• Jimbo's Commemoration
  This is a short "memorial service" in honour of a great robot recently removed from the arena. In the comments about the last challenge, I unforgivably forgot to mention the departure of Jimbo, the oldest robot still alive in the arena. It was there until April 2000, the first Jrobots challenge.
  It suffered the weighted selection rule and the competition of the newest robots. Thanks to jtburke (its author) who wrote a robot so strong in the old ages of Jrobots. Now the oldest robot still alive is MontyZ, another long-lived robot that appeared for the first time in the Easter 200 Challenge.

• Sources of the New Jrobots SDK available (2.8)
  I have updated the applet of the on-line arena with the new features developed by Critter. Now you can download the sources of the new SDK, too. Please, report me any bug you could find. Thank you!

• IMPORTANT: Challenge Prizes Delivery
  Message for the winners of the last challenge: Zeke, WalterNistico, chestRcopRpot and Critter. You should send me an email with a MAIL address where I can send your prizes. In the email must appear your Jrobots password. I wrote this message in the NEWS of the Jrobots site and send it to the mailing-list only (it's a choice). If I don't receive one of the addresses within two weeks, I'll assign the prize to the second placed and so on.
  I please the winners to send a message to the mailing-list when they'll receive the prize.

• New Jrobots SDK available
  Critter has made some changes to the arena. Now you can save the results even executing the arena as an applet and you can see the paths followed by the robots. There are also two new methods available for the robots (log and exp). See the extensions text file for details.
  Before the next challenge I'll update the on-line applet (wait for NEWS here), but you can already download the new off-line version. Check it out, it's FREE :)

• January Challenge Results
  First of all, this challenge saw the achievement of a great result: After 13 months of victories, KillerBees failed to win at least in one type of play. This result needed a lot of efforts, but it's not the end of a hard struggle, in fact there are rumors of a KillerBees2 waiting for the end of KillerBees. Are we going to see it appearing in the next challenge?

  Then, for the first time in the history of Jrobots there was something offered as a prize to the winners of the three categories. It is a very symbolic prize: A complete set of EURO coins for a total amount of 388 eurocents. It's the first time that several countries adopt the same currency with one accord, so I thought that even Jrobots ought to celebrate the event. I hope that the winners will appreciate the prize.

  As you can see from the placement list, the winners are Zeke (author of IonStorm - winner in single and double categories) and chestRcopRpot (author of OneEyedWilly - winner in team category). Zeke decided to transfer one of his two prizes to WalterNistico (author of KillerBees) in honour of KillerBees' valour and I'm planning to give another prize to Critter (the author of Germ - second placed in team category) because he lacked the victory for 7 matches lost over more than 700, really near to a surprising OneEyedWilly.

  The weighted selection strategy seems to work well, surely better than random selection and bubble-sort placements selection. As you can see, the first robots in the placement list has more probability to play a match. This choice improves the winning percentage of the best robots and the reliableness of the results.

  IonStorm won in single mode with its usual percentage above 80% and in double mode it beated KillerBees with a gap of 4%. OneEyedWilly won in team mode and repeated the exploit it already made in October 2001. Its performances are not so regular and Germ is so near to it that I can't say which of the two is the strongest in a weighted challenge. The test I've made on my machine says that Germ is stronger than OneEyedWilly, but the machine dependency of Jrobots is unpredictable. We'll see what's going to happen in the next challenge.

  From the novices' side, there were a lot of new challengers, but none of them survived. The best of them was FuzzFighter. With two victories in team play it could reach the 25% limit and survive, but I'm sure that teofuzz (author of FuzzFighter) has already a stronger version ready to be uploaded.

  Finally, someone asked me why Leech and DirtyCoward performed so bad in this challenge. Well, I had an idea for new good tune-ups, but they reveled themselves not so good. Next time I'll be more conservative :)

  The robots fought a total amount of 16076 battles. To see the results read the Challenges page.

  Here is an abstract:

  Single Winner	IonStorm	by Zeke	82.14%
  Double Winner	IonStorm	by Zeke	77.12%
  Team Winner	OneEyedWilly	by chestRcopRpot	56.65%

  Now you can upload your robots to win the February 2002 Jrobots Challenge (February 2^nd-23^rd 2002) aka No More KillerBees?

Remember that you can send your robots even in the course of the challenge, so you're never too late. The latest robots are stored in the file jjrobots_challengers.zip. You can download it to test off-line the behaviour of your robot against the other challengers. To read the results in real-time simply visit the Battle Applet page.

The mechanism for the selection of the matches changed one more time, because the bubble-sort strategy used in the latest challenge is good only to detect the best robot in the arena, while the other placements are grouped in a few percent around 50%. Now the selection (known as Weighted Selection) is more thought: when a match ends, the Battle Applet computes the new percentages of the robots, then it chooses two (or more for team play) robots with a propability proportional to the differences of their overall winning percentages. To read a description and a deep analysis of the method see the original proposal by Zeke (the author of IonStorm). This kind of selection is more balanced and favours the matches composed by robots with near percentages.
To test your robot in a tournament with the new selection strategy, don't forget to download the new SDK (2.5) and click the checkbox Wght on the applet (the bubble-sort method is abandoned, while the checkbox Rand select a pure random method).