Quote from: WilliamLie2 on September 02, 2014, 01:56:03 PM

If this bot is better than humans not only at searching for map but at actually playing the game (and it seems so, given how fast it can solve these non-trivial maps) then requiring more work to generate a map will not make humans more competitive. It may slow sown him but he will still be superior to humans.

This depends on his actual ratio of map search to solving time, relative to that of humans. Really, though, if he is doing any more map iteration than humans the anti-warp should significantly impact him.

Let's look at some hypothetical numbers to quantify this.

Let's say that the bot iterates 500 maps and takes virtually zero time at all to solve it once found. (Probably he iterates more like 2000+ maps and takes more like 5-10 seconds to solve, but...) This means that when the anti-warp settles into a target, it will take an average of 0.12 seconds to generate each map. Let's say a human iterates only 10 maps (pretty reasonable assumption, I think) on average to find a solvable one, which will take them 1.2 seconds. This leaves them 58.8 seconds to solve by hand before the bot finds its map.

On the other end of the spectrum, let's say the bot only had to iterate the same 10 maps that human does, and still solved instantly. This would create a situation where (after anti-warp target settles, again) each map would take 6 seconds to generate. Now a human has only 1/10th the chance of the bot on any given block, assuming the human can still solve within the 54 seconds remaining. This is, of course, less than ideal but also still a significant improvement over our prior scenario where the humans were left with virtually no chance at all. (Also, this is probably more or less our "worst case scenario" under the new configuration.)

So right now, with the bot still requiring large nonce iterations, we can expect the system to balance itself out, even though it is able to solve the traversal so impressively quickly, simply because humans can still "out intuit it" on map selection. When the bots can significantly out-perform humans on both tasks simultaneously then humans will be cut out entirely, again, but that doesn't appear to be the case just yet.

Quote

We need to change game itself, not only map validity criteria, to fight these kinds of bots.

What changes, in particular, would you propose?

Quote

I must say that this bot is pretty impressive, old bots look like idiots compared to him. This annealing thing or whatever else he uses is pretty good.

Yes, this bot is much more along the lines of what I had expected to see come out of the Moto competition. This bot is much more of what I would call a "real AI" than what we saw before.

Quote

The one who made this bot probably put a lot of effort for very low reward.

It is hard to really estimate what his "reward" will be until after the fact.