I briefly talked about this in the last 2 streams I did, so maybe I should actually write about it. I don't think anyone's actually talked about it before,
and it's a great way for me to procrastinate on my essays.
To people who are new to the game, Dominion often looks like multiplayer solitaire, where everyone races to the best deck with little interaction. Of course, once you play for a while, you realize this isn't true. The right cards to buy and right cards to play can depend heavily on your opponent's deck, even in games with no attacks.
When you're playing Dominion, you're not playing a wall, you're playing a person. So, you shouldn't do anything that makes it easier for your opponent to win the game. This article focuses on a small aspect of this: minimizing information.
(As a foreword, this applies only to Dominion Online. Information in face to face Dominion is very different.)
Here's a simple example. Suppose you have 6 Coppers in hand. You would prefer buying Lab over Gold. What most people would do (and what I would do) is click the play all treasures button, and buy a Lab. If you do this, you're actually leaking information: your opponent now knows you value Lab more than Gold. You had the money for Gold, but bought Lab instead. Depending on your opponent, this might not help them. If they agree that Lab > Gold for you, they didn't learn anything new. But, if they disagree, you've given them a reason to reassess their beliefs. If Lab > Gold for you, is Lab > Gold for them? Maybe it is, and if they decide that's true, they're going to play a bit better going forward.
Here's a more extreme example. You draw a starting hand of 5 Platinums. You play your hand and buy a Colony. Now, your opponent knows all 5 of your Platinums collided, and you're much less likely to hit $11 in later hands this shuffle.In Dominion, every card you play, every card you discard, every card you gain, and every card you buy is public information.
This will all show up in the text log. If you buy Herbalist for $11, your opponent knows you've just bought Herbalist for $11.
That realization comes with a corollary: if you can hide information about your deck and/or strategy, you should do so
. Giving your opponent more information can only help them play better, so you should give off as little information as possible.
Now, this comes with limits. Take the 6 Copper example. If I planned to buy Lab over Gold, I wouldn't play 5 Coppers one by one, because I don't want to play treasures one by one until I hit exactly how much I need. For me, getting a faster game by always clicking "Play All Treasures" is worth whatever information I leak about myself. Even in the 5 Platinum case, I'd play all 5 Platinum. The only times I put information hiding into practice is when it's around as fast as "normal" play.
(This is the big reason this applies mostly to online play. In real life, playing 5 Coppers is no slower than playing 6 Coppers, and I'll hold back treasures with no hesitation.)
Here are the scenarios where I explicitly do things to hide information about my deck. I won't share logs, because this is something that's best visible in streams and videos. For what it's worth, a lot of high-level streamers do this already. It's something a lot of people teach themselves.
- A player has 1 buy, more than enough to buy a Province, and only needs to play 1 Treasure to hit $8. For example, they have $5 from actions, and their hand has 1 Gold and 2 Silvers. Instead of clicking "Play All Treasures", they should click just the Gold and buy Province. Same speed, but they reveal fewer cards from their hand. (Aside: in the Making Fun implementation, you can tell if people are doing this from the text log. Clicking just the Gold logs "plays a Gold". Clicking "Play All Treasures" with 1 Gold in hand logs "plays 1 Gold".)
- A player is building an engine. They've drawn their deck, have actions left over, and none of them have any benefit. (For example, they're all Villages, but there's no more cards to draw.) They should end the action phase and go to buys, because it hides that they have leftover actions. It takes longer to play all the extra actions anyways - why bother? Note this also means you should never play Ruined Village, unless you need it for Conspirator or Horn of Plenty. (Or you're trying to make it miss a reshuffle, but let's not get into edgecase land.)
The last scenario is big enough to deserve it's own section, and in fact is the reason why I wrote this in the first place. In many scenarios, you can hide information in the opening.
First, here's an example without a board. Suppose Ambassador is on the board, and these are the only openings on 3/4 that you think are reasonable.
Let's further say you think Amb/Silver is better than Amb/Amb, and you think your opponent is good enough to open Ambassador. Then on 3/4, the correct turn 1 buy is always Ambassador, not Silver.
If you buy Silver, and your opponent models you as someone who will buy an Ambassador, then they know you're going to open Amb/Silver. Whereas if you had bought Ambassador, your opponent still doesn't know if you plan for Amb/Amb or Amb/Silver. This is one of the rare cases where being player 2 is helpful - if you buy your Silver after your opponent commits to opening Amb/Amb, it's too late for them to change their mind.
If you have 3/4, and your opening is two cards costing $3 or less, you should think briefly about which card appears in more openings and buy that card first. This maximizes your opponent's uncertainty about you.
Here's another example: Chapel/Silver vs Chapel/Workshop. The first is better at hitting $5 early on, the second could gain more engine pieces. You should always buy Chapel turn 1, to hide how much you value an early $5. (If Witch is in the game, Chapel/Silver has a lot of merit.)
Finally, here's the most recent example I have, from a Base-only game.http://gokosalvager.com/static/logprettifier.html?http://dominion-game-logs.s3.amazonaws.com/game_logs/20160229/log.0.1456727556775.txt
Cellar, Chapel, Chancellor, Village, Workshop, Feast, Gardens, Moneylender, Smithy, Market
The reasonable openings are
- Chapel/Silver, with the plan to trash down and get some early economy (and possibly get an early +Buy source from Market)
- Chapel/Workshop, with the plan to trash down and gain Villages + Smithies before buying treasures
- Workshop/Workshop, with the plan for a Workshop/Gardens rush
I judged the engine as stronger than the rush, and Chapel/Workshop as the clear best opening. Accordingly, I bought Chapel turn 1 and Workshop turn 2.
But, this was actually the wrong opening order. Anyone who thinks the engine is best is going to open Chapel/Workshop, because they'll be a good enough player to immediately write off Chapel/Moneylender or Chapel/Silver as good openings. The correct turn 1 buy is Workshop, because this hides whether you're going for the engine or going for Gardens.
Finally, if you're patient enough, you can see information hiding in the linked logs. On my turn 8, I draw my deck and don't play Chapel because I want to keep all my treasures. On turn 10, I draw my deck, play Workshop gaining Smithy, played Village to draw it, then Chapelled away a Copper. This is a mistake. I had no reason to play Village, and should have played only Workshop and Chapel.
(Feel free to share any notable examples from your games. I'll stop here, I've procrastinated enough.)