The Two Squares Puzzle | Hacker News Reader

The Two Squares Puzzle

futilitycloset.com

39 pointssohkamyung16 comments

An interesting challenge is to try to generate this puzzle, or one like it, once you know the answer.

Given any existing grid, you can implement a "step" function in code that generates the next grid describing it. You'll always find a cycle by stepping over and over (the numbers can never get too high, so the space is finite, so you'll always converge to a cycle rather than diverging). The goal, then, is to find a cycle of length 2 (or, better yet, a fixed point that maps to itself).

I just coded this up to try it out, and one simple starting point gave a cycle of length 27. I tried randomly generating starting points and I can generate a 3-cycle but haven't hit any 2-cycles yet. So maybe there's another trick to get the configuration from the puzzle.

Edit: nevermind, I misinterpreted the puzzle. All letters in the grid count, so "FOUR F'S" has two Fs and an S. Accounting for that, a randomized approach quickly finds a 2-cycle. But I guess it's interesting that a relatively minor tweak like that has a different outcome.

http://yawn.io/2019/08/28/pangrammatic-autograms.html was recently posted here and uses a SAT solver for a similar problem. It seems like that might be applicable here too.

I gave up, and boy am I glad I did.

Felt a bit like a typical job interview.

I wish puzzles would indicate whether it's something language/decimal-notation based, vs something fundamentally mathematical based, before spending time on it

And people say security through obscurity doesn't work! As you can see, it's a very effective time waster that can be layered on top of existing security.

For me it's more about avoiding the kind of puzzles I don't like. I would be disappointed if I spent hours on this only to find out the solution is language-based.

Security through obscurity is not effective in the sense that dedicated attackers, or those with large amounts of resources can _eventually_ crack the code. It gets such bad attention because people NOT in the security field (or aware of security methods) tend to view it as a good line of defense.

I've seen implementations where the obscurity was client-side JavaScript that modified the POST. The coder thought this would protect the server from malicious payloads.... Unfortunately the obscurity is lost when you provide all visitors with the function to do the obfuscation.

I don't think this would be considered "security through obscurity". Security though obscurity is a security mechanism that relies on a secret only the implementer knows about. Like hiding a key to your house under a specific rock in the yard that only you knows the location of and hopefully isn't discovered by someone else. A puzzle is solvable and would only protect you from people who aren't good at puzzles.

I think it is. You are making the puzzle more difficult to solve by obscuring what the object of the puzzle even is. If you don't even know what the goal is, how can you start solving it?

If I gave you an AES-encrypted cipher and told you to crack it, you would at least know what the end goal is (break AES). But if I just said, "crack the following cipher":

a13908x91827p93892a...

Where would you even start? You could randomly try things, but the layer of obscurity wastes your time such that you don't even know what goal is other than some vague, nebulous "find the hidden message" goal.

Then I reveal that the cipher is actually just AES but with a layer of obscurity on top of it using a quirk of the Afrikaans spoken language to transform the cipher into a baffling string.

These squares don't describe each other because you need external information, namely of the English language.

It's like saying that 56 describes a house. Yes, page 56 of some architecture book might show a house. But saying that 56 describes that house is just wrong.

The puzzle is flawed.

I would go further and say it’s not really even a puzzle. It’s an interesting construction but in no way is this a meaningful puzzle. There’s so much knowledge that isn’t captured or really discoverable via the puzzle itself. “Oh, it’s encoded English. Oh, the first letters are just excluded because it’s convenient.”

This is basically not solvable unless you’ve been given a bunch of constraints that aren’t actually part of the puzzle.

I don't believe anyone anywhere could solve that.

That is certainly correct. The name of the blog is "futility closet", so...

This “puzzle” reminded me of Nick Land’s essay “Qabbala 101”, which remarks on patterns found in “primitive numerization” schemes such as this. http://files.eshkolot.ru/C1_Nick_Land.pdf

I think this is a form of autogram: https://en.m.wikipedia.org/wiki/Autogram

The article even provides a pair of mutually-describing sentences. Hofstadter has described these in at least one of his books, but I don't recall if he discussed how to generate them.

Reply on news.ycombinator.com