Print

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

[theory]

On BGG, you wrote:

It is common in games in the real world to use the randomizer of "let's take all the randomizer cards, shuffle them, and take 10 off the top". The reason this randomizer is popular is because it is very easy to implement. Any other randomizer that tries to bias things is a bear to implement in real life.

But because it is so easy to implement (and consequently because isotropic has decided to use it default), there's a substantial part of the community that now treats it as gospel, as the One True Way of playing Dominion.

In real life play, while I am willing to use the unbiased randomizer as a basis for generating a kingdom, I won't commit to its results. If the resulting output kingdom contains Militia, Goons, Cutpurse, and Mountebank, I'm going go "hey guys, I think there are too many similar Attacks here, how about we swap out a couple of them?" Most players will respond to my question with "Yeah, that makes sense". But some players will react with unease, as if I had just suggested that the God they believe in was not omniscient and omnibenevolent (which in some sense is exactly what I did).

That generally works well in real-life. But online, it's very hard to discuss such things, when everyone just wants a quick play.

Online play offers a rare opportunity that can't be done in real life -- having the ability to use randomizers that use unbiased random as a starting point but biases away from those crazy kingdoms.

The unbiased randomizer has its uses. It's great for tournaments and professional games because it makes serious players need to be well-rounded and serious players don't mind adopting non-flashy strategies or being hit with the long whiplash of the variance curve. A tournament player who gets a kingdom where everything costs $4 or less doesn't have any problem just rolling with it. But I think most casual players will go "man, this set ain't any fun."

1. You don't win anyone over to your side by suggesting that their opinions are the product of cognitive biases.

2. The point of random sets is to learn something.  If every set you throw in cards that you "know" work together, you never get that feeling of discovering something new.  Maybe you believe that you've exhausted everything new, but suppose your rubric had been in place pre-KC/Goons/Masq: according to this set generator, maybe you would have never seen it.  You would think KC is too swingy.  Goons is too high variance or dominating.  So you never discover interesting combos because you've already limited yourself.

3. I don't think it's a cognitive bias to suggest that a player's default expectation is going to be a randomly chosen set of 10.  It's really misleading that when you choose "random set", you almost will never get Throne Room or King's Court, and almost always get Bandit Camp and Border Village.

4. I understand that this is eventually going to be somewhat subjective, but wow, I did not realize the extent to which your personal views were being injected into this generator.  What the heck is the point of playing Dominion without King's Court?  Platinum is more swingy, as DXV likes to say, and no one bans that.  You can't please everyone, but right now it seems like there is a very very wide gulf of opinion on what is "fun" between your set generator and what I thought to be general consensus.

5. I understand also that you are invested in this company, literally, but you gotta be able to take a step back and realize when you're not helping.  I mean, for real, you put in a set generator that stops KC from being on the board.  Obviously you didn't mean to do this, and that you're going to bump it up, but we're concerned that your personal views on how you'd like to play Dominion are dominating at the expense of both casual and experienced players.  If in the end, you come up with a set generator that nevertheless actively discourages KC, that's not the Dominion I enjoy playing.  It's some weird form of "onigame Dominion" where a lot of the variance is removed.

[theory]

Here's a card-playing analogy: suppose I make a Bridge site, the only official online place to play Bridge.  Except, because I have strongly held views on what constitutes "interesting" and "not interesting" hands, I'm going to make everyone play with my Bridge deck shuffler, which only creates hands that I deem "interesting". 

Do you see how this might raise objections?  The whole point of Bridge is to play it full random.  That's where the joy of the game comes in.  There is value in occasionally picking a deal, but in general, the idea of Bridge is supposed to be that you master a set of techniques to be applied to a randomly sorted deck of 52 cards.  Not only certain techniques relating to certain sorts of those cards.

[Beyond Awesome]

Here, I am going to post my thoughts on select cards that I think are too high or too low. Please note that these are just my opinions, and anyone who disagrees may say so.

Too high:
Festival   
954
238.50%
Market   
943
235.75%
Laboratory   
921
230.25%

Don't get me wrong, these are all great cards. But, out of 1000 games, they will almost always show up. Heck, I haven't even played 1000 games on Goko yet. I would be pretty frustrated if I kept seeing these three cards over and over.

Too low
Smithy   
137
34.25%

This is an important card in a lot of games. Surprised to it so low.

Too high
Harem   
966
241.50%
Duke   
963
240.75%
Great Hall   
906
226.50%

Harem does not add enough value in my opinion. I do like Duke games, but it should not be so dominating that every game of Intrigue has it show up. For Great Hall, it does not add enough value to show up in every 1 out of 10 games.

Too low
Nobles   
149
37.25%
Masquerade   
145
36.25%
Steward   
144
36.00%
Swindler   
108
27.00%
Minion   
99
24.75%
Courtyard   
84
21.00%

Most of these are key power cards from Intrigue and yet, they are missing from a majority of games. I find something wrong with that.

Too high
Merchant Ship   
918
238.68%
Conspirator   
852
213.00%

Merchant Ship is a good card, but I don't think any one card is so good that it should be allowed to show up every 9 out of 10 games. Also, I do enjoy me some Conspirator chains but 85% of the games is just too much imo.

Too low
Cutpurse   
155
40.30%
Pirate Ship   
146
37.96%
Tactician   
139
36.14%
Lookout   
133
34.58%

I think a lot of new players overvalue Pirate Ship, and enjoy playing it, so you probably want it to show up a bit higher. Tactician is a pretty cool card and should show up higher, and it is always good to have some trashing going on, so Lookout should also be higher.

Too high
Grand Market   
930
232.50%
Peddler   
869
217.25%

Again, no card should be showing up 93% of the time. And, Peddler is cool, but it doesn't need to show up in every kingdm that Worker's Village shows up in. I assure you that's going to get old very fast.

Too low
Mint   
155
38.75%
Loan   
141
35.25%
King's Court   
73
18.25%

These are all interesting cards that people like playing and KC is one of the most popular cards of all-time. And, also a card that a lot of players consider fun (some do consider it unfun though). But, nothing gets a casual player more excited when they string together a major combo turn and realize they can do that with King's Court. IMO, KC is the benchmark of fun cards when it comes to casual Dominion. I might be wrong this, but it seems theory shares a similiar opinion to mine in htis respect.

Shockingly for Hinterlands, nothing is too high. Border Village seems about right. However, Hinterlands does have a few cards that imo, show up at too low of a frequency.

Too low
Oracle   
176
45.76%
Fool's Gold   
176
45.76%
Inn   
139
36.14%
Develop   
124
32.24%
Embassy   
81
21.06%


All of these cards are interesting. Embassy offers amazing draw. Inn is a village that also sifts through your deck. Develop is a trasher. Fool's Gold is just awesome and Oracle is pretty decent drawing. I think the case can easily be made for having these cards show up a little more common.

Too high
Bandit Camp   
999
349.65%
Forager   
908
317.80%
Beggar   
901
315.35%
Vagrant   
835
292.25%

Wtf at Bandit Camp. No card should show up 99.9% of the time, especially in a set with 35 kingdom cards!!! Forager is fun. Don't get me wrong, but it is still a little too high. For Beggar, I actually really like the card, but most often, you can ignore buying it and a lot of players I play against never touch that card. Ditto for Vagrant. Unless you're playing a Ruins/Cursing game, it is essentially a sucky Pearl Diver and Pearl Diver already sucked to begin with.

Too low
Pillage   
163
57.05%
Sage   
162
56.70%
Cultist   
159
55.65%
Death Cart   
156
54.60%
Graverobber   
151
52.85%
Ironmonger   
143
50.05%
Knights   
142
49.70%
Rebuild   
136
47.60%
Hermit   
134
46.90%
Feodum   
110
38.50%
Wandering Minstrel   
89
31.15%
Storeroom   
86
30.10%
Squire   
77
26.95%
Procession   
72
25.20%
Rogue   
72
25.20%
Scavenger   
66
23.10%
Poor House   
48
16.80%


DA seems to have a huge imbalance. Poor House is actually pretty fun. Sometimes, you can never use it, but it shouldn't be so low. To be honest, Graverobber and Procession might be placed right. But, I am saying that because I almost never play with them. It could very well be possible that I just don't know how to make them work. Anyway, Wandering Minstrel is a pretty awesome village. Scavenger can do neat tricks. My opinion of Sage is dropping, but people seem to like buying it. Overall, though, I think a lot of cards show up a lot less than they should and a handful of cards show up too much.

[Donald X.]

This is definitely a big surprise to me too, especially Squire.  Right now I suspect I probably just didn't type in the right flags for Squire.

I bet Bandit Camp is part of it. If you have Bandit Camp then you down-weight Squire, since it's another village, and as it happens you always have Bandit Camp. You can manage to have Squire by picking it before Bandit Camp, or taking it even though the odds are against it.

[Donald X.]

What the heck is the point of playing Dominion without King's Court?

Another thing is, that you bought Prosperity, you spent money, and you would like to play with those cards you bought. Even if you decide to add King's Court to a veto list (which would be nice to have), you would still like to see the next-least-likely card make a decent number of appearances. I mean every purchase makes every card show up less often, but, you know.

I think having card selection algorithms is fine; I could see having a choice of algorithms somewhere, something like

1) random (with a veto list for casual games)
2) onigame's algorithm
3) onigame's algorithm but with cards appearing at the normal frequency
4) wacky - like, five cards with one functional theme, five with another
5) some other algorithm that someone proposes
6) "set of the day" - a set of 10 someone picked out

Incidentally I think #3 is possible, and onigame said he might try it out (it preserves frequency but does mean it would be more likely to have certain cards show up with/without certain other cards).

Divide the number of cards the player owns by 10. Prepare to generate that many sets-of-10. Pick a random starting card for each one. Now go through and pick a 2nd card for each set out of the cards remaining, using your existing algorithm. Then pick a 3rd card for each set out of the cards remaining, etc. Eventually you will have used every card, while doing a certain amount to increase the chance of "fun" via how the cards are grouped. Now just pick one of the generated sets at random and there you go. You can use the other generated sets later or not.

If the number of cards they own isn't a multiple of 10, then for perfect results we would have to generate ten times as many sets, using ten copies of each card but of course not allowing duplicates in a set. There is a chance that you will end up with a duplicate left at the end - the card that has to go in this slot is already in this set. I don't know how common that would be, but there are things you can do there, depending on how much work you want to do vs. how much you care about a small amount of compromising card frequency or fun.

I see now that this doesn't account for Young Witch, so possibly banes would just have to show up slightly more often; I do not think the math works out there.

[onigame]

Here's a card-playing analogy: suppose I make a Bridge site, the only official online place to play Bridge.  Except, because I have strongly held views on what constitutes "interesting" and "not interesting" hands, I'm going to make everyone play with my Bridge deck shuffler, which only creates hands that I deem "interesting". 

Do you see how this might raise objections?  The whole point of Bridge is to play it full random.  That's where the joy of the game comes in.  There is value in occasionally picking a deal, but in general, the idea of Bridge is supposed to be that you master a set of techniques to be applied to a randomly sorted deck of 52 cards.  Not only certain techniques relating to certain sorts of those cards.

Yes, I see how this might raise objections.  Let me make a different analogy.

Suppose I make an online Bridge site, the only official online place to play Bridge.  I have a section of the site for seasoned Bridge players and it's pretty much everything they want, with a randomly sorted deck of 52 cards.  There's a different section of the site for novice Bridge players, where instead of a randomly sorted deck, there are custom decks that make it easier to for novices to learn Bridge.  Most hands perform well to early techniques such as "draw trumps and then collect winners," and there aren't any hands where you have to perform squeezes or end-plays to make the contract.  The seasoned Bridge players see this section and are completely aghast.  They feel that the site sullies the fundamental nature of Bridge!  Bridge is supposed to be played with a randomly sorted deck of 52 cards!  That's the whole *point* of the game!  These novices might, oh my gosh, end up actually having fun playing these false "novice games" without ever enjoying the pleasure of a well-executed squeeze play or a French Coup!  I learned Bridge by going to my local Bridge Club and getting berated by little old ladies and how dare this site make something more accessible and friendly and call it Bridge!

Now, does *that* analogy make any sense to you?

(The sad thing is that, to some extent, I think this actually *is* happening in the Bridge world.  Very frequently I meet intelligent game players who say "well, I tried Bridge once, but there was all this entrenched knowledge about how to bid and all this complicated terms they used for playing and I just kind of lost interest.")

[Donald X.]

Suppose I make an online Bridge site, the only official online place to play Bridge.  I have a section of the site for seasoned Bridge players and it's pretty much everything they want, with a randomly sorted deck of 52 cards.  There's a different section of the site for novice Bridge players, where instead of a randomly sorted deck, there are custom decks that make it easier to for novices to learn Bridge.  Most hands perform well to early techniques such as "draw trumps and then collect winners," and there aren't any hands where you have to perform squeezes or end-plays to make the contract.  The seasoned Bridge players see this section and are completely aghast.  They feel that the site sullies the fundamental nature of Bridge!  Bridge is supposed to be played with a randomly sorted deck of 52 cards!  That's the whole *point* of the game!  These novices might, oh my gosh, end up actually having fun playing these false "novice games" without ever enjoying the pleasure of a well-executed squeeze play or a French Coup!  I learned Bridge by going to my local Bridge Club and getting berated by little old ladies and how dare this site make something more accessible and friendly and call it Bridge!

We have novices covered though. They can play endless carefully chosen sets of 10 (well, they can once the old base set campaign gets replaced with the new one). We could have an option to play rulebook recommended sets - 7) recommended set. They don't need that much of that to be ready for random.

[onigame]

1. You don't win anyone over to your side by suggesting that their opinions are the product of cognitive biases.

True.  But you can win over the undecideds, whom you're not accusing of having cognitive biases. 

2. The point of random sets is to learn something.  If every set you throw in cards that you "know" work together, you never get that feeling of discovering something new.  Maybe you believe that you've exhausted everything new, but suppose your rubric had been in place pre-KC/Goons/Masq: according to this set generator, maybe you would have never seen it.  You would think KC is too swingy.  Goons is too high variance or dominating.  So you never discover interesting combos because you've already limited yourself.

I can learn better stuff by taking classes at a community college.  The point of playing games is to have fun; learning something is secondary and in many cases not actually a necessary component of having fun.  (Learning about KC/Goons/Masq has increased my enjoyment of Dominion in the same way that learning about wedgies has increased my enjoyment of high school gym class.)

Yes, it's true that playing with more restricted sets limits what you can get out of a game.  But as long as you're having fun, does it matter?  Once you stop having fun, you remove some of the limits and there are now new things to discover.  Let players go to the next level when they feel they're ready.

3. I don't think it's a cognitive bias to suggest that a player's default expectation is going to be a randomly chosen set of 10.  It's really misleading that when you choose "random set", you almost will never get Throne Room or King's Court, and almost always get Bandit Camp and Border Village.

Yes, that would be very misleading.  Which is why I do NOT want to call it "random set".

4. I understand that this is eventually going to be somewhat subjective, but wow, I did not realize the extent to which your personal views were being injected into this generator.  What the heck is the point of playing Dominion without King's Court?  Platinum is more swingy, as DXV likes to say, and no one bans that.
You can't please everyone, but right now it seems like there is a very very wide gulf of opinion on what is "fun" between your set generator and what I thought to be general consensus.

There's a difference between (1) what my personal preference for games are, (2) what I want my generator to output (by default), (3) and what my generator currently outputs.  Please don't confuse the three.

(3) is 18.25% as of LastFootnote's last stats run.
(2) is somewhere between 95% and 120%.
(1) is somewhere around 0.3%.

5. I understand also that you are invested in this company, literally, but you gotta be able to take a step back and realize when you're not helping.  I mean, for real, you put in a set generator that stops KC from being on the board.  Obviously you didn't mean to do this, and that you're going to bump it up, but we're concerned that your personal views on how you'd like to play Dominion are dominating at the expense of both casual and experienced players.  If in the end, you come up with a set generator that nevertheless actively discourages KC, that's not the Dominion I enjoy playing.  It's some weird form of "onigame Dominion" where a lot of the variance is removed.

I understand your points and your concern.  I will vow right now that, in any version of software that I approve of, playing a game with 10 cards selected by a "pure unbiased random" generator will be always available and easy to get to.  No more than 5 taps/clicks from the home screen.

Honestly, I didn't know until yesterday that my generator was *sooo* dismissive towards KC.  I announced this thing, what, 37 days ago, and asked for feedback then?  In the last 1 day I've gotten more useful feedback and analysis on my generator than in all 36 days preceding it.  (I'm not sure if I'm learning the right lesson from that.)

[onigame]

Thanks to some help from LastFootnote, you can now generate your own stats similar to what he posted earlier in the thread.

Please go to:  http://dominionsetgenerator.com/stats.html

Based on some help from this tool, I have changed many of the underlying parameters for the generator, and some of the extremes are more tempered now.  For example, Bandit Camp is no longer in 99% of all Dark Ages games -- it's in 51% of them (which is still a lot more than pure random, 28.6%).  Some of the more interesting changes I made:

* Victory cards originally tended to cluster; i.e., if the set had at least one Victory card, the generator would try to favor adding more of them.  This made some sense when mixing expansions but had a side-effect when doing Intrigue-only decks, which is that you're very likely to get at least one Victory card in your first 5 picks and as a result the game often chose all three of them.  Upon reflection I realized that clustering Victory cards wasn't really doing much good and removed that.

* I had two attributes that were closely correlated: "Carddrawers" (things that draw cards) and "Selfmillers" (things that can potentially skip past cards in your deck).  Both of these had a recommended range.  However, cards like Smithy and Moat would often get a double penalty for going past the range of both of these.  (Council Room too, but it got a boost both for being cost $5 and for having a +Buy).  I increased the viable range and decreased the penalty to compensate.

Here are the current top and bottom 5 cards for each expansion based on a 1000-trial run:

Base Set top 5:
Market    732    73.2%    183%
Festival    636    63.6%    159%
Laboratory    595    59.5%    148.75%
Council Room    543    54.3%    135.75%
Remodel    527    52.7%    131.75%

Base Set bottom 5:
Throne Room    290    29%    72.5%
Gardens    286    28.6%    71.5%
Thief    266    26.6%    66.5%
Bureaucrat    259    25.9%    64.75%
Chancellor    218    21.8%    54.5%

Intrigue top 5:
Tribute    606    60.6%    151.5%
Trading Post    593    59.3%    148.25%
Mining Village    545    54.5%    136.25%
Upgrade    528    52.8%    132%
Baron    519    51.9%    129.75%

Intrigue bottom 5:
Harem    278    27.8%    69.5%
Minion    273    27.3%    68.25%
Courtyard    266    26.6%    66.5%
Secret Chamber    245    24.5%    61.25%
Coppersmith    235    23.5%    58.75%

Seaside top 5:
Bazaar    701    70.1%    182.26%
Treasury    654    65.4%    170.04%
Explorer    603    60.3%    156.78%
Wharf    557    55.7%    144.82%
Island    489    48.9%    127.14%

Seaside bottom 5:
Pirate Ship    237    23.7%    61.62%
Merchant Ship    219    21.9%    56.94%
Embargo    208    20.8%    54.08%
Sea Hag    190    19%    49.4%
Cutpurse    188    18.8%    48.88%

Prosperity top 5:
Peddler    710    71%    177.5%
City    666    66.6%    166.5%
Grand Market    620    62%    155%
Forge    600    60%    150%
Expand    564    56.4%    141%

Prosperity bottom 5:
Talisman    253    25.3%    63.25%
Royal Seal    247    24.7%    61.75%
Mountebank    224    22.4%    56%
Counting House    215    21.5%    53.75%
Quarry    205    20.5%    51.25%

Hinterlands top 5:
Stables    666    66.6%    173.16%
Highway    652    65.2%    169.52%
Border Village    589    58.9%    153.14%
Margrave    544    54.4%    141.44%
Spice Merchant    540    54%    140.4%

Hinterlands bottom 5:
Embassy    242    24.2%    62.92%
Silk Road    242    24.2%    62.92%
Noble Brigand    240    24%    62.4%
Oracle    237    23.7%    61.62%
Fool's Gold    231    23.1%    60.06%

Dark Ages top 5:
Bandit Camp    507    50.7%    177.45%
Cultist    458    45.8%    160.3%
Count    439    43.9%    153.65%
Fortress    401    40.1%    140.35%
Junk Dealer    399    39.9%    139.65%

Dark Ages bottom 5:
Feodum    176    17.6%    61.6%
Armory    170    17%    59.5%
Rogue    166    16.6%    58.1%
Poor House    154    15.4%    53.9%
Marauder    154    15.4%    53.9%

I'm certainly not claiming that things are perfect now, but hopefully people will find these extremes more palatable.  As always, I welcome comments.

[hsiale]

I don't know if this is due to the set generator but I just played a game at Goko in single player mode where every single card except the bane was from Cornucopia.

Started another one now and I got some more distribution - 5 Prosperity, 3 Dark Ages, 1 Seaside, 1 Base.

[Donald X.]

Seaside bottom 5:
Pirate Ship    237    23.7%    61.62%
Merchant Ship    219    21.9%    56.94%
Embargo    208    20.8%    54.08%
Sea Hag    190    19%    49.4%
Cutpurse    188    18.8%    48.88%

In the older sets I mean expansions, the idea was that 1 in 5 cards would be an attack, with a sprinkling of 1-2 additional non-attack interactive cards like Council Rooms (we will use your term for them, "helpers"). This holds true for Alchemy/Cornucopia considered as a single expansion. I wanted to have player interaction and attacks were where you got that (and helpers, and, to a lesser degree, VP cards like Gardens). For Prosperity though I downplayed attacks in favor of helpers, to make it easier to get to Colony. Then it seemed like people preferred it that way, so Hinterlands and Dark Ages are both a little lighter on attacks, but not on interaction - one attack slot replaced by a helper (and obv. IGG counts for an attack slot).

You seem to be aiming for maybe one attack card per set-of-10? But that's half what the expansions themselves have. So here's the Seaside bottom 5, with three attacks plus a "helper" (okay Embargo does not "help" much, but you know, it's non-attack interaction). The Hinterlands bottom 5 has two attacks, two helpers, and a VP card.

So, I think player interaction is being downweighted too much.

[werothegreat]

When the set generator was announced, my thought was that the implementation was going to be pretty much exactly what was put up on Wei-Hwa's website - you CHOOSE exactly what you want put into the kingdom.  Now, I realize this may be too much for novice players, so why not have three options for casual games:

1) Pure random
2) An average implementation of the generator, maybe called "Well-rounded kingdom"
3) "Advanced" - which lets you set all the parameters of the kingdom

I would like a pure random option in casual because, personally, I like to reserve my pro games for when I'm making videos, and I don't always feel like making a video, so if I just want to play a game with someone, I'll do casual, and I sometimes want to play a purely random kingdom, as much as I complain about there not being Villages in a kingdom.  I may dislike when that happens, but it encourages me to think more about what to do in a kingdom.

[Polk5440]

Do you see how this might raise objections?  The whole point of Bridge is to play it full random.  That's where the joy of the game comes in.

I disagree with this. And I find this comment surprising given the "build a kingdom" competition for the final round of the tournament....

This analogy overlooks the ways that Bridge deals with uninteresting games. Because of the bidding round, uninteresting games are either played quickly or not at all. Sometimes Bridge players even re-deal games if the winning contract is 1 of something (i.e. too boring) when playing casually. If a contract is easily made by a team, often play stops, and a certain number of tricks will be immediately conceded, so the game is often not played out or is played out quickly.

The opposite tends to happen in Dominion online with full random. There is no way to speed up or pass on the uninteresting games. In addition, because of unresolved shuffle luck people rarely concede (unless very far behind or are shut out of the needed points) because there is still a chance they can win.  A biased randomizer is a proposed solution to do so (by making uninteresting kingdoms come up less often), and one I really like.

Backgammon deals with this issue by having a doubling cube. There is a lot of luck and variance to Backgammon (because of dice rolling) and strategy (how to move your pieces) so it is a decent comparison (although, the starting position is the same every time). If a player gets out to an early lead (think flipping a Sea Hag with his Sea Hag, twice in a row), then doubling the stakes forces the player who is behind to concede or to commit to mounting a comeback relying on some luck. This pressure of the doubled stakes makes people more likely to concede when losing and ends uninteresting games quickly (lopsided games are uninteresting, too). I would be interested to know what a doubling cube solution would look like for 3 or 4 player games. Or what a doubling cube style solution would look like for 2 players in casual play with different opponents (where stakes are not easily defined).

In addition, in duplicate Bridge tournaments, everyone plays the same set of starting hands, not full random. The point is to play better than others, not to play random deals. Also, you can learn more from duplicate tournaments. If one team sees something interesting that many others overlook, people can then go back and say, "wow! I learned something!" You can learn from how the winners played your exact deals. Your learning is not dependent on playing every single deal all the way through. The fact is, many deals are uninteresting no matter how you slice it. Much like in Dominion. It's a great game, but some kingdoms are just more interesting than others. Very few people want to think about EVERY kingdom and EVERY game all the way through equally seriously. They shouldn't be forced to.

[LastFootnote]

onigame, I want you to know how much I appreciate your apology. It was very good of you to take the time to address my concerns. Even this part:

Why are your opinions so much more important than mine?

I feel I have to address this even though you may not enjoy the answer.  My opinions are "more important" than yours because Goko has a lot more of my money than they have of yours (as far as I can tell; I don't actually know for sure that you also aren't one of the investors).  I invested in the company rather early, when it was more hope than actual implementation, and a lot of the work I'm putting in is motivated by making that investment be profitable.

That was me being petulant. Just the fact that you were a Dominion playtester and helped create the recommended sets of 10 in many of the expansions (which I also appreciate!) would put a lot of weight behind your opinions, and I was aware of that. It was kind of you to respond anyway.

I know that the company can't succeed without having customer loyalty, but also that it can't succeed without a wider, casual, audience.

We all have invisible biases towards our own experiences.  I enjoyed isotropic when it first came out, but soon got tired of it, because I felt I was just playing too many games where the kingdom was a one-strategy kingdom and that to do well I had to invest a lot of time to sense and execute tiny improvements in strategy and that having a good time seemed to be more about winning games rather than actually having fun with different game mechanics (which is the whole reason I enjoy playing Dominion).  I don't feel that a completely faithful clone of isotropic would ever be monetarily successful.

It's very hard treading the line between getting new users and satisfying old users.  In this case a confluence of events, some of which I was responsible for, tipped the ship too far in one direction.  I'm doing my best to course-correct.

It's never been about getting *my* code into Goko.  It's been about doing what I feel is important to make a niche game appeal to a wider audience.

I understand. From the beginning, I've also want Goko to succeed and for Goko Dominion to be friendly to casual players. At no time have I wanted Goko to be a faithful isotropic clone.

Let me dispel another assumption you may (or may not!) have about me. I'm not a pure-random zealot, to coin a phrase. I, too, have designed my own randomizer that I believe makes Kingdoms more fun. It's a Java Swing app and whenever I play in real life, it's what I use to create sets. Even when I played on isotropic, I rarely played pure random. I would bias my first game toward the base set, my second game toward Intrigue, and so on down the line. Seeing more of the interesting card interactions between cards within the same expansion is one of the things I consider fun.

My concerns are that your randomizer isn't effective at its stated goal (making games that are more friendly to casual players) and that it's going to have lots of negative side effects.

I can't speak for everybody, but for me, the long-lasting appeal of Dominion is its sheer variety. Even if you iron out the severe biases that your randomizer has, the fact that it always creates sets with 'a little bit of everything' removes a huge chunk of variety from the game. If all games felt so similar, I think I would have tired of Dominion really quickly. I'm guessing that with your randomizer many new players will have that very experience. If your goal is to keep them interested long enough to buy an expansion or two, then drop off to make space on the server for new players, maybe this is perfect. (I'm not claiming this is your motivation, but if somebody's sole concern were revenue, this might be attractive.)

Theory and Donald have already put forth many of the concerns I have, such as the fact that if I buy a new pack of cards, I want to see all of them a good portion of the time, or I don't feel I'm getting what I payed for. I'll try to avoid just repeating what they've said, but know that I agree wholeheartedly.

2. The point of random sets is to learn something.  If every set you throw in cards that you "know" work together, you never get that feeling of discovering something new.  Maybe you believe that you've exhausted everything new, but suppose your rubric had been in place pre-KC/Goons/Masq: according to this set generator, maybe you would have never seen it.  You would think KC is too swingy.  Goons is too high variance or dominating.  So you never discover interesting combos because you've already limited yourself.

I can learn better stuff by taking classes at a community college.  The point of playing games is to have fun; learning something is secondary and in many cases not actually a necessary component of having fun.  (Learning about KC/Goons/Masq has increased my enjoyment of Dominion in the same way that learning about wedgies has increased my enjoyment of high school gym class.)

Yes, it's true that playing with more restricted sets limits what you can get out of a game.  But as long as you're having fun, does it matter?  Once you stop having fun, you remove some of the limits and there are now new things to discover.  Let players go to the next level when they feel they're ready.

I think theory's point is that for many players, learning a new strategy or card interaction is fun. I don't like the KC/Goons/Masq pin either and I don't think it was a very good example. It's sort of a fluke.

3. I don't think it's a cognitive bias to suggest that a player's default expectation is going to be a randomly chosen set of 10.  It's really misleading that when you choose "random set", you almost will never get Throne Room or King's Court, and almost always get Bandit Camp and Border Village.

Yes, that would be very misleading.  Which is why I do NOT want to call it "random set".

But in your GetSatisfaction thread, you do advocate calling it a "Basic Game" and calling pure random a "Pro Game", which is just as misleading. There's nothing basic about your randomizer.

In fact, that's what's so wrong. Your randomizer is the very definition of inelegant. If I had to give someone an example of an inelegant software solution to a perceived problem, I'd just point them to dominionsetgenerator.com. When I said that I liked playing with niche cards and your randomizer discouraged them, your solution was to add a 'niche card' weighting. When I read that, I shook my head in disbelief. You're missing the entire point! The point is that, for many cards, what categories it fits under isn't cut and dried. 'Niche' is the perfect example of that. Which cards are niche is not pinned down, and it changes both as more cards are released and as people become more familiar with using the cards in different situations (which they won't be able to if they're using your generator)!

To put it plainly, the set of categories available is arbitrary, the cards that fit into each category are arbitrary, and how much each category is weighted is arbitrary. The default settings are all dependent on what role you feel each card fits and which roles are important to you! You will never fix this randomizer! It is fundamentally flawed.

All that said, it's easy to criticize without suggesting any solutions. In a later post today, I will put forth my own ideas about how to address these issues in Goko.

[onigame]

I don't know if this is due to the set generator but I just played a game at Goko in single player mode where every single card except the bane was from Cornucopia.

Started another one now and I got some more distribution - 5 Prosperity, 3 Dark Ages, 1 Seaside, 1 Base.

It most likely is due to the set generator.  The generator (on default settings) tries to cluster cards from the same expansion together.  The weight of the cluster is stronger for the small expansions (Cornucopia, Alchemy, Guilds). 

[WanderingWinder]

1. You don't win anyone over to your side by suggesting that their opinions are the product of cognitive biases.

True.  But you can win over the undecideds, whom you're not accusing of having cognitive biases. 

I think you're more likely to lose them by appearing to just insult anyone who disagrees with you. I mean 'my opponents have cognitive biases' comes across as 'my opponents are mentally challenged' or 'my opponents are stupid', and this is not a very endearing thing to say. Of course, it doesn't actually mean that, but it can come across that way, and, having cognitive biases is something that is just inherently true of everyone - you have them just as much. So do I. So as an argument, it's a non-starter.

2. The point of random sets is to learn something.  If every set you throw in cards that you "know" work together, you never get that feeling of discovering something new.  Maybe you believe that you've exhausted everything new, but suppose your rubric had been in place pre-KC/Goons/Masq: according to this set generator, maybe you would have never seen it.  You would think KC is too swingy.  Goons is too high variance or dominating.  So you never discover interesting combos because you've already limited yourself.

I can learn better stuff by taking classes at a community college.  The point of playing games is to have fun; learning something is secondary and in many cases not actually a necessary component of having fun.  (Learning about KC/Goons/Masq has increased my enjoyment of Dominion in the same way that learning about wedgies has increased my enjoyment of high school gym class.)

Yes, it's true that playing with more restricted sets limits what you can get out of a game.  But as long as you're having fun, does it matter?  Once you stop having fun, you remove some of the limits and there are now new things to discover.  Let players go to the next level when they feel they're ready.

3. I don't think it's a cognitive bias to suggest that a player's default expectation is going to be a randomly chosen set of 10.  It's really misleading that when you choose "random set", you almost will never get Throne Room or King's Court, and almost always get Bandit Camp and Border Village.

Yes, that would be very misleading.  Which is why I do NOT want to call it "random set".

4. I understand that this is eventually going to be somewhat subjective, but wow, I did not realize the extent to which your personal views were being injected into this generator.  What the heck is the point of playing Dominion without King's Court?  Platinum is more swingy, as DXV likes to say, and no one bans that.
You can't please everyone, but right now it seems like there is a very very wide gulf of opinion on what is "fun" between your set generator and what I thought to be general consensus.

There's a difference between (1) what my personal preference for games are, (2) what I want my generator to output (by default), (3) and what my generator currently outputs.  Please don't confuse the three.

(3) is 18.25% as of LastFootnote's last stats run.
(2) is somewhere between 95% and 120%.
(1) is somewhere around 0.3%.

But the point isn't that this is your definition of what is fun, the point is that it's some definition that doesn't come from them and is being imposed by you. Whether it's what you actually enjoy or not is not super relevant. Now, it's true that uniform random isn't from them either - but it is a null position, objective except insofar as it was decided by the people who made the game.

5. I understand also that you are invested in this company, literally, but you gotta be able to take a step back and realize when you're not helping.  I mean, for real, you put in a set generator that stops KC from being on the board.  Obviously you didn't mean to do this, and that you're going to bump it up, but we're concerned that your personal views on how you'd like to play Dominion are dominating at the expense of both casual and experienced players.  If in the end, you come up with a set generator that nevertheless actively discourages KC, that's not the Dominion I enjoy playing.  It's some weird form of "onigame Dominion" where a lot of the variance is removed.

I understand your points and your concern.  I will vow right now that, in any version of software that I approve of, playing a game with 10 cards selected by a "pure unbiased random" generator will be always available and easy to get to.  No more than 5 taps/clicks from the home screen.

Honestly, I didn't know until yesterday that my generator was *sooo* dismissive towards KC.  I announced this thing, what, 37 days ago, and asked for feedback then?  In the last 1 day I've gotten more useful feedback and analysis on my generator than in all 36 days preceding it.  (I'm not sure if I'm learning the right lesson from that.)

The thing is, if you don't know this stuff, you should probably not be pushing it out. People are unlikely to take your 'vow' seriously if that's the way it actually is right now, which may not be your fault, but if you don't have control over this, people are actually going to care when someone who does have that control says something.

The reason you're hearing about it now is that it's having an effect on people now. Before, if they didn't like it, there was no reason to point out problems they have with it - maybe someone else likes it, and well, you generally just don't tell somebody that their work is bad if there isn't a reason; it's impolite. But now, this is being pushed on people - really forcefully apparently, which I understand isn't your particular intention, but it does get people's attention. And even if your scenario were put in place, it would be the default, which still is going to carry some weight.

One more point:
The point about not graduating people to full random until they're ready seems at odds with the concept of card selection you're talking about, that this is supposed to get rid of boring sets where there's just one clear dominant strategy - those should be EASIER to learn; if you're graduating people, it should be FROM them, not TO them (assuming your original argument were right, which I'm not convinced of).

tl;dr It's not that uniform random is better, it's that people don't always agree on what 'fun' is, and uniform random is the one objective null point inherent to the game.

[onigame]

But in your GetSatisfaction thread, you do advocate calling it a "Basic Game" and calling pure random a "Pro Game", which is just as misleading. There's nothing basic about your randomizer.

Coming up with short descriptive names for complicated things is hard.  I want a name that means "generator that uses a complicated algorithm to create a kingdom with the goals of making a kingdom that is well-rounded in strategy space intended for casual players who want something comfortable (and not expert players who want the chance to explore rare and specialized strategies)."

Maybe "Onigame's Casual Setup" would be a better short name.  It feels so vain though!

[hsiale]

I have just looked at generator page. I don't even understand what half of the settings mean. And this looks like the biggest problem for me. How should a new player use this generator? As an opposite example, let's look at http://inprogressgaming.com/dominion-card-picker/ - a generator by Toskk I use when playing with RL cards. It has customization, but the options are easy to understand to anyone who knows game rules.

I don't know how would I code a randomizer, as I don't know how to code. But here's how I would like it to work (note: I am not a very experienced player, I play Dominion for 3 months, with my games divided more or less 50% between Goko and Iso. My current Iso level is 20). For first games, there should be no randomizer at all. There are (and from what I've heard they will be improved) adventures with lots of fixed board to learn the game. If a new player doesn't like adventures, we have recommended sets of 10. There are 5 of them in Base Game rulebook and I guess it's not a problem to make 10-15 more. Throwing a card from some other expansion into each of them, one that works well within the set (to make people want to buy more sets and to learn them rules introduced in them). And each of them should have a short article available directly at Goko, hinting at the possibilities in the set while still leaving interesting things to explore on your own. This should cover beginners.

Now we have a player that has learned something. But not a lot still. But enough to understand what the game is about. And enough to start playing random sets. This is where the randomizer should step in. A customizable one, though the default should still be full random. And there should be options. Many of them, allowing to customize towards any kind of deck. If someone chooses to customize towards engines all the time, then so be it. But give everyone a choice, they may want to play something else from time to time. For example, I would love to have a generator giving me rush sets, as I enjoy those games but often fail at them, so having a chance to practice several games in a row would help me a lot. Other people may want to practice anything else.

Engines are not everything in the world. And many of them definitely are not where the world should begin. There's nothing more frustrated than playing an engine board when you have screwed up building and your opponent did not - not only on your turns you can't do anything interesting, but also you have to sit through long turns of your opponent, full of magic tricks most of which you don't understand and having to do something only when a discard attack is fired at you. And beginners are likely to get into such situations, as engine building is very often a high level skill. I still fail on this most of the time. If every game sends you unprepared in the middle of such hell, chances are you leave the game before learning to survive there.

So, there should be choices. Full random, full engine, full money, other possibilities, mixes between them. And a good manual. So that people can learn using the generator the way they prefer and bias the set in whichever direction they like. And of course it should be done the way that no card appears too often or too seldom - I think no card should appear more than twice as often as any other card. And for cards with rare uses, a nice thing would be to bias towards those uses whenever they appear - so that for example 30-40% of games with Chancellor also use Stash (on the other hand definitely not 100%, there are other uses of this card as well).

--------------------
And by the way, I think a set generator is not the most important thing Goko needs to make new players enjoy the game, stay and buy cards. A good help system would do much more. All you get at Goko is a pdf containing Base Set rules and a link to RGG saying "learn more about offline Dominion". I think the mistake people at Goko make is trying to create an online implementation of a board game, while they should create an online game (which happens to already have a paper implementation, a very popular one, and this is good as there is a player base to begin with). What Goko needs is a true online rulebook:
- big, covering lots of topics (because storage space on the server is cheap while printed rulebooks are not),
- hyperlinked, both internally and to external sources (this is the internet, links are the very point of creating webpages),
- clearly divided into beginner, medium level and advanced content, so that people read what they need and can come back for more when they're ready.

Of course this is not something that a Goko employee can do in half a day (I guess this is how the current "rulebook" was created). I think no single person could create something like this, unless being a high level player given a month or two of full time job on this. If Goko feels they need something I describe, what they need is to come here and ask people to help. There definitely are people willing to help here, especially as Goko has improved a lot since summer. Everyone will benefit on lots of casual players appearing and learning the game - Goko will have money when they buy sets (if the game grows big enough it can bring money perpetually at low maintenance cost) and players needs lots of people playing casually so that new high skill players appear to make top level online Dominion interesting. I think a good help system should be Goko's top priority after all cards are implemented. If you agree, then make them come here and ask us to help.  I will help and I'm sure many others will as well.

[werothegreat]

But in your GetSatisfaction thread, you do advocate calling it a "Basic Game" and calling pure random a "Pro Game", which is just as misleading. There's nothing basic about your randomizer.

Coming up with short descriptive names for complicated things is hard.  I want a name that means "generator that uses a complicated algorithm to create a kingdom with the goals of making a kingdom that is well-rounded in strategy space intended for casual players who want something comfortable (and not expert players who want the chance to explore rare and specialized strategies)."

Maybe "Onigame's Casual Setup" would be a better short name.  It feels so vain though!

Which is why I'm wondering why everyone is ignoring my suggestion.  On the game creation page, if you select Casual, just have 3 options:
1) Pure random
2) Basic implementation of set generator (call it "Well-rounded kingdom" or whatever)
3) Advanced - let's you pick and choose the parameters of the generator.

Doesn't that seem self-explanatory enough?

[onigame]

I think you're more likely to lose them by appearing to just insult anyone who disagrees with you. I mean 'my opponents have cognitive biases' comes across as 'my opponents are mentally challenged' or 'my opponents are stupid', and this is not a very endearing thing to say. Of course, it doesn't actually mean that, but it can come across that way, and, having cognitive biases is something that is just inherently true of everyone - you have them just as much. So do I. So as an argument, it's a non-starter.

Consider these two arguments:
(A) "Everyone who disagrees with me is an idiot."
(B) "There are some people out there who believe X.  People who believe X are idiots.  If you're reading this and you haven't made up your mind about X, please don't believe X."

On some level, these two arguments are actually the same.  Yet if you go out in the world you'll see that (B)-style is actually used quite frequently -- in political campaigns, in public-service announcements, in activist activities.  Presumably they wouldn't use it if it wasn't at least working to some extent.

The reason you're hearing about it now is that it's having an effect on people now.

Absolutely.  I knew this.  My cognitive dissonance isn't because I didn't know why this thread is suddenly so popular; it's because I can't tell whether the events of the last three days was a good thing or a bad thing.

On some level, it's a bad thing because I did stuff that made people unhappy.  But on some level, it's a good thing because I'm actually getting some very good opinions from people on this thread and I feel that both my generator and Goko's eventual implementation of Kingdom selecting will be much improved because of it (but who knows, that might just be unrealistically optimistic on my part).  So maybe a couple of days of unhappiness was worth it.  I guess time will tell.

One more point:
The point about not graduating people to full random until they're ready seems at odds with the concept of card selection you're talking about, that this is supposed to get rid of boring sets where there's just one clear dominant strategy - those should be EASIER to learn; if you're graduating people, it should be FROM them, not TO them (assuming your original argument were right, which I'm not convinced of).

This is an absolutely beautiful point, and I'm glad you made it, because it's definitely a game-changer in forming how I think about players and what my goals are in trying to design a good interface for Goko.

Unfortunately I'll need some time before I can articulate my new thoughts fully on this... Other projects beckon and I won't have a free weekend for a while.

[WanderingWinder]

I think you're more likely to lose them by appearing to just insult anyone who disagrees with you. I mean 'my opponents have cognitive biases' comes across as 'my opponents are mentally challenged' or 'my opponents are stupid', and this is not a very endearing thing to say. Of course, it doesn't actually mean that, but it can come across that way, and, having cognitive biases is something that is just inherently true of everyone - you have them just as much. So do I. So as an argument, it's a non-starter.

Consider these two arguments:
(A) "Everyone who disagrees with me is an idiot."
(B) "There are some people out there who believe X.  People who believe X are idiots.  If you're reading this and you haven't made up your mind about X, please don't believe X."

On some level, these two arguments are actually the same.  Yet if you go out in the world you'll see that (B)-style is actually used quite frequently -- in political campaigns, in public-service announcements, in activist activities.  Presumably they wouldn't use it if it wasn't at least working to some extent.

There's a difference between telling someone they're wrong and calling them idiotic.

[onigame]

Which is why I'm wondering why everyone is ignoring my suggestion.  On the game creation page, if you select Casual, just have 3 options:
1) Pure random
2) Basic implementation of set generator (call it "Well-rounded kingdom" or whatever)
3) Advanced - let's you pick and choose the parameters of the generator.
Doesn't that seem self-explanatory enough?

It's a good suggestion.  So good that it was what I expected to happen when I gave them my code a week ago.  Unfortunately what actually came out was
1) To get "Pure random", select "Pro Leaderboard".
2) To get "Well-rounded kingdom", select "Casual Leaderboard".
3) Advanced -- we'll implement this later

I can certainly understand why someone at Goko might have thought this would have been a reasonable midpoint milestone to get to the original idea.  The devil is in the details!

[onigame]

I think you're more likely to lose them by appearing to just insult anyone who disagrees with you. I mean 'my opponents have cognitive biases' comes across as 'my opponents are mentally challenged' or 'my opponents are stupid', and this is not a very endearing thing to say. Of course, it doesn't actually mean that, but it can come across that way, and, having cognitive biases is something that is just inherently true of everyone - you have them just as much. So do I. So as an argument, it's a non-starter.

Consider these two arguments:
(A) "Everyone who disagrees with me is an idiot."
(B) "There are some people out there who believe X.  People who believe X are idiots.  If you're reading this and you haven't made up your mind about X, please don't believe X."

On some level, these two arguments are actually the same.  Yet if you go out in the world you'll see that (B)-style is actually used quite frequently -- in political campaigns, in public-service announcements, in activist activities.  Presumably they wouldn't use it if it wasn't at least working to some extent.

There's a difference between telling someone they're wrong and calling them idiotic.

True, but that's missing my point.  Imagine I wrote this; maybe it gets my point across better:

(A) "Everyone who disagrees with me is wrong."
(B) "There are some people out there who believe X.  People who believe X are wrong.  If you're reading this and you haven't made up your mind about X, please don't believe X."

[Donald X.]

Maybe "Onigame's Casual Setup" would be a better short name.  It feels so vain though!

Call it "Trisha's Casual Setup." Then it sounds good because there's a girl involved. Or "Engine-Heavy."

I feel like "pick a random 10, count the strategies" has got to be the best way to make sets where the goal is to have multiple strategies. Go directly to what you want and you'll do the best job of maximizing it. Each step away hurts how well you do this. Focusing on categories of effects is a step removed and those categories aren't even a measure of fun, it's not "fun cards," it's "+1 buy" and there's an idea that games have more possibilities if you throw that in. Which I mean is true. But Woodcutter isn't directly fun or even directly a strategy. Saying "let's always include these six things" means you'll have sets that always include those six things; it may tend towards three things to do but it's making the sets more samey in other ways having nothing to do with how many things to do there are. But count the strategies and there will be three things to do and +buy won't be remotely mandatory.

I think for a lot of people a good approach to "fun" would just be, let them have a list of cards they don't like, automatically left out of games they generate. Bam, good times. For new players who aren't playing recommended sets / campaigns, a good first pass for "make sure there are multiple things to do" would just be "include three $5's."

Even something like "be less likely to include Counting House when there isn't a combo for it" is I think overall *less* fun. It's not so bad to have a 9-card game with Counting House sometimes. When you get to go for Counting House, part of the joy of it is those other games where Counting House was useless. If every time it shows up you've got a combo for it available, it stops being special.

I think it's fine to have an engine-heavy option, and totally good to have multiple options for how the set-of-10 is generated. And certainly your approach is related to how I made the campaign lists. It's like, most of the time I want a village because it lets you do more, ditto +buy, gotta have some interaction, need some $5's. A big difference is I was specifically trying to use the cards about equally often.

[Donald X.]

So, there should be choices. Full random, full engine, full money, other possibilities, mixes between them.

I broadly categorize decks by how you deal with the "one action per turn" rule:

1) only play a couple actions
2) play with actions with +1 action, or special treasures
3) play with villages
4) play with ways to make use of dead cards, broadly categorized as Remodel and Vault
5) play a strategy that tolerates having dead cards, such as Gardens

And of course you can combine these.