42
$\begingroup$

I want to alert puzzle writers to a potential cognitive bias I'll call the many-eyes effect. Many people try to solve your puzzle, but only those who succeed post. It's tempting to imagine the posters as typical solvers and forget the many invisible people who tried and got stuck. As a result, your puzzle seems easier and better-clued than it is.

It's exciting that through the magic of the Internet, lots of people see your puzzle, and the crowd can out figure things that any one person would be unlikely to. This does justify an increase in difficulty, and we get some impressive solutions that way. But I do think puzzles should be made to be enjoyed by many solvers, not just an few particularly skilled or lucky ones.

This especially affects underclued one-step puzzles where you have to try loads of things with no indication which one is right. With so many readers, inevitably someone stumbles on the correct thing, maybe on their first try. Then, other solvers say, "Geez, I would have never seen that, how did you figure it out?" The puzzle poster, if they don't keep the many-eyes effect in mind, might respond, "No, the puzzle is fine. See, this person figured out the right thing to do." So, the writer doesn't get the feedback they should.

Relatedly, I think puzzle-writers should not be disappointed if their puzzle is solved quickly. That just means one person of many got it fast. It's entirely possible that the difficulty is just right, but there's enough variance that one person of many found it easy. Also, succeeding is fun. Given that puzzle writers already tend to think their puzzles are easier than they are, it's easy to fall into confirmation bias by putting too much weight on a single unrepresentative data point.

$\endgroup$
  • 10
    $\begingroup$ I need to keep this in mind when writing puzzles (eg. I felt my last one was too easy as it got solved quickly, but that's not necessarily true). It also highlights a broader psychological issue that I, and likely others, have, where you tend to only focus on solving the unsolved. So there's a great swathe of awesome puzzles from the past that don't get looked at because "what's the point?". Therefore, puzzle creators likely aim at creating difficult puzzles to keep them "alive" longer and feel like they've "failed" if they get solved quickly. $\endgroup$ – Alconja Apr 24 '15 at 5:32
  • 3
    $\begingroup$ Imaginary points motivator: easy puzzles tend to get many more upvotes. Like twice as many. $\endgroup$ – Lopsy Apr 24 '15 at 10:10
  • 6
    $\begingroup$ @Lopsy I feel this is partially because if the puzzle is easy, the solution tends to be short/intuitive and posted quick, and people look at the answer and go '"Haha, that's clever / That's a neat trick" and upvote. If one solves a 2 month old behemoth of a puzzle, with 50 steps to a solution, or a whole lot of complex math that isn't accessible to most people, it's harder to judge whether the puzzle was a good one or not just by reading the answer, and hence it doesn't get as many upvotes. $\endgroup$ – Tryth Apr 24 '15 at 10:14
  • 1
    $\begingroup$ @Lopsy If a puzzle is really easy, it'll probably get downvotes as being not challenging enough. There's a happy medium for puzzles that are relatively easy to understand (as Tryth said) and yet still quite difficult, often unless you spot the right trick. The highest-voted puzzles are probably those that stay unsolved for a few hours, maybe with a few incorrect answers before the correct one. $\endgroup$ – Rand al'Thor Apr 29 '15 at 22:12
  • $\begingroup$ @randal'thor Still, I think there's a vote premium on single-step find-the-trick puzzles over more involved ones. $\endgroup$ – xnor Apr 29 '15 at 22:34
  • 1
    $\begingroup$ @xnor I think we all agree that democracy is not the best way to rate a puzzle. But hat is how SE is designed, so there's no point debating it. And, by the way, that is a slightly different topic from what is asked. $\endgroup$ – ghosts_in_the_code May 4 '15 at 16:09

You must log in to answer this question.

Browse other questions tagged .