I've thought about it again, and I guess computers could optimise certain rooms faster than humans. Take something like
The Domain of the Grinfidel:First:1N1W. A computer would probably make short work of this. Why? The optimising goal is simple. Start killing roaches as soon as possible, kill one on every single move, and end on the edge of the room. That reduces the number of moves to consider drastically. You typically have only one way to kill a roach, or at most two. What's more, a lot of search paths will be eliminated quickly. As soon as you're in a position where you can't kill a roach, there's no point in continuing.
(Note: It took me about 10-15 minutes to find an optimal solution for this room. I mention this for two reasons:
-Setting up a program, waiting for it to do it's search, and entering the move sequence would take at least as long, and that's assuming you already have a program written, which is a big job.
-The first time I played this room for highscores, about a year ago I think, I didn't come anywhere near optimal. What I'm getting at is that there is such a thing as learning and improving. The feeling I get - and I apologise if I'm wrong - is this: You're new to optimising. You think certain scores must have been obtained by computer because *you* still find it hard to equal them. Maybe if you played for another year or so you wouldn't find them so amazing?)
Rooms where you can't kill a monster on every turn, but nearly every turn, will also qualify. WW rooms are often like that. Add "
overhead"
, such as first having to get to an orb to release the monsters, and I guess a computer could very well optimise some rooms requiring over 100 moves.
But it's becoming pretty clear that you know all this already, halyavin. That's why you're so adamant. I wonder why you didn't say so? Where you're going wrong is that you lose all sense of proportion. There are still only a small number of rooms that can be solved by computer. You aren't going to get a lot of #1s, or get into the top 20 because you cheat on the 100 or so rooms a computer can solve. Plus it's pointless to write such a program and use it to cheat on a few rooms. Anyone who wrote such a program would surely rather tell us about it. Right? It would also be suspicious (see below). What everyone's been telling you - that computers can't solve DROD rooms - is true for 99% of rooms. You're seeing only the 1% of rooms that computers can solve. You even forget about the other 99% while accusing players of cheating. You started this thread because of a couple of demos that could have been obtained by computer. I'll bet you that the same players have equally good demos that clearly could not have been obtained by computer. (I really wish you would give examples, so I can go:"
If X did *this*, which a computer could never do, is it so hard to believe he did *that* without a computer?"
)
Which brings us to the patterns you're looking for: I don't see how it's possible to tell the difference between a solution obtained by computer and one obtained by hours of effort. But one thing I can think of is that it would be suspicious if someone uploaded very good scores for rooms that look optimisable by brute force, while otherwise they do poorly. It could be coincidence, but that's probably what you should watch out for.
/dy
.
