67 replies to this topic

[octopuppy]

Password accepted, system shutting down.

Oh, no, the former Evil Mastermind has cleverly regained control of the system and stopped it.
The rest of the states are safe (... for now).

P.S.: However, in order to reprogram the system, he needs to explain the algorithm so that the system recognizes its master.

This I can't yet do. There's a couple of [boolean] states that depend on something I haven't figured out (leaving 4 possible passwords, of which DudleyDude guessed one and I did the other 3). Still pondering...

[octopuppy]

Well, hell with it, I'll share what I know with the other hackers and see who can get to the bottom of it:

Spoiler for Solution so far:

Each letter is defined differently:
1) 3rd letter of command
2) 1 if [unknown 1], otherwise 0
3) Last letter if command length is odd, otherwise 2nd to last
4) Number of vowels in command
5) 1st letter in command if [unknown 1], otherwise 2nd letter
6) Length of command mod 5
7) 4th letter in command if [unknown 2], otherwise 3rd letter
8) Vowel count if [unknown 2], otherwise vowel count + 1

Here's the values of unknowns 1 and 2 for various commands

NEVADA: true, false
MAINE: false, false
SOUTHCAROLINA: false, false
HAWAII: true, false
IOWA: false, true
LOUISIANA: false, false
NORTHDAKOTA: false, false
SOUTHDAKOTA: false, false
NORTHCAROLINA: false, false
NEBRASKA: false, true
MARYLAND: false, true
NEWYORK: false, true
OHIO: true, false
COLORADO: true, false
INDIANA: false, false
CONNECTICUT: false, true
IDAHO: true, true
UTAH: true, false
OKLAHOMA: false, true
OREGON: false, true
TEXAS: true, true
ABORT: false, true

As to what those unknowns might be, I'm stumped...

[Cherry Lane]

Well, hell with it, I'll share what I know with the other hackers and see who can get to the bottom of it:

Spoiler for Solution so far:

Each letter is defined differently:
1) 3rd letter of command
2) 1 if [unknown 1], otherwise 0
3) Last letter if command length is odd, otherwise 2nd to last
4) Number of vowels in command
5) 1st letter in command if [unknown 1], otherwise 2nd letter
6) Length of command mod 5
7) 4th letter in command if [unknown 2], otherwise 3rd letter
8) Vowel count if [unknown 2], otherwise vowel count + 1

Here's the values of unknowns 1 and 2 for various commands

NEVADA: true, false
MAINE: false, false
SOUTHCAROLINA: false, false
HAWAII: true, false
IOWA: false, true
LOUISIANA: false, false
NORTHDAKOTA: false, false
SOUTHDAKOTA: false, false
NORTHCAROLINA: false, false
NEBRASKA: false, true
MARYLAND: false, true
NEWYORK: false, true
OHIO: true, false
COLORADO: true, false
INDIANA: false, false
CONNECTICUT: false, true
IDAHO: true, true
UTAH: true, false
OKLAHOMA: false, true
OREGON: false, true
TEXAS: true, true
ABORT: false, true

As to what those unknowns might be, I'm stumped...

Well, I was left with the same unknowns as you, plus a problem with the third character.

Spoiler for you see,...

I missed the last/second to last bit. Instead I was working with: is the seventh character for every input of at least seven letters, and some other character if less than seven. But there was no consistency for those less than seven. Interesting that for every input of at least seven characters, the seventh one ended up being correct.

[araver]

Well, hell with it, I'll share what I know with the other hackers and see who can get to the bottom of it:

Spoiler for Solution so far:

Each letter is defined differently:
1) 3rd letter of command
2) 1 if [unknown 1], otherwise 0
3) Last letter if command length is odd, otherwise 2nd to last
4) Number of vowels in command
5) 1st letter in command if [unknown 1], otherwise 2nd letter
6) Length of command mod 5
7) 4th letter in command if [unknown 2], otherwise 3rd letter
8) Vowel count if [unknown 2], otherwise vowel count + 1

Here's the values of unknowns 1 and 2 for various commands

NEVADA: true, false
MAINE: false, false
SOUTHCAROLINA: false, false
HAWAII: true, false
IOWA: false, true
LOUISIANA: false, false
NORTHDAKOTA: false, false
SOUTHDAKOTA: false, false
NORTHCAROLINA: false, false
NEBRASKA: false, true
MARYLAND: false, true
NEWYORK: false, true
OHIO: true, false
COLORADO: true, false
INDIANA: false, false
CONNECTICUT: false, true
IDAHO: true, true
UTAH: true, false
OKLAHOMA: false, true
OREGON: false, true
TEXAS: true, true
ABORT: false, true

As to what those unknowns can be, I'm stumped...

Spoiler for Small response from the system to the algorithm proposed vs. the original algorithm

Original algorithm for character 3 is not that simple, but it is indeed impossible to distinguish from the proposed algorithm in the values seen so far.

There are only 7 states/commands (which have not been played) which actually contradict the proposed rule: "3) Last letter if command length is odd, otherwise 2nd to last".

However, it may be impossible to guess without seeing these states so I assume there is no way the original hash algorithm could be reconstructed - Occam's Razor always favors simplicity.

However, taking into account that:
-for WISCONSIN the third letter of the password is S.
-for WASHINGTON the third letter of the password is G.
can you deduce the original algorithm for the 3rd letter?

Spoiler for Response to the rest of the algorithm proposed

The rest is correct, or more clearly dependencies of the characters in the password to the unknowns is correct.

Again as few states have been played, it might or might not be possible to correctly distinguish unknown1 and unknown2 (as is the case with the 3rd character-spoiler above).

IMHO, this was generally the case in classic cryptography: being able to read at least some of the encrypted messages may count as breaking an encryption algorithm for practical purposes, but does not necessarily be enough to reveal the actual algorithm.

[dawh]

Well, hell with it, I'll share what I know with the other hackers and see who can get to the bottom of it:

Spoiler for Solution so far:

Each letter is defined differently:
1) 3rd letter of command
2) 1 if [unknown 1], otherwise 0
3) Last letter if command length is odd, otherwise 2nd to last
4) Number of vowels in command
5) 1st letter in command if [unknown 1], otherwise 2nd letter
6) Length of command mod 5
7) 4th letter in command if [unknown 2], otherwise 3rd letter
8) Vowel count if [unknown 2], otherwise vowel count + 1

Here's the values of unknowns 1 and 2 for various commands

NEVADA: true, false
MAINE: false, false
SOUTHCAROLINA: false, false
HAWAII: true, false
IOWA: false, true
LOUISIANA: false, false
NORTHDAKOTA: false, false
SOUTHDAKOTA: false, false
NORTHCAROLINA: false, false
NEBRASKA: false, true
MARYLAND: false, true
NEWYORK: false, true
OHIO: true, false
COLORADO: true, false
INDIANA: false, false
CONNECTICUT: false, true
IDAHO: true, true
UTAH: true, false
OKLAHOMA: false, true
OREGON: false, true
TEXAS: true, true
ABORT: false, true

As to what those unknowns might be, I'm stumped...

I'm at the same point you and CL are at, but one thing I did notice was that for "Unknown1":

Spoiler for What I noticed

If true, no consonants will be consecutive to each other. It's not an equivalence because MAINE is false, so there's obviously more to it than that. But if you look, every time the consonants and vowels alternate, it's a 1. Then OHIO and HAWAII preclude that from being the solution.

[Cherry Lane]

Spoiler for Small response from the system to the algorithm proposed vs. the original algorithm

Original algorithm for character 3 is not that simple, but it is indeed impossible to distinguish from the proposed algorithm in the values seen so far.

There are only 7 states/commands (which have not been played) which actually contradict the proposed rule: "3) Last letter if command length is odd, otherwise 2nd to last".

However, it may be impossible to guess without seeing these states so I assume there is no way the original hash algorithm could be reconstructed - Occam's Razor always favors simplicity.

However, taking into account that:
-for WISCONSIN the third letter of the password is S.
-for WASHINGTON the third letter of the password is G.
can you deduce the original algorithm for the 3rd letter?

Ahh, for those (unplayed) states,

Spoiler for my observation

that for inputs of at least seven letters, use the seventh letter as the 3rd character

is still true. But it doesn't help me get to the ABORT solution any better.

Edited by Cherry Lane, 27 October 2010 - 10:13 PM.

[octopuppy]

Spoiler for Small response from the system to the algorithm proposed vs. the original algorithm

Original algorithm for character 3 is not that simple, but it is indeed impossible to distinguish from the proposed algorithm in the values seen so far.

There are only 7 states/commands (which have not been played) which actually contradict the proposed rule: "3) Last letter if command length is odd, otherwise 2nd to last".

However, it may be impossible to guess without seeing these states so I assume there is no way the original hash algorithm could be reconstructed - Occam's Razor always favors simplicity.

However, taking into account that:
-for WISCONSIN the third letter of the password is S.
-for WASHINGTON the third letter of the password is G.
can you deduce the original algorithm for the 3rd letter?

Looks like Cherry Lane was right after all!

Spoiler for Response to the rest of the algorithm proposed

The rest is correct, or more clearly dependencies of the characters in the password to the unknowns is correct.

Again as few states have been played, it might or might not be possible to correctly distinguish unknown1 and unknown2 (as is the case with the 3rd character-spoiler above).

IMHO, this was generally the case in classic cryptography: being able to read at least some of the encrypted messages may count as breaking an encryption algorithm for practical purposes, but does not necessarily be enough to reveal the actual algorithm.

Feel free to give us a few more examples if you think it would help. I'm still stumped!

[araver]

Well, I was left with the same unknowns as you, plus a problem with the third character.

Spoiler for you see,...

I missed the last/second to last bit. Instead I was working with: is the seventh character for every input of at least seven letters, and some other character if less than seven. But there was no consistency for those less than seven. Interesting that for every input of at least seven characters, the seventh one ended up being correct.

Spoiler for Your hypothesis ...

cannot constitute a full rule since there is a small number of states with less than seven letters. It is true by pure coincidence (seven and last or second to last characters coinciding in states)

Why it contradicts the Game Rules: According to the game rules I was not allowed to use anything that does not appear in at least 50% of the cases. There are only 12 commands with less than 7 characters (4,5,6). So I cannot choose one rule for more than 7 characters and another rule for 6 characters or less - that would be cheating the rules. I must use the same deterministic rule for all cases.

BTW, If the situation were reversed, if one thought he had a rule for less than 7 or less than 8 characters, one could always try it on ABORT and see the result. Doesn't hurt

[araver]

Looks like Cherry Lane was right after all!

Yes,

Spoiler for to a point she was closer on commands

between 7 to 13 letters. If different states were played (for example no states with 11 or more characters), her rule would have been flawless ... except for commands with 4,5,6.

[octopuppy]

I'm at the same point you and CL are at, but one thing I did notice was that for "Unknown1":

Spoiler for What I noticed

If true, no consonants will be consecutive to each other. It's not an equivalence because MAINE is false, so there's obviously more to it than that. But if you look, every time the consonants and vowels alternate, it's a 1. Then OHIO and HAWAII preclude that from being the solution.

Maybe double vowels are only allowed at the end? OREGON (false) remains as a counterexample, but I bet you're not far off the answer.