Dominion Strategy Forum

Please login or register.

Login with username, password and session length

Show Posts

This section allows you to view all posts made by this member. Note that you can only see posts made in areas you currently have access to.

Messages - Titandrake

Filter to certain boards:

Pages: [1] 2 3 ... 81
Variants and Fan Cards / Re: Offering, the Heirloom that time forgot
« on: April 05, 2018, 03:20:45 am »
Timmy plays games for the experience. They favor big, flashy effects.

Johnny plays games as an outlet for creativity. They favor weird cards that can be used in innovative ways.

Spike plays games for the competition. They favor powerful cards that let them outplay their opponent.

Hunting Grounds is a Timmy card because it draws a ton of cards.
Rats is a Johnny card because it reads weirdly, but can be very powerful with the right enabler.
Margrave is a Spike card, because it's very powerful, but doesn't completely simplify the game.

Players can be a mix of archetypes and cards can appeal to multiple archetypes. For example, Margrave could also be seen as a Timmy card.

Variants and Fan Cards / Re: Offering, the Heirloom that time forgot
« on: April 03, 2018, 07:42:06 pm »
Black Market is swingy, but it ends up being a high-skill card because of all the choices it introduces - when do you buy your first Black Market, when do you play your Black Market, and do you choose to buy any of the cards.

Offering also looks swingy. I don't think it's as swingy as Black Market or Swindler though.

It looks cool, my main criticism is that there's basically no choice involved. You add the top Boon to your Offering, and then picking the order for the Boons you set aside should usually be straightforward. Have you tried revealing the Boon and having the set-aside be optional? It'll add some action paralysis, but it could introduce some neat play patterns.

Dominion Articles / Re: Castles
« on: March 30, 2018, 04:21:31 am »
I made some edits.

Re the 1-3 turns thing: after thinking about it I agree that you usually start greening a bit later just because there's more VP, so I made that clearer in the heuristic.

Re Small Castle: I also find that I usually don't keep my Small Castle all the way to King, but I do find it useful to trash it in quickly to gain one of the more expensive Castles. In more engine-like games I've also found that having Small Castle in your deck helps exert a lot of pressure over the Castle pile.

I'm pretty sure in a BM game, mostly-Castles-and-some-Provinces beats Provinces-first-and-Castle-sniping-late. This should be reasonably simulatable because it's Big Money, I'd be curious to see the numbers.

Re Humble Castle: I mostly find the card very confusing, because the card text suggests that you want to pick it up early, but the payoff is so small that it's rarely worth buying early. I wasn't sure how to explain that, so I wrote something completely different instead. Changing it to say what I wanted to say.

Agreed that Haunted Castle is a pretty good VP card, added a note for this.

Dominion Articles / Castles
« on: March 28, 2018, 02:40:07 am »

Despite having lots of different effects, Castles aren't that complicated. Just think of them as having a 2nd pile of Provinces in the game. If it's the kind of game where that's appealing to you, then imagine when you'd start greening in a game with 2 Provinces, and start buying Castles about 1 shuffle or 1-3 turns before that.

If you do the math, buying 8 Castles gives you 45 VP, plus whatever VP you gained from buying Grand Castle. 8 Provinces gives 48 VP. In principle, this shows that buying only Castles in competitive with buying only Provinces. In practice, no one will ever let you buy all the Castles uncontested. At minimum, expect competition on the $8+ Castles. Sprawling and Grand are almost always worth more VP than a Province, and players will buy King just to deny VP from the player who has more Castles.

Having more VP in the kingdom leads to all the follow-on effects you'd expect. More VP makes the game go longer, and longer games favors building your deck more. Many of the Castles help with this - all the Castles between $4 and $7 give you more than just VP. The two Action castles, Small Castle and Opulent Castle, are particularly important. Small Castle can quickly upgrade itself into a better Castle, whereas Opulent is important because it can produce a lot of money when players start greening. This makes buying either Crumbling Castle or Haunted Castle a bit of a risk, because it gives your opponent the first chance at buying Small Castle and Opulent Castle respectively. Of course, how big a risk this depends on how long you think the game is going to go. The sooner the game end, the less the Action castles matter. Buying Haunted Castle also makes it less likely your opponent hits $7. I often find that the risk of revealing Opulent Castle is worth the 2 VP + Gold + topdeck attack that Haunted Castle gives.

As for Humble Castle, it's a little weird. It gives VP for every Castle, and there's only 1 Humble Castle, which can make it look like it's worth fighting over. However, it's still a genuinely bad card to buy. At the end of the game, Humble Castle is usually worth 3 VP to 5 VP, and although that's above-average rate for $3, buying it early on is a huge opportunity cost. There are two cases where it's worth buying Humble Castle early. One is Shelters games, to turn your starting Hovel into a Copper. The other is Keep. In a Keep game, Humble Castle is always worth at least 6 VP. Even in this scenario, I don't like opening Humble Castle. 6 VP for $3 is very cheap, but opening buys are very, very important. On most boards I'd gladly give my opponent 6 VP if they were forced to open Copper.

Besides these considerations, Castles are mostly a tactical decision, not a strategical one. They aren't like Gardens, or Duke, or Fairgrounds, where you have to plan a bit to get VP out of them. Castles are just there, and whether the VP is worth it or not depends on where you think your deck and the game are at.

Other Games / Re: Celeste
« on: March 14, 2018, 02:01:45 am »
Celeste is excellent. I didn't try it right away because it's been compared to Super Meat Boy and I didn't really like that, but turns out I like Celeste anyway. Not sure what's the difference. (Maybe it's mostly that I don't like the SMB aesthetic.)

I haven't played Celeste yet, but the distinction I've heard is that Super Meat Boy is a precision platformer and Celeste is a momentum platformer. Precision platformers are about hitting very small jump windows, but your movement is very consistent and it depends very little on how you reach the jump point. Momentum platformers give you larger timing windows, but in exchange your movement state matters a lot more, and the game is more about maintaining speed and understanding the movement system.

Dustforce is a momentum platformer and so far everyone in the Dustforce Discord recommends Celeste.

Other Games / Re: In defense of Monopoly
« on: March 07, 2018, 05:10:41 pm »
Someone rename this thread:  "The Necro Wars".  Then everyone immediately shut up and let the games begin.  See you in 2 years.

I shall reclaim my title.

*marks March 7th 2018 on Google Calendar*

pacovf, you joked about doing it.

I actually did.

Dominion Articles / Re: What's stopping AI from mastering Dominion?
« on: March 02, 2018, 10:16:35 pm »
What is a realistic target skill level to shoot for, given the resources available?

My understanding is that the current Dominion AI projects are from a group of grad students working on it in their spare time.

These predictions have a habit of being very wrong, but I think a group of grad students working in their spare time have a solid chance of hitting Base-only level 50 pretty quickly, assuming that money strategies are enough to get there. If they work on it for months, they have a reasonable (~50%) chance of hitting level 55, which is about the point where I'd expect you'd need to learn non-trivial engines (the kind that only beat Smithy-BM by 1-2 turns.)

Provincial was able to get pretty far with genetic algorithms to learn buy rules + hardcoded play rules, and that included Intrique and Seaside, so I do think it's definitely possible.

Does Dominion change too much and too rapidly for an AI to have time to develop? Or rather, could a neural network be built to be able to learn completely new mechanics that hadn't yet been programmed into it? My hunch is not generally, depending on how different the new mechanics are. There's 2 extremes: (1) "$4 | Action | This is just like Village but with some vanilla add-on", (2) "$4 | Action | Play a game of Go with your opponent; if you won, +$3". If needing to handle such drastically new mechanics requires a complete overhaul of the AI, then the problem is more than there not being enough motive for people now to build an AI; there is also the problem of having to update the AI every 12 months or so, which may or may not require a complete start-from-scratch overhaul, and may or may not require more than 12 months. Chess, Texas Hold Em, etc. have the luxury of a static ruleset.

This is another argument in favor of sticking to Base-only for a proof of concept. In principle, it is possible that a bot can get good at evaluating new cards. I believe Provincial showed this - the author of that bot implemented a custom card, and showed the genetic algorithm learned how to incorporate it into its strategy.

It just depends on how close it is to existing cards. You see a similar thing in card evaluations. In Cornucopia previews, people thought Jester was going to be nuts, but it turned out it was just good, not insane. So there was a lot of error there, but Jester was doing something very different from the things before it. In contrast, everyone knew Junk Dealer was going to be a good card, because they had experience with Upgrade, and Junk Dealer is pretty close to Upgrade effect-wise.

Dominion Articles / Re: What's stopping AI from mastering Dominion?
« on: March 02, 2018, 04:27:29 am »
Okay, so I've seen a few threads about Dominion AI, which I have tried to ignore, but I don't think I can anymore.

I'm doing machine learning stuff right now, and in the past have messed around with game AIs. If you want you can read some blog posts I've written about the subject, here and here and here and here. None of those posts will be necessary, I'm linking them just to prove I have thought about this stuff before.

This is a subject where I can probably rant for a long time, and I'm not looking to rant, so my hot takes are

1. CouncilRoom has logged all games from Isotropic days, if you're interested I'm sure you could asked the current runner of CouncilRoom for the database.

2. I'm not that concerned by the number of starting states. It makes your problem harder, but is not a deal-breaker by itself. Chess / Checkers / Go only have 1 starting state, but it quickly branches into several different situations. If you take a start-of-the-art and throw it in the middle of a random Chess / Checkers / Go game, we'd expect it to perform well. A good AI should generalize to different game states - a good Dominion AI should do the same. That being said, it is definitely easier if you stick to a fixed board, but it isn't as interesting of a problem.

If anything I'm concerned more about the number of unique things cards do. Stuff like Young Witch or Landmarks or Events, where their mere existence drastically changes the landscape, and each one is super different. Game difficulty is guided by a mix of branching factor and ability to generalize across states.

3. Randomness is definitely an issue, but this doesn't have any relation to whether a Dominion bot can calculate an exact win chance. To be pedantic, it can, it's just not guaranteed to be accurate or to have low-variance estimates. But sometimes that's fine, as long as the right move is given higher value than all the wrong moves. Although this does affect search trees, I think credit assignment is a more important problem - it's a lot harder to attribute winning / losing moves when the game is random.

4. I'm not sure you need to learn all the small choices. I could see some handcoded heuristics taking you very far for that.

5. Basically I think 5 is the same as point 2.

AFAIK, the project Dan is on is focusing on Base-only. To me, that seems doable but also non-trivial. Superhuman performance in full Dominion sounds really, really hard. I think difficulty-wise, full Dominion is harder than multiplayer Texas Hold'Em, because there's so much fiddly stuff / uniqueness in all the different cards. I don't think this is a "12 ML researchers for 1 year" project. I think full Dominion is more like a 30-50 ML researchers project, and even with that many people, they only have about a 20% chance of doing it in a year.

Edit: realistically, the biggest thing stopping it is that there aren't enough people interested. The Computer Go community wasn't that big, but was around for long enough to make several very fast Go engines (~thousands of games/sec on a single CPU thread running highly optimized C) + hold Computer Go tournaments + get servers to support letting bots play against humans. The Dominion community is a lot newer and not as many people care about the problem.

Game Reports / Re: Dear My Opponent: I am Sorry
« on: February 24, 2018, 03:57:49 am »
In a game against Cave-o-sapien, with Scrying Pool, Swindler, no way to play more than 1 terminal a turn, and no +Buy.

Step 1: Start building.
Step 2: Hit Potion with Swindler, turn it into Death Cart.
Step 3: Start buying Pools while Cave doesn't have a Potion, using Pool's attack to force a Copper topdeck so that I can guarantee Swindling into Curse.
Step 4: Win Pool split 7-3, then use my Pools to hunt for Cave's remaining Pools to Swindle them into nothing.

I ended the game with 6 Pools. Cave ended with 1 Pool. Oh, and I ended up turning 5 of Cave's Coppers into Curses.

It was not pretty.

Dominion General Discussion / Re: Semi-Interesting Dominion Moments Thread
« on: February 23, 2018, 12:44:00 am »
In a game against Lord Rat, I Inherited Page, then messed around with Goons and Groundskeeper.

On my final turn, I played a bunch of GKs and 3-piled by emptying the Estates. Then I auto-played the Traveller upgrades, which accidentally returned all Estates I'd played that turn to the pile. Had to play 2 more turns to actually end the game.

If I wanted to, I could have gotten infinite VP off Groundskeeper and Goons by buying and returning Estates indefinitely.

General Discussion / Re: Random Stuff Part IV
« on: February 20, 2018, 03:56:27 am »
Them's Fightin' Herds is coming out in a few days.

Dominion General Discussion / Re: Top 20 kingdom cards
« on: February 19, 2018, 09:29:55 pm »
Do you mean top 20 as favorite or top 20 as most powerful?

Dominion Articles / Re: Band of Misfits and Overlord
« on: February 19, 2018, 07:28:26 pm »
Nice article!

I do wonder what the difference between your two bold highlighted phrases are? If the second and third paragraphs were deleted, I think the article would read just fine. That would leave you a little room to take Chris's suggestion:

I would devote a whole paragraph to the mechanic of an empty pile being uncopyable. Itís the most important drawback of the card and ultimately it determines if itís worth going for, so significant time should be spent figuring out what piles it will be able to emulate at which phases of the game, strategic pile emptying, etc.

and maybe talk a little bit about Overlord in the opening or clarifying why Overlord is such a power card when BoM is often just "meh".

Maybe turn the "final points"/rules clarifications into bullets?

They mean the same thing.

Second and third paragraph = defining what contextually powerful means. 4th paragraph = explaining how this relates to BoM / Overlord. I agree there's a lot of overlap but I'm not sure you can simply remove those paragraphs. Also think some repetitiveness is okay, when it's repeating the core argument for how you should view the "play as another Action" cards.

Dominion Articles / Re: Overdrawing
« on: February 18, 2018, 02:57:41 am »

Game Reports / Re: Dear My Opponent: I am Sorry
« on: February 18, 2018, 01:57:08 am »
Game #11777769 against stoogebag.

I did Scrying Pool for draw + Lost Arts on Black Market, and oh my god did my turns take forever, because I was continually playing Black Market, evaluating whether I wanted the buys, and redrawing bought Black Market actions with Scrying Pool. And then sometimes the card I bought from Black Market was itself gaining cards that I wanted to redraw, and that isn't even getting into me gaining new Black Markets midturn, and it was just a complete mess.

Dominion General Discussion / If you could only have 1 opening
« on: February 14, 2018, 03:47:06 am »
Context: there was a recent Reddit thread about the balance of the opening hands, and some discussion on the subject in the markusin vs Stef World Cup match, where Stef had to open Amb/-- on a 5/2 split.

If you had to pick between only 5/2 or only 4/3 for all future games, which would you rather have?

To address pedantic points:
* The order is still random, i.e if you pick 5/2 it's still random whether you get $5 on turn 1 or turn 2.
* If Heirlooms are in the game, they replace a random Copper.

To the surprise of very few, it is possible to string together English words into sentences that form a coherent argument for an incorrect point. Gotta admit that I don't see why 26 examples were necessary.

Dominion Articles / Re: Witch, revised for 2018
« on: February 10, 2018, 03:54:05 pm »
"But Mountebank is in the kingdom!" - Actually, after you play Mountebanks a few times, they block it often enough that you probably want to buy the Witch to hand out the rest of the Curses.
Just get a second Mountebank. Or (n+1)th. Playing two Mountebanks in one turn is worse for your opponents than one Mountebank and one Witch.

There is the old saying that nobody ever got fired for buying IBM. I now propose: nobody ever lost Dominion by playing King's Court on a Mountebank.

Sure, but that's assuming your in a game where you can expect to play two Mountebanks a turn. If you only play 1, there's a tipping point where Mountebank gets blocked too often. Witch drawing cards can also make it a bit easier to draw your other Actions, although this does get balanced out by the Coppers you're gaining.

I do think you get Mountebank first, and that your deck is probably going to end with N Mountebanks and 0 or 1 Witch. I don't think it's obvious that it should always be 0.

Dominion Articles / Witch, revised for 2018
« on: February 10, 2018, 03:47:41 am »
It's still good.

You should still buy it.

If you don't, you should still have a good reason, and I mean a REALLY good reason.

"But there's trashing!" - No, every Curse they trash is a Copper or Estate they didn't get to trash.

"But the trasher competes at the $5 price point!" - You may want to buy the trasher first, but if you do, you should probably still buy the Witch later.

"But Mountebank is in the kingdom!" - Actually, after you play Mountebanks a few times, they block it often enough that you probably want to buy the Witch to hand out the rest of the Curses.

"But Moat is in the kingdom!" - Having a Witch still threatens handing out Curses. They only have to fail to have Moat once to get punished for it. And if they buy a lot of Moats, then you've hurt them in a different way: they spent a bunch of Buys on Moats.

"But Lighthouse is in the kingdom!" - Okay. The Moat principle still applies - they only have to miss it once to get hurt. However, it's a lot more okay to have lots of Lighthouses, and in some games it's even beneficial (i.e. Minion games). So sometimes, they do have enough Lighthouses to make buying Witch redundant. But if you notice they've gotten complacent and are starting to miss Lighthouse turns, maybe you should reconsider that Witch...

"But Ambassador or Masquerade is in the kingdom!" - These do counter Witch the most regularly. It sucks when they give you back a Curse.

"But I'm drawing my deck, and I have a trasher, and I'm out of things I want to trash with it! I'll just trash the Curse if I get Witched!" - Sure, this is true for now. It's easy for this to stop being true, as you add payload or VP to your deck. Much like the Moat argument, if they're playing Witch every turn, you only have to fail to draw your trasher once to fall behind on having a thin deck. Once you fall behind the first time, it's more likely you fall behind the next, and soon your thin deck has 3 Curses in it. It's a classic recipe for beating people who are overconfident in their deck's reliability.

A lot of new players don't like Witch, because it slows down the game. Most people's idea of fun is getting to play a bunch of cards and produce a bunch of money. Witch gets in the way of that. But I like Witch. When I give out Curses, I think to myself "yes". When I get Curses, I think to myself "no". Even if you don't like it, you have to respect it.

Sometimes, you can skip Witch, and you get to be a Hero. Maybe even a Champion. Just be careful out there. Curses make the world a scary place.

Tournament + Grand Market + Ball

Get to Prizes first, gain Princess first.
Play Princess.
Pay $5 to Ball for 2 Grand Markets (don't even need to avoid having Coppers.)
No other cost reduction is in the Kingdom.
On my next turn, I am about to Ball for 4 Grand Markets.
Opponent resigns before I get to do so.

Dominion General Discussion / Re: help! i'm horrible at lurker
« on: February 02, 2018, 11:48:45 pm »
Play 100 (or more) games with Lurker in the kingdom. Try using it in different ways (no Lurker, double/single Lurker opening, odd or even amounts etc.) to see what is best when. Observe how other people use it. I remember it taking me a very long time to work through and develop a better understanding of Scrying Pool and Apothecary(first online expansion).

100 games is a lot of games...I only have about 500 games lifetime on the Shuffle IT client.

One thing good to know about Lurker is that it's a form of indirect interaction between players (i.e. this was part of the intended design if you look at the Secret History for the card.) If your opponent buys Lurker, there's a good chance you need to buy Lurker too, but you may not need to buy as Lurkers as they do. That doesn't mean it's overpowered, or that you always need to buy Lurker first, but it does mean that its power level is really, really dependent on what the current game is like (and not just based on the other cards in the Kingdom.)

Dominion General Discussion / Re: Forced Wins Training Sessions
« on: February 02, 2018, 11:10:48 pm »
Here's what I did in the original game.

Down by 27 VP. Two Provinces is 12 VP. 15 VP from Keep + any extra VP is enough for the win.

Currently have 0 Silvers to 1 Silver, and 2 Treasure Troves to 2 Treasure Troves. I didn't bother tracking the Copper or Gold splits, because Treasure Trove really makes those confusing. The first 2 Silver gains are each worth 5 VP (first gains me 5 VP, 2nd makes them lose 5 VP). The first Treasure Trove gain is worth 5 VP. If I gain 2x Silver and 1x Treasure Trove, that's the 15 Keep VP I need.

I first need to get more gains --> Stonemason WW into 2 Stonemason.
Graverobber gains Silver.
Play City Quarter, draw rest of deck.
Stonemason Gold --> Treasure Trove + Silver.
That's all the treasures I need, and after playing Treasure Trove I'm going to have lots more Gold so I don't need to worry about Gold count --> Stonemason Gold --> 2x Duchy to get the extra VP needed to win.
City Quarter draws gained Treasures.

I will play 3 Treasure Troves this turn, which will gain 3 Copper. So I can trash up to 3 Copper without losing VP from Keep from the Copper split.
3x Forager trash 3 Copper.
4th Forager trashes 5th Forager.

That leaves plenty of money to buy the last 2 Provinces.

It's almost always just for fun, and to see what machine learning can do. Sometimes it's interesting to inspect the model after it's trained to see what it's learned to care about.

I think there was a recent paper that learned how to generate Amazon reviews with positive or negative sentiment - you can imagine how some people might find that useful...

Dominion General Discussion / Re: Forced Wins Training Sessions
« on: January 31, 2018, 02:39:07 am »
Here's a neat one from a match I played 11 days ago.

There's actually a lot of leeway here, but you can guarantee a win even if you only produce enough money to buy 2 Provinces.

Your last 2 cards aren't necessary - assume they're blank.

Dominion Articles / Trash-for-Benefit and Stored Value
« on: January 27, 2018, 08:55:24 pm »
Stream of consciousness article, may clean up later.

I should start by defining trash-for-benefit.

A trash-for-benefit card is a card that trashes other cards, then gives you a benefit based on the cost of the trashed card or cards.

This excludes cards like Spice Merchant, Forager, and Trade Route. All of these cards trash cards, and they all give some benefit, but the benefit isn't tied to the cost of the card.

Why do I bother making this distinction? All trash-for-benefits that care about cost hit upon the same concept: the idea of stored value. I believe MicQ was the first to coin this term. It's a pretty simple principle: whenever you buy a card, you "store" the money spent that turn into the card you bought. You can later "cash-in" that value with a trash-for-benefit card.

Once you start thinking in terms of cost, you can derive several synergies.

Trash-for-benefit + on-gain cards: The cost of some cards is tied to their on-buy or on-gain effects. Border Village and Farmland are classic examples, and Skulk is a more recent example. Once you've gained the card, you've already gotten a lot of the value, so feeding those cards to a trash-for-benefit lets you cash-in more value than a typical card of that cost.

Trash-for-benefit + cards that get worse over time: Spice Merchant is a good opener, but gets worse as you run low on Coppers to trash. Witch is a good card, but once the Curses are gone, it's just a Moat. A trash-for-benefit card can let you get rid of the Spice Merchant or Witch after it's done most of what you wanted it to do.

One of my favorite synergies is Butcher + Spice Merchant. Normally, you don't buy that many Spice Merchants because they quickly run out of treasures to trash. Butcher lets you buy more Spice Merchants because you know you'll be able to cash in your extra Spice Merchants once your deck is thin.

Trash-for-benefit + cards that gain other cards: Magpie lets you gain a ton of Magpies, but from a trash-for-benefit perspective, it's more important that it gains you a bunch of $4 costs. Rats is similar - it converts junk cards in your deck into junk that costs $4. That isn't great, unless you have something that cares about the cost of cards, like a trash-for-benefit...and then suddenly it's ridiculous. Port gives you two $4 costs per buy.

Often, treasure gainers let you gain high-cost cards for less than their true cost. Think how Bandit (a $5 cost) gains you Gold (a $6 cost) every time you play. Treasure gainers can do this because treasures are usually worse than Actions, but this opens up neat lines of play if you can feed the treasure-gain to a trash-for-benefit.

A final note: it's usually not worth going too far out of your way for these trash-for-benefit synergies. One of the big newbie mistakes is to believe that you should always spend all your money each turn. This is still a mistake, even when trash-for-benefits are in play. The one exception is when you have a very consistent engine, and it's very important to have a card of a specific cost. The last example I can think of is an old game where Procession, Adventurer, and King's Court were in the Kingdom. On that board, I definitely Processed Adventurer a few times to gain King's Court. But that was very much the exception, and not the norm. More commonly, you add these trash-for-benefit synergies to an existing strategy, and let those synergies nudge you in the direction of buying extra Spice Merchants, or knowing you can buy something other than Gold because you're going to Remodel Witch into Gold. Not enough to sustain a deck by itself, but certainly enough to make an existing deck better.

Pages: [1] 2 3 ... 81

Page created in 0.148 seconds with 18 queries.