Print

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

[Catalytic]

We spend a lot of time thinking about various combos and the relative strengths of cards, but we do not spend as much time thinking about how to approach evaluating strategies when presented with a supply.  We could know all of the combinations and various card strengths, but our game still lacks if we do not know how to approach a supply and evaluate the various options offered by a random Kingdom.

How do we think about the various possibilities?  One way is to approach the problem is a piecemeal approach seeking out a combo that we like and if we see no obvious reason not to play it then we proceed to play it.  But, I think that we can conceptualize how to think about the supply in a more organized fashion.

I am certainly not the best Dominion player, but I have begun constructing a decision tree to think my way through the various possibilities offered by any supply.  Below is the beginnings of that decision tree.  Hopefully, the decision tree is fairly self-explanatory, but I want to briefly go through some of my thoughts.

I begin with the question of whether a viable alternative VP or rush strategy is present since these strategies fundamentally change the way the game can be played.  You could build the best deck in the world, but if it is not capable of capturing most of the provinces in a timely manner a sound alternative VP strategy will win.  So, if such a strategy is available then we must consider whether we can build a deck capable of taking, if not all, almost all of the provinces.  If we cannot build such an engine then my thought is that you must play the VP strategy.

If there is no alternative VP strategy then the question of fast trashing is next since fast trashing (or pseudo-trashing a la Native Village or Ambassador) also changes the game.  Engines become viable that would not be viable in lieu of such trashing (conspirator chains, bridge combos, HoP combos, etc.).  Since these engines can create mega-turns that can end the game without warning, a slow deck has very little chance against them.  So, if an engine exists then we need to find it before considering alternatives.  This is not necessarily decisive since various attacks or alternative strategies might be faster or create too many problems for an engine deck (e.g. good Saboteur or Swindler decks for instance) but in order to understand the possibilities we must search for them and see if they are viable in any given kingdom.

In absence of strong trashing, engine decks are often too slow (often being the key word).  Then, finally, we begin thinking about the various attack possibilities.  Big Money + BM enabler is usually the correct path if strong trashing and strong attacks are absent.  If strong attacks are present, then we have to do the work of figuring out how the attacks change the game and how we might counter them.  Donald X's article (http://dominionstrategy.com/2011/12/27/your-guide-to-beating-attacks/) is a great beginning in thinking about how attacks alter the game and the possibilities of hard and soft counters.

Anyway, I digress.  A couple of quick notes:

1. This decision procedure is only intended for a Province game.  Colony games both increase and decrease the possibilities (increase in that more complicated strategies are feasible but decrease in that fewer strategies are feasible if you want to reliably get to $11).

2. I do not present this decision tree as decisive in deliberations.  The infinite subtlety of the game is what makes Dominion fascinating and an option that appears lower on the decision tree is not ruled out since various combinations on the board and certain enablers can make some strategies more dominant.  This is merely presented as a first pass way to think about the various possibilities that any Kingdom offers.

I, of course, welcome thoughts from everyone here especially those with far more experience than myself.  Thank you for reading this and may you never draw your terminals dead.

[tlloyd]

Great work--this is a very interesting idea. I think that the initial question (alternate VP rush strategies?) needs some work though. Because there are actually two distinct questions here: rush strategies and alternate-VP strategies.

The first question is whether a rush strategy exists. Often these will involve alternative sources of VP like Gardens or Silk Roads, but they don't have to. As you point out, IGG often creates a rush strategy involving only Duchies. A rush strategy is generally not designed to score more than a Province strategy would over the long-term, but is instead designed to (1) accelerate the end-game by emptying piles and (2) grab more VP along the way than a longer-term strategy will have accumulated by that point. So the first question is whether a rush strategy is viable.

The second question is whether there are alternative VP strategies. These are not necessarily rush strategies. The clearest example is a Goons engine, where the goal is to create an engine that generates VPs without ever getting bogged down in green cards. Oftentimes these games end on piles, but that rarely happens until after both players have set up elaborate engines. Bishop and Monument can also work for this, but they are more commonly used in conjunction with a standard province-gaining strategy. Of course, alt-VP strategies aren't limited to VP-tokens. Fairgrounds or Vineyards can be worth as much or more than Provinces in a deck that is built right--so you might plan all along to aim for those alternate VPs instead of Provinces. And alternate VP strategies don't have to exclude Provinces outright--for example you can decide to buy Harems as part of a BM strategy aiming for Provinces. Your economy builds slower than a competing gold-based strategy, but you are amassing VP as you go. Obviously there are lots of examples, but the basic point is that after ruling out a rush strategy, you still have to ask yourself where your VP is going to come from in the end.

[mharris717]

Great post.

To the engine section, I would add something like this

-------------

If there is strong trashing available and you plan on building an engine, you should classify what kind of engine you're building.  There are two main types.

1. An engine that will start acquiring provinces 3-4+ turns before the end, will likely acquire no more than 1 province per turn, and will win essentially by getting four provinces before your opponent

2. An engine designed to get one mega turn.  If a mega turn cannot get a minimum of 3 provinces, it is probably not worth doing.  You must evaluate whether you can get your mega turn set up before it is too late.

------------------------

[Catalytic]

Thank you for the thoughts tlloyd.

I was thinking of the rush/alternative VP decks as presenting the same problem: if you are not going to play that strategy then you need to be able to hit all 8 provinces (or near enough) reliably and fairly quickly to counter them.  The two strategies that I mentioned in the notes were Gardens/Workshop and IGG/Duchy which are both rush strategies.  But, I was thinking that strategies such as Duke/Duchy, Silk Road or Gardens (w/o a Workshop/Ironworks enabler) and Fairgrounds decks are not truly rush strategies, but they will outscore a typical Province centered deck unless the Province deck can hit all of the Provinces fairly quickly which usually requires a reliable engine.  So, the response to both rush decks and most alternative VP strategies is the same: either play the alternative strategy or build a solid engine to outscore them or outrun them with Provinces.

I was not thinking that Bishop decks or Goons decks as truly alternative VP strategies in this sense; I think of them fitting more comfortably as Engine decks.  Not every deck with alternative VP is an alternative VP deck in the sense I suggested as you rightly pointed out (which means I certainly need to be clearer about).  For instance, using Monument as a BM enabler involves alternative VP, but the strategy focuses on hitting $8 and trying to win the Province battle.  Other examples, as tlloyd pointed out, would be using Harem in a Big Money Deck, using Bishop as a trasher to enable other strategies, etc.  So, merely using alternative VP is not an alternative VP strategy in the required sense and I need to be clearer about that.

So, the first question might take on of two forms to account for your suggestion:

1) It can be split into two separate questions and decision procedures:

Is there a rush strategy on the board (e.g. Gardens/Workshop, IGG/Duchy)?

Yes? - So, is there a viable alternative strategy to grab enough Provinces to outscore the rush strategy? - No? Play rush strategy. - Yes? Choose between the rush strategy and the Province grab strategy.

No? Proceed to next question.

Is there a non-Engine alternative VP strategy that will outscore a Province centered deck (e.g. Duke/Duchy, Silk Roads/Gardens without enablers, etc.)?

And then proceed down the tree.

2)  Just make the first question more specific to alleviate worries about what counts as an alternative VP strategy:

"Is there a rush strategy or non-Engine VP strategy that will either outrun or outscore a non-Engine (or maybe "slow" or "typical") Province centered deck?"

The first version is more specific which is a bonus, but it also adds a sub-tree that is extraneous if I am right that the same question must be asked in the face of both types of decks (i.e. can I grab enough Provinces fast enough?).

The second version is cleaner, so it would be good if the difficulties suggested by tlloyd can be averted by merely being more specific about the type of decks that we should be concerned about at that level.

Anyway, thank you greatly for your thoughts tlloyd!

[Mean Mr Mustard]

OP: I think you are getting warm. 

I am not a student of the simulators, but I certainly see that they provide an avenue to deeper understanding of the game for even the best players.  And while they can teach improved build orders and can help a player discriminate between similar cards and different simple strategies, the simulators do not encompass all the strategic Kingdom possibilies; they cannot.  It is far too complex.  Any flow chart that a student of the game designs will have the same  shortcomings.  That is not to say that this kind of exercise is futile.  Like the simulators, I think there is a place to discuss and explore this idea.

Okay, the thing to remember is that the simulators are backed by mathematics, and a decision tree isn't necessarily so.  Without a strong input from the sim crowd a decision tree is basically made from analogies, imperfect experience and what we can scrape from game data.  We are firmly in the realm of opinion.  Almost all of what we discuss in this forum is situational to the Kingdom.  A master of the Sim, like Geronimoo, has a great edge against an intuitive player like myself, no doubt.  But in reality the game does not operate with a three or four card kingdom.  I have doubts about the possibility of a master AI for this reason as well.

So, again backing out to a general stance, as we always do here, my decision tree is a bit different in places, and is often a process of elimination.

First, I assess my VP strategy.  I am not going to rehash this.  You guys basically got it right.  I'll just say that well played Alt VP strategies usually make four Province in thirteen move strategies look stupid.

Number two: what disruptive cards are present?  I am under the opinion that I should always try to throw the first punch.  Do unto others before they do unto me.  Any counter that would make me shy away from attacking better be fucking good, and I'll tell you why.  The attacker has the initiative.  The attack is good every time it is drawn while the counter must be present at the right time.  This is why cards like Menagerie are so powerful.  They fill the role of attack counter while still adding potential energy the deck by themselves.

Next, how will I deal with my opponent disrupting me?  I seek out the tools to alleviate what will probably happen to me.  Also now is the time to mentally discard strategies and cards that will be destroyed by the present attacks.  For example, think about the difference between a Menagerie and an Oasis in a hand disruption game.  Cellar, Warehouse, Inn, Oasis; these cards suck in three card hands.  There are a million examples to cite, and gaining an ever deepening insight into card interactions lead to continual improvement.

Now, taking into account the above it is time to plan an economy.  I am looking at trashing, engine possibilities, treasure ramp, strong combos, anticombos, strong single cards and beefy openers.  Look at the $5 cards first.  If they look sexy, my opening buys should reflect my desire to gain these as early and as often as possible.  In a way, there are three steps to take in building a VP purchasing economy: ramp, engine, payload.  The early ramp buys the engine that produces the payload that gains VP. You want a simple recipe for Dominion strategy, there ya go.  It can be as simple as Masq/Venture/Gold and be very good, or the best strategy may be much more subtle.  But always, the clock is ticking.

[Catalytic]

Thank you for your thoughts mharris.

This section of the tree needs plenty of work.  The "Search for Engine Possibilities" portion of the tree is likely a tree within itself and needs to connect with the attack section since certain attacks will make setting up different engine decks more problematic.

But, it is also nice to leave that part of the tree open since that is where knowledge of engines, how to construct them, how they fair in the face of attacks, etc. is where the creativity and knowledge of truly fantastic players shows itself.  I don't know that there is a decision tree to figure it out, but more specificity might not be bad.

So, we might add a layer under "Search for Engine possibilities":

1. Search for reliable $8 engines (e.g. Minion decks, Lab chains)
2. Search for megaturn engines (HoP, Bridge or Highway engines, etc.)
3. Search for alternative VP engines (Goons or Bishop engines)
4. Search for "Remodel" engine decks (e.g. Governor decks)

And more types of engines that I am forgetting.

Heck, part of the decision tree might require a question like "Is King's Court on the board?".  No? Proceed down the tree Yes?  Spam it and figure it out from there.

I would like to generally avoid focusing on specific cards or strategies (other than BM since it is the baseline strategy that any other strategy must beat) and keep the tree somewhat general.  The point here being that I like the openness of the tree since it leaves in tact the notion that the truly excellent player possess something that no decision tree or algorithm can possibly replicate.

Thank you greatly for your thoughts mharris and I will certainly keep your ideas in mind if I decide to expand that part of the tree.

[Catalytic]

Thank you for your thoughts Mean Mr Mustard; they are greatly appreciated.

First, as my response to mrharris suggested, I am equally skeptical that there is an algorithm to Dominion success (or at least an algorithm that a non-supercomputer can figure out).  The great players (whether they are simulator masters a la Geronimoo or great intuitive players such as yourself) possess something that no decision tree can capture and that lower level players such as myself lack.  The thought behind the tree is merely an organized way to search for the possibilities on a given board.  Your success identifying winning strategies will require a great deal of experience and "feel", but the problem for most lower level players is even being able to find the alternatives.  My hope is that a decision tree can at least get the dominant strategy (strategies?) on the table instead of missing it due to either unorganzied thinking or tunnel vision (I see Goons and must play it no matter what!).

Second, your decision procedure is, I think, what must come after understanding the various options on the table and this is where the magic happens.  Once we have identified the possibilities, the other cards on the table effect how well we can pull off our strategy.  For instance, I see a great engine possibility, but the King's Court/Sab combo is most likely going to blow it up before you can pull it off.  And so on.  I begin this at the end of the tree when I consider attacks, but I think it is a mistake to think that any part of the tree is independent of those questions.

So, to address your thoughts, think of the decision tree as creating a set of possibilities and each possibility must then be evaluated given all of the other strategies on the board.  But, if the decision tree works correctly, all of the viable strategies are on the table, so, from there, we can begin the really heavy lifting that separates the great players from the decent players.  Then, your decision procedure concerning economy/ramp, engine, payload must be addressed concerning each possibility that the decision tree offered.

Again, thank you for your thoughts.

[lympi]

Thank you for reading this and may you never draw your terminals dead.

This is a great thread, but this is my new favorite thing. It's like saying Grace but for Dominion.

I'd be interested to see other players make their own decision trees. I like engines, so usually the first thing I look for in a kingdom is if there are any +Actions.

[tlloyd]

I think it is worth considering rush strategies and alt-VP strategies separately. You don't need all 8 Provinces to counter a rush strategy -- you need to grab a few and (this is key) do it faster than you generally would. A rush strategy doesn't try to outscore a standard Province strategy, it tries to jump to a quick lead and then end the game before the Province strategy really kicks into gear. So this question really boils down to "how much time do I have before the game ends?"

The same is not necessarily true of alt-VP strategies. If I build a deck that makes Vineyards, Fairgrounds, or Dukes worth as much or more than Provinces, then I can take my time buying them, because my opponent will take quite a while buying all the Provinces. And since these alt-VP cards are generally cheaper to buy than Provinces, my opponent is going to need more than half the Provinces to win. So this question boils down to "how many points will I need to win the game?" If the game is limited to standard VPs, you need roughly half the Provinces plus. In a game with Harem/Island/Monument/Bishop, half the Provinces will likely not be enough. If there is a plausible Vineyards/Fairgrounds/Duke strategy, you may need almost all of the Provinces. If there is a plausible Goons strategy, all the Provinces may not be enough. So this question is not about pace, but rather about the amount of VP that will be necessary for the win.

EDIT: Duke is really a hybrid strategy, since it generally requires you to rush the Duchies in order to make the Duke's value comparable to Provinces.

[Catalytic]

I have made some modifications to the tree in light of comments by other users.  Thank you everyone for your input and I of course welcome further thoughts.

[DStu]

I wouldn't say that fast thinning is the only way or most important way to get engines going.

First, you of course can replace it by Warehouse/Cellar. Also some weak trasher (or none) can be enough, if you have some other conditions satisfied. These are: strong Attacks, (again) alternative VPs, or the engine being reliable against dillution.

You can really play Fishing Village/Torturer without caring on a trasher, because it includes a strong attack, and the duration effect of the Fishing Village stabilizes the chain. Wharfs or Schemes or Alchemists also stabilize.
Alternative VPs again for me are important in the decision of building an engine, because they, like attacks, give you the time to built up. Even if you don't get Fairgrounds or Gardens to 6, or 100 tokens from Goons, they often bring enough VPs to not getting nervous if your opponent buys their 3rd Province and you are still building your engine.

More important usually is the +buy, because you can neither build an engine fastly, nor get enough profit from it if their is no +buy, and there is not much that can replace it. Ironworks or Haggler can help building it but nor so much in scoring, but Horn of Plenty is maybe the only card that can really get you many VPs at the end. OK Outpost.
The other aspects maybe all just mix into a pool of:
Will the game last long enough for my engine and how strong is it once it is running: attacks/payload, alternative VPs, how strong are the Villages/Drawers/trashers/+buy, reliability, how strong is BM (to half of the VPs) on this board?

[Catalytic]

Thank you for the comments DStu.

I agree that this is the weakest part of the tree (and admittedly my own game).  Maybe the question needs to be split into a series of conditions to begin engine searching such as:

1. Are there cards to quickly amass cards (either through +Buy (Wharf, Worker's Village), gain on gain (Border Village) or gain on play (Smugglers, Haggler))?

2. Is there strong trashing (Chapel, Ambassador) or strong drawing (Warehouse, Lab)?

3. Are there attacks that will give you time to build an engine?

4. Are there +Actions?

And so on.  Maybe it is as simple as replacing "Is there fast trashing" with "Search for Engine Possibilities" and what that entails is consideration of these types of questions.  This would likely be a set of conditions (as above) to seek out Engine possibilities.  I will have to think about it.

Thank you kindly for your thoughts DStu!