In case you haven't heard, a bug was discovered in the tournament code that significantly affected the results of edition 1. Because of this, the results are invalid and the tournament will be repeated.
No action is required on your part. The same snapshot of bots from the original tournament will be used to run the tournament because that part of the code was unaffected. I will be taking down the existing results soon and will be moving the tournament's state back to "scheduled".
I'll re-run the tournament either tonight or tomorrow. I'll also make sure that per-bot results are available from the start to help give confidence in the validity of the results. If all goes smoothly and folks are generally satisfied with the output, sometime after that I will give out trophies and other awards.
Sorry about this everyone. As I mentioned on the bug thread, I take pride in the quality of this game and this is probably the most glaring issue bots4 has ever had so far in its 4.5 year stretch. It was just subtle enough to be able to trick an untrained eye during testing (read: me), but was obvious enough to anyone fully investing their time and energy into this game (read: you).
|
|
Oh damn I missed a tournament
|
Bugs are good, they force us to learn stuff :)
|
Alright new edition 1 results are now up along with per-bot results. Things look reasonable to me, particularly with category 1, but please let me know if anything looks out of whack.
|
thats more like it :) 26 wins
|
Question: How do you deal with ties? Sorting by ID is my guess. Just looking for confirmation.
Question: In cat 11 I had 1 fight between Howard Wolowitz and Prosperity where we both ended up with 0% HP left. I´m assuming I ended up below 0% but the fight log only shows it as 0%?
|
love love love the per bot results :D
Great job Wiggin!
& ye I mentioned that last time Jans, but it got buried in the other issues. It's not by id, so I'm guessing the it must be something to do with hp.
|
oh my in my excitement I only saw yellow moving things in the logo & confused Benny with Jans!
Ooops sorry guys ^_^
|
Repeating what Benny posted; how did you deal with ties Ender? And can you explain the 0%-0% results? From what I can tell ties were broken according to ID, which isn't the fairest way to split bots.
Everything else appears legitimate, but I'll have a thorough look tomorrow. Lucky winners everywhere, but there always will be in a 1 fight tourney (And as the person who would benefit most from an increased number of fights I don't support it, despite my disdain at losing).
|
yay, redone tourney - being able to view the individual fight results is very cool :)
and silly Rose, confusing Benny's yellow oompa loompas with my suave dance crew..
|
The 201 still needs work though.
|
If by work, you mean being divided into more cat´s then yes :)
The results in the cat looks okey.
|
What benny said xd didn't word it good sorry. Moooooo
|
Question: How do you deal with ties? Sorting by ID is my guess. Just looking for confirmation.
Good question. The story for ties isn't great right now:
- Bots - level (lower first), then id (lower first)
- Clans - id (lower first)
I'm going to change the first tie-breaking criteria for each as follows though::
- Bots - average remaining HP percent (across all battles for bot)
- Clans - average rank (across ranking (i.e. >0 clan points) bots in clan)
That's more performance-focused and will more fairly break ties. And I'm actually going to go ahead and apply this to edition 1. This is straightforward to do because the ranking and battling processes are completely independent steps. There might be a few small changes in the overall results from doing this, but I don't expect anything to change drastically. This seems more sane than only doing this starting with edition 2 and having to always explain that ties were broken differently in edition 1.
I'll update this thread when the re-ranking has taken place.
Question: In cat 11 I had 1 fight between Howard Wolowitz and Prosperity where we both ended up with 0% HP left. I´m assuming I ended up below 0% but the fight log only shows it as 0%?
Another good question. In this case, a number close to 0 got rounded to 0 for the winner's remaining HP percent. I intentionally did not use precise values like 12.3456% because having enough of these for a bot would probably be enough to reverse engineer a bot's total HP. From there it's then fairly trivial to reverse engineer a bot's complete build by attacking it a view times and seeing what equipment it's wearing (via damage messages). ma made use of this kind of tactic in bots2.
I can think of some ways of dealing with this (e.g. show "~0%" instead of "0%"), but the existing code structure actually doesn't lend itself well to this change, so let's see how often it comes up where it confuses people and work from there. Maybe I'm also overstating how important it is to keep bot's builds private (or the effort people will put forth to reverse engineer them).
|
|
I thought the 1st tier-breaking criteria could be the direct result of the bots.
1st example: Bot A and Bot B have the same amount of wins in the tournament. If Bot A won against Bot B, it should be ranked first because in direct competition it was better (read probably luckier :P).
If there are more than 2 bots, this is more complex and another tier-break criteria may be applied.
1st scenario:
- A loses against B but wins against C. (1 win)
- B wins against A and C. (2 wins)
- C loses against A and B. (0 wins)
- Rank: B A C
2nd scenario:
- A wins against B but loses against C (1 win)
- B wins against C but loses against A (1 win)
- C wins against A but loses against B (1 win)
- Rank: it's necessary more tie-breaking criteria.
So, in more than 2 bots tied, the ranking criteria would be the number of wins between the tied bots. The ones that continue tied, a 2nd criteria would be applied. This could be the average of HP left against just the tied bots or against all the bots in the category.
|
I actually suggested this to Ender last night on mIRC Esv. Your second scenario is what Ender fears the most with this type of tie breaker. He wants something uniform throughout, and I agree. He should have a set of rules that he'll need to explain every time a "tie" occurs.
|
im intending to repeat my result next month, hopefully, be prepared, its seems dex, str, won it from overconned bots this month
|
so when do i get my new trophy?
|
Yeah, I'd prefer to keep the tie-breaking algorithm simple both in terms of code and in terms of players understanding. Doing it differently based on the number of bots involved in the tie sounds confusing for both.
Awards (trophies, exp, kudos, etc.) will come within the next few days, sorry for the delay.
|
All that needs to be done as and when the tournament finishes.
|
It will in the future. I didn't consider it critical for the launch though.
|
Ok is difficult to implement straight away? Or is there some thinking towards it first?
|
Nope, just a matter of finding the time to do it.
|
That´s okey. Don´t really want the exp now anyway :P
|
Oh and the other thing I should mention about not having done awards at the time of the tournament is that it would have made it much, much harder to redo the tournament. Making sure everyone was satisfied with the results was also part of that decision, so it ultimately turned out to be a good one.
|
I can accept that :D but now everything's sorted when you do get the time make sure you implement it :) because we can't let the good people get away with this :p
|
isin't it a idea to have 201plus in several cat's, every 25 levels instead of every 10, this will also give ppl the chance to build bots specially for that category, now it does not make sense build a 225 or 250 bot if its going to compete against 300 plus bots, so till level 300 a cat for every 25 levels, then above 300 a cat for every 50 levels?
Thoughts?
|
I agree with TheCause. If is too much cats, it should definitively be at least 201-250, 251-300 and 301+.
|
i think till 301 plus it should be 25 levels per cat, gives more variation in builds, plus makes it more exciting to actually take the time to build one
|
I doubt it would be much more variations for builds. Maybe for 201+ cats it could be. But definitely not in lover levels. Everyone would go as it always was, just best they can think of, at max level of the category.
I just wish it would be lover price to join bots into future tournaments. For someone who isnt active it could be quite hard to get kudos to join.
|
not talking about lower levels, they are fine, was referring to 201 plus cat's, instead of competing with 300 plus, create 4 cat's 25 levels each till 300, foor 300 plus you can make 1-2 cat's
|
Just found tourney cats from bots2 on my pc. I think these bots2 cats made more sens, but still it didnt have anything past 200. It wasnt problem back in the day?
cat. 1 : lvl 200+ XP-prize = 5G
cat. 2 : lvl 170+ XP-prize = 1G
cat. 3 : lvl 146+ XP-prize = 350M
cat. 4 : lvl 133+ XP-prize = 160M
cat. 5 : lvl 120+ XP-prize = 70M
cat. 6 : lvl 107+ XP-prize = 34M
cat. 7 : lvl 96+ XP-prize = 15M
cat. 8 : lvl 88+ XP-prize = 11M
cat. 9 : lvl 82+ XP-prize = 7M
cat. 10 : lvl 76+ XP-prize = 4M
cat. 11 : lvl 70+ XP-prize = 2500K
cat. 12 : lvl 65+ XP-prize = 1600K
cat. 13 : lvl 60+ XP-prize = 1000K
cat. 14 : lvl 55+ XP-prize = 640K
cat. 15 : lvl 50+ XP-prize = 400K
cat. 16 : lvl 45+ XP-prize = 280K
cat. 17 : lvl 40+ XP-prize = 100K
|
Every one of those numbers not ending in a zero or five made me cringe.
|
Okay, the re-ranking is done now. The winners for categories 2, 3, and 11 changed. The only interesting (top 10) clan changes were that ranks 2 and 3 swapped, as did 8 and 9.
To recap, here's how ties are being broken now:
- Bots - average remaining HP percent (across all battles for bot)
- Clans - number of bots from clan in tournament (note that this is different than what I originally proposed above)
Expect awards to land sometime tomorrow night. Sorry again for the delay. This will all be automated for edition 2.
|
Yay! I got another win ðŸ˜.
And trophies are coming soon? Also awesome 😊.
|
Are levels coming from 1 tourney? I can see myself not liking the rewards
|
I don't understand how levels don't count in the ties.
|
I don't understand how levels don't count in the ties.
Except for cat-1, i think it should count.
|
I think you should gain at least half a level, don't see how one or a few bots should dominate the same cat because exp is so shocking
|
in my opinion, levels matter in every category, and more as you go DOWN in levels. for 2 reasons:
1) a lvl 30 bot that beats a lvl 31 is more impressive than a 31 beating a 30
2) if you have 2 bots that tie in # of fights won... then the level is a good choice for determining which of the 2 bots should be above in the ranking
still if 2 or 3 or 4 bots win the same amount of fights... and are all at the same level... a tiebreaker should be held between them. in the case of 4 bots, it would be 3 fights more for each of them. and this process would loop for a 2nd round if 2 end up with another tie for 1st.
|
edit: think of it like a winner bracket, loser bracket, consolation finals for 3rd place if need be.
|
shouldn't the first deciding factor be the direct match between the tying bots? and if there are 3 or more bots tied (each won against another) than a second factor should step in (level for instance)
|
There is no level-adjusted formula for points, so I don't see why level should be used as a tie-breaker for tied bots. I think it's fair the way it is - the strongest bot overall will be ranked higher.
|
The trophy awarding stuff is mostly wrapped up, but it's late here now and I'm sleepy, so I want to look at it with a fresh mind before running it. I'll be traveling this weekend, so the earliest that will happen is Sunday night, but Monday night is probably a safer bet.
|
Also, I see a ton of suggestions/questions in this thread, some of which have been addressed and re-asked. In the interest of keeping things organized, it would probably be better to start separate suggestion threads for each one with supporting reasons etc. instead of piling onto this thread which is about the re-running and awards of tournament edition 1.
|
trophies would be nice, also which date is this going to be run each month?
|
Trophies have been given out for edition 1, so that should wrap things up. Let me know if anything looks out of the ordinary. Thanks for everyone's patience and help in working out all the kinks with the first tournament. Edition 2 should go much more smoothly.
Note that I did not give exp or kudos awards. Given the variance in the results due to the one-fight-per-matchup setup, the aim of the exp "award" (it was going to be intended as a forcing function to level bots out of categories in the interest of avoiding staleness) seems less important now and most bots don't want exp. Kudos seems like a minor thing, at least to me.
As mentioned above, if you have strong opinions about changing any aspect of the tournament, please start a suggestion thread with supporting reasons and so on.
|
which date is this going to be run each month?
Going forward, tournaments will open for registration at noon (server time) on the 20th of the month and will run at noon (server time) on the 25th of the month.
|
:( was hoping for some xp, even 10% of current level is huge
|
Geez no exp given? What! A lot of people don't want it because they want there bots to dominate cat they are in for ages
|
here its the other way around around give me xp because i want to dominate some more
|
|
As mentioned above, if you have strong opinions about changing any aspect of the tournament, please start a suggestion thread with supporting reasons and so on. I started a new thread for the discussion about exp gain:
http://bots4.net/forum/4/6469
|