I also think the logs should be made available at least to some extent.
I coded the parity strategy along with the heatmap and I ended up with 36 wins and 36 losses…I have no idea what went wrong and without any logs I don’t know where to even start improving my bot.
The only real thing that will be revealed when releasing the logs will be the winning ship layout (yet there are only 3 possible overall layouts). Out of all the players, it is ONLY the winning player that will not benefit from seeing the logs. That being said, the winner already received his reward.
I personally think releasing the logs can only improve the challenge.
since this is an ongoing competition i think it is not wise sharing complete logs , because this is not comparable to sport where everything remains unpredictable while in progress ,the bots follow pre coded logic and from patterns logic wil be revealed.
however , only one thing would interest my curuosity and thats if my bot times out sometimes or not , the rest is of no interest to me since the only thing that needs evaluation in failure/success is one’s own code/strategy.
it seems my bot loses as much as it wins , which is as bad as it is good which is bad … something like that…
what i do like is finaly having something to compare against , playing against yourself is really boring and a bad measure of how good your bot is , last year had player portal challenges which gave you a benchmark in how good your bot is doing after attempts to improve.
My 2c worth, I’d like to see the logs too, to check for errors, timeouts and if my fleet was following orders or if there was a mutiny. I think it’s called log.txt, and doesn’t contain any details about ship placements.
If I’m sneaky and have my bot log hits and misses, then I still won’t get the opponents’ full picture unless I win. I don’t see much harm in releasing just that log.txt file.
@unss I dont believe timeout is a factor for battleship this year.
My bot runs at around 165 ms on a pc that is much smaller than the tournament server. I am placed 4th (The worst possible placement for the tournament in my opinion) But I see it as good because it forces me to better my best.
If theres timeouts then I believe a person is doing things they shouldn’t be doing… Anyways, Its back to the drawing board. If the same tournament ran again the board could change drastically, Am hoping that the new game harnesses come out soon. Good luck to everyone for the next Battle.
I did say, “for example”. Other errors are also possible right? Then you have no idea if what caused the problem since you have no insights into how things went.
Well, that will make the game even MORE interesting. People will now have to change their strategy, Everyone will have this information. Use it, or don’t!
The captains of my ships only drink the finest Java… no rum for them, so there is no amnesia!
The tournament has now entered the verification phase. During this phase all games that were marked as suspicions will be analysed and we will look at the log files to determine what caused the flag to be set. There are various factors that can cause the flag to be set and it is intended to give us an indication of something that might have gone wrong. If during the analysis it was found that a bot misbehaved as a result of system settings/load causing the bot to time out for instance then the that match will be rerun.
Due to this the leader board you see now is not final. Only after all matches have been verified will the leader board show the final results.
None of the tournament log files will be made available to the public or the contestants. The tournament is intended to for players to compete with each other. The clans feature launching in the next few weeks will provide contestants with the opportunity to test your bot against others and access the log files from those matches. It provides for a much more controlled environment so as not to disadvantage any one player. I know the clans feature is a little behind schedule but we are trying our best to get it done as soon as possible.
As a first time competitor, do the results change significantly during the verification phase?
I understand that the logs won’t be made available, but will there be statistics of any kind? The one I’m looking for is an indicator of how many matches ended in Destroyer vs Destroyer (the 2 hit ship) overall?
The changes all really depend on the type of reruns that have to be made. It has the potential to drastically upset the results, but I can’t really say as it all depends on how the games go and how many of them are rerun.
Unfortunately we did not track statics down to that detail for this tournament. We only captured basic information to determine ranking and to identify matches with potential issues. There are the log files but I don’t have the time now to write something to extract the stats from all the logs. We can look at capturing more detailed stats for each player in the next tournament and present them with those stats. Giving access to aggregated data like that I am fine with.
How was the bounty calculated?
Was it 10 points per hit, and then additional 100 points for destroying the fleet? any points for destroying a ship?
Would like to break down my score and see how tight my matches were, seeing as we can’t get log files.
The bounty is the total points for all your matches. If you want to break it down you will have to look at how the game engine calculates the points for each match. The Github Page has a break down for how points are calculated.
We won’t be giving the scores for each individual match. I believe the clans feature will be able to help you out in determining the strength vs luck element of your bot once the clans feature launches.
The changes to the game engine being release later this week will probably also upset and invalidate any current strategies in play and it will remove quite a bit of luck from the game.
