I don't see your logic. Rot13(Jul jbhyq gur yvne unir gb cbvag gb gur qvssrerapr xrl?)

@Sinh rot13(Gur pbeerpg nafjre gb gur dhrfgvba "Juvpu xrl bcraf lbhe [gur yvne'f] ybpx?" vf bar cnegvphyne xrl: yrg'f pnyy guvf xrl N. Fb gur yvne jbhyq tvir bar bs gur bgure gjb nafjref: xrl O be xrl P. // Fb vs V vafgrnq nfx gur ulcbgurgvpny "jung jbhyq lbh cbvag gb vs V nfxrq [gung dhrfgvba] lrfgreqnl?", gur pbeerpg nafjre vf xrl O be xrl P: gubfr jbhyq or gur xrlf gung gur yvne jbhyq unir cbvagrq gb. Fb gur yvne zhfg tvir na vapbeerpg nafjre, juvpu zhfg or xrl N.)

If I understand it right rot13(Lbh ner gelvat gb znxr qbhoyr artngvir dhrfgvba gb znxr gur yvne fnl gur gehgu. V qba'g guvax lbhe dhrfgvba znxr gung pyrne.)

But... You didn't know which one is the truthteller and liar right? It leaves you the two possible answers. However, you can just give one lock each to them and finding the key to that lock with the same question (even if both locks are indistinguishable)! cmiiw

Maybe the liar always answers lies deterministically. For example, if the choices are enumerated "Key for my lock, Key for other lock, Key that does nothing", he always answers the last in the list. So when asked yesterday, "which key will open your lock", he would always answer "key that does nothing". Knowing this, when asked today what he would have said yesterday, he can now say "Key for the other lock".

Would this work for any number of keys?

@DrXorile Yes, this should work with any number of keys. (It would also work with any number of locks, provided there were as many guards as locks -- and with any combination of truthtellers and liars.)

@Trenin Good point! That can be fixed, at the cost of being an even more cumbersome question than before.

@Deusovi I can see you're fixed on this logic of thinking, while I not say my intended is the best but it's definitely not that complex. Here is a little hint to start rot13(Jul gurer ner 3 xrlf?)

@Sinh This may not be your intended answer, but it's completely correct. I'm not interested in "guess which solution the questioner is thinking of".

It's not my intention at all, and I not say your answer is wrong, just wanted to let you know there's another way. I mean your current answer is very complex & cumbersome. And I also hope to see a answer better than mine

Would have +2'ed simply for being correct despite not giving the asker their intended answer. Sometimes those are the best kind of riddle answers.

Can someone explain the gibberish comments above? I'm new here. What on earth is that?

@temporary_user_name Oh, that's Rot13, a common cipher used to hide spoilers without explicit spoiler formatting. Just copy-paste the text into that site to decode it.

@SayedMohdAli No, that's not true - yesterday, the liar could have pointed to either key B or key C. So if the liar answers B or C, they have made a true statement (that that is one of their possible answers). So they must answer A. [...]

[...] There was no question actually asked yesterday. It's a hypothetical that has two possible 'true' answers. The liar must lie, and must give the impossible option, which is the answer we were looking for.

[...] It's true that giving a correct answer would be more misleading. But they are not a misleading guard - they are a liar, which means they give false answers to questions.

@Deusovi wrote "a liar will always give an incorrect answer"... actually, I know a liar who does occasionally give a correct answer. It really throws us off when he does, because we are not expecting it to be the truth. ;)

You're just assuming the liar can't/won't "double lie" while making his choice.

@kbtzr What do you mean by "double lie"? I am assuming that the liar will give a false statement, yes, because that's the premise of the puzzle (and of all truthteller-vs-liar puzzles). My question allows for both of the two wrong options to be the truth, so the only way the liar can lie is to give the right option. (It's true that giving a wrong option would be more misleading, but to mislead and to lie are different things.)

I think "could've pointed to" might be a little more solid than "might have pointed to". They may answer extraneous key on even days and wrong key on odd days as the lie.

@John That's exactly what that extra clause is for.

Ah. Okay. I kind of glossed over the algorithm part. Thanks.

Maybe the OP is looking for, "what key will you point to". That's the same thing and fewer words but idk.

Hi, I proposed a modification to OP that makes your answer wrong. Not that I didn't like it, my upvote is unfortunately the only justice I can do to it.

@PatJ As far as I can tell, that wasn't a loophole - OP seemed to say that my answer was correct. Please do not edit questions to change their meaning.

@Deusovi I think you should move all this conversation in the chat :P...

@Deusovi sorry for the wrong edit, I'm new here and knew that my edit would only be allowed if accepted by reviewers

(or OP)

@PatJ no worries! generally we do less editing of people's posts here, because there can often be clues hidden in the formatting or precise wording of a puzzle

Most of the time, if I see an edit that should be done, I'll ask OP first to make sure it doesn't break anything. With some puzzle types, it's clearer when edits are fine (pure logical deduction puzzles in particular come to mind), but I'm pretty hesitant to just do non-tag edits in most cases

(It's completely possible that I may be wrong about what the OP intended, though - I'm not sure, which is why I would hold off until I asked them directly)