How can we optimize for pearls?

To become a go-to site for great puzzles, we need to highlight our best content. We should make a good impression to a first-time visitor by having our finest puzzles be a click away.

As is, I feel like the gems are getting lost among a sea of puzzles that are good but not exceptional. How can we make our best puzzles stand out?

The voting system, unfortunately, falls short. Upvotes are largely proportional to views. You upvote a fantastic puzzle just as you upvote a good puzzle. (There's favorites, but they are not searchable, and some people use them value-neutral bookmarks.) So, someone looking for the best puzzles on the site won't necessarily see them in the obvious place to check, the top-voted questions.

The activity page also creates weird incentives that disfavor some excellent puzzles. If a puzzle doesn't need editing, is clear enough not to get quick wrong answers, and takes thought and time to solve, then it takes a while to return to the top of the activity page and is unlikely to reach HNQ. At the same time, high-activity puzzles stay visible whether or not the activity is good activity.

People have put forth ideas like making a puzzling digest and voting for the best puzzles of the year, but they lost steam. What can we do?

(I'm aware that Jeff Atwood's blog post Optimizing for Pearls, Not Sand is on the relative value of questions and answers, not on highlighting the best questions, but I like optimizing for pearls as a general metaphor.)

• +1 and thanks for the reminder! :) – A E Apr 14 '15 at 16:31

The Stack Exchange platform is telling us that we should set the bar for good puzzles higher. Voting culture is tricky to change, but I think it is one of the more effective routes to do so.

Imagine you're a paper-pusher, acting as a content filter for P.SE's questions. Eleven questions land on your desk a day, and you've been asked to pick five to receive the most attention.

Which do you pick?

Here are the criteria I would use, in order of importance:

• Originality. Is the question something new? Or, is it something we've seen tons of in the past? Does it contain new ideas and new thoughts?

If it does, I'd be far more inclined to let it pass. New ideas are the most important.

• Clarity. If the question isn't original, is it at least clear?

Clarity is a spectrum. Ultimately, the clearer what a question wants is, the better the question will be for the site.

• Complexity. Good questions can be too simple to be worth time. While this also fails originality, a simple example might be:

Which number comes next: 2, 3, 5, 8, __, 21

It's clear, but it's not complex enough to be interesting. By way of contrast, we've seen a number of puzzles that are good and original, but may be too complex to maintain interest.

This is going to inevitably be touchy, since it starts to strike at what "good puzzle" means. But, when faced with the challenge of sifting for gold, this is what I'd do.

We should be more stringent with our votes. Don't let everything through. "Not original/interesting enough" should be a legitimate reason to downvote.

There are probably a number of ways to handle this, but voting is the main knob we can turn for quality control.

• Your criteria are OK, but not quite the ones I'd use. The best kind of puzzle IMO is one that looks as though it can accept multiple possible answers, but when you find the intended solution, it fits everything so perfectly that it's obviously right. So you get a few wrong solutions and then the right one comes along and question and answer are massively upvoted. I don't like to blow my own trumpet, but here's an example of such a puzzle. All good riddles (I stress "good") also fit this criterion. – Rand al'Thor Apr 15 '15 at 0:31
• Another, perhaps more important, point: calling for people to change their voting habits is all well and good, but impossible to enforce. Be realistic - how many people are likely to alter the way they vote (which is anonymous) because of a post like this? The regular users are too set in their ways to even think of changing their habits, and the new users don't come to meta anyway! I downvoted this answer because I don't think your suggested solution is going to be effective. – Rand al'Thor Apr 15 '15 at 0:35
• @rand Just because it's difficult doesn't mean we shouldn't try - nor is this answer where the discussion ends on this is topic. Also, these criteria are mostly my own, so I'd encourage being much more strict with your own criteria if they differ. – user20 Apr 15 '15 at 0:43
• I admire your dedication - I really do. But it's not just difficult either to force people to vote in a certain way or even to know whether they're doing so - it's impossible! All you can do is encourage them to do so, and they have no incentive. – Rand al'Thor Apr 15 '15 at 0:48
• I'm not forcing anyone, just suggesting. The incentive is to alleviate the above issue. – user20 Apr 15 '15 at 0:54
• I'm a new user and this convinced me to vote this way. – Daniel Apr 16 '15 at 16:45
• A puzzle that is not interesting to you, may very well be interesting to others. Is your idea of a perfect site one where all puzzles are interesting to you? If you think it is, then you do not understand the power of contrast. Without the contrast, you would not know what you want, what you like. I downvote only when a puzzle makes no sense after many attempts of the OP to explain, or is very poor English, or in cases where the OP has abandoned the puzzle and has not responded to questions. – JLee Apr 17 '15 at 3:05
• @JLee By interesting, I mean unique, not 'I like it.' Should I edit to clarify that? – user20 Apr 17 '15 at 3:11
• When you point to downvote, the message that appears says: "This question does not show any research effort. It is unclear or not useful." It does not say "I don't like this kind of puzzle or I think it is not interesting." Unfortunately, it seems many people ignore that message. – JLee Apr 17 '15 at 3:12
• @JLee Unfortunately, we're stuck with that standard message from all SE's, which is unfortunate. It make sense for a Q&A site but not a challenge-based site. – xnor Apr 17 '15 at 8:41

Upvoting/downvoting is just not working well, in my opinion.

I have an idea. It may not be the perfect solution, but here goes.

There are puzzles of many varieties here. That is one of our huge strengths! There are puzzles that are intended to be solved quickly, and are not so difficult, all the way up to puzzles like A code from the Lines, which are so tough!

I think there's a way to organize our puzzles a bit better in order to have them targeted at the users that want to see them, with the end result being like targeted advertising.

Some users want ONLY short, quick, easy puzzles, or puzzles of only 1 or 2 kinds. Other users want only super challenging puzzles. Still others want something different every day.

So, my idea is to add functionality so that each user can select, based on tags, what kind of puzzles they enjoy and want to see, and also what kind of puzzles they never want to see. For example, if I hate Sudoku, then I can uncheck the Sudoku box, and those puzzles don't show up on the main page for me. Maybe, adding a "difficulty-level" would be another good thing to be able to sort on, and a user can select which difficulty(ies) (or approx. time required to solve) they want to be able to see.

(Side note: Also, I want to be able to follow puzzles, and be notified when they are edited or answered, or a hint is added, but I don't think this functionality exists yet, and I usually forget to return to them.)

• I agree that more functionality to filter puzzles by type, length, and difficulty would be fantastic. I don't really know how we could implement these though. SE is not going to make anything custom for a beta site, and a user-added solution would require a lot of buy-in from the community to take off. – xnor Apr 17 '15 at 3:35
• @JLee - re your side note, puzzles can be followed but it is a little cumbersome and there are no notifications. Just click the star beside a favorite question. Later you can go to your favorites via your Profile > Activity > Favorites. When you are done with a question, it can be unstarred. – Len Apr 17 '15 at 5:17
• @xnor They plan on implementing some of that functionality - at least allowing users to save their favorite tags and only showing questions with those tags: meta.stackexchange.com/questions/251095/… – pacoverflow Apr 17 '15 at 5:23
• @Len Thanks I will do that! – JLee Apr 17 '15 at 14:02
• @pacoverflow Thanks for the info. That is recent, and I am excited to see these changes. I hope it will make navigation and user experience as easy as I imagine it can be. – JLee Apr 17 '15 at 14:08

Here is an answer that expands upon thoughts from both Emrakul and xnor.

Set the bar for good puzzles higher by being more stringent with our votes

This chart shows voting data for a number of high-rep users on this site.

One observation is that 75% of these votes come from 40% of these users. A more important observation is that, between users, upvotes vary from ~50% to 90%. This large divergence in voting suggests a large divergence in site objectives. More specifically, this suggests that some users want more Q&A's and less quality than other users or vice versa.

I am not judging what is right or wrong because I do not know the answer. But according to Area 51 criteria, 10 questions per day is healthy for a beta site. And puzzling.se has been hovering around 10 questions per day for the last number of months. (Yes, I am aware that Area 51 criteria are soft but they are better than no criteria.)

So, although I agree with Emrakul's proposal, I am wondering whether this site can maintain its activity with more stringent voting at this time. Whatever the answer is, I am quite willing to "move the bar" on my voting to help to improve this site.

How can we optimize for pearls?

Highlighting this site's best content is an excellent objective, not only for first-time visitors but for existing users also because there are so many different perspectives on what is a great puzzle.

Just as great wines are selected by qualified sommeliers, great puzzles should be selected by excellent puzzlers. So I am suggesting that a few dedicated individuals select a few great puzzles on a regular basis (personally, I would nominate xnor, rand al'thor, and Emrakul to select the great puzzles).

However, it seems that there is no good place on this site to maintain a list of great puzzles. Perhaps the selected puzzles could be tagged with "great puzzle" but that seems inadequate.

• I honestly think the site would be fine with more stringent downvoting. We're already one of the most active betas by a ton of various metrics. As a side note, Area 51 stats don't tend to be as meaningful as we'd like, because one of the main things the community team looks for in graduation is quality - not just traffic and other stats. Interesting post, though! – user20 Apr 15 '15 at 9:31
• I agree about the activity. I'm one of the most active users of the site, and still I find that puzzles come in at a faster rate than I can or want to solve. The Area 51 metrics are for a typical Q&A site and don't account for puzzles taking longer to consume than Q&A's. – xnor Apr 15 '15 at 9:48
• There is a place on this site to maintain a list of great puzzles: this meta post. It hasn't been very active so far, as @xnor mentioned in the OP, but that doesn't mean it can't be. Nominating a few puzzles there is high up on my todo list - I just have to decide which ones! BTW, thanks for nominating me as a selector :) – Rand al'Thor Apr 15 '15 at 10:31
• @Emrakul - OK. I will shift my voting a little more. Any suggestions for encouraging others? Perhaps one-on-one queries to some of the more regular users who are under-utilizing their votes? – Len Apr 16 '15 at 6:44
• @Len Mm. I'd suggest commenting to explain why you're downvoting - I think this is the best way. Obviously not obligatory, but when people see why others aren't as much a fan of a post, it will nudge people to reconsider how they vote, too. Voting is finnicky - we don't want to tell people how to vote, because it's their choice, but we do want people to understand how to vote to help the site. If a discussion comes up, definitely feel free to chime in, but going out of your way to start them might not be advisable. – user20 Apr 16 '15 at 8:34
• @Emrakul: How stringent do you think is stringent enough? My upvote ratio appears to be about 78%, so I'm somewhere in the middle of that graph. – Joe Z. Apr 16 '15 at 15:22
• @JoeZ I'm reluctant to set a particular threshold. If we get ten high quality questions in a day, definitely upvote them. That may make my answer somewhat unhelpful, but "whatever feels right" is probably the best answer to that. – user20 Apr 16 '15 at 15:52