Print

Please login or register.
1 Hour 1 Day 1 Week 1 Month Forever
Login with username, password and session length

[Geronimoo]

I think my scripts will end up more complicated than WW's entries for the pairs, and that's even after making some very broad generalizations. There are just too many conditions to cover. Some suggestions for improving condition clauses will be going Geronimoo's way!

This probably means you're trying to win mirrors, which means you have a clear best deck, which means I should be tuning mine to death as well, which means I won't bother anymore.

[WanderingWinder]

I think my scripts will end up more complicated than WW's entries for the pairs, and that's even after making some very broad generalizations. There are just too many conditions to cover. Some suggestions for improving condition clauses will be going Geronimoo's way!

This probably means you're trying to win mirrors, which means you have a clear best deck, which means I should be tuning mine to death as well, which means I won't bother anymore.

Not necessarily. Lots of my tuning was to win the mirror, but lots of it was to win against different things that might come up (i.e. the fool's gold/wharf decks I expected, other things that scrambled for provinces, whatever). And here, there's lots of different things to tune against.

[DStu]

I think my scripts will end up more complicated than WW's entries for the pairs, and that's even after making some very broad generalizations. There are just too many conditions to cover. Some suggestions for improving condition clauses will be going Geronimoo's way!

This probably means you're trying to win mirrors, which means you have a clear best deck, which means I should be tuning mine to death as well, which means I won't bother anymore.

I'm not sure if that means it. I have a quite simple script which wins against everything I could imagine, but there are lots of situations I can imagine that it would not deal well with. I just can produce no bot that gets it into such a situation, but it might exist. Maybe some bot uses the same key cards, but an completely different approach. Anyway, it would be a good idea to defend against it. But every simple change will let me loose the mirror.
So not the mirror is the problem, but some unknown situations that may happen, and in order to win both I would have to bloat the bot.

[Geronimoo]

One thing that isn't clear yet: does the person who has the best average record for the 3 bots win, or the person with the 1 bot that beats the rest or something else?

[DStu]

One thing that isn't clear yet: does the person who has the best average record for the 3 bots win, or the person with the 1 bot that beats the rest or something else?

I don't even have an idea for a second bot...

[DG]

I've got too many ideas. I'm just going to have to take a punt on a couple of themes and develop them. Unfortunately you can't really get a measure of some decks without getting the vp purchasing/gaining right and that can be hideously complicated.

[WanderingWinder]

One thing that isn't clear yet: does the person who has the best average record for the 3 bots win, or the person with the 1 bot that beats the rest or something else?

Bots are judged individually. So if you enter three, you can get first, 5th, and 8th (assuming there are eight entries).
You also don't have to enter more than one.

Finally, if you guys want it enough, I might extend the deadline.

[blueblimp]

While testing the mirror match of one of my strongest bots, the game would very often end up with a pin where one player gets his hand reduced to 0 and 5 cards of his deck gone each turn. Totally not meant to do it, but still cool to see this happen.

Very cool, I'd love to see this bot once the submissions are in. I gave up on trying any strategy that intends to trash all non-actions because the play rules make it awkward.

[DG]

Is anyone else finding that some heavy swindler/jester/ambassador decks are impossible to run for a number of simulations? The hourglass comes up and win percentages clear but that's it. You can continue working in the simulator and open sample games but that simulation never completes. Maybe that's the key to making an unbeatable deck.

[Geronimoo]

that's weird. The simulations don't run on a separate thread so you shouldn't be able to do anything else while it's still running. Could you send me your script (pm)?

[WanderingWinder]

Also, having received only one submission at this point, I'm going to extend the deadline... indefinitely at the moment. Person who submitted (you know who you are), feel free to tweak and resubmit. Everyone, let me know if you actually want to stick something in here, or whether your interest is theoretical at best.

[DG]

I'll submit some decks. The problem is that there are so many possible decks that you can't rule anything out and say a script is finished. The decks are also a bit extreme and put the simulator play rules under a lot of pressure. Finding the irregularities takes time too. I don't mind if the deadline remains at tomorrow.

[DStu]

Also, having received only one submission at this point, I'm going to extend the deadline... indefinitely at the moment. Person who submitted (you know who you are), feel free to tweak and resubmit. Everyone, let me know if you actually want to stick something in here, or whether your interest is theoretical at best.

I'll of of course let the others decide, but I'm not sure if extension is neccesary. If I wasn't quite sure that I have no time this evening and tomorrow to play with the script, I wouldn't have submitted it yet, although the main part was ready last week.
But as DG says, there's so much you can try and tweak, that there always can be some idea for an improvement, it was just that this morning it felt more likely to forget or have no time to submit than to improve, so I sent it.

But of course there is not much point in just having two scripts, so if no one else is this far, just extend it.

[Geronimoo]

I'd rather you keep the deadline. Some people will otherwise keep tweeking their bots to infinity and I'm not sure we're going to learn all that much from that. And there's only a handful of players who do these more complex simulations (it's not like a big money deck is going to win this) so I wouldn't expect many entries anyway.

[WanderingWinder]

Well, I say indefinitely now... but probably it would be ~ a week. If nobody suggests interest in submitting... well then it's moot. If people do, they should do it quickly (before tomorrow!), when I guess I'll make my decision.

[DG]

I think for this challenge anyone can have a go. Most decks will have an Achilles heel and it might be fun to find them.

[blueblimp]

I have 2 I'm going to submit. I was holding off on submitting until I made a 3rd and checked that they all work reasonably against each other.

[Galzria]

I thought about it. I love playing around in the simulator, but just the thought of not only producing a solid 5 card deck, with all contingent buy/play rules (well, play rules within the simulator's ability), but ALSO adding in code to account for the times when my opponent's bot may choose to share a card or two with me is just mind numbing. Assuming that you can gain more than 5 of any given card on a pre-created board of semi-strong cards (or at least high synergy) seems like a set-up to failure to me ... and some** of the strongest setups only work when amassing the card in question.

I'll be very interested in seeing the results once they're posted however, since the challenge was (is) very intriguing.

[blueblimp]

I love playing around in the simulator, but just the thought of not only producing a solid 5 card deck, with all contingent buy/play rules (well, play rules within the simulator's ability), but ALSO adding in code to account for the times when my opponent's bot may choose to share a card or two with me is just mind numbing.

Just don't worry about it. There are enough powerful cards that you will face some bots that you don't overlap with. I'm not too worried about fine optimization since it's just for fun anyway.

[WanderingWinder]

Okay, so since I've actually gotten a flood in today, I'm going to probably limit my extension to whenever I go to bed Sunday evening (about the time anyone in Europe might be waking up, if they're early risers, on Monday). Unless somebody really feels like begging...

[WanderingWinder]

Entries closed. I'll try to get results sometime around this coming weekend.

[WanderingWinder]

Okay. So some good news and bad news. Good news first. There are several quite different bots, and it's pretty darn interesting stuff. Nice variety, nice matchups, it's good. Yes. Huzzah.
Bad news. We've got some rules violations (I don't mean to imply that this was purposeful cheating - I'm sure they were oversights of some form or another). Of course, I missed them until I was most of the way through checking things. One competitor who submitted multiple entries did so specifying start states. I.e. he specifies getting a 4/3 split in one of his bots, specifies getting 5/2 in another. Second thing is that one of the competitors used two of the same cards in two of his different bots, i.e. he used both transmute and thief in bot A, and he used both transmute and scout in bot B. Third problem is that two of the matchups broke the simulator.
Now, the person who specified the start states- this is going to be a clear advantage. One of his bots was super super strong, and definitely is medalling (but not winning) if we allow it. But that doesn't really seem fair. Another one of his bots ticked what is otherwise the winning bot for its only loss. Also, this one is the one that has both matchups that are breaking the simulator.
What I could do is try to re-do these bots, to take out the start state restrictions, and re-run it. But that's going to probably push back the results be a week, apart from being a big pain for me personally. I also doubt it will solve the problem of the simulator breaking.
On the person with two bots using too many of the same cards - the thing is one of these two was one of the two worst bots in the contest, and the other was middle-of-the-pack. So I could toss them both out, toss the weaker one out, or leave them both in. I'm inclined to do the latest one, since it doesn't really make a difference, and this way I don't have to re-tabulate all the results, AGAIN. Also, it should be noted that, despite the presence of two of the same cards, the two bots play pretty differently.

Finally, there's a question of what to do about tiebreakers. Places, I had imagined, would go to whoever does best (i.e. wins the most matches) in the round robin. But how to break ties? There are a few ways I can think. Head-to-head matchup, or overall win% across games. Doesn't make much difference to me.


So, what do you guys want me to do here?

[DG]

I specified the start states. Just remove those and the good script should continue anyway with a 4/2 start. I'll withdraw the script that breaks the simulator if need be. The third one was only a curiosity anyway (although it would have been stronger if I'd submitted the version that I knew broke the simulator with the other one!).

[RaVeNLoRD]

This is all very interesting, Can't you post the bots in the mean time until you have the results?

[DStu]

At the moment, I'm more interested in the results than in fairness.