Print

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

[pacovf]

He recruits villagers for you. Which you then use to man your smithies and militias and whatnot. The location he recruits from doesn’t like his aggressive recruiting techniques, so they secede from your Dominion.

[silverspawn]

  well I guess that does make sense

[spineflu]

do you think superman, having been raised by jonathan and martha kent, thinks applebees is fancy

[GendoIkari]

Maybe as a kid he did; but he would have gotten more cultured once he moved to Metropolis.

[Tables]

In my dream last night I somehow created a Pandemic role that almost seems plausible as a real card.

Janissaries
When you treat disease, you may first place a supply token in this city, or move a supply token from another city into this city.
When any player treats disease in a city with at least one supply token, they treat one additional cube of any colour.

I assume there would be something like 8 supply tokens.

Asides from the absurd name, I feel like it would be a weak but actually viable role. It would work similar to the Medic, but with less immediate benefit in exchange for a bit of extra long term power. Unless dreams count, I have not playtested the role.

[Kuildeous]

I'm going to create a new Pandemic roll called Karen. You have to add an infection cube in your current city, but you get a sick haircut.

[silverspawn]

I've just had a conversation with my phone about the importance of word(s). I feel misunderstood.

I must say, it is ironic that's you do not understand verbs. Out of all all the birds that I would like you to understand I think the word Verde is pretty far up there karma at least much farther up Thunderbird bird.

[2.71828.....]

I just realized the "Interview with Donald X." thread has reached 100 pages (at least for all of us who use the superior 50 posts per page view format)

Anyone think it will go the way of random stuff 1, 2, and 3?

Coming soon to a thread near you: "Another interview with Donald X."

[silverspawn]

I CAN'T FIND MY PHONE

I just decided to stop searching for a while and do some nonproductive stuff but I WANT MY PHONE FOR THAT



Edit: I found it (wtf I don't know how it got there) but man it's so scary not to have it. I don't even need to use it but it needs to be there. Otherwise, something is missing.

[LostPhoenix]

I CAN'T FIND MY PHONE

I just decided to stop searching for a while and do some nonproductive stuff but I WANT MY PHONE FOR THAT



Edit: I found it (wtf I don't know how it got there) but man it's so scary not to have it. I don't even need to use it but it needs to be there. Otherwise, something is missing.

I've been living without a mobile phone since January. It's amazing how much money can be saved by not having the monthly bill for a while.

[GendoIkari]

https://www.reddit.com/r/dominion/comments/hp1do3/i_would_understand_if_my_girl_left_me_over_this/

[Straw]

do you think superman, having been raised by jonathan and martha kent, thinks applebees is fancy

Probably not IMO.

[shraeye]

I'm trying to put together a small project, I'm hoping for native (or near native) speakers of non-English languages to record the phrase

"One, Two, Three; this is a test.  Thank you"

in their target language.  Bonus points if it's a reasonably common language.

Any takers?

[Awaclus]

I'm trying to put together a small project, I'm hoping for native (or near native) speakers of non-English languages to record the phrase

"One, Two, Three; this is a test.  Thank you"

in their target language.  Bonus points if it's a reasonably common language.

Any takers?

Like this? https://dl.dropbox.com/s/sip800vlv2kludv/not_a_serial_killer.wav

[shraeye]

Exactly like that.  Thank you sir!

[LostPhoenix]

Does anyone have the like to the Dominion Discord? I'd like to join, but can't seem to find it.

[Donald X.]

Does anyone have the like to the Dominion Discord? I'd like to join, but can't seem to find it.

https://discord.gg/wcGE9PfG

[silverspawn]

How hard is dominion to learn for AI? Easier than chess? As hard as chess? As hard as Go? Harder?

By learning dominion, I mean something like 'program an AI that, given a random kingdom, plays about rating 65 with about 15 minutes of total thinking time'.

I find this question quite interesting because, on first glance, I really have no idea. Neither 'way easier than Chess' nor 'way harder than Go' would surprise me that much. Any thoughts?

The only concrete idea I have here is, maybe each individual dominion board is so easy to solve that you can solve the general problem by building an AI that solves each individual board within the 15 minute limit. That might work. I think AlphaZero only played for about 8 hours against itself to learn Chess, and a single dominion board could be much easier. Then again, even that isn't obvious.

[faust]

How hard is dominion to learn for AI? Easier than chess? As hard as chess? As hard as Go? Harder?

By learning dominion, I mean something like 'program an AI that, given a random kingdom, plays about rating 65 with about 15 minutes of total thinking time'.

I find this question quite interesting because, on first glance, I really have no idea. Neither 'way easier than Chess' nor 'way harder than Go' would surprise me that much. Any thoughts?

The only concrete idea I have here is, maybe each individual dominion board is so easy to solve that you can solve the general problem by building an AI that solves each individual board within the 15 minute limit. That might work. I think AlphaZero only played for about 8 hours against itself to learn Chess, and a single dominion board could be much easier. Then again, even that isn't obvious.

I guess this can be approached in several ways. In chess, a player makes about 37 decisions throughout the game. I think it's fairly obvious that a typical game of Dominion has way more of those. Of course, a large number of those will be obvious, but I feel like overall the complexity is going to be higher. Go apparently averages at about 211 moves, and this seems to come closer to a typical game of Dominion. But I think the decision space for each of these decisions is significantly bigger than with Dominion. So I'd assume that Dominion is probably harder than chess and easier than Go.

This is not taking into account different Dominion boards, which of course also add complexity. On the other hand, an AI has an inherent advantage over a human Dominion player compared to a human chess player - in chess/Go, all relevant information is always openly available. deck tracking is trivial for an AI but tricky even for a good human player, and tracking everything that your opponent does is even harder. From that perspective, I would expect a human Dominion player to be "worse at the game" compared to a human chess player (though that is difficult to define), meaning that an AI doesn't have to be as sophisticated to beat most humans.

[LastFootnote]

On the other hand, an AI has an inherent advantage over a human Dominion player compared to a human chess player - in chess/Go, all relevant information is always openly available. deck tracking is trivial for an AI but tricky even for a good human player, and tracking everything that your opponent does is even harder. From that perspective, I would expect a human Dominion player to be "worse at the game" compared to a human chess player (though that is difficult to define), meaning that an AI doesn't have to be as sophisticated to beat most humans.

On the flip side, having all information always openly available allows a computer to look several moves ahead, potentially planning out all possible permutations of moves for the rest of the game. This isn't possible in Dominion due to randomness. I don't know that this would give a human player an advantage, but I'm not sure it wouldn't.

[silverspawn]

I'm not sure how strong the link between numerical complexity and difficulty really is. It's true that Go has a larger decision space than Chess and is harder, but this could be coincidence. Starcraft has an astronomically larger decision space than either of those games, and I don't think it was that much harder.

The difficulty of evaluating a game state may be a better metric. I think that would be easier in dominion and Starcraft than chess, and I don't know how hard it is in Go.

[infangthief]

On the other hand, an AI has an inherent advantage over a human Dominion player compared to a human chess player - in chess/Go, all relevant information is always openly available. deck tracking is trivial for an AI but tricky even for a good human player, and tracking everything that your opponent does is even harder. From that perspective, I would expect a human Dominion player to be "worse at the game" compared to a human chess player (though that is difficult to define), meaning that an AI doesn't have to be as sophisticated to beat most humans.

On the flip side, having all information always openly available allows a computer to look several moves ahead, potentially planning out all possible permutations of moves for the rest of the game. This isn't possible in Dominion due to randomness. I don't know that this would give a human player an advantage, but I'm not sure it wouldn't.

I read somewhere (admittedly a long time ago, so I don't know if it is still the case), that humans could improve their chances against chess computers by aiming for closed games and making positional moves whose benefit was beyond the computer's analysis horizon.
I suppose it is the strategy vs tactics distinction.
In Dominion terms, a human may have a better 'instinct' of things like deck composition compared to a computer. Some boards would favour this sort of thing more than others.

[silverspawn]

On the other hand, an AI has an inherent advantage over a human Dominion player compared to a human chess player - in chess/Go, all relevant information is always openly available. deck tracking is trivial for an AI but tricky even for a good human player, and tracking everything that your opponent does is even harder. From that perspective, I would expect a human Dominion player to be "worse at the game" compared to a human chess player (though that is difficult to define), meaning that an AI doesn't have to be as sophisticated to beat most humans.

On the flip side, having all information always openly available allows a computer to look several moves ahead, potentially planning out all possible permutations of moves for the rest of the game. This isn't possible in Dominion due to randomness. I don't know that this would give a human player an advantage, but I'm not sure it wouldn't.

I read somewhere (admittedly a long time ago, so I don't know if it is still the case), that humans could improve their chances against chess computers by aiming for closed games and making positional moves whose benefit was beyond the computer's analysis horizon.
I suppose it is the strategy vs tactics distinction.
In Dominion terms, a human may have a better 'instinct' of things like deck composition compared to a computer. Some boards would favour this sort of thing more than others.

I don't think this is up to date anymore. I've watched analyses of AlphaZero vs Stockfish, the best conventional chess program, and AlphaZero went for a closed game with ridiculous positional plays, like sacrificing several pawns and then moving the king one square in a super closed position before it launched the attack. It may have been true while Stockfish was the best chess program (which already plays better than humans).

[blueblimp]

How hard is dominion to learn for AI? Easier than chess? As hard as chess? As hard as Go? Harder?

Harder for sure.

First issue is how you deal with each card introducing its own rules. Arguably, a fully legit Dominion AI would be able to play with a new card without having to re-train the model. That's mega hard, so realistically you probably would just re-train the model.

Some strategies in Dominion need all the components in place or they are garbage, e.g. in some kingdom, if you try to play an engine without +buy it may be garbage, and with +buy the best strategy. You can't really just do random stuff and stumble into a decent strategy the way you can in chess and go. That's a problem because that's how self-play training works.

Dominion has hidden information, most notably your opponent's hand. It's like there's a little bit of poker embedded in the game. AIs can play superhuman poker now, but the techniques used for that are completely different from what's used for other games.

Randomness means search isn't going to help much, unlike in chess and go. Randomness isn't itself such a big problem; IIRC computers were playing superhuman backgammon before superhuman chess. But it does diminish the opportunity to out-calculate humans.

Strategies in Dominion are highly diverse: what's good in an engine is totally different from what's good in BM+X. This is different from chess and go, where, despite some minor variation, you're doing essentially the same thing every game. I can't off-hand think of any game with diverse strategies where computers are superhuman. I don't know if that's because it's hard or just because it hasn't been seriously tried, but it means you're breaking new ground.


I'm not an AI researcher, but if I were trying to develop a Dominion AI, I'd start by making some massive simplifications. Singleplayer only (with the goal being something like buy 5 provinces as fast as possible) and limited subset of cards (maybe just first kingdom). This already sounds pretty hard. Try to get it to discover the first game engine, for example. (Although that has Militia, so maybe it's not even good singleplayer...)

Starcraft has an astronomically larger decision space than either of those games, and I don't think it was that much harder.

AFAIK, computers are still garbage at the strategic part of Starcraft.

[Awaclus]

Randomness means search isn't going to help much, unlike in chess and go. Randomness isn't itself such a big problem; IIRC computers were playing superhuman backgammon before superhuman chess. But it does diminish the opportunity to out-calculate humans.

Doesn't that also work the other way round? Humans are notoriously garbage at calculating probability (especially at guesstimating it when the math gets too complicated to do on the fly), but for computers, it's trivial. Working out chess or go decision trees in your head is still impossibly hard to do even nearly as comprehensively as computers can because you just don't have the brain capacity to do it, but I don't think it's an inherently difficult skill for humans.