Talisman: Digital Classic Edition

Talisman: Digital Classic Edition

Προβολή στατιστικών:
Unmatched Choice Error= Fix the AI cheat engine
I see these errors constantly. I am wondering if it's in regards to the AI cheat engine not working properly? We are all well aware this game is not true random, for that you'd need Static noise random.

This game is just pseudo random PRNG with guided play. It's that last part that I think is causing the crashes. The game, like all games, goes through a series of playbooks and anything off the playbook menu causes an unmatched choice error. In other words the game doesn't know how to handle non predicted behavior. So the ai manipulates the rolls to make the game as stable as posible and as predictable as possible while creating an illusion of random.

http://boallen.com/random-numbers.html

http://www.random.org/randomness/


Perhaps you should just scrap the pseudo random and use a static random system?

The fact it's a standard psuedo random system doesn't bother me so much, as the game makers stating that this is true random. I'd rather they just be honest, I think this adapataion is excellent, mostly. The major down fall is and has been the AI dynamics along with it's terribly obvious non randomness.
Τελευταία επεξεργασία από PinkEyeSteve; 26 Απρ 2014, 21:23
< >
Εμφάνιση 1-5 από 5 σχόλια
AI cheat engine?

And while you are right that a PC's default random number generator is not truely 100% random, for something as easy as shuffling card decks and rolling six sided dice, it suffices if you use a different seed number for every single game, for example, by using the system clock at the start of the game as the seed number. Unless you start a game down to the same millisecond, no two games will be alike.

Also, I've had those errors come up with games involving only human players, so I fail to see how this is AI related.
I've also had various unmatched_choice errors, both in games with AIs and with all human players.

In my case, they only ever happened when there was a game patch ready to go that I hadn't downloaded yet. Which to me, in an arcane way, made perfect sense. Elements of my game client weren't matching up with the most current client trying to be run, so... "unmatched".
NomadRob  [δημιουργός] 27 Απρ 2014, 6:24 
1. There is no playbook (whatever that means)
2. The AI does not cheat - although of course we could make it cheat.
3. Unmatched choices are when one or more players get out of sync - while this might have something to do with the AI taking over a player's turn mid-turn because they leave or are disconnected, it has nothing to do with the AI cheating (see 2). The next patch to come out will address this to a degree by restarting the players turn when they leave

The link you posted to the blog post cites the Mersenne Twister RNG. We use WELLRNG512, developed by the inventors of Mersenne, which you can read about here http://www.lomont.org/Math/Papers/2008/Lomont_PRNG_2008.pdf, and which is also linked to and discussed here http://stackoverflow.com/questions/1046714/what-is-a-good-random-number-generator-for-a-game
I just come from playing the game. AI Player 2 got consistent good numbers and cards in everything. When I attacked him successfully and managed to hit him twice, he healed those two points IN THE NEXT TURN.

After an hour and a half of that, I had to quit. I feel hugely ripped off.
< >
Εμφάνιση 1-5 από 5 σχόλια
Ανά σελίδα: 1530 50

Ημ/νία ανάρτησης: 26 Απρ 2014, 21:16
Αναρτήσεις: 5