanks, those are relevant too, or could be refined to be, if I learned how to do it.. $\endgroup$ – Glen_b Sep 10 '13 at 10:59
  • 1
    $\begingroup$ I overlooked the queries. They did not provide filter for date (so, they do not answer your question). I do not know how moderation tools look like, but it is probably something like search for: is:answer score:5 (for answers with score >= than 5), and filter by newest, right?. $\endgroup$ – Andre Silva Sep 10 '13 at 11:14
  • 1
    $\begingroup$ FYI the data explorer is only updated on a weekly basis, so depending on your definition of recent this may or may not be sufficient. $\endgroup$ – Andy W Sep 10 '13 at 12:52
  • 3
    $\begingroup$ You can search using the terms: is:answer score:__ (using whatever threshold score appealed to you), & then sort by newest. $\endgroup$ – gung - Reinstate Monica Sep 10 '13 at 13:44
  • $\begingroup$ Andre, @gung thanks. I clearly have many things to learn. $\endgroup$ – Glen_b Sep 10 '13 at 13:55
  • 2
    $\begingroup$ Upvotes by whuber are likely to be informative ones :). Generally, though, your approach, whatever it will be, will suffer from the problem of the lack of external validity. CV is a self-contained community, and by using queries to it, you can only improve the fit to the existing data. What would be more interesting is to see whether the external world (Wikipedia, Google, PNAS, etc.) gives references to CV, and that is arguably much harder to figure out. $\endgroup$ – StasK Sep 10 '13 at 14:06
  • 1
    $\begingroup$ And by StasK, no doubt. However, I don't believe there is any way to determine who upvoted a post, or what posts a user has upvoted, unless they leave a +1 comment. For more strategies along the lines of my comment above, it may help to read my answer here: How should a new user navigate Cross Validated to learn more about statistics? $\endgroup$ – gung - Reinstate Monica Sep 10 '13 at 14:16
  • 3
    $\begingroup$ Could you clarify what you mean by "good"? I think many of us are familiar with the experience of working hard and creatively to solve a difficult question and posting an answer we are proud of, only to find that it is neglected: little upvoted, perhaps not even accepted, and with few pageviews. You won't find such answers through the search engine, although arguably these may be among the very best. If this is the kind of hidden gem you're looking for, about all I can suggest is to ask community members to volunteer links to answers they think are good: in short, conduct a poll! $\endgroup$ – whuber Sep 10 '13 at 19:17
  • $\begingroup$ BTW, nobody in this community has access to detailed voting records. You, and only you, can see your personal record. Moderators can obtain some summary statistics, but because the purpose is to identify certain sneaky forms of site abuse, I may not reveal more than that. That's it. $\endgroup$ – whuber Sep 10 '13 at 19:27
  • 4
    $\begingroup$ @whuber thanks for clarifying that point on seeing votes... As for what I mean by "good" - at bottom it is unfortunately the rather impossible-to-convey standard of 'what I think is good', but that seems to be highly correlated with what other expert users think is good. I'm especially interested in finding more underappreciated 'good' answers but at a loss as for suitable ways to identify them aside from happening on them almost at random. $\endgroup$ – Glen_b Sep 10 '13 at 23:01
  • 2
    $\begingroup$ One insight I can share from viewing the voting summary statistics is that even experts are not entirely in agreement or consistent: each has a particular set of interests that likely constrains the scope of the questions and answers they read. Thus, if you are interested in a less than popular subject, many of the good answers you seek may be undervoted. (There may be some statistical ways to ferret these out by controlling for the pageviews or voting rates associated with each tag.) $\endgroup$ – whuber Sep 11 '13 at 14:11
  • $\begingroup$ @whuber - yes, thanks. I was certainly aware of the fact that the range of my own interests reduce the scope of what I vote on; I assumed this was typical. While I cover a fairly broad range of tags (perhaps not quite as many as you, but up in that region), I know there are a number of quite popular topics I barely touch. That recognition was part of my motivation for asking in the first place. $\endgroup$ – Glen_b Sep 11 '13 at 22:04
5
$\begingroup$

Non-traditional methods for mining good and/or recent answers:

1- Search for:

is:answer score:5

and filter by newest. It will appear all the newest replies with score equal or more than 5 (the threshold 5 was just one example).

To better understand how the search-by-score syntax works see this thread.

2- Look for the lowest score responses from users with high reputation per number of answers ratio. The answers are very likely to be good ones.

3- Investigate replies from new users sorted by reputation and I think good answers will pop up. Some of them already have statistics background and their contribution take some time to reach a good visibility.

| |
$\endgroup$

You must log in to answer this question.

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