Camp Ivanhoe

[Young Writer]

Are you certain that still works if we just know THAT they are in the correct position and not which numbers they are?

[Young Writer]

That was at Richie

[Richie Tenenbaum]

Are you certain that still works if we just know THAT they are in the correct position and not which numbers they are?

Logical deduction on the iterations narrows down which ones were correct until you lock them in

[Oracle]

My problem with 99994 is that they may correctly deduce that that was the highest sum of digits we could reach, whereas 44444 doesn't really give any deductions and it's highly unlikely for the tiebreaker to come into play.

---

For our guessing, the strategy I've come up with is:

First guess; +1 to all digits
(For example, if we receive 37966 guess 48077)
Then -1 to all digits
+2 to all digits
-2 to a digits
+0 to all digits
+3 to all digits
+4 to all digits
+5 to all digits

Because we know that the differences must sum to 15 we may be able to skip one or more of those. What this does is gets us to at most 120 combinations left after at most 8 guesses, because we will know the amount slots need to change by, we just need to arrange these changes in the correct order.

Suppose our changes our A, B, C, D, and E; Our next guess would apply them in order ABCDE

If we have 5 match we're done (9 total).
4 match is impossible.
If we have 3 match it takes at most an additional 5 guesses (14 total)
If we get 2 match it takes at most an additional 6 or 8 (idk which) guesses (15 or 17 total)
If we get 1 match it at most an additional 8 guesses. (17 total)
If we get 0 matches we than submit BCDEA and go to that many matches (18 total)


Which means unless I messed up the numbers or misunderstood something it will take at most 18 guesses.

The fact I'm having a hard time following this probably means if you are certain it's likely a good strategy.

[Anthony Adams]

What if we do 79978 5+3+4+2+1

Well, that's still 40, so 40 is prolly the max for this set of numbers?

Yeah, I think so

[Richie Tenenbaum]

I think it's a mistake to sleep on the possibility of a tie in the solve so doing one of these scrambles that sums to 40 is the right move

[Young Writer]

The real question is that while (and I'm trusting Samuel here) that guarantees 18 at the most, is a standard logical deduction/mastermind strategy while also making all guesses differ by 15 better?

[Young Writer]

Also the one that begins with a 7 seems quite good.

[Oracle]

Also the one that begins with a 7 seems quite good.

I don't hate 79978

[Richie Tenenbaum]

79978 is good by me

[Anthony Adams]

No objections here

[Richie Tenenbaum]

The big issue with having the first two guesses be a simple 12345 then 67890 is you learn almost nothing that can inform your third guess

[Oracle]

My only fear with that is what if they somehow figure out the 54321 pattern with it, but at this point it's no worse than 99976 to me seeing as I don't know how they'd logic that out in any way shape or form, patterns in our coding scares me though.

[Richie Tenenbaum]

My only fear with that is what if they somehow figure out the 54321 pattern with it, but at this point it's no worse than 99976 to me seeing as I don't know how they'd logic that out in any way shape or form, patterns in our coding scares me though.

It's 5 3 4 2 1

[Anthony Adams]

My only fear with that is what if they somehow figure out the 54321 pattern with it, but at this point it's no worse than 99976 to me seeing as I don't know how they'd logic that out in any way shape or form, patterns in our coding scares me though.

Actually, it's 53421, no?

[Young Writer]

Anyway, as for the strategy, if we get a combo of numbers that has 3 correct, at worst it's likely like 5 iterations before all 3 are locked in, and then it's just a matter of moving in pairs of the remaining 5 numbers until you have the right ones. ANd if one set of numbers only has 1 correct answer, then the other set likely has 3 or 4 correct numbers. Of course numbers repeating do make it more difficult but it feels more efficient than the doing 8 guesses to limit the pool and still maybe needing 18.

[Oracle]

Oh yeah haha. Okay sorry my bad.

[Sam Shakusky]

There's two types of information from guesses to extract, what numbers are in the code and what the differences are. By making our code all 4's we would make those two sources of information the exact same, which I think minimizes the amount of information they can get per guess.

Have to leave now so submit whatever for the shuffle (even if it is summing to 40) but I'd vote for not summing to 40.



@Writer that increases our chance of hitting on a given guess from 0/100000 to 1/100000, I think knowing how many of each of +1, +2, etc there are is much more valuable because it limits our worst case scenario drastically.

[Sam Shakusky]

The above is referring to #406

[Young Writer]

I mean to be fair, if we want to test strategies, it's rather easy to randomly generate a 5 number sequence and scramble it here. We need not start working on deducing the combination immediately.

[Oracle]

My one fear with 44444 is if they guess 99999 and get 2 right, they know once they place those 9s properly the other digits are only a total of 5 off limiting the amount of guess potentials for the other 3 numbers that they may stumble along the way to place.

[Oracle]

My vote is for 79978.

[Richie Tenenbaum]

My vote is for 79978.

[Young Writer]

Same, someone put it in then?

[Oracle]

Where do we put it?