Dominion Strategy Forum
Dominion => Dominion General Discussion => Topic started by: faust on October 27, 2017, 06:08:06 am
-
With all the excitement over Nocturne previews, it's easy to forget our thrid most favorite passtime: Making
brackets lists.
In order to justifiably call the next cards lists a 2017 one, voting for it should at least start in 2017. There's not terribly much time left.
Nocturne release is currently scheduled for mid-November, starting the cards list before that seems like a bad idea.
Since Avist indicated he likely won't have the energy to compile and post all the lists in the nice exhaustive format we've had so far, I thought it would be a nice idea to turn this into a community effort. So, after polling, we would have 1 person responsible for the writeup of each list. That way, we could have all the results online quicker and don't risk burnout.
My suggestion would be to use all of December for voting and present the results starting January.
Notcurne also opens some questions on new things to be ranked. Some suggestions:
- rankings for Boons and Hexes. This is pretty obvious I assume. What's the best Boon to get? What's the worst Hex to dish out? I think it would offer nice insights into what small vanilla-ish bonuses are "worth".
- starting cards ranking. With Heirlooms and Shelters, there are now 12 different cards you could have in your starting deck. Would be interesting to see what's the strongest and what's weak. We'll finally be able to answer whether an Overgrown Estate is cooler than a reuglar one.
- non-supply card rankings. So far, we only have a list for Prizes that isn't terribly interesting. Nocturne offers a whole bunch of new non-supply cards. I could imagine doing something along those lines.
So what does the community think? Is there interest in some collaboration?
-
Some random thoughts:
- Would it be easier to not do the whole cards ranking? With so many cards that seems like it would take forever, maybe in the 4+ hour range. Splitting it up into just the separate costs might be easier for most people. I don't think the complete card list is very helpful anyway.
- We're definitely removing the 12 ditched cards from the list, right? I think it's pointless to include them.
- The cards will be up online as soon as the expansion gets released, so people will be playing a lot more games in the first weeks then they normally do. Starting ranking around Thanksgiving (and ending January/February) should give plenty of time for everyone to form an opinion.
- If we could post all the results in a shorter timespan that would be nice. I know Qvist got burned out last year, but even at full pace, it took weeks to get through one list. I would prefer it if we could have all the results out in the period of about one month. Your suggestion of different people doing different lists is a great idea, I wouldn't mind taking a smaller one myself.
- I'd like to keep the arrows that show the increase/decrease from last year, but the other things could probably be cut, especially for the smaller lists.
Also, here's a list of what lists I think we're doing:
- (http://wiki.dominionstrategy.com/images/thumb/5/5d/Coin0.png/16px-Coin0.png)-(http://wiki.dominionstrategy.com/images/thumb/3/3d/Coin2.png/16px-Coin2.png)
- (http://wiki.dominionstrategy.com/images/thumb/3/32/Coin3.png/16px-Coin3.png)
- (http://wiki.dominionstrategy.com/images/thumb/2/2a/Coin4.png/16px-Coin4.png)
- (http://wiki.dominionstrategy.com/images/thumb/7/7d/Coin5.png/16px-Coin5.png)
- (http://wiki.dominionstrategy.com/images/thumb/6/6f/Coin6.png/16px-Coin6.png)+
- Boons
- Hexes
- Knights
- Landmarks
- Events
- Ruins
- Card Art
Notably, I don't think it's worth having the "favourite cards" list. A starting card list would be interesting that had Copper, Estate, Shelters, and Heirlooms and you ranked which ones were best to start with.
-
Good idea overall.
Hexes and Boons definitely should get their list.
Ranking Starting cards is kinda funny, but in all honesty kinda wonky the way you put it. We all know which Shelter is #1. Ranking Heirlooms is funny, even though their position should be weighed in ranking their Kingdom cards.
Non-supply cards ranking is probably too messy to be any good.
We should make a list for Castles though. They were probably too hard to understand for last years ranking.
-
Good idea overall.
Hexes and Boons definitely should get their list.
Ranking Starting cards is kinda funny, but in all honesty kinda wonky the way you put it. We all know which Shelter is #1. Ranking Heirlooms is funny, even though their position should be weighed in ranking their Kingdom cards.
Non-supply cards ranking is probably too messy to be any good.
We should make a list for Castles though. They were probably too hard to understand for last years ranking.
I second all of this
-
Ranking Starting cards is kinda funny, but in all honesty kinda wonky the way you put it. We all know which Shelter is #1. Ranking Heirlooms is funny, even though their position should be weighed in ranking their Kingdom cards.
I mean ranking them all together. So you have a list of Copper, Estate, Necro, Hovel, Overgrown Estate, Goat, Cursed Gold, Haunted Mirror, Lucky Coin, Pasture, and the 2 unknown heirlooms and you'd rank them in the order you'd want them as starting cards. It seems no weirder than it is to rank the Landmarks.
-
I would personally argue against having Events as a separate list. It seems enough to do them at their respective price point.
-
Ranking Heirlooms is funny, even though their position should be weighed in ranking their Kingdom cards.
I disagree that the Heirlooms should be considered in ranking their connected cards. That's like weighing the position of Dark Ages cards by the fact that they might bring Shelters into play.
-
Yeah, I don't think a kingdom card should get more points just because it has a powerful Heirloom. You should take synergy/anti-synergy with the Heirloom into account when making your own rankings though, since it's guaranteed to be present every time.
-
Yeah, I don't think a kingdom card should get more points just because it has a powerful Heirloom. You should take synergy/anti-synergy with the Heirloom into account when making your own rankings though, since it's guaranteed to be present every time.
Right. Rate Shepherd knowing you'll have an extra Victory card in your deck to start with, don't rate it as "wow my Estates are worth 1 extra (http://wiki.dominionstrategy.com/images/thumb/9/92/VP.png/16px-VP.png)!"
-
Yeah, I don't think a kingdom card should get more points just because it has a powerful Heirloom. You should take synergy/anti-synergy with the Heirloom into account when making your own rankings though, since it's guaranteed to be present every time.
It absolutely should get more points for a powerful Heirloom. That’s supplied with the card every time, and a consequence of the card’s presence in the kingdom. The same way Baker is made better by giving you a free Coin token. I don’t think the model some people use of “how good is it to buy this card” as a proxy for total strength is really the full picture of a card’s strength anymore. I think “how much stronger is the average kingdom given the inclusion of this card” is a lot more powerful of a metric.
Look at Events. Events that you only buy once or never, for a marginal effect at the time, can have a huge impact on how the board is played, and we don’t say the event is “weak” just because it’s influence isn’t explicitly borne out in every game it’s present in.
-
“how much stronger is the average kingdom given the inclusion of this card” is a lot more powerful of a metric.
How does one evaluate the "strength" of a kingdom? What does that even mean?
-
It absolutely should get more points for a powerful Heirloom. That’s supplied with the card every time, and a consequence of the card’s presence in the kingdom. The same way Baker is made better by giving you a free Coin token. I don’t think the model some people use of “how good is it to buy this card” as a proxy for total strength is really the full picture of a card’s strength anymore. I think “how much stronger is the average kingdom given the inclusion of this card” is a lot more powerful of a metric.
But it's redundant to score the Heirlooms separately and also add their scores to their respective kingdom cards' scores.
-
“how much stronger is the average kingdom given the inclusion of this card” is a lot more powerful of a metric.
How does one evaluate the "strength" of a kingdom? What does that even mean?
This is as subjective and blurry of a question as “how does one evaluate the strength of a card”.
-
It absolutely should get more points for a powerful Heirloom. That’s supplied with the card every time, and a consequence of the card’s presence in the kingdom. The same way Baker is made better by giving you a free Coin token. I don’t think the model some people use of “how good is it to buy this card” as a proxy for total strength is really the full picture of a card’s strength anymore. I think “how much stronger is the average kingdom given the inclusion of this card” is a lot more powerful of a metric.
But it's redundant to score the Heirlooms separately and also add their scores to their respective kingdom cards' scores.
Scoring them separately and also adding them is totally fine, because the Heirloom list is gonna be separate from the Supply list (see: Prizes). Death Cart makes you gain two Ruins, and you don’t “not consider that” just because the Ruins are separately scored...
-
Death Cart makes you gain two Ruins, and you don’t “not consider that” just because the Ruins are separately scored...
Because that depends on you buying Death Cart.
-
So first a disclaimer: I never totally understood how the previous ranking system worked, but for me, it was a big turn off that it took sooooo long to rank all the cards. Is there any way it could be split into more manageable sections? I would be totally on board if it was, but frankly I find it hard to justify spending an unreal amount of time ranking the cards. Maybe instead of rating a bunch of cards against each other, people could just arrange them from least to most powerful.
I realize this would be a big change to the ranking system, I just wanted to throw this out there.
-
I would be totally on board if it was, but frankly I find it hard to justify spending an unreal amount of time ranking the cards. Maybe instead of rating a bunch of cards against each other, people could just arrange them from least to most powerful.
You could already do that in the past rankings. Rating the cards against each other was just an alternative (default) way to rank them, but it would still ultimately give you an order from least to most powerful. It was also possible to combine the different methods so that you would do most of the work in one mode and then do the fine tuning in the other.
-
Just an idea, do we need to rate every single card now in the game? What's wrong with like a top 25 or top 50 in each of the proposed categories? I know how much we like being completionists but the amount of work required to make these lists is immense, even for a team. So my suggestion is to cut the workload.
-
I strongly, strongly recommend whoever is making the list to try out MaxDiff sorting.
Basically, humans suck at ranking long lists of things. Is Cutpurse the 47th best card? Or the 48th?
So instead, we create a system where people are given randomly-selected sets of 4 cards, and asked to rank the best and worst out of those 4. It's easy, and ends up producing a ton of data for very little mental cost.
http://www.mdrginc.com/market-research-best-practies-can-improve-college-football-ranking-system/
-
I strongly, strongly recommend whoever is making the list to try out MaxDiff sorting.
Basically, humans suck at ranking long lists of things. Is Cutpurse the 47th best card? Or the 48th?
So instead, we create a system where people are given randomly-selected sets of 4 cards, and asked to rank the best and worst out of those 4. It's easy, and ends up producing a ton of data for very little mental cost.
http://www.mdrginc.com/market-research-best-practies-can-improve-college-football-ranking-system/
Unless I missed something on that link, lots of methodology details aren't given.
1) How many total teams were in the set? All ~130 teams in the FBS?
2) How many rounds of voting were there per participant?
2000 participants is a lot more than we would get here, I imagine (the 2015 list had ~100 participants?) and we have many more cards to rank than there are FBS teams.
With the aim of making the analysis more tractable, I think we should ask everyone what they think the top X cards are, then take the union of those cards as the set to rank.
-
The number of participants is not important. The question is how many times each person is willing to vote. Basically you show people, over and over again, a random set of 4 cards, asking them what is the best and what is the worst. After enough iterations you have some solid Bayesian data on how those cards are tiered.
-
We could definitely get some more people to vote by putting it on the blog, BGG, and maybe even on ShuffleIT's forum.
-
The number of participants is not important. The question is how many times each person is willing to vote. Basically you show people, over and over again, a random set of 4 cards, asking them what is the best and what is the worst. After enough iterations you have some solid Bayesian data on how those cards are tiered.
As I see it, there are two key considerations to the number of people:
1) Spreading out the work of ranking.
2) Obtaining data that is meaningful for your question.
I'd argue that N is important for both, but that doesn't mean it has to be large.
-
Yeah, I don't think a kingdom card should get more points just because it has a powerful Heirloom. You should take synergy/anti-synergy with the Heirloom into account when making your own rankings though, since it's guaranteed to be present every time.
It absolutely should get more points for a powerful Heirloom. That’s supplied with the card every time, and a consequence of the card’s presence in the kingdom. The same way Baker is made better by giving you a free Coin token. I don’t think the model some people use of “how good is it to buy this card” as a proxy for total strength is really the full picture of a card’s strength anymore. I think “how much stronger is the average kingdom given the inclusion of this card” is a lot more powerful of a metric.
Look at Events. Events that you only buy once or never, for a marginal effect at the time, can have a huge impact on how the board is played, and we don’t say the event is “weak” just because it’s influence isn’t explicitly borne out in every game it’s present in.
Do people actually take the extra coin token into account when evaluating the strength of Baker? That seems really odd to me. It has nothing to do with your decision whether to buy the card or not.
-
I strongly, strongly recommend whoever is making the list to try out MaxDiff sorting.
Basically, humans suck at ranking long lists of things. Is Cutpurse the 47th best card? Or the 48th?
So instead, we create a system where people are given randomly-selected sets of 4 cards, and asked to rank the best and worst out of those 4. It's easy, and ends up producing a ton of data for very little mental cost.
http://www.mdrginc.com/market-research-best-practies-can-improve-college-football-ranking-system/
Well that would require someone to implement such a thing. The current system has the distinct advantage that it already exists.
-
Just an idea, do we need to rate every single card now in the game? What's wrong with like a top 25 or top 50 in each of the proposed categories? I know how much we like being completionists but the amount of work required to make these lists is immense, even for a team. So my suggestion is to cut the workload.
I like seeing the bottom of the list as well. As a beginning player the complete lists helped me a lot. It was hard for me to believe that pirate ship was bad, and without the list, it would have taken me a while to learn.
Now that I'm not a beginner, it is still tons of fun to see a whole list.
-
Ranking the cards doesn't take nearly as long if you base it on your list from last year, and I think last year's lists may still be available.
I have no interest in ranking knights, landmarks, ruins, or art.
-
- Ranking Hexes and Boons will be fun I think
- Ranking Heirlooms doesn't make any sense as we didn't rank Shelters either. I mean they are not meant to be equal in strenght anyway. Goat is likely to be the strongest anyway.
- Ranking Non-Supply cards makes even less sense, given different cost and ways to get them. Like Madman is one of the strongest without doubt.
I like theory's idea, but I won't code that because I don't have the time. I have already 3 different methods to rank the cards. One of them should fit the needs for everyone. As there is still confusion, you don't need to do the compare method. You can also just assign numbers for strength. I will try to make that clear this time.
-
Oh and regarding presenting the results. Having different people present them is actually a neat idea. I am still against presenting them all at once because I like the speculations and the discussion that rise from that. But raising the parts to like 15 or 20 cards at one and having different people presenting each part sounds pretty good.
-
I like comparing 2 cards with each other. It might be bias, but I don't think I'll like comparing 4 cards at once.
-
Well wero, you can just choose the ranking/rating option you prefer, an additional wouldn't hurt.
Anyway, really like the idea now that I thought about it, to let different people present the results.
Feel free to apply if you are interested in the #cardlists channel in discord (preferred) or PM me here.
The more people apply, the less cards each person might have to present, I am aiming for about 15 per person.
-
I like Theory's idea. Simply put, there are so many cards now, we need a faster more efficient way to rank cards, especially if we are going to try and get a strong voting population.
-
I like Theory's idea. Simply put, there are so many cards now, we need a faster more efficient way to rank cards, especially if we are going to try and get a strong voting population.
Like I said, you can for now just assign numbers to each card and that is a decent and fast way to rank them, faster than any other method, just not as precise.
-
I like Theory's idea. Simply put, there are so many cards now, we need a faster more efficient way to rank cards, especially if we are going to try and get a strong voting population.
I like theory's idea a lot. It seems pretty fun and efficient, and doesn't require a huge commitment from any individual. I think it also requires a little more thoughtfulness than just 2 cards at a time.
-
I think ranking the Boons and Hexes would be interesting, even if it's pointless strategy-wise because they're random. We already rank the ruins even though those are random.
I don't see the point in ranking Heirlooms though, or even how you could do it. They're all at different price points, but you get them for free. Doesn't this mean the most expensive one is automatically the best? Though I admit that I'm curious about people's opinions on Hovel vs Overgrown Estate. Hovel has the decency to get rid of itself, but Overgrown Estate has the on-trash bonus and the Victory type...
-
I think ranking the Boons and Hexes would be interesting, even if it's pointless strategy-wise because they're random.
I don't think it's pointless strategy-wise for the Boons; it's relevant to when it's a good decision to buy Druid or trash Pixie.
-
+1 for just making a top-X list and bottom-X list for each pricepoint.
It comes back to what these lists are actually used for. When I first started looking at them, pretty much the only takeaways I remember having were 'What are the best cards that tend to dominate whatever Kingdoms they're in?', and 'What are the worst that should I should avoid unless I have a good reason to get them?' ... "Huh, there must really be something to this Ambassador card - guess I'll try it out." "Wait, having a Duchess is usually worse than having nothing at all?" Etc.
Do I really care if Cutpurse is better than Poacher or if Knights are better than Bridge Troll (whatever "better" even means in this context)? Like, how will that information in a vacuum ever be useful when evaluating a Kingdom with those 2 cards, especially to a beginner or intermediate player looking at these lists for the first time? EDIT: I realize you could make this argument about any 2 cards. But blindly buying Mountebank instead of Harvest every time you have $5 is not a terrible rule to follow - I would not be able to say the same about the middle-100-or-so cards.
Oh and it would also make the ranking easier.
-
We should only rank things we buy. That means cards and events.
Heirlooms, boons, hexes and landmarks are not things we buy, they either are just there in set up or are given randomly.
Events should be combined with cards of the same price point, since they compete directly with those cards for your buys.
-
I am okay with any ranking system, as long as it doesn't involve me doing thousands of comparisons. (You need about 2000 pairwise comparisons to sort 300 cards.) I think the person who actually writes the code is the one who gets to decide on the details. I don't want to make demands for other people to do more work than they're willing to do.
For stuff like events, landmarks, heirlooms, boons, etc., I am in favor of ranking them all. Honestly I don't think ranking heirlooms makes much sense, but the thing is, ranking 7 cards takes a tiny fraction of the time that it takes to rank the $5-costs. So if there's even a small fraction of people in the dominion community who would enjoy the ranking, why not do it? It's just so easy.
-
I strongly, strongly recommend whoever is making the list to try out MaxDiff sorting.
Basically, humans suck at ranking long lists of things. Is Cutpurse the 47th best card? Or the 48th?
So instead, we create a system where people are given randomly-selected sets of 4 cards, and asked to rank the best and worst out of those 4. It's easy, and ends up producing a ton of data for very little mental cost.
http://www.mdrginc.com/market-research-best-practies-can-improve-college-football-ranking-system/
Reminds me of this site for MTG:
http://www.draftaholicsanonymous.com/p1p1-ixalan/
-
The all cards list was always optional. You can just choose not to do it, as I’ve been doing. I don’t really have any interest in doing it with this many cards.
The other lists are more manageable as is.
I’d recommend Theory’s method for the all-cards list and the existing methods for the others, unless they are linked somehow and this isn’t possible.
-
The all cards list was always optional. You can just choose not to do it, as I’ve been doing. I don’t really have any interest in doing it with this many cards.
The other lists are more manageable as is.
I’d recommend Theory’s method for the all-cards list and the existing methods for the others, unless they are linked somehow and this isn’t possible.
For the all-cards list, I did that by first doing the rankings for all the individual costs, then popping from the top of those lists onto the next slot in the full list. You compare maybe 4-6 cards (each the current top card of their respective lists) each time this way, picking the best one to go next.
But the all-cards list is optional anyway.
-
So, is there something happening here?
-
So, is there something happening here?
Qvist is updating the database and will post on FDS in a couple days.