Print

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

[Davio]

I have this idea in my head to create an advanced kingdom selector.
I doubt whether I'll actually make one, because I'm very good on starting new things and very bad on completing them, but it might make a fun experiment.

There are purists among us who say "Every Dominion kingdom should be random!". Well, I'm not a purist and I'm not entirely on the other side, so I don't really care.

To me, a kingdom should be interesting. It should provide multiple options. The plethora of ways one could go about a kingdom were what sucked me in when I started playing. It's fun to discover new things! I'd rather succeed greatly or fail horribly with a Rube Goldberg engine than toss a coin on an uninteresting mirror board.

I actually feel kind of bad when I fire up my Torturer chain or play my 7 Hunting Parties + 1 Terminal over and over and over and over. These games are not fun.

So, in an effort to make kingdoms fun (and provide multiple roads to victory) I'd like to see if we can find some partners for otherwise sad pandas.

Our mission: Remove dead cards from the kingdom!

There is no point in having a card in your kingdom that's never going to get bought, you might as well play with a stack of napkins. It is an inevitable aspect of having so many cards to choose from, but still...

A simple example is Tunnel, it has a reaction to discarding, so the obvious thing would be: If we select Tunnel from our random list, select a discarder, like ... (it's a long list).

We could even add weights to the list. Warehouse is a better discarder than Spy, so they could be given different weights. And maybe we don't want super obvious combos like this, so we could tweak it some. You have to watch out for the really dominant combos.

The point is that some cards need partners to be considered for a strategy and that's what we're trying to achieve, they should be at least considered by both players.

It could also be tuned for 3p and 4p (no Pirate Ship, Tom Vasel?).

We could also let users fill in their own preferences. Torturer likes villages combowise, but players don't necessarily like having Torturer always come with a village.

So the partnership doesn't have to be bidirectional per se. I mean, a card that reacts to attack cards is probably more useful with attack cards around than without them, exceptions noted. On the other hand, attack cards don't really need reaction cards.

At this point I'm pretty much at the end of what I came up with at this time, but I'd like for you guys to keep spitballing along and seeing where it goes.

[LastFootnote]

It's an interesting idea, but impractical to my mind. There are just too many variables and the power of combos is in the eye of the beholder to a certain extent.

When I randomize, I usually pick 2 sets randomly and pull 5 random cards from each, sprinkling in promo cards here and there. That way we usually have a pretty good mix of card types (villages, terminal draw, attacks, etc.) and we see a bit more of the intra-set synergies. Are there still dead cards? Sure, but not as often as pure random, I'll wager.

[jonts26]

I think the idea is to have more like a selector app which creates interesting kingdoms instead. I know lot of people who use app selectors.

As for the original idea, it sounds somewhat complicated but I could get behind something like this. I prefer playing engines, though I don't go out of my way to build them when strong big money alternatives exist. So maybe something like you give cards certain properties (attack, strong trashing, weak trashing, +buy, village, etc) and you increase the weighting of certain cards based on the already selected cards. So if you've selected 5 cards with no villages yet, you become more likely to get a village. This basically makes it more likely that you get a good mix of engine components like attacks, trashing, card draw, maybe alt VP etc.

[LastFootnote]

As for the original idea, it sounds somewhat complicated but I could get behind something like this. I prefer playing engines, though I don't go out of my way to build them when strong big money alternatives exist. So maybe something like you give cards certain properties (attack, strong trashing, weak trashing, +buy, village, etc) and you increase the weighting of certain cards based on the already selected cards. So if you've selected 5 cards with no villages yet, you become more likely to get a village. This basically makes it more likely that you get a good mix of engine components like attacks, trashing, card draw, maybe alt VP etc.

But some people (like me) prefer many or even most of their games to be non-engine games. I don't think there's a one-size-fits-all solution here.

[jonts26]

Fair enough. You could just have certain qualifications then. You can choose between several selection criteria. Engine, Big Money, Alt VP, Pure Random, etc. Options are the key.

[Insomniac]

I think it would be key to have multiple different routes to the same goal exist that are viable.

For example sure Tunnel+Warehouse is good. But if another strategy is a torturer chain then thats where this game is going.

[eHalcyon]

There was a thread a while back about ways to do this.  Something about pulling CR data and reverse engineering a tree of synergies and anti-synergies based on win-rate-with percentages or something.  The idea was to put together cards that combo together as well as cards that counter those ones, so that there are interesting decisions to be made about whether the combo is worth it, etc. etc.

Anyone else remember that thread?

[Toskk]

Hi Davio (and all),

Well, the features are a little more simplistic than the weighted combo method you were musing about, but I've been developing a Dominion card picker with a variety of options for artificial Kingdom selection for a while now. For example, one of the options specifically attempts to make sure that 100% of the features of all Kingdom cards present will be 'useful' (e.g. if there's a Moat there will always be an Attack card present, or if there is a Tunnel, there will always either be a discarding attack or a way to discard cards). For those interested, that project can be found here:

http://inprogressgaming.com/dominion-card-picker/

I'd definitely welcome suggestions and feedback on how to better eliminate 'dead cards', too.

[ednever]

There was a thread a while back about ways to do this.  Something about pulling CR data and reverse engineering a tree of synergies and anti-synergies based on win-rate-with percentages or something.  The idea was to put together cards that combo together as well as cards that counter those ones, so that there are interesting decisions to be made about whether the combo is worth it, etc. etc.

Anyone else remember that thread?

That was me. If you search by threads started it should be findable (I'm on an iPhone sobincant do it). I still like the idea. Especially since its not perfect.

A general thought:

One of the real appeals to dominion (for me) is seeing non-obvious things. A lot of what people are talking about is eliminating those opportunities. There was a great thread that tallied about someone's (I think it was Marin) engine they built with festivals and moats.

How cool is it to figure out and succeed buying moats on a board with no attacks?

Equally fun is recognizing when a card should NOT be purchased. I think some of the old kingdom design contest sets were like that - they put in cards specifically designed to be traps.

Ed

[werothegreat]

I've actually played the First Game setup enough where I can use every card in it.  Except Woodcutter.  Woodcutter has absolutely no business being in that setup.  Why would I buy a Woodcutter when there are Markets?  I would rather finish off my chain with a Militia.  Every other card in that setup has a purpose.  Village/Smithy to give you your engine, Market to do its little thing (aiding the engine, giving +Buy and +Coin), Mine to turn Copper into Gold and Remodel to turn that Gold into Provinces, Workshop to get a head start on the Village/Smithy acquiring, a couple Militias to slow down your opponent and a Moat or two to prevent your opponent from doing the same thing, and Cellar to keep the heartbeat going during greening (although once you've got the Mine/Remodel thing working, greening won't last very long).  Cellar also is a shield against bad shuffle luck - I'm perfectly willing to discard Militias and Mines if I know I'll draw Villages and Smithies, allowing me to draw all that back again.  But Woodcutter is just... there's no reason for it to be there.  Honestly, I'd rather have Chancellor.

[popsofctown]

There was a thread a while back about ways to do this.  Something about pulling CR data and reverse engineering a tree of synergies and anti-synergies based on win-rate-with percentages or something.  The idea was to put together cards that combo together as well as cards that counter those ones, so that there are interesting decisions to be made about whether the combo is worth it, etc. etc.

Anyone else remember that thread?

That was me. If you search by threads started it should be findable (I'm on an iPhone sobincant do it). I still like the idea. Especially since its not perfect.

A general thought:

One of the real appeals to dominion (for me) is seeing non-obvious things. A lot of what people are talking about is eliminating those opportunities. There was a great thread that tallied about someone's (I think it was Marin) engine they built with festivals and moats.

How cool is it to figure out and succeed buying moats on a board with no attacks?

Equally fun is recognizing when a card should NOT be purchased. I think some of the old kingdom design contest sets were like that - they put in cards specifically designed to be traps.

Ed

Humans have a nasty tendency to value what they already have in hand over something they could potentially gain.  http://en.wikipedia.org/wiki/Loss_aversion.
This is the main thing holding people back from advanced kingdom generation methods.  Yeah, you lose that one cool Festival Moat game.  But what if you gain so much more?

[ConMan]

Dominion Shuffle for Android has a lot of flexibility in what you can set, and while sometimes I want a purely random kingdom sometimes I tweak it for something more interesting. You can set any individual card or category of cards to have a minimum and maximum number of cards in the Kingdom, including conditionals of the form "If there is an X in the Kingdom, have a minimum of Y of this card/category" (which I usually set for Potion-cost cards so that there's always at least two if there are any of them, but you could also have guaranteed Reactions when there are Attacks and so forth). There's no "Discarder" category, but there are a few interesting ones in there.

[ftl]

Humans have a nasty tendency to value what they already have in hand over something they could potentially gain.  http://en.wikipedia.org/wiki/Loss_aversion.
This is the main thing holding people back from advanced kingdom generation methods.  Yeah, you lose that one cool Festival Moat game.  But what if you gain so much more?

But... what do you gain? Any kingdom that is possible with fancy selection is possible with random selection, but the reverse is not true.

[popsofctown]

This is a problem in perception.  There are more kingdoms than atoms in the universe or something like that.  If advanced kingdom selection goes from all possible random kingdoms to 1% of the possible kingdoms, then your new set of kingdoms is just a subset of the old one, and it's a subset that retains the property that you will ne-ver play the same kingdom twice.

[ConMan]

This is a problem in perception.  There are more kingdoms than atoms in the universe or something like that.  If advanced kingdom selection goes from all possible random kingdoms to 1% of the possible kingdoms, then your new set of kingdoms is just a subset of the old one, and it's a subset that retains the property that you will ne-ver play the same kingdom twice.

And, as long as your rules of selection are half-way decent, they should still give you a good selection of interesting kingdoms. Actually, if your rules are good enough, they should give you a *better* selection of kingdoms than average.

[Davio]

This is as much a thought experiment as it is a concrete approach.

What I would like to achieve is for every card in the kingdom to be considered. Whether they end up being bought is less important, that's up to the players.

That's why I thought about creating subsets of cards.

So for instance, if Scout is in the kingdom, put Mystic in there with some probability. It's better than having a dead Scout.

What makes a good or bad or interesting kingdom is hard to quantify as it's probably different for different people. And even for the same people, it can be different with other playing groups. I don't like to play a Torturer chain with my parents, but I don't mind KC-ing a couple of Wharves, they actually like seeing me do these megaturns.

I don't think the Councilroom correlation data is a good starting point, probably because people will stick to what they know a lot of the time.

And maybe only a small selection of cards actually need partners. Cantrips or +Actions cards are useful a lot of the time. Even Spy might be a halfway decent purchase after a couple of Silvers.

[PenPen]

I'm actually be more interested in a card selector (app) that generates a kingdom set the first time, and for your next game generates a kingdom set that doesn't include ones that you already played.

I'd still prefer a random draw so players can discover combos or methods on their own. Or a random draw with veto.

[popsofctown]

Also, look, there are at least a few advanced selection rules you can make that won't cause any loss of possibility at all.  Scout without support will not work, you can list the support cards one by one.  No weird combination of 2 or 3 other cards will enable it on its own, because the way the card works is how it works. 

But it'd be even better to use aggressive rules that might remove some hipster combos.  There will still be milliions of hipster combos left.  It's okay.  Think about it this way.  In your lifetime you will play X kingdoms.  They will have like 8, 9, or 10 cards (a few will have 1 card named Jack).  In the 8, 9, or 10 card kingdoms you will find and play Y hipster combos.

Now what if you play those same X kingdoms with Advanced kingdom selection.  Perhaps half of your hipster combos will get removed.  But you'll play 9 and 10 card kingdoms.  That increases the odds of a hipster combo being available by at least 1*8, more if you think about 3 card combo combinatorics.  So you'll actually play more than Y/2 hipster combos in the games that got affected - you'll actually play 4Y.

If Dominion was just the base set, worry about deleting something interesting would be valid because we could exhaust the possibilities.  But as it is, they are impossible to exhaust.  There are quatrillions iirc.  Cutting down to 1% is still trillions.  You cannot exhaust.

[DG]

There are some problems with this approach. If there are villages, mining villages, and farming villages in a kingdom then a selector might throw some out as unwanted cards. It could be the case that this particular mix of villages can enable some strategy that is otherwise unlikely, or perhaps the right mix of villages or the right choice of village at the right time is important. Perhaps having similar cards are differing costs can enable a strategy as well.

There is also an argument that even unbought cards in a kingdom can alter strategies, even if it only rewards a commitment to buy key cards from the stronger supply piles.