In just a few days, on Friday, April 19th 5PM UTC, a new puzzle will be made available to the community.
During the first 3 weeks upon releasing the game, an unofficial contest will be run by the game designers. Players have until Sunday, May 12th to submit an AI on the puzzle’s public leaderboard to enter the contest. At the end of that period, the organizers will award 1st, 2nd and 3rd place virtual rewards for the contest winners.
Based on a physics engine, you will compete against other players to arrest as many prisoners as you can. Gotta catch 'em all!
All leagues up to Legend will be made available from the beginning of the contest.
As a tiny teaser:
Ranking
Since this is an unofficial contest, CG will neither provide support to calculate a fair ranking nor supply prizes to the unofficial winners.
Unofficial winners will be chosen by the contest organizers as per the procedure described below:
On Sunday, May 12th at 5PM UTC, a snapshot of the CG leaderboard will be made. Be sure to have submitted long enough before the deadline: if your score is still computing we will use the one that appears in the snapshot!
The 10 best AIs according to that snapshot will play 100 games against each other in round-robin format, using the tool CGBenchmark
The Top 3 AIs according to the procedure above will be considered winners of the contest. Winners will subsequently be recognized with their name integrated at the top of the puzzle statement and within the puzzle viewer.
Just to ask the obvious : why making it an unofficial event rather than official ?
Guess at answer: to allow yourself to participate ? Or to use your own ranking method rather than CG’s one ?
Why an “unofficial” instead of an “official” event ?
The puzzle was submitted to CG staff with a proposal to make it a contest. It was refused on the basis that it did not meet Codingame’s current criteria for contest design.
The main criterion seems to be that there should be a viable progression to make a decent AI based on heuristics. While a very simple heuristics bot will possible, people will soon need a minimum of simulation to improve their bots.
As we felt we already had a solid game design, we decided not to alter the puzzle and propose an “unofficial” contest. We hope to attract players in a common time-frame to increase the level of competition
Creators are mainly Illedan & I, with input from Agade when we had a first prototype working.
I guess Illedan & I won’t be allowed to be in the ranking, because we made (strong) AIs to test and balance the gameplay.
Agade will be allowed to appear in the ranking since he hasn’t (yet) written an AI capable of beating the wood 2 boss.
While we’re on the topic of those bosses, prepare to spend time getting out of wood (it will already be the full game)
We’ll use the special ranking method described in the first post because it solves most Trueskill shortcomings in our context.
Clarification regarding the ranking procedure at the end of the contest:
If Illedan’s or my AI are in the top10, our AIs will be included in the round-robin phase where the top10 AIs play 100 games against each other.
It is after the round-robin phase that Illedan’s and my AI will be excluded: we will not appear in the final “top 3 ranking”.
Players with alternate accounts in the top10 are expected to come forward to the organizers. Those alternate accounts will be excluded from the “top 10 round robin phase” and leave room for an actual player.
(it will already be the full game)