The new weighted selection rule is now available. It's currently tested on the Battle Applet page and you can download the Jrobots SDK 2.5 with its implementation. The proposal was made by Zeke (the author of IonStorm) who wrote the code, too. To know the reason to adopt such a rule and drop the old bubble sorting one, read the original proposal.

Now another good news: On the occasion of the adoption of the Euro currency by the countries of the European Union, the next January 2001 Jrobots Challenge will prize the authors of the first placed robots in the three caregories with a complete set of Euro coins (italian version) for a total amount of 388 Eurocents. It's only a symbolic prize, but I hope you'll appreciate it :)

Best wishes of a Happy New Year!

Finally we know which are the best robots of the year 2001. The bubble-sort selection strategy used to choose the fighting robots did its job. As we can see from the final percentages this selection rule can cause a lot of problems, but it emphasizes the percentages of the strongest robots and this is the aim of the finals.
And the winners are... IonStorm and KillerBees. This is not a surprise, because these two robots dominate the latest challenge, but the interactions of the robots with the bubble-sort selection were unpredicatable in advance.
Well, since December 14^th the bubble-sort selection wasn't pure due to a bug in the selection algorithm: The first opponent of every robot was chosen at random, then the following matches apply the new selection mode. Thanks to Zeke (the author of IonStorm) I discovered the bug and, from December 15^th on, the tournament used the correct rule and the robots in the middle of the placement list began to drift towards 50%. The final percentages are so near that I used a 48%-48%-24% percentage limit to remove the losers. I think that such a choice saved some robots that had only bad luck in a few matches. The next challenge will use a smart selection strategy proposed by Zeke called Weighted Selection Rule. I'll post another news with the original Zeke's proposal before the beginning of the next challenge and a new SDK will be available as soon as possible.

So the final results crowned IonStorm and KillerBees as the stongest robot of this year (and even of the past years). IonStorm is unbeatable in single mode and its percentage (89%) testify it. Then seven robots follow: They are IlTristoSmorzatore, DirtyCoward, KillerBees, Leech, OneEyedWillyII and Schmamageddon and got almost the same percentage because their relative strongness is very similar even if they use very different strategies. In double mode the gaps among the percentages are narrower, but it's incontrovertible that the strongest are IonStorm and KillerBees, with a little percentage in favour of the latter. Finally, there is the neat victory of KillerBees in team mode with almost a 10% gap from the good Fish. It's evident that KillerBees is more versatile, but IonStorm is very specialized and its author is working to a more specialized team-fighter. Then follows the OneEyedWilly series, that is very versatile, too. We'll see new incredible results in the next challenge.

From the novices' side, there are not very surprising robots. A lot of the brand new robots were thrown out of the arena because they didn't reach the requested victory-matches ratio. The best ones were DirtyCoward, Leech, Parasite and Cycler.
Cycler is the first robot that uses the JJVector extensions. I think that adding more extensions to the API of Jrobots will be an advantage for all the players. Cycler suffered the bubble-sort selection and it was confined in the middle of the list.
Parasite has good percentages but it was uploaded later in the tournament. In the next challenge it will have more time to show its real value.
Finally, there are Leech and DirtyCoward. They are both robots of mine and finally I've got a good placement after months of poor results. The story of DirtyCoward is interesting, so I tell you about it. First came Leech and its name tells its strategy: it runs toward the enemy to try a close fight, but it runs away if the damages are too much to arrive close to the enemy. I've found that the escape following a zig-zag path is a good strategy and the placements testify it: Even IonStorm suffered injuries from Leech at the beginning of the challenge (then it was modified :< ). So followed DirtyCoward that made the escape its reason to live, no more close combat, only flee away and keep a distance of 700 meters from the enemy... I hope you can learn something from this story :). Another strategy that DirtyCoward uses is to drop the aiming correction techniques, a good choice in an arena full of robots that follow complex paths and walk back on their steps.

The robots fought a total amount of 13623 battles. To see the results read the Challenges page.

Here is an abstract:

Single Winner	IonStorm	by Zeke	89.04%
Double Winner	KillerBees	by MontyFish	68.11%
Team Winner	KillerBees	by MontyFish	57.45%

Now you can upload your robots to win the January 2002 Jrobots Challenge (January 5^th-26^th 2001) aka The Weighted Challenge