UNplausible deniability

[lostdummy]

I got idea for this because on one of threads here I had impression that problem author had different solution in mind when he initially stated problem, compared to what he posted at end.

Not related to that specific issue, nor even if my impression was correct, what I started wondering was:

Is there any convenient way for problem authors to state solution at same time when they state problem, in such way that they can not change it later on, but at same time to prevent possible solvers to see what is that solution?

Solution to this should be:
1) 'convenient': very easy for author to do it, and even easier for others to check when later author post solution.
2) 'irreversible': very hard for solvers to guess correct solution only based on what is initially posted
3) 'protected': prevent from someone maliciously changing posts/data needed to verify that author did it correctly


Although I'm sure it could be done better, my current best solution for this is:

Author post salted hash of solution, where 'salt' comes from some finite set with unknown number of members.

Example rule:
1) author select name of any book from Amazon as 'salt'
2) they use exact name as appears there, but are free to change upper/lower case on any letter
3) they append ':' and their solution to this salt
4) they calculate SHA512 hash on above, for example using sites like www.tools4noobs.com/online_tools/hash/
5) they post resulting hash together with problem

Reason for salt is that without it, it would be very easy for solvers to guess solution in problems that have small set of possible solutions (they would just have to calculate hashes for all possible solutions).

Reason for salt to be from "finite set with unknown number of members" it that if we would allow ANY salt, than author could later change solution and find appropriate new salt to result in same hash - it IS hard to do with good hashes, but it is doable. But if we limit author to salts like I suggested (book names), then it is practically impossible they will find other book to match same hash. On other side, "unknown number' was needed because if we limit salt options too much, solvers could use all possible salt values with all possible results to find which one fits. So salt should be selected in such way that 'all possible salt values' are not known, and yet that it is limited.

I believe selection of book names (limited, but generally unknown set), with combination of upper/lower cases ( increase number of salts millions of times, yet still keep it very limited for author to find new salt+result that make same hash).

Also, above principle can work for large solutions too, not only numbers. For example, if I didn't post this wall of text as suggested solution already, and if i use above principle, I could have posted following (for text in first spoiler):

in first post where i stated problem, i could post hash of solution:
50054074ac3d5de8bcc9f42a54c28127e104cf35214c3a41227035767fc0595457599a54fb0adf31a69f2bbd344c75b0f3cc4093907da72a55a0a9d817c672ae

and in last post where I reveal solution I would post what I posted in first spoiler, and would add that my 'salt' was RINGWORLD.


That would make it easy for people to check if I really had this same solution in mind from start. Of course, its even easier when solution is something like single number ;p

*EDIT* added 3rd 'requirement'

[Zag]

This sounds a little like a solution in search of a problem. I mean, have we had a big problem with puzzle writers claiming a different solution was correct than the one they originally intended?

We do have a rather nice mechanism for solvers to check if their answer is correct without the puzzle author being around to confirm. We use it here, for instance, and here.

(Can you guess where those links go without following them?)

On the other hand, I could see your solution being used for a "Whoever guesses my number first wins $1 million" sort of game.

[Trojan Horse]

If the author of the puzzle that inspired this puzzle comes by and attacks you for doubting his honesty, you won't be able to plausibly deny that you were referring to his puzzle.

[lostdummy]

Zag,

I did not post this as idea to be used on forum, but as interesting problem - reason why we post majority of puzzles here.

So, while its nice if solution to this problem could be used anywhere (even if thats $1milion game), its finding best solution that was my intent, not using it ;p

BTW, links that you posted do not solve this problem. Because it allows author to change his intended solution later on - he just need to change password too.

[Nsof]

Commitment scheme
_________________
Will sell this place for beer

[lostdummy]

[Elethiomel]

I think the most convenient solution is to introduce a trusted party.

[Suspence]

[Jack_Ian]

I used MD5 here to allow users to verify their own answers.
It's not very safe for single words, but should be suitable for complete structured sentences. It would certainly be very difficult to construct a coherent sentence after the fact that had the same digest code.

[lostdummy]

[lostdummy]

[Jack_Ian]

If answer is 100, give MD5 hash X for "The answer to my wonderful puzzle is one hundred. Hope you enjoyed it." or some other long-winded phrase.

[lostdummy]

[Suspence]