9
$\begingroup$

Question

What advice, strategies, keys, or elements would you suggest for creating challenging puzzles that aren't (reasonably) computer soluble?

Details

I'm a programmer, so I can appreciate the argument that writing code to solve a puzzle is as enjoyable as solving it directly; don't take this question as any slight against computational solving. But I want to write good puzzles that people enjoy struggling with for a day or two; problems that can be solved via program typically lack this, in my experience.

Sometimes, as in the case of this puzzle, the inclusion of computers seems to take some of the fun out of it (as well as overshadow manual answers that deserve recognition) but there's no hard-and-true way to differentiate computer-generated solutions from manual ones.

Now, normally I'd say that riddles (e.g. What am I?) are the go-to solution for this problem, but they can get old very quickly (as evidenced here, based on some of the discussion). Another is cryptograms, but those also feel a bit tired sometimes.

Do any veterans of the site have useful tips or advice in this area?

$\endgroup$
5
  • $\begingroup$ Also, +1, perceptive question! $\endgroup$
    – A E
    Aug 27, 2015 at 15:49
  • 1
    $\begingroup$ Just make it so they have to include hardcore math proof lol. $\endgroup$
    – warspyking
    Aug 28, 2015 at 16:34
  • 1
    $\begingroup$ @warspyking I had considered this, but wasn't sure where the line is drawn on appropriateness of 'mathy' problems on PSE $\endgroup$
    – NeedAName
    Aug 28, 2015 at 16:41
  • 1
    $\begingroup$ @NeedAName Have you seen this canonical meta post re mathsy on-topicness? $\endgroup$ Aug 28, 2015 at 17:06
  • $\begingroup$ @randal'thor I have now. Thanks for the referral $\endgroup$
    – NeedAName
    Aug 28, 2015 at 17:32

6 Answers 6

6
$\begingroup$

I've posted a link to this article elsewhere, but I think it might be especially relevant to your question here.

It discusses some things to consider when trying to design a good puzzle, and I think all of the example puzzles in the article would be nigh-impossible to solve with a computer. (At the very least, a computer wouldn't provide any significant advantage.)

I think there are two particular aspects of the puzzles that give them this quality:

  1. A visual aspect. When a puzzle involves looking at a picture of some sort and deriving clues or somehow altering or rearranging it, that's something that is very difficult to do with a computer. Note that I'm not suggesting that you take a simple computer-solvable puzzle and publish it in an image format; I'm suggesting that the image be an integral part of the puzzle.
  2. An "intuitive leap". This, I think, is what is referred to by the tag on Puzzling.SE. It means that the means or method of solving the puzzle is not immediately apparent. You look at a cryptogram and immediately know that you need to find a series of letter substitutions that come out to plain English. With an , you see a series of words/pictures/symbols etc., and must first figure out what is being asked of you before you can go about solving the puzzle.

Basically, I think what it boils down to is that creating a good, un-brute-forceable puzzle (if that's a word) requires a lot of time and effort.

If you can create a cryptogram in 5 minutes by running some text through a website, you can't really expect that someone else won't be able to solve the same cryptogram in 5 minutes using a computer as well...

Here are a few examples of puzzles on this site that I think embody this answer well:

Perplexing Fireworks
Sunset Parliament Sacrificing Posers
What is the central color?
A treasure hunt for the magic artefact

$\endgroup$
2
  • $\begingroup$ +1 for the feedback and links. I hadn't really considered using images before $\endgroup$
    – NeedAName
    Aug 27, 2015 at 16:28
  • 1
    $\begingroup$ @NeedAName Note that using images does make your puzzle less accessible -- to those with visual impairments, or to those using browsers that can't properly load images (or perhaps aren't large enough to view them (e.g. smartphone)). $\endgroup$ Aug 27, 2015 at 16:30
6
$\begingroup$

the tag honestly, to me, seems like a crutch. If a puzzle is designed in such a way that its main appeal comes from not using a computer, then I feel people will enjoy going machine free to solve it.

However, if the puzzle is nothing more than an exercise in frustration until a computer gets involved, then it's more of a failure of design. That's not an insult; we're all guilty of making mistakes in our lives and posting bad puzzles is as important as posting good ones, because it builds our library and gives us a more impressive array of skills that are necessary for making and solving puzzles.

Programming, at its core, is literally problem solving. Just be aware it's only natural for programmers to want to try and use those skills to solve puzzles.

$\endgroup$
5
$\begingroup$

I'm going to come back to this and extend it later, but I'll start off now anyway.

Every puzzle falls into one of 3 categories:

  1. someone posts a computer-assisted answer
  2. it can't be solved using a computer
  3. it has a tag (i.e. it could be solved using a computer but the OP specifically asks people not to do this)

For point 3, I created the tag 5 months ago and it's only had 25 questions. For point 1, there aren't that many questions answered by brute-forcing on this site - people with those sorts of questions tend to go to PPCG instead. So the vast majority of the puzzles here are in point 2, i.e. of exactly the sort you're looking for.


OK, that doesn't quite answer your question :-) You want tips on how to construct such questions. So let's go through it tag-by-tag, starting from the top.

  • As you say yourself, this is the easiest way to write a puzzle that can't be done with the help of a computer. And they'll never get old: they're works of art and people can always come up with new artistic ideas. If exactly the same idea gets flogged like a dead horse (hello spaghetti-parties/rebus-puzzles/warspyking), then that can get old; but riddles in themselves? Never.

  • There are whole books full of mathematical puzzles, most of which computers can't help you with. Computers can do calculations, sure. They can solve linear equations. But anything mathsy that could be solved by a computer would count as a maths problem and be off-topic anyway. Maths puzzles generally need an unexpected approach and/or an 'aha!' moment, both of which require a human brain.

  • Again, there are whole books full of these. All the old classics (knights-and-knaves, Monty Hall, hats of different colours) have been thrashed out already, so it's now a matter of writing variants. Boring variants can be solved using exactly the same method as the original, and a computer might be able to do that. Interesting variants need new ideas. That said, there are a lot of logic puzzles that can be brute-forced, like this one or this one.

to be continued

$\endgroup$
1
  • 3
    $\begingroup$ I'm famous. I really don't care why. But I'm famous. $\endgroup$
    – warspyking
    Aug 28, 2015 at 20:32
3
$\begingroup$

One other aspect - often hard to do for the puzzle author!: If you manage to create a puzzle which can be solved by deduction (to a large extend), then you can make it a condition of the solution to provide the arguments and not only the solution.

Yes, people may still brute-force it, but the computer will not tell them why one can come to this conclusion.

$\endgroup$
1
$\begingroup$

In my experience, no-computers is for puzzles with some amount of computation, that a computer script would trivialize. For example, how many zeroes does 100! end in? What is the solution to [insert grid logic puzzle]? Usually the point of the puzzle is finding a cleverer way than brute force to solve it.

For these puzzles, you can either try to make the puzzle infeasible for computers (e.g. by asking about 1000000000! instead) or add a no-computers tag. The first approach is more elegant when possible. Unfortunately, some puzzles cannot be made infeasible for computers without either killing the clever approach or making the clever approach annoyingly hard for humans.

$\endgroup$
2
  • $\begingroup$ So, how many zeroes does 100! end in ? ;c) $\endgroup$
    – BmyGuest
    Aug 29, 2015 at 18:40
  • 1
    $\begingroup$ @BmyGuest 24 zeroes. And 1000000000! ends in 249999998 zeroes. $\endgroup$
    – Rohcana
    Sep 11, 2015 at 20:29
1
$\begingroup$

I'd say we can take this question as a nice example: Nerds, Jocks, and Lockers

The OP did not specify (I'm not entirely sure if that was a tag then). However when me and somebody else provided programs/ scripts to solve this, they were cast aside as "inelegant".

Instead he accepted the answer which provided math proof.

So I'd say the best way is to simply ask for a piece of pencil and paper (just a bit of math that can be done easily on paper) proof. No need to overcomplicate things. A computer shows you the answer, not the mathematical steps to get there. You should filter out the programmers (such as myself) from the true mathematical.

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge that you have read and understand our privacy policy and code of conduct.

Not the answer you're looking for? Browse other questions tagged or ask your own question.