1
Dominion General Discussion / The Dominion Cards Lists 2018 Edition: Landmarks
« on: February 24, 2019, 05:22:36 pm »The Most Impactful Landmarks
Only 13 people voted for the landmarks this year (compared to 18 voters last year). I'm guessing that's low enough that many of the ranking anomalies could be due to noise. ("Noise" includes, for example, someone voting Labyrinth as the best landmark, and someone voting Wall as second-to-worst. :swashbuckler: :parody:
)Obligatory comment about how "no one knows how to rank landmarks". Actually, I think it's the opposite: landmarks have only one metric that makes sense, whereas for cards, lots of different criteria are possible. For example, does how often you buy the card in optimal play matter? (This favors events like Save, Alms, Travelling Fair, and Borrow, and it favors King's Court over Donate.) Or is it just a matter of how strongly you should consider buying it at least once? What if it's a card you only buy when you're already winning or already losing (Salt the Earth, Duchy)? How should VP be ranked relative to non-VP? In contrast, for landmarks, the only metric that really makes sense is "how much does this change your strategy", or more concisely, "impact".
Assuming that impact is the metric, here are the results from least impactful to most impactful (countdown order).
 | #21 =0 Baths Weighted 2.79% / Unweighted 7.31% (21) / Median 0% (21) / Std. Dev. 21.09% (15) In the Weighted and Unweighted average scores, Baths is in its own bottom tier, very far below everything else. All except one voter put it last or second-to-last. It's not uncommon to pick up points from Baths, but usually it's when doing something you would have done anyway. Deliberately taking the points instead of building is comparable to discarding your hand against a Torturer engine rather than taking a couple of curses, which is to say you might as well resign. Baths has some combo potential, but it's limited in scope: compared to not-buying (e.g. Hermit and Mission), not-gaining is a very steep activation cost. So you need a strong Event or Project to buy, but which doesn't gain you any cards, and right now that's just Salt the Earth. However, as most games use only 0-2 landscapes, the chance of Baths and Salt appearing together is low. |
 | #20 ▼1 Aqueduct Weighted 29.72% / Unweighted 31.92% (18) / Median 25% (18, tie) / Std. Dev. 16.24% (19) The next five landmarks have a similar weighted score, so it's not too surprising that they jump around a bit. Aqueduct holds the distinction of being the worst landmark which isn't "put 12 VP here, take 2 VP when X". It's less impactful than Battlefield because in high-level 2-player play, Silver and Gold aren't gained in very large quantities. If each player gets maybe 1 silver and maybe a few Golds later, that doesn't amount to many points. Combos with Trader and Masterpiece, if you can set it up fast enough. |
 | #19 ▼2 Arena Weighted 30.07% / Unweighted 30.77% (20) / Median 25% (18, tie) / Std. Dev. 22.17% (13) In my experience, Arena points are usually depleted, since all you have to do is over-terminal a bit in the early game, or build enough to have villages or draw leftover in the middle game. Getting multiple terminals and risking collision is often good anyway, and now you get a 2VP consolation prize if you draw poorly. Building to have villages and draw leftover is also often good anyway. Free VP -- but not very impactful. |
 | #18 ▲2 Basilica Weighted 30.56% / Unweighted 31.54% (19) / Median 35% (14, tie) / Std. Dev. 18.64% (17) Basilica hasn't risen significantly this year. Occasionally it incentivizes getting cheaper cards, like Pixie, Druid, or Market Square. If you aren't mirrored and you can drain most of the 12VP quickly, the point swing could matter. |
 | #17 ▼1 Labyrinth Weighted 32.76% / Unweighted 35.77% (16) / Median 25% (18, tie) / Std. Dev. 30.94% (2) The most amusing thing about these landmarks is watching top-level players right-click them for the reminder. Like Basilica, Labyrinth incentivizes buying multiple cheaper cards, except it can't stack. On the other hand, Labyrinth also incentivizes gainers. |
 | #16 ▼1 Obelisk Weighted 33.38% / Unweighted 34.23% (17) / Median 40% (13) / Std. Dev. 22.6% (11) Obelisk is the last of this group of 5 with a similar weighted average score. It fell 5 places last year and continues to drop this year. Although tacking 2VP on a card can be a real bonus (see Nobles), this generally requires the card to be spammable. Nobles is, but something like Militia or Noble Brigand isn't. And if the spammable card is also strong (the only village, or Grand Market), you were probably going to spam it anyway. Still, Obelisk is 20VP, which is more than any previous landmark on this list. |
 | #15 ▼1 Palace Weighted 38.14% / Unweighted 41.15% (12, tie) / Median 45% (12) / Std. Dev. 24.43% (8 ) Damn it, I already trashed my coppers, and now I have to buy 3 stop cards for each 3VP? Effectively-unbounded VP is something, but in practice it's difficult to make Palace scale to more than 2-3 sets. Palace is also lacking insane combos, as far as I know. You need to gain lots of Coppers, Silvers, and Golds all at once. Treasure Trove comes close. |
 | #14 ▼2 Defiled Shrine Weighted 38.5 / Unweighted 41.15 (12, tie) / Median 30% (17) / Std. Dev. 15.08% (20) A lot more points than Aqueduct, and they almost always eventually accumulate to unignorable levels. Worse players often buy the Curse too early, but once it gets to a Province or better (for 0), it can be a scary point swing. Plus, you can trash the Curse and keep the VP. |
 | #13 =0 Battlefield Weighted 39.93% / Unweighted 39.62% (14) / Median 35% (14, tie) / Std. Dev. 18.34% (18) This will make my Ironworks-gain-estate rush strategy work again, right? Probably insane with Lurker+Hunting Grounds; I hope I never see that game. |
 | #12 ▲6 Colonnade Weighted 40.78% / Unweighted 37.31% (15) / Median 35% (14, tie) / Std. Dev. 27.01% (5) Our first huge winner this year. Rising 6 places, Colonnade outscored all the other 12VP pile landmarks. I'm not sure what to make of this. Have I been sleeping on Colonnade? Perhaps the rise is due to favoring the engine; you don't get many Colonnade points if you mainly want Silvers and Golds with a few actions. Or maybe it's due to favoring the first player to hit a large price point with multiple buys. |
 | #11 =0 Fountain Weighted 41.2% / Unweighted 49.23% (10) / Median 55% (9, tie) / Std. Dev. 31.55% (1) Although Fountain keeps its place in the rankings, it remains as controversial as ever with the highest standard deviation of any landmark. It had several votes near the bottom, as well as a vote for best, a vote for second-best, and several votes in the upper half (5th to 9th place). Fountain also has the highest absolute difference between its weighted and unweighted scores (8.03%). It's fair to conclude that worse players are significantly overvaluing it. The 15VP is tempting, but a strong engine still wins against some sad strategy which buys coppers. Don't skip strong copper trashing for Fountain points. More often, you just want to build enough +Buy to pick up coppers near the end of the game. |
 | #10 ▼3 Tower Weighted 52.64% / Unweighted 48.46% (11) / Median 55% (9, tie) / Std. Dev. 23.07% (10) A big jump in weighted average from the last group of landmarks, Tower falls from last year but remains far above Obelisk. Superficially similar to Obelisk, it plays differently: rather than picking up one card for the points, you look for a pile-out with a free VP lead at the end. |
 | #9 ▼1 Triumphal Arch Weighted 59.52 / Unweighted 54.23 (9) / Median 50% (11) / Std. Dev. 23.85% (9) For some reason, I always have to remind myself what this does. Triumphal Arch maxes out at a lot of points (30 VP). In practice it's more likely to get 12VP or 18VP. I like looking at the board and planning which two cards I can go for; it's not always obvious in advance. Like Tower, it can easily enable a pile-out win. |
 | #8 ▼2 Orchard Weighted 65.46% / Unweighted 63.85% (7) / Median 65% (8 ) / Std. Dev. 20.21% (16) The theme of many high-end landmarks it that they allow a ton of alt VP for the player who builds longer. You only get points for actions; well that's great, because that's how you build. You also only get points if you diversify your deck; well that's also great, because you need to do that anyway to avoid piling, and now you're playing with more of the board. As far as gameplay goes, I like Orchard (and Museum, later) because they make more cards on the board relevant. Watch out for split piles. |
 | #7 ▼3 Wolf Den Weighted 65.71 / Unweighted 61.54 (8 ) / Median 75% (3, tie) / Std. Dev. 30.78% (3) I don't understand the theme of this landmark. Wolves will eat you unless your workers come in groups of at least two?? Wolf Den falls somewhat from last year; perhaps people have figured out how to manage the bookkeeping and play around it. Step 1, build your engine from not-way-too-many components; step 2, buy an extra copy of one-of components at some point when you would gain a Duchy, or before that if they're good for your deck anyway. Works out very nicely with Remodel variants and other trash-for-benefit. Serious counter-synergy occasionally, like with Black Market. |
 | #6 ▼1 Tomb Weighted 65.97% / Unweighted 68.46% (5) / Median 70% (6, tie) / Std. Dev. 24.45% (7) This landmark sucks. Getting points for trashing? Who cares, you were going to trash anyway if possible, and if the trashing is weak (or nonexistent), Tomb probably won't make it good. Tomb + Trade Route is not a combo. Given the rankings, what I said in the above paragraph must be wrong. I'll try again. Tomb is unbounded VP, and it favors the player who builds longer. Remodel variants can build up a lot of Tomb VP without requiring extra gains. Gaining copper and trashing it to Forager or Chapel also works well. Unlike unbounded VP from something like Palace, you don't have to add a bunch of junk to your deck to get it. |
 | #5 ▲5 Bandit Fort Weighted 66.29 / Unweighted 66.15 (6) / Median 70% (6, tie) / Std. Dev. 30.01% (4) You heard it here first: Silver and Gold are good cards again. In a surprising reversal, Bandit Fort regains the 5 ranks it lost last year. High standard deviation, though. |
 | #4 ▲5 Mountain Pass Weighted 69.72 / Unweighted 71.15 (4) / Median 75% (3, tie) / Std. Dev. 21.68% (14) The second huge winner on the upper half of the list. Mountain Pass feels less like a Dominion card and more like an exercise in game theory. Which is disorienting, but a very unique effect and a welcome strategic challenge. "Strategic challenge" is an understatement. There way too many numbers between 1 and 40. Normally in Dominion you just click to play your cards and then buy whatever you think is best for the amount of money you have; now you have to think about your exact chances of winning. Overall, Mountain Pass probably deserves its high spot on this list. |
 | #3 =0 Museum Weighted 72.84% / Unweighted 72.69% (3) / Median 75% (3, tie) / Std. Dev. 12.19% (21) Museum has the lowest standard deviation and comes in the same rank as last year. Usually, Museum is just tons of extra VP; in contrast to Triumphal Arch or Orchard, you don't have to buy multiple cards to get it. It's almost always worth diversifying for Museum points. |
 | #2 ▼1 Keep Weighted 86.88% / Unweighted 82.31% (1) / Median 95% (1) / Std. Dev. 22.5% (12) Keep has by far the most first-place votes (5), but loses its spot at the top in the average. My personal pick for most impactful landmark, Keep is like Duchy-dancing on steroids. Sadly there is no nice equilibrium where you both get the points and move on; treasure piles don't usually come close to running out. So you just keep fighting back and forth. Sometimes if you pick up a lead early enough (e.g., a few Silvers), your opponent will give up on that particular treasure and only fight for the other ones. Usually, though, it's very careful bookkeeping of how many treasures you each have of each type. Keep loves cards that can gain treasures easily, like Artisan, Bandit, Soothsayer, Hero, Tragic Hero, Treasure Trove, Charm, and Haggler. More often than not the point swing in the last turn is so large that it decides the game. Also, thank god for the VP counter. |
 | #1 ▲1 Wall Weighted 87.15% / Unweighted 81.15% (2) / Median 90% (2) / Std. Dev. 24.66% (6) Finally, we have Wall in first place, regaining its spot from the 2016 rankings. It has only 2 first-place votes, and the raw weighted score is not significantly different from Keep. However, there is no doubt that Wall has a serious impact on every board. Wall is the definition of game-warping. It feels like your deck is working against you, and everything you learned about Dominion is wrong. The negative points are often just enough that if you build more, and try to catch up on VP later, you just can't quite close the lead. Strategically, I usually think of Wall as "-1 VP for every card in your deck". Other versions of it were tested (see secret history). But I doubt it would be that different if the first 15 cards mattered as well. |
Summary: Landmarks Tiers
Since there were only 13 votes, it makes sense to also present the landmarks in tiers, by weighted score. Landmarks which are in the same tier had similar weighted scores; the cutoff points are chosen such that there is a significant weighted score gap between one tier and the next.
The tiers range from S (completely game-warping) to F (completely ignorable), with A, B, C, D, and E in-between.
Tier S (86-88%): Wall, Keep
Tier A (65-73%): Museum, Mountain Pass, Bandit Fort, Tomb, Wolf Den, Orchard
Tier B (59-60%): Triumphal Arch
Tier C (52-53%): Tower
Tier D (38-42%): Fountain, Colonnade, Battlefield, Defiled Shrine, Palace
Tier E (29-34%): Obelisk, Labyrinth, Basilica, Arena, Aqueduct
Tier F (2-3%): Baths