mvnForum Homepage


Print at Dec 7, 2019 6:36:06 AM

Posted by Roleni at Mar 25, 2012 8:25:05 PM
Re: Botting
DeepNine wrote: 
Roleni wrote: 
I did some more math regarding the carpentry; if we figure a really good bot could even manage something like 100,000,000 positions a second, that's still only enough to consider 100 possibilities per move and look four moves ahead

The possibility of a bot that could beat a human at carpentry existing isn't really a question, even on crummy computers. The most basic heuristics cut out well over 90% of possible moves per move, e.g. not considering moves that overlap when nonoverlapping moves are still available. Make a few more tweaks and you have a bot that gets at least MP^16 or better damn near every time. This is before you consider that bots move near-instantaneously - and we're talking finishing all of its holes before you finish your first six. With how important speed is, the margin by which humans get crushed is really quite staggering.

A hundred possibilities per move is already discarding a good 90% (or more) of the technically possible moves. ;) The point is, the numbers work in an exponential manner that eventually puts a limit on computational power; human brains don't have that same limit, because they cut out far more than 90% of the possibilities, which is both an advantage and a disadvantage to having a brain.

Anyway, I'll cede the issue of speed - but that's also something that OOO should be able to detect, and enforce. So for a bot to really be good enough to count as legitimately good, it should be able to perform well at a humanly possible speed.

And a quick Google/YouYube finds that what ready examples of bots are out there...can't. Sure, improvements would help, but we're talking MASSIVE improvements. Much much more than "a few tweaks". I was actually impressed with a few of the bots - one in particular was quite good whenever it had a putty bucket - but they still have multiple fundamental errors. I could list a good half-dozen or so, but I'd rather not help any botters improve their program. :P

Besides, to be honest, if I treated carpentry like grandmasters treat chess, I would *expect* to hit MP^16 damn near every time. (Well, actually, I'd go with MP^15; it's better to get that and finish with another masterpiece than try to force the sixteenth and instead end up with two craftmanships.) Most of the time that I don't, it's because I've made a blunder, which I often notice if I'm focusing well.

BobJanova wrote: 
I'd like to see the secondary PoE market attacked with a vengeance. It should be straightforward to inspect PoE streams and see who has a vastly greater amount of PoE given to them than makes sense for the reverse trades, and who gives away most of their PoE and gets nothing in return. No-one who cares about the game enough to be making millions would just give it away on a regular basis if they weren't trading it out of game.

OOO should also have agents sign up to the big PoE-selling sites and spend a few months building up a ban list, and then killing everyone. It is relatively little work and if you can destroy confidence in those sites then you will do a lot of damage to the illegal market.

Bots are a real problem too but probably harder to prove, although the 'port, wait, deport, turn' test is fairly conclusive and should be enough to ban on the basis of. (Remember to wait long enough though, as porting legitimately doesn't reset your indicator.) But I think that by attacking PoE-selling you could do a lot to restrict botting in game as well.

This. In effect, my skepticism of the prevalence and usefulness of bots and other cheating mechanisms shouldn't really matter; I'd prefer the rules simply be enforced better.


@Wok: I do think your comment on my skeptical tendencies is a legitimate one. I'll get back to it some time that I don't have multiple homework assignments to work on. :P
----------------------------------------
Thalatta & others

Hera tells ye, "You got me so flustered by the spanking, I booched my chat response!"
Hera tells ye, "Purple is the most awesome colour ever!"

Puzzle Pirates™ © 2001-2016 Grey Havens, LLC All Rights Reserved.   Terms · Privacy · Affiliates