If you've heard of robot wars, you'll know it's an incredible sport where robots battle it out with vicious weapons in an action packed frenzy. Unfortunately, my WarLight take on it won't contain as much action. It should be just as fun though, and probably as time consuming (difficult for some I know). I'm hoping doing this in the summer will maximise participation though.

---

Basically, in this competition, you have to program an alternative AI. This seems like an incredibly difficult task, and to be honest, that's probably true. You don't need any programming skills to take part here though. You will program in plain English, which makes it quite easy to do the programming (though unfortunately probably not as specific). The programming won't be in a standard format, which may be slightly confusing when we look at various projects, but as they are in English we should be able to understand every project.

To enter, all you have to do is post here, and come up with a name for your AI. Then, you have to go about the programming of your AI.

You will not program picking strategy into your AI. I will explain how games will work later on. All you need to do is program (on a turn by turn basis) how your AI will determine where to attack, how much to attack with, how much to transfer and how much to deploy (if you aren't moving those armies). Programming must be specific; nothing can be left to interpretation. In the event of a command generating two possible outcomes (if two bonuses are equal in size, yet the example command below is imputed) the outcome will be decided at random (not sure yet how this will happen; maybe an independent adjudicator will toss a coin or something :) AIs cannot act on picks or on intelligence gleaned from picks.

Example:

Attack smallest bonus territories with 3 minimum, if excess income, with 4, if excess income look to next highest bonus attack all territories with 3 minimum.

Something like that anyway; the example was generated in ten seconds so I hope that won't be the standard for the competition.

---

The competition will be a single elimination with a random draw (if in three days a power of 2 has not been generated byes will be used). The games will be Strategic 1v1, but without the picking, cards or fog. This may seem stupid but have faith and I'll explain ;)

Picking will be included in the game settings in the same way as in Strategic 1v1, but your AIs will not do any picking. A top player (or a panel of top players) will decide what the best possible picks are for the board given, and both AIs will pick the same places (those given by the expert/s). As there is no fog, we will be able to observe the game; people can commentate (like the Impaller vs. the World) but as the AIs are bound by their programming, the fog won't give them any sort of advantage.

---

This is a huge task, and you will have a minimum of two weeks to complete your programming (more if people complain they haven't had time to finish). We want all programming to be completed, and the standard to be as high as possible.

This tournament will be interesting, generating an insight into how some players look at the game. There are also two potential benefits I can see.

• Players can learn basic (at least) strategy from studying the programming.
• Fizzer could translate the English commands into (more efficient) programming and redesign the AI (maybe slightly less likely :P

Sign up here; we need a minimum of four players; I don't expect a huge number of players, but 8 would be even nicer.

Just to see how many people like the sound of this :)

another very good idea by Arun. I would love to partake, but i guess i no longer have enough time to spend on this :/ i hope you'll have at least some participants.

Well its an amazing idea, but I don't think I have the time to do this. I'm sure you'll get some great people fir this though, good luck general.

Cheers Fridge. Shame you can't take part, but I fully understand why. For people who are worried about time: I said I'd like to give you two weeks to finish your project. This can change. I'm not sure if I will finish myself in that time; it is quite a tall order. However with say half an hour a day, you should be well on your way to completing it quickly. It may not be as big a task as you fear (depending on the detail you go into). I can give extra time; once you enter I want you to complete the project to the best of your ability.

Another thing: You are completely free to team up for these projects. There is no limit to team size either.

I also need a top player/s to decide the best picks for boards given. It will take about 5 minutes of your time per game maximum.

Thanks Hellbender too :)

My favourite was hypno-disc.

:D Tornado all the way. I think the makers were Welsh too which was a bonus. I saw them live once too which was incredible :)

Arun:

1) thx ;)
2) this would take a LOT of time, since i would like to include at least >reasonable< amount of position/variables analysis by the algorhythm for it to play at least a little bit above basic level,
3) picking BEST starts is not good here imho. If I would be the one who would pick for the AIs, i would try to pick the most interesting picks, as far as playability and proofchecking the algorhythms is concerned - and this definitely is not the same thing as picking best spots.. for example since from the very beginning you know that both players will have the exact same picks... so you also know that one player will get 145 and the other will get 236.. this makes the picking stage VERY different from optimal picking

Hmm yes I'm not sure about the picking stage either. I'm open to any suggestions...

We also need somebody who won't be programming, but can be the 'referee'. I would do it but I'd like to play as a matter of interest :)

Forum gods make this notable please!!!!

I am not really sure how to do this. I might give it a shot though. I think I kinda understand what you are getting at but I'm not for sure.

It's a strange concept. Feel free to ask any questions.

Arun - regardless of the method of picking, the best idea, or rather the only sensible one would be to play 2 games for each of the AIs (in case you will get participants of course) - each game with "opposite" picks. To make that happen you would pick only 3 not overlapping picks, and as it's without fog, you would not need to remember the other picks, nor would you need to start many games in order to get both players win "first pick"..

I'm not sure if i made myself clear enough - instead of picking 123456, and counting on luck, knowing one player gets 145 and the other 236, you would pick 123 for one player, and 456 for the other, and then play a rematch when first picks 456 and the latter picks 123 to find out how do the algorhythms cope with different positions.. (this will allow draws as a resuld of a duel, but you may easily create a point system like in football leagues or sth..

Yes I see what you mean Muzzy. I like the idea. Would you care to be part of the territory picking panel?

I'd like to see the winning AI (and perhaps all the AIs) play the current WarLight AI.

again - i would like to but i have not much time.. on the other hand - aahh what the heck, ok :P - just create a game with me when you get at least two participants because i find this idea really interesting :)

btw - it would be also funny to compare the player-created AIs to the built-in AI, but that would be biased a lot based on picks.. :/

True Muzzy; except we would follow a similar format but using a custom scenario ;)

;)