<< Back to Warzone Classic Forum   Search

Posts 431 - 450 of 565   <<Prev   1  2  3  ...  12  ...  21  22  23  ...  25  ...  28  29  Next >>   
WarLight AI Challenge: 2014-06-04 22:38:23

Ikalgo
Level 50
Report
Game 1: http://theaigames.com/competitions/warlight-ai-challenge/games/538c74984b5ab23cee6dbcf4 DRAW

Like GreenTea mentions... just painful. My bot acts silly in the start but recovers. In the end, my bot is overly concerned with the enemy holding 2 regions in Europe. I was aware of the static nature.


Game 2: http://theaigames.com/competitions/warlight-ai-challenge/games/538c74ad4b5ab23cee6dbcf5 DRAW

We have to admit Dalek sucks here. My bot basically expects enemies when he arrives in North Africa. I have not had enough experience with Europe when choosing a region in SA (which slows down the enemy a lot). I know that I should have my bot now to be waiting until it meets a big stack in SA, then problems solved. He was now able to just walk into Europe. Fortunately, my walking around in Europe made this a Draw (it happens when big stacks are 2 regions apart, Ad uses this as well)

Game 3: A painful round 53 - http://theaigames.com/competitions/warlight-ai-challenge/games/538c74be4b5ab23cee6dbcf6 DALEK

This is where I refer to with my incomplete last strategy. It wasn't implemented well and now instead my bot gets to "I win - just destroy any enemy near you" rather than try to destroy Australia.
Also, it's painful to lose in round 99, let's not add this one to the "funny games" thread -

Game 4: http://theaigames.com/competitions/warlight-ai-challenge/games/538c74ce4b5ab23cee6dbcf7 DALEK

Very solid of Norman. My bot does quite badly generally when enemy holds North Africa, it's difficult to capture SA and protect of course

Game 5: http://theaigames.com/competitions/warlight-ai-challenge/games/538c74dd4b5ab23cee6dbcf8 BLENDER

Standard succesfull Europe game

Game 6: http://theaigames.com/competitions/warlight-ai-challenge/games/538c74ed4b5ab23cee6dbcf9 BLENDER

Standard succesfull Europe game

Game 7: http://theaigames.com/competitions/warlight-ai-challenge/games/538c74fe4b5ab23cee6dbcfa DALEK

Dalek is too slow, my bot should instead wait/build up for the big stack in SA like mentioned for Game 2. This really sucks, it would be so simple ;/

Game 8: http://theaigames.com/competitions/warlight-ai-challenge/games/538c750c4b5ab23cee6dbcfb BLENDER

Standard succesfull Europe game.

Game 9: http://theaigames.com/competitions/warlight-ai-challenge/games/538c751e4b5ab23cee6dbcfc DALEK

This is again watching a repeat. My bot's last strategy is not implemented well and my bot just does ridiculous things in Asia while being ahead.

Game 10: http://theaigames.com/competitions/warlight-ai-challenge/games/538c75304b5ab23cee6dbcfd DALEK

Well done by Norman, again North Africa attacking SA is such a great move like in Game 4.

---

Perhaps now you want to do your version of these games, Norman?

Edited 6/5/2014 08:22:07
WarLight AI Challenge: 2014-06-05 00:58:49


Norman 
Level 59
Report
Hello

OK, I knew that the match against Blender was closer than SupremeDalek's record against him pointed out. Also I knew that fighting against Blender would make my bot look stupid since Blender fights at different terms.

However if you look at the low ranked pairup Muli <--> SupremeLeader you clearly see especially SupremeLeader having no idea how to expand properly. So it should be obvious that you can't say that there are rock paper scissors bots and since most bots are rocks the scissors bots lose most of the time. Those low ranked bots are just weaker, plain and simple.

Same goes for the low ranked pairup Speedy <--> DamianWarlight. Taking a bonus with maximum army usage (given current armies on board and armies to expand) is a mathematical problem that my bot algorithmically solves in linear time. Most bots (including Blender) however can't solve this problem. (It's possible to construct situations where my bot doesen't find the best solution but I never saw my bot running into such a case).

SupremeDalek <--> Blender (game 1):
Blender plays to passive. Once you have income advantage + stack advantage finish the opponent off by killing the maximum amount of his armies with your attacks. Not much to say about Blender hiding his income since this only works on this map. Don't defend your bonuses with more than necessary.

SupremeDalek <--> Blender (game 2):
Some like dead piggy say that they can define a bot being able to beat a top player while others say that it's far to complicated since there are billions of different situations. The truth is probably in the middle. With his computing power a bot can play far more precise than any human could do. Furthermore I believe that you can model solid warlight play with a reasonable amount of parameters and if you stick them together then the whole is more than the sum of it's parts. My bot consists just of a couple simple heuristics that I never minded fine tuning but working together they result in relatively solid play. However the downside of a bot is that he has no common sense. If you specify good warlight play not precise enough then your bot will stick to your specification even in situations where a human sees without problems that this is nonsense. I specified my bot in a way that if he has no chance defending a certain spot he doesen't even try. However in a situation where you have a huge income advantage + stack advantage and the opponent is facing elimination you have to play differently than when the game is still open.

SupremeDalek <--> Blender (game 3):
Here you see the difference of a hardcoded bot and a bot who is really dynamically analyzing the map. Round 53 Blender should have gone strong towards Australia. Round 54 he didn't bother hiting Australia strong.

SupremeDalek <--> Blender (game 4):
No, this wasn't good play of SupremeDalek. Round 10 SupremeDalek left South America. This is because all the hatred inside SupremeDalek leads him to be an always attack when possible bot. That North Africa spot was the only spot that he was capable of attacking so he attacked this spot although he knew that this spot is unimportant.

SupremeDalek <--> Blender (game 10):
Hiting South America isn't a hardcoded strategy. SupremeDalek knows that South America is generally a good bonus (not many territories) and that there are 2 neutral spots and 2 Blender spots there. That's why he ranks Brazil as a better attack target than the Africa spot. Since he is bordering Blender in two spots he has no idea where Blender will place his defence armies so he doesen't bother about Blender defending. This was one of the last things I changed in my bot about a month ago. The old SupremeDalek would have attacked that Africa spot since he would have assumed that Blender defends the more valuable South America spot.


Edit: OK, maybe SupremeDalek hasn't got the most sophisticated AI of all bots. But I completely stopped updating him over a month ago and he is basically the same bot as almost two months ago. This was in the old days where there wasn't any real competition. I was alredy specifying a big update a month ago to make him play at human level but then I ran out of time and lost interest a bit since this was the time where all those hardcoded bots began to rise.

Edited 6/5/2014 01:15:00
WarLight AI Challenge: 2014-06-05 02:21:02


Norman 
Level 59
Report
Although pedrito and GreenTea have good AI they fail in many situations that SupremeDalek plays out properly.

http://theaigames.com/competitions/warlight-ai-challenge/games/538c766e4b5ab23cee6dbd16
Here you see pedrito going after the far away Australia bonus while it was far more important to protect the entrance to his South America bonus.

http://theaigames.com/competitions/warlight-ai-challenge/games/538c76f74b5ab23cee6dbd24
Here you see GreenTea playing bad facing the opponent in multiple spots from the start. Instead of trying to win in one of the important spots (South America or Australia) he tries to defend all 3 of his spots. In fact looking at the games GreenTea <--> Gadzbot Gadzbot could maybe have pushed GreenTea to the loosers bracket if he only had a more conservative / better stalemate detection.

Edited 6/5/2014 02:54:53
WarLight AI Challenge: 2014-06-05 03:25:13


Doushibag 
Level 17
Report
However the downside of a bot is that he has no common sense. If you specify good warlight play not precise enough then your bot will stick to your specification even in situations where a human sees without problems that this is nonsense. I specified my bot in a way that if he has no chance defending a certain spot he doesen't even try. However in a situation where you have a huge income advantage + stack advantage and the opponent is facing elimination you have to play differently than when the game is still open.


I realized this flaw in my own bot and gave it some conditions to try to solve it. My bot has an attackratio variable it uses in its attack checks to see whether or not to attack such that I only attack when it is advantageous to do so (generally). However I realized my bot would be stupid and I'd see games where I'm surrounding an enemy with stacks of armies. Enemy is pretty much defeated, but none of my stacks attack because they're all to small individually to attack to individual advantage even though if they all attacked together they could take the spot. So I actually coded in some conditions to change the attack ratio such that it's substantially lower in the main case which then allows my bot to make all the attacks it can with minimal check for strict advantage. I called it 'panic attack' mode. I was thinking 'I'm winning but haven't won, hurry up and attack everything and kill it quick!' Not sure how many games that has come into effect, but it's one of the things I was glad I had time to put in.

My bot has a lot of hardcoding, but it also has a fair amount of general coding that would work regardless of map or spot. It deploys randomly most of the time too in a controlled randomness. I use various means to define which spots can be deployed to and then it randomly deploys among them with +2s/1. Although I also added in some checks late that checked threat level for standoff situations to determine when I need more armies to keep a spot so you'll occasionally see larger deployments. Also there is a check to see when I can spare armies to take a bonus. Although I hardcode which bonuses to go for given various situations. Which basically doesn't tell it to attack the bonus, but allows it to.

I think random maps will make things a lot more challenging and interesting and people will be forced to program more general behaviors if they're going to do well in the next competition (assuming it's setup well). I was too much of a noob programmer and had too little time to make my bot I figured hardcoding a lot was my only chance of getting something reasonably decent up in time. But I built it around that random core of the starter bot. Not sure if I'll participate in the next one, but I might. I'm trying to learn how to program in the mean time for other reasons which may then help me if I do decide to do it.

I was thinking after the next competition Randy should get someone to 'donate' their bot to Warlight and make it an official AI to play against in the game. It could finally be the challenge after insane that I've wanted. For the final (new) single player challenge, you must beat this evil SupremeDalek bot! And then he will devour them all and make them cry and things will be right with the universe. You'll have your true victory then! Because really, what is SupremeDalek if it's only killing bots? It needs some non-bots to truly release its hatred upon the world.
WarLight AI Challenge: 2014-06-05 09:16:45

GreenTea 
Level 60
Report
"Here you see GreenTea playing bad facing the opponent in multiple spots from the start."
This if feature of scoring function in v63, which try to find income distribution with max average score by all regions. In current version (which I have locally) bot tends to find income distribution which maximize scores reather for one region, then for all. Except this I fixed many bugs in v63 (almost all strange moves, visible in lost games), but to late to update..

I belive that will not win this finals, because you guys had created very good 'Small earth' bots which exploite 2-3 effective strategies on this map. But when organizers will implement random maps.. Im 99.99% sure that any of current top bots will not be even close to GreenTea ;)

Edited 6/5/2014 09:25:21
WarLight AI Challenge: 2014-06-05 10:31:02


ps 
Level 61
Report
blargh, facing mweb again, ok, i'm dead, this sucks, losing against a hardcoded bot with no proper troop managment :(

i'm amazed at Muli, was not thinking it would make it to the finals.
WarLight AI Challenge: 2014-06-05 13:05:55

Ikalgo
Level 50
Report
To the guys who are so proud of their "true AI": this is a competition in which some form of hardcodedness can certainly outperform general AI. It is just a too restricted map. I have always more seen it as a collaboration of me and my bot... combining the strengths of a human with the strength of the machine!

I'm not sure if I will compete in the next Warlight competition (first poker, then I'll see), and agree that there will be fewer hardcoded bots (none perhaps?). However, the competition won't be less fierce. I (and probably other hardcoded bots) also have complex algorithms in place, and I also really like the excitement of writing a more general smart AI. It's just that this competition favors a little bit of hardcoding here and there.

Edited 6/5/2014 13:07:02
WarLight AI Challenge: 2014-06-05 13:20:15


125ch209 
Level 58
Report
why such hate toward hard coded bots?
WarLight AI Challenge: 2014-06-05 14:29:41

GreenTea 
Level 60
Report
No hate, just making general AI requires more efforts from coder, and then it sad that bot was beaten by some simple hardcoded strategy, on which your opponent coder spend far less time and efforts, then you on your general AI.
WarLight AI Challenge: 2014-06-05 16:00:56


Muli 
Level 64
Report
i'm amazed at Muli, was not thinking it would make it to the finals.

To be honest i'm quite surprised too and don't think I deserved it, there are far better bots around.

I was actually missing some time to update my bot, so the Version that is playing is from 10th of April and not updated since then. The bot is quite bad and basically just spends his armies attacking territories with lower IDs first because they appear first in the 'foreachs'.

Edited 6/5/2014 16:59:53
WarLight AI Challenge: 2014-06-05 18:05:39


Norman 
Level 59
Report
I belive that will not win this finals, because you guys had created very good 'Small earth' bots which exploite 2-3 effective strategies on this map. But when organizers will implement random maps.. Im 99.99% sure that any of current top bots will not be even close to GreenTea ;)

To sum this up: Every bot that acts smarter than GreenTea (in some situations) must have this specific behavior hardcoded? I believe there were also some other warlight players and computer scientists putting thoughts and a reasonable amount of time into how to formalize good warlight play. I myself have nothing against hardcoded bots. I just lost interest as I realized that all those thoughts I put into my bot could be countered by just sitting me out in North Africa. As I designed the fighting behavior of my bot I had messy situations in mind where there are a lot of important spots to attack and other important spots to defend.

In fact if I look at the leaderbord then among the top 10 the only hardcore hardcoded bots I see are Herz (rank 7) Gazdbot (rank 10), (in all likelihood) Blender (rank 9) and to a lesser amount Trogabot (rank 5).
http://theaigames.com/competitions/warlight-ai-challenge/games/538c6f424b5ab23cee6dbc74
Turn 6 you see SupremeDalek moving strong towards Australia instead of taking Africa. This is the same code that makes him move strong into Africa / South America from Australia. There is no Australia excpetion, but code telling him to move towards the opponent when losing sight.
WarLight AI Challenge: 2014-06-05 18:46:48

GreenTea 
Level 60
Report
Norman: ok maby Im not right, lets see how Dalek will play on random maps. In games vs GreenTea Dalek always put 5 to Australia in first turn, then always 5 in Australia and attack. I thought this is hardcode, but after link you posted
http://theaigames.com/competitions/warlight-ai-challenge/games/538c6f424b5ab23cee6dbc74
I see that second turn Dalek not always put 5 in Australia..
WarLight AI Challenge: 2014-06-05 20:05:06


Doushibag 
Level 17
Report
While I have many general algorithms I also have a fair amount of hardcoding or a combo of the two in most cases. Given I didn't have the know-how or experience with programming to do anything fancy and the less than 3 weeks I had to get it all done I saw doing a fair amount of hardcoding as the only way to make a reasonably competent bot in time. And with the highly limited nature of the competition there was no good reason not to. If you can get to an answer elegantly that's great, but if I can get to the same answer with a little thinking and hardcoding well then I might actually be better to do it that way. It might be both faster to run and faster to code leaving more time for other things. Competition and thus fitness is always dependent on the environment. Look at nature and you'll see that all around, that things that are better and more efficient in some environments can still be outcompeted in others. The more generally you program your bot though the easier it will adapt to new environments. So many of us just programmed for the only environment our bot would ever see whereas you can adapt yours to the new competition.

If I were to do the new competition I would likely start over from scratch pretty much. Maybe salvaging some sections of the code that were general, but my focus from the start would be very different for the new highly variable environment being the primary first challenge to a competent bot. I'm sure in the next one if they set it up properly you'll see all the top bots being ones that are good at reading and adapting to the map and hardcoding will be much harder to do effectively and thus will be employed much less.

So if you don't like losing to a hardcoded bot you basically have a problem with the limited environment that prevents your bot from showing its full potential. I remember this effect in playing with other Warlight players as well where you considered yourself the better Warlight player, but when you have a competition on a single map and they've played nothing but that map for thousands of games they may rightly have advantage over you where you'd beat them 9 out of 10 times on any other map. That's the whole specialization thing. When you have a stable environment specialization tends to win out. Which also gets to that whole chess comparison thing in another topic. Chess is super specialized so people with lots of experience on the map do better than what may be from the start an easily smarter person. Whereas Warlight in its full glory is highly variable in many ways and you can't simply memorize the map to employ tried and true strategies.
WarLight AI Challenge: 2014-06-05 20:36:58


Norman 
Level 59
Report
Hello

There are no hidden secrets in my bot that I haven't revealed in this thread (even as the competition was still open, I'm not that a competitive guy). The reason SupremeDalek fights turn 2 for South America in this particular situation is because South America has 1 neutral / 3 player territories while Australia has 2 neutral / 2 player territories. The main reason my bot is lacking hardcoded moves isn't because I dislike them but because I didn't knew how to add them without breaking the other logic of my code. In the evaluation stage my bot evaluates which regions are worthwile and in the move stage he performs his moves by just looking at the integer results of the evaluation stage. Hardcoded moves have no place in the moves stage. What I do admit is that my bot evaluates South America and Australia as equal if each player has one spot there and he would usually try to fight in South America since the South America spots have lower IDs. Since I want my bot to fight for Australia instead I gave Australia one extra point in the evaluation stage.
WarLight AI Challenge: 2014-06-05 22:05:28


Doushibag 
Level 17
Report
Blender beats Dbagbot [5/3/1]
Speedy2014 beats Assimilator [5/2/0]
mweb beats psWarlighter [6/4/0]
TheeYo beats Webmartino [5/2/0]
SupremeLeader beats HouplessBot [5/0/0]
Gadzbot beats KewKareban [5/1/0]
WarLight AI Challenge: 2014-06-05 22:07:36

{rp} pedrito 
Level 48
Report
I'd really like to know how the last bracket of the losers pool will play out... it seems to me that there will be 3 bots left for 2 places in the final.
WarLight AI Challenge: 2014-06-05 22:15:46


Doushibag 
Level 17
Report
"And yes, all 3 bots will play once against each other, so that's three games in total. The bot that loses both games is out of the tournament. It's possible that all bots win once, then the bot that takes the most rounds to win his games is out." -- Jim
From: http://theaigames.com/discussions/warlight-ai-challenge/538ba3815d203cf6398b4569/semi-finals-have-started-/1/show
WarLight AI Challenge: 2014-06-05 22:16:18


ps 
Level 61
Report
Match Winner Rounds Score
1 mweb 21 0-1 Review match
2 psWarlighter 12 1-1 Review match
3 psWarlighter 31 2-1 Review match
4 mweb 42 2-2 Review match
5 mweb 33 2-3 Review match
6 psWarlighter 34 3-3 Review match
7 mweb 23 3-4 Review match
8 psWarlighter 25 4-4 Review match
9 mweb 25 4-5 Review match
10 mweb 69 4-6 Review match

goddamnit, so close and still managed to lose it! :(
very disapointing not to make it to the finals :(
WarLight AI Challenge: 2014-06-06 09:20:39

GreenTea 
Level 60
Report
Doushibag: While I have many general algorithms...

Thanks for post. Yes currently environment is limited by simple fixed map. And people tends to go on path of least effort - and this is normal. Maby 1 map is good because the threshold of entering into competition is much lower then in case of random maps. Now people can use their previous experience to try more ambicious task - to make bot which is not coupled with one map.

Edited 6/6/2014 09:28:23
WarLight AI Challenge: 2014-06-06 13:28:46

Ikalgo
Level 50
Report
GreenTea, Douchibag: I completely agree with both your statements.

Edited 6/6/2014 13:29:27
Posts 431 - 450 of 565   <<Prev   1  2  3  ...  12  ...  21  22  23  ...  25  ...  28  29  Next >>