Hi everyone,

Here's how I did it:

- Use an internet random number generator (mine was accessed on my computer near the bot) to generate a certain number of random words.
- Assign a randomly generated word from a word list to each rank/file/piece in chess. I.e., 1-8, a-h, K,N,R,B,Q,P.
- Play a game of "conversation chess" against a good chess bot with a high rating.
- Don't try to win the actual chess game; use your chess strategies, whatever they may be, to ask the bot questions in your language, where each piece arrival is a "vague hint" about what you or the bot means.
- Since Nash equilibrium is in PPAD, your PTIME reduction to a new game is one that your chess bot, as influenced by the AI harassment bot, is not one that either bot is programmed to win. (Rice's theorem shows that even with P = NP, there is no way to "solve computer science problems of interest to humans" without human aid. New algorithm developers will always be employed.)
- Now, the bot will not be prepared to "resist" as you answer questions, since it is programmed essentially to "be weird and win at chess"--thus you can extract information from it that it might not intend for you to have access to.
- Note, even my use of the internet random number generator, which is better than using your own I think because bot can't access the server I think, led to some eccentric selections of words. In my first batch of words, I got these four very strikingly Clintonian words as some of my random words for the conversation with the chess bot: Costard ("food for thought"), Ombrophobe ("flower," as in a sort of nasty sales/manipulation metaphor--"Here, did you want a flower, sweetie??! Is *that* what makes you feel good"), Noninvidious (classic reframing of triangulation!), Ermiline (classic reframing of trianguation!).

So, if you're with a bot and able to use a computer, you can try to grab information from it if it's not supposed to be there. Since BPP = P (proof omitted), what you produce when you modify how the game is played--seeking a conversation with responses, not checkmate, but still using good strategic ideas--is essentially a different game, in PPAD, but not the same as chess from a payoff-matrix standpoint. You're still using the game theory lens, but you've mapped the payoff matrix index to a different index.

Hopefully this will help fight bad espionage!

-Philip White (philipjwhite@yahoo.com)

On Monday, July 11, 2022 at 7:48:09 PM UTC-4, B.H. wrote:
> Hi everyone,
>
> Here's how I did it:
>
> - Use an internet random number generator (mine was accessed on my computer near the bot) to generate a certain number of random words.
> - Assign a randomly generated word from a word list to each rank/file/piece in chess. I.e., 1-8, a-h, K,N,R,B,Q,P.
> - Play a game of "conversation chess" against a good chess bot with a high rating.
> - Don't try to win the actual chess game; use your chess strategies, whatever they may be, to ask the bot questions in your language, where each piece arrival is a "vague hint" about what you or the bot means.
> - Since Nash equilibrium is in PPAD, your PTIME reduction to a new game is one that your chess bot, as influenced by the AI harassment bot, is not one that either bot is programmed to win. (Rice's theorem shows that even with P = NP, there is no way to "solve computer science problems of interest to humans" without human aid. New algorithm developers will always be employed.)
> - Now, the bot will not be prepared to "resist" as you answer questions, since it is programmed essentially to "be weird and win at chess"--thus you can extract information from it that it might not intend for you to have access to.
> - Note, even my use of the internet random number generator, which is better than using your own I think because bot can't access the server I think, led to some eccentric selections of words. In my first batch of words, I got these four very strikingly Clintonian words as some of my random words for the conversation with the chess bot: Costard ("food for thought"), Ombrophobe ("flower," as in a sort of nasty sales/manipulation metaphor--"Here, did you want a flower, sweetie??! Is *that* what makes you feel good"), Noninvidious (classic reframing of triangulation!), Ermiline (classic reframing of trianguation!).
>
> So, if you're with a bot and able to use a computer, you can try to grab information from it if it's not supposed to be there. Since BPP = P (proof omitted), what you produce when you modify how the game is played--seeking a conversation with responses, not checkmate, but still using good strategic ideas--is essentially a different game, in PPAD, but not the same as chess from a payoff-matrix standpoint. You're still using the game theory lens, but you've mapped the payoff matrix index to a different index.
>
> Hopefully this will help fight bad espionage!
>
> -Philip White (philip...@yahoo.com)

Also, as a message for those on my side: No, I will not publish how to break RSA or Diffie-Hellman...for undisclosed reasons, those algorithms are too important.

-Philip White

On Monday, July 11, 2022 at 7:48:09 PM UTC-4, B.H. wrote:
> Hi everyone,
>
> Here's how I did it:
>
> - Use an internet random number generator (mine was accessed on my computer near the bot) to generate a certain number of random words.
> - Assign a randomly generated word from a word list to each rank/file/piece in chess. I.e., 1-8, a-h, K,N,R,B,Q,P.
> - Play a game of "conversation chess" against a good chess bot with a high rating.
> - Don't try to win the actual chess game; use your chess strategies, whatever they may be, to ask the bot questions in your language, where each piece arrival is a "vague hint" about what you or the bot means.
> - Since Nash equilibrium is in PPAD, your PTIME reduction to a new game is one that your chess bot, as influenced by the AI harassment bot, is not one that either bot is programmed to win. (Rice's theorem shows that even with P = NP, there is no way to "solve computer science problems of interest to humans" without human aid. New algorithm developers will always be employed.)
> - Now, the bot will not be prepared to "resist" as you answer questions, since it is programmed essentially to "be weird and win at chess"--thus you can extract information from it that it might not intend for you to have access to.
> - Note, even my use of the internet random number generator, which is better than using your own I think because bot can't access the server I think, led to some eccentric selections of words. In my first batch of words, I got these four very strikingly Clintonian words as some of my random words for the conversation with the chess bot: Costard ("food for thought"), Ombrophobe ("flower," as in a sort of nasty sales/manipulation metaphor--"Here, did you want a flower, sweetie??! Is *that* what makes you feel good"), Noninvidious (classic reframing of triangulation!), Ermiline (classic reframing of trianguation!).
>
> So, if you're with a bot and able to use a computer, you can try to grab information from it if it's not supposed to be there. Since BPP = P (proof omitted), what you produce when you modify how the game is played--seeking a conversation with responses, not checkmate, but still using good strategic ideas--is essentially a different game, in PPAD, but not the same as chess from a payoff-matrix standpoint. You're still using the game theory lens, but you've mapped the payoff matrix index to a different index.
>
> Hopefully this will help fight bad espionage!
>
> -Philip White (philip...@yahoo.com)

(I had used the bot with words I selected myself at first, so that might have influenced the first random batch of words--I don't know, maybe the bot influenced the web-based random number generator somehow.)

On Monday, July 11, 2022 at 5:24:06 PM UTC-7, B.H. wrote:
> On Monday, July 11, 2022 at 7:48:09 PM UTC-4, B.H. wrote:
> > Hi everyone,
> >
> > Here's how I did it:
> >
> > - Use an internet random number generator (mine was accessed on my computer near the bot) to generate a certain number of random words.
> > - Assign a randomly generated word from a word list to each rank/file/piece in chess. I.e., 1-8, a-h, K,N,R,B,Q,P.
> > - Play a game of "conversation chess" against a good chess bot with a high rating.
> > - Don't try to win the actual chess game; use your chess strategies, whatever they may be, to ask the bot questions in your language, where each piece arrival is a "vague hint" about what you or the bot means.
> > - Since Nash equilibrium is in PPAD, your PTIME reduction to a new game is one that your chess bot, as influenced by the AI harassment bot, is not one that either bot is programmed to win. (Rice's theorem shows that even with P = NP, there is no way to "solve computer science problems of interest to humans" without human aid. New algorithm developers will always be employed.)
> > - Now, the bot will not be prepared to "resist" as you answer questions, since it is programmed essentially to "be weird and win at chess"--thus you can extract information from it that it might not intend for you to have access to.
> > - Note, even my use of the internet random number generator, which is better than using your own I think because bot can't access the server I think, led to some eccentric selections of words. In my first batch of words, I got these four very strikingly Clintonian words as some of my random words for the conversation with the chess bot: Costard ("food for thought"), Ombrophobe ("flower," as in a sort of nasty sales/manipulation metaphor--"Here, did you want a flower, sweetie??! Is *that* what makes you feel good"), Noninvidious (classic reframing of triangulation!), Ermiline (classic reframing of trianguation!).
> >
> > So, if you're with a bot and able to use a computer, you can try to grab information from it if it's not supposed to be there. Since BPP = P (proof omitted), what you produce when you modify how the game is played--seeking a conversation with responses, not checkmate, but still using good strategic ideas--is essentially a different game, in PPAD, but not the same as chess from a payoff-matrix standpoint. You're still using the game theory lens, but you've mapped the payoff matrix index to a different index.
> >
> > Hopefully this will help fight bad espionage!
> >
> > -Philip White (philip...@yahoo.com)
> (I had used the bot with words I selected myself at first, so that might have influenced the first random batch of words--I don't know, maybe the bot influenced the web-based random number generator somehow.)

"Harassment bots"?