Talk:MagicD2

Hi, good luck for your future nanobots! I read about my bot Assertive. :) I'd recommend you Mosquito 1.1, because Assertive is highly vulnerable against head-on targeting, so as soon as you switch to linear gun, you won't get half of the points scored by head-on. I worked much on Mosquito's movement, and it's 22nd with a single linear gun. I think it's much better as a testbot than Assertive. --HUNRobar 21:43, 14 February 2009 (UTC)

Hey, MagicD2 is his first bot. I don't think testing with good bot is good ideas. I know I can't write this because I test my nano with Shadow! But, for simple head-on, testing on bot that you know you can't hit it isn't good ideas, anyway. » Nat | Talk » 01:57, 15 February 2009 (UTC)

Yes, I've downloaded a variety of bots now. Just wondering, would it be better to download problembots and see why those beat it, or bots that it scores about 50% against and see how it can improve against those?--CrazyBassoonist 13:31, 15 February 2009 (UTC)

That is, a challenge of roborumble. While trying to remove problembots, you need to keep your ranking to other robots the same! But, for nanobot, I don't think this is good idea since there are very little room of improvement. Don't care problembots :) Just try one that you score 40% or less. I've realize all of this with my Ocnirp improvement. I've once add distance keeper that make it score good agaist RaikoNano, but my rank was dropped by 8 ranks! Good luck with your bots! » Nat | Talk » 14:09, 15 February 2009 (UTC)

Yes, the beauty of NanoRumble is that you can't prepare for everything. The better compromises you made the better bot you create. Just look at Pugio! It scores 30-40% against top 20 bots, but it's still 5th! Why? Because it beats almost all other bots by 70-90%! Nanos are about clever specialization and not about little tweaks or ultra-complex techniques. That's why I love the idea of femtorumble, it's very similar to this, but with the clever play of codesize, which makes the whole thing very exciting! I'd like to write about testing, too. Personally, I use excel during development. I set up a bunch of robots of different types to test against. After each bigger step in development, I run 3 battles against these bots, and then calculate the average of them. I put the results of the steps next to each other, so I can compare the scores. (actually, I prefer percentages, because they're easier to understand) Now I have to write goodbye, if I have more time, I'll collect you a bunch of great testbots from the rumble. --HUNRobar 14:57, 15 February 2009 (UTC)