Print

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

[Toskk]

The Goko card picker, as is, will throw up Monument super often, almost always has about three Attacks in it, and almost never has Pirate Ship.

Why would the Goko kingdom picker bias anything? It's specifically billed as 'true random' kingdom selection, so it would have to be a failure of their random number generator, if there's any kind of bias there. I'm a firm believer in artificial kingdom selection (thus all of the artificial selection options in my card picker program on inprogressgaming.com), but in my opinion it is critical that *all* players understand (and agree on) the selection criteria that will be used for any (rated) match. Goko should/would need to be very open about how they are selecting kingdoms *if* they were to be intentionally using anything other than true random.

[Donald X.]

Onigame has readily admitted that his set generator includes some cards more often than others. He has no plans to fix it. It's an integral part of the algorithm, apparently.

EDIT: Here's the relevant post.

I feel like he said the opposite there. More directly, he said to me that he might change the program to de-emphasize cards that are ubiquitous as it stands (Festival being an example). This would just be a matter of weighting that card by default, which is trivial (the entire thing is based on generating weightings for how likely it is to pick a card).

[onigame]

It depends on what the actual problem really is. If the problem is the generator returning some cards more often, then you're right (and some big scale testing is needed to verify this, find the reason and fix). But it may well be that the problem is users seeing some card (most often a card they don't like, and Familiar is not a popular card) as appearing very often while in reality it doesn't. I also sometimes feel that Iso is throwing Possession or Mountebank on me every second game.

Onigame has readily admitted that his set generator includes some cards more often than others. He has no plans to fix it. It's an integral part of the algorithm, apparently.

EDIT: Here's the relevant post.

You're misinterpreting my response.  I certainly intend to figure out which cards are chosen more often than others and adjust the algorithm so that the respective probabilities of individual cards getting selected are not egregiously out of proportion.  For example, right now I believe Governor gets chosen waaaay too much, and it's because the algorithm thinks it's a very versatile card that improves a lot of layout.  This tells me that it's okay for Governor to be chosen a bit more frequently than the other cards, but not to the extent to which the generator is doing so.

What my point was in the thread is that I'm not going to enforce a hard limit that constrains each card to appear exactly the same number of times as any other card.  I want to have cards roughly chosen equally, yes, but I'm not going to sacrifice playability and fun for what I consider to be an arbitrary restriction.

The goal is to have the generator create fun sets.  It's pretty clear, from popularity rankings and such, that "fun" is antithetical to strictly making everything precisely fair.

The hard part is that I want to affect these biases by changing the underlying parameters to the algorithm, rather than just have a blanket statement like "you're choosing Governor too much, so put a special case for Governor that makes it get chosen less."

[LastFootnote]

Ah, I see. Sorry for misinterpreting your response and thanks for clarifying.

[onigame]

The Goko card picker, as is, will throw up Monument super often, almost always has about three Attacks in it, and almost never has Pirate Ship.

Why would the Goko kingdom picker bias anything? It's specifically billed as 'true random' kingdom selection, so it would have to be a failure of their random number generator, if there's any kind of bias there. I'm a firm believer in artificial kingdom selection (thus all of the artificial selection options in my card picker program on inprogressgaming.com), but in my opinion it is critical that *all* players understand (and agree on) the selection criteria that will be used for any (rated) match. Goko should/would need to be very open about how they are selecting kingdoms *if* they were to be intentionally using anything other than true random.

Right now the plan is that "Pro" games on Goko will use uniformly random, with the rules as published, but Casual games and single-player games against the computer will use some form of my generator's algorithm.  As for being open, my code is open source and freely available if anyone wants to decipher how it selects "kingdoms".  Also, the "Deck Builder" option to fill out the rest of the deck with random cards will use my generator initially, but I'm lobbying for a setting that lets you adjust the settings for that, including changing
it to "uniformly random" if you want.

Personally, I don't see why having something be rated necessarily means that the kingdom card selection has to be uniformly random.  You could have every game be the "First Game" recommended set of 10, and the ratings would still be valid -- they'd just be ratings for how well a player plays with that specific set.  I think for a rating to be completely fair, the necessary attributes should be that an individual player has no control over what cards are picked, and that a player can't refuse to play a game without it counting as a resignation.

Right now, on isotropic, a player gets to look at the kingdom before deciding if they want to play a rated game with that set.  Suppose a player hates Saboteur and chooses to never play a rated game if Saboteur is in the set.  In that case, isn't his rating based on a non-uniformly random selection?

I feel that anyone who insists on uniformly random being some sort of "true basis" of Dominion skill is, to some extent, deceiving themselves.  For example, they'll play a random setup where it's clear that the Chapel-Witch opening is going to win and suffer through a boring game and justify to themselves that at least it's "fair".  Or some setup where Big Money is 70% likely to win and it's just a matter of hoping you get $6s and $8s instead of $7s.

But, because of the large following on isotropic, it's clear that there are plenty of intense Dominion players who enjoy the challenges that come from having a uniformly-random chosen set, with strict rules in play.  So I think it's okay to have the "Pro" section provide them what they want.  I just don't see any reason why casual players have to be confronted with the occasional crazy-but-not-fun setup where everything is an Attack or there aren't any +Actions or everything costs $4 or less of whatever (unless they want to).

[Beyond Awesome]

I agree with you on your points. I am not a fan of veto mode on isotropic, and I will play any kingdom regardless. On iso, I used to play veto mode a lot since most players play that way. I do like how on Goko it is pure random, and in pro games you have no clue what the kingdom is until you play it.

[Donald X.]

Personally, I don't see why having something be rated necessarily means that the kingdom card selection has to be uniformly random.  You could have every game be the "First Game" recommended set of 10, and the ratings would still be valid -- they'd just be ratings for how well a player plays with that specific set.  I think for a rating to be completely fair, the necessary attributes should be that an individual player has no control over what cards are picked, and that a player can't refuse to play a game without it counting as a resignation.

I think it would be fine to rate games with only very broad limits on what cards are used - specifically, letting you pick "use X cards from expansion Y" seems fine (maybe X can't be more than 5 for the small sets for it to be rated?). People would use that setting to get games they thought they had an edge in, but I don't see it screwing things up. I mean it screws it up for people who want to play with Alchemy but always force 10 Prosperity because they think they do the best there. But people like to play with the new cards, newly purchased or new for everyone, and it seems sad to push them out of that if they care about being rated on the pro board. I guess they are already pushed out by not wanting to take risks with new cards.

If you can pick specific cards then you can either do the King's Court / Masquerade / Goons trick that one guy did on isotropic when such games were rated (or some new thing if there ever is one), or you can just play Duke games against the bots until that's fixed (or whatever card people find out the bot is bad with). So I think those shouldn't be rated for sure. It would make the "casual" rating way more meaningful if picking specific cards didn't get you rated there either.

Using the uh "fun set of 10" generator with default settings seems like fair game for rating. Engines will be more common I guess? But like, let's say Joe uses the generator because he wants engines that he's good at, and Sam says all-Prosperity for me (in this new world of that not forcing Colony), and Fred says who cares, shut up and deal. I don't feel like now the pro board is a joke (unless the rating system sucks as some people think, but that's a separate issue). You have to be pretty good to crush people consistently in whichever of these formats.

[ftl]

I think I'd dislike having games with this set generator be the default. Dunno, maybe I'm wrong, maybe I'd like it, but I'd be worried if this became the default that everyone plays.

Variety is fun - variety in what sorts of games happen. Some games are engine-y and some are BM, sometimes you use all the cards and sometimes you just have to realize that you only need one of the 10, sometimes the games are ugly and sometimes they're clever, sometimes everyone picks different strategies and sometimes there's one obvious strategy with minor tweaks that everyone does, sometimes there's no attacks and sometimes every other card seems to be an attack.  This seems to be aiming for balance within a single game - make sure you have a balanced set! - instead of balance over a bunch of games (sometimes you'll get balanced sets, sometimes you'll get ones which are unbalanced in various ways).

Maybe it's just trying to solve a problem I don't really have. I just don't find "sometimes it's a Witch-BM set" to be a problem. Play that one, then play again and it'll be something different, I don't really want to exclude them. I'm not "suffering through" those, there's nothing wrong with a straightforward game. Or with a crazy one. As long as they don't happen too often - which they don't, even in full-random! - they enhance the game and don't detract from it.

It's not even that I have a particular attachment to "plain random". But I don't think I like the design philosophy of this set generator...

(And it's sort of strange and more than a bit rude to just decide that everyone who disagrees with you is just deceiving themselves. Seriously now...)

[Beyond Awesome]

I think I'd dislike having games with this set generator be the default. Dunno, maybe I'm wrong, maybe I'd like it, but I'd be worried if this became the default that everyone plays.

Variety is fun - variety in what sorts of games happen. Some games are engine-y and some are BM, sometimes you use all the cards and sometimes you just have to realize that you only need one of the 10, sometimes the games are ugly and sometimes they're clever, sometimes everyone picks different strategies and sometimes there's one obvious strategy with minor tweaks that everyone does, sometimes there's no attacks and sometimes every other card seems to be an attack.  This seems to be aiming for balance within a single game - make sure you have a balanced set! - instead of balance over a bunch of games (sometimes you'll get balanced sets, sometimes you'll get ones which are unbalanced in various ways).

Maybe it's just trying to solve a problem I don't really have. I just don't find "sometimes it's a Witch-BM set" to be a problem. Play that one, then play again and it'll be something different, I don't really want to exclude them. I'm not "suffering through" those, there's nothing wrong with a straightforward game. Or with a crazy one. As long as they don't happen too often - which they don't, even in full-random! - they enhance the game and don't detract from it.

It's not even that I have a particular attachment to "plain random". But I don't think I like the design philosophy of this set generator...

(And it's sort of strange and more than a bit rude to just decide that everyone who disagrees with you is just deceiving themselves. Seriously now...)

Everything that ftl just said. To be honest, I don't mind cursing games or games without +actions/lack of villages. You need to learn to adapt. My biggest concern is that this set generator will lead to almost every game being enginey, and to me, that would be a bit boring. There are many strategies to Dominion, and it would be unforuntate to have these almost never show up. And, for the record, I enjoy engine games.

[Davio]

Set generators bias promos way too much anyway.

But there's two sides to this coin: On the one hand it's cool that playing online enables access to cards that have a limited amount of physical copies. You could also say that because there aren't many physical copies and they're promo items that exist a bit outside of the regular Dominion space, people would not have come into contact with them as much. And the generator should reflect this.

I just think it's weird that with Dominion's online play possibilities the promos have become an integral part of Dominion space. I know the strength of BM-Envoy and I don't even have that card at home! I will also be the first to recognize that online play isn't the same as offline play and it's cool to leverage this kind of added value.

I'm still a bit clueless as to my own point of view, but I thought it would be at least good to let you guys think about it.

[Lashof]

I've found that when I try to play "pro" games on goko, I can't seem to get it to randomize shelters/plat-colony.  Is there anything special I need to do, or have I just had a weird string of province/estate games (I don't own all sets, just prosp + dark ages + 1/2 seaside, so one of the two should come up fairly often)

[Beyond Awesome]

I've found that when I try to play "pro" games on goko, I can't seem to get it to randomize shelters/plat-colony.  Is there anything special I need to do, or have I just had a weird string of province/estate games (I don't own all sets, just prosp + dark ages + 1/2 seaside, so one of the two should come up fairly often)

You can just click on the shelters or colony button and that should do the trick. But, I would say about 2 out of every 10 games for me are shelter games and 1 out of 15 are Colony games (maybe because Prosperity is a smaller set than DA), and about 1 out of 40 games seem to be both Colony and Shelters. If you play more games and pure random, you will see Shelters and Colonies.

[onigame]

(And it's sort of strange and more than a bit rude to just decide that everyone who disagrees with you is just deceiving themselves. Seriously now...)

I didn't decide that everyone who disagrees with me is deceiving themselves.  Just those people who disagree with me on whether imbalanced sets are fun for Dominion players in general.

I don't have a problem with playing the occasional game where BM is clearly dominant.  I don't even have a problem with playing many games where BM is slightly dominant.  I also don't have a problem with playing the occasional ugly game where you need only one of the 10 cards to win ... the problem is that with a wide pool of cards to choose from, those ugly games show up too often!

Roughly I would characterize kingdoms (sets of 10 cards) into these categories:
A. There's three or more viable strategies, with approximately equal likelihood of winning.
B. There's two or more viable strategies and maybe some that are obviously inferior to
 an experienced player but not to a casual player.
C. There's only one viable strategy for an experienced player.  Casual players who don't
 see it will lose.  There are some nuances to executing the strategy that will distinguish
 the experienced player from the expert player.
D. There's only one viable strategy for the expert player.  The winner will come down to
 whomever is luckiest with the shuffle (and maybe seating order).

What I would call a "fun" set is a set that is in the A or B categories.  I certainly see the C category as personally interesting, but I would only be willing to play that with experienced players -- C category games are not fun against casual players.  I don't understand why anyone finds D category games fun, but perhaps expert players will claim that D category games don't exist and that every game is really a C, I'm just not expert enough to see the differences.

When I called some players "delusional" I'm referring to people who believe that it's important to have D category games to get a full experience of Dominion, and that if you don't include them "you're not really playing Dominion."

To make an analogy, I enjoy playing Super Mario Brothers going through every level and collecting every coin.  Most casual players, including most (but not all) of my friends would not find that fun.  If I insisted that my way of playing was the only way to fully enjoy Super Mario Brothers and that everyone should play that way to get the full experience, I think my friends would be fully justified in calling me delusional.

On the other hand, I would be very sad if it was impossible or very hard to play Super Mario Brothers the way I enjoy playing it -- I feel the opportunity should be there for experienced players who do enjoy playing things to a much thorough detail than a casual player would.  This is why my set generator has a master "balance" setting now -- put it at zero and you get an unbiased uniformly-random set.  Put it at -1 and you're more likely to get C and D sets and fewer A and B sets.  The default setting is 1 (more A and B, fewer C and D) because I think that's what casual players will enjoy the most.  And I do believe that "Pro" games on Goko should use it at zero.

It's possible that I might be wrong about that assumption (what casual players like).  Another thing that I am trying to get Goko to do is to have a way to get user feedback on kingdoms.  A simple way to start would be, after a game, let the player give a simple "thumbs up / thumbs down" on whether they liked the kingdom they just played.  I'd then adjust the generator to compensate.

Having said all that, I do believe that my current implementation of the Set Generator weights engine games too high and I am trying to find ways to lower it.  There are many fun sets that don't have an engine and I don't think my generator is finding them frequently enough.  (It's also worth noting that since my generator is probability-based, every possible set can be theoretically generated -- so I'm really tweaking frequencies, not absolutes.)

[ftl]

... the problem is that with a wide pool of cards to choose from, those ugly games show up too often!

I would disagree with you there. I just went to iso and flipped 10 sets of full-random cards, I got just one "ugly" set out of those...

Roughly I would characterize kingdoms (sets of 10 cards) into these categories:
A. There's three or more viable strategies, with approximately equal likelihood of winning.
B. There's two or more viable strategies and maybe some that are obviously inferior to
 an experienced player but not to a casual player.
C. There's only one viable strategy for an experienced player.  Casual players who don't
 see it will lose.  There are some nuances to executing the strategy that will distinguish
 the experienced player from the expert player.
D. There's only one viable strategy for the expert player.  The winner will come down to
 whomever is luckiest with the shuffle (and maybe seating order).

What I would call a "fun" set is a set that is in the A or B categories.  I certainly see the C category as personally interesting, but I would only be willing to play that with experienced players -- C category games are not fun against casual players.

Man, A and B category games aren't any fun against casual players either. You pick a strategy and win, they flounder with strategies and don't put together anything reasonable.

D is the only way to make a big skill disparity fun, because otherwise the game is boring, easy wins aren't any fun.

C is a lot of fun too - it feels good to be able to make very small changes and see a difference in results. (I remember at one point on Goko I was testing what the bots did, and played some games where I deliberately set up there to be nothing to do except smithy-money. It was pretty cool to see how just by being slightly better with my duchy-dancing how I could get a noticeably above 50% winrate even on a board where literally all either of us bought was a smithy, maybe a second smithy depending on draws, money, and green.)

I don't understand why anyone finds D category games fun, but perhaps expert players will claim that D category games don't exist and that every game is really a C, I'm just not expert enough to see the differences.

I'm no expert, you're probably better than me. But at the very least, D category games are fun in the same way that Fluxx or gambling or straight-up games of chance are fun. That with a dash of knowing you can raise your chances a bit in the long run with experience and playing better.

Dominion can be either a game of basically pure skill, basically pure chance (assuming both players are familiar with basic strategy), or anything in between depending on the board, and that's pretty cool. 

And in addition, the presence of C-D games MAKES A-B games better. If you've a priori excluded "Boring optimal strategy both people go for" from the strategy space, there's no longer a decision to be made of "hmm, is this a set where I should get clever or a set where I should buy courtyards and that's it". Games where the thinking goes "Hmm, I thought it was a D game but it turns out you can do this clever thing" are infinitely more satisfying than "Hmm, looks like a boring set with nothing going on; but since those are effectively excluded by the set generator, I'm clearly missing something."

When I called some players "delusional" I'm referring to people who believe that it's important to have D category games to get a full experience of Dominion, and that if you don't include them "you're not really playing Dominion."

I think the presence of category D games makes Dominion a better game than without them; so yes, I think people who miss out on them ARE missing "the full experience of Dominion".

I mean, obviously they're still playing Dominion, any method of picking sets is fair game.

To make an analogy, I enjoy playing Super Mario Brothers going through every level and collecting every coin.  Most casual players, including most (but not all) of my friends would not find that fun.  If I insisted that my way of playing was the only way to fully enjoy Super Mario Brothers and that everyone should play that way to get the full experience, I think my friends would be fully justified in calling me delusional.

On the other hand, I would be very sad if it was impossible or very hard to play Super Mario Brothers the way I enjoy playing it -- I feel the opportunity should be there for experienced players who do enjoy playing things to a much thorough detail than a casual player would.  This is why my set generator has a master "balance" setting now -- put it at zero and you get an unbiased uniformly-random set.  Put it at -1 and you're more likely to get C and D sets and fewer A and B sets.  The default setting is 1 (more A and B, fewer C and D) because I think that's what casual players will enjoy the most.  And I do believe that "Pro" games on Goko should use it at zero.

See, I'd think the analogy is flipped. Playing every game with the set generator is going through every level and collecting every coin. You're making every single game a clever game with lots of strategies to do. Playing full-random would be playing some levels collecting every coin, but playing some levels just breezing through doing the minimum to get to the end of the level.

[Toskk]

Using the uh "fun set of 10" generator with default settings seems like fair game for rating. Engines will be more common I guess? But like, let's say Joe uses the generator because he wants engines that he's good at, and Sam says all-Prosperity for me (in this new world of that not forcing Colony), and Fred says who cares, shut up and deal. I don't feel like now the pro board is a joke (unless the rating system sucks as some people think, but that's a separate issue). You have to be pretty good to crush people consistently in whichever of these formats.

Yes, essentially all of the artificial selection methods I've seen/used tend to at the very least slightly prioritize 'engine' games, or at least the presence of a +2 action card in the set (even if there isn't any great use for it with the specific other cards). Many of them also tend to prioritize a +buy or gain x card. I agree that these two options alone aren't likely to give any specific player much of an artificial advantage.. certainly less than being able to select specific expansions would, anyway.

If Onigame is correct, and Goko is going to implement artificial selection options, I'd strongly recommend that they allow them for all rated game types, along with combining all matches generated in this manner into a single rating system (i.e. combine 'pro' games with 'casual' games when their kingdom selection method is the same as that used for 'pro' games).

As for the Pro board.. currently it is indeed a 'joke', simply due to the ability of players to disconnect instead of being counted as a loss. With that fixed (and of course all ratings reset), the second piece comes down to just how meaningful the rating system itself is.. which I've been arguing is a problem for quite some time now. The problem, as I keep trying to explain, is the range/accuracy measurement that accompanies the actual rating number. In short, as total number of matches increases, the overall range/accuracy of player ratings is not decreasing adequately, at least not enough to mask the fact that some percentage of games are going to come down to luck rather than skill (i.e. the win/loss had no meaning, regardless of player rating). For example, my current rating is approximately 7200 on Goko, with roughly 900 matches played, all against bots using true-random kingdom selection. A win vs. a bot will give me anywhere from 7-20 points, and a loss will cost me 195-210 points. This means that over the span of 30 matches, my rating could be ~6500 or it could be ~7500. ~95% of the time it'll stay within a range of 6800-7400. So a 600-point rating interval could well have no meaning whatsoever when it comes to determining win/loss probability vs other players, yet the number of points awarded/lost for each player will vary a considerable amount based on that 600-point swing, when it should vary essentially none (once a sufficient number of matches have been played).

[Rabid]

A. There's three or more viable strategies, with approximately equal likelihood of winning.
B. There's two or more viable strategies and maybe some that are obviously inferior to
 an experienced player but not to a casual player.
C. There's only one viable strategy for an experienced player.  Casual players who don't
 see it will lose.  There are some nuances to executing the strategy that will distinguish
 the experienced player from the expert player.
D. There's only one viable strategy for the expert player.  The winner will come down to
 whomever is luckiest with the shuffle (and maybe seating order).

I would be really interested to see a few example type A sets please?
I think they are much harder to make than you think.

[jonts26]

A. There's three or more viable strategies, with approximately equal likelihood of winning.
B. There's two or more viable strategies and maybe some that are obviously inferior to
 an experienced player but not to a casual player.
C. There's only one viable strategy for an experienced player.  Casual players who don't
 see it will lose.  There are some nuances to executing the strategy that will distinguish
 the experienced player from the expert player.
D. There's only one viable strategy for the expert player.  The winner will come down to
 whomever is luckiest with the shuffle (and maybe seating order).

I would be really interested to see a few example type A sets please?
I think they are much harder to make than you think.

I agree. Most likely, you'll get game C. There tends to be one good overall strategy, but there's a lot of wiggle room on how to play it, what support cards you get and what you prioritize, etc. That's where skill is separated out.

Occasionally I see two strong yet viable options, but three is incredibly rare.

[Davio]

I'm pondering: Does the set generator decouple from "engine" or "BM" terminology and just asses the number of strategies picking either one?

I'll try to explain what I mean.

Stef posted an interesting topic the other day where Courtyard-BM was very competitive against the engine he would build (although the jury is still out on that one) and it even depended on what the opponent would do. So there you have a B-type game with both engine and BM as competitive strategies.

Just because a game has a possible BM strategy that can go toe to toe with an engine doesn't make it boring. Correctly assessing whether BM will beat an engine or not is also part of being an expert player.

[Beyond Awesome]

It's possible that I might be wrong about that assumption (what casual players like).  Another thing that I am trying to get Goko to do is to have a way to get user feedback on kingdoms.  A simple way to start would be, after a game, let the player give a simple "thumbs up / thumbs down" on whether they liked the kingdom they just played.  I'd then adjust the generator to compensate.

I do want to say, I like your user feedback idea.

[onigame]

A. There's three or more viable strategies, with approximately equal likelihood of winning.
B. There's two or more viable strategies and maybe some that are obviously inferior to
 an experienced player but not to a casual player.
C. There's only one viable strategy for an experienced player.  Casual players who don't
 see it will lose.  There are some nuances to executing the strategy that will distinguish
 the experienced player from the expert player.
D. There's only one viable strategy for the expert player.  The winner will come down to
 whomever is luckiest with the shuffle (and maybe seating order).

I would be really interested to see a few example type A sets please?
I think they are much harder to make than you think.

I would say that most of the Recommended Sets of 10 that come with the game are type A sets.  Obviously depending on where you set the bar for "experienced" player is going to change whether a set is A, B, or C, but as a rough guess, I mentally set the bar at around a level 15 on isotropic.

And yes, absolutely they're hard to make.  I remember spending a lot of time testing the sets for Alchemy, Prosperity, Cornucopia, Hinterlands, and Guilds.  I got a bit lazy during Dark Ages and let the other playtesters do more of the work   But in general I'd consider them the pinnacle of "fun" sets -- all of them are sets that I'd be happy to play many games with.

[Polk5440]

A. There's three or more viable strategies, with approximately equal likelihood of winning.
B. There's two or more viable strategies and maybe some that are obviously inferior to
 an experienced player but not to a casual player.
C. There's only one viable strategy for an experienced player.  Casual players who don't
 see it will lose.  There are some nuances to executing the strategy that will distinguish
 the experienced player from the expert player.
D. There's only one viable strategy for the expert player.  The winner will come down to
 whomever is luckiest with the shuffle (and maybe seating order).

I would be really interested to see a few example type A sets please?
I think they are much harder to make than you think.

That is why for those of us who are interested in seeing more kingdoms from set A (or B) find a tool that can give them to you with higher likelihood than uniform random draws so valuable.

I am happy to see this generator being implemented in some fashion in Goko.

[WanderingWinder]

A. There's three or more viable strategies, with approximately equal likelihood of winning.
B. There's two or more viable strategies and maybe some that are obviously inferior to
 an experienced player but not to a casual player.
C. There's only one viable strategy for an experienced player.  Casual players who don't
 see it will lose.  There are some nuances to executing the strategy that will distinguish
 the experienced player from the expert player.
D. There's only one viable strategy for the expert player.  The winner will come down to
 whomever is luckiest with the shuffle (and maybe seating order).

I would be really interested to see a few example type A sets please?
I think they are much harder to make than you think.

I would say that most of the Recommended Sets of 10 that come with the game are type A sets.  Obviously depending on where you set the bar for "experienced" player is going to change whether a set is A, B, or C, but as a rough guess, I mentally set the bar at around a level 15 on isotropic.

And yes, absolutely they're hard to make.  I remember spending a lot of time testing the sets for Alchemy, Prosperity, Cornucopia, Hinterlands, and Guilds.  I got a bit lazy during Dark Ages and let the other playtesters do more of the work   But in general I'd consider them the pinnacle of "fun" sets -- all of them are sets that I'd be happy to play many games with.

If that's your bar, there's almost nothing of the witch-chapel sort you complain about.

[Beyond Awesome]

I know some people might disagree with me here, but I do think people should choose what sets get chosen for the randomizer. As it is, someone can buy Prosperity and have just Prosperity and base come up. In real life, many people only play with two sets at a time. So, I think that the option for full random or choosing which expansions should exist. Even in real life tournaments, often the kingdoms are from two expansions and are not full random of every expansion ever created. But, for pro ranked games, I am against the idea of being able to decide what kinds of cards or which cards come up. But, limiting the sets to specific expansions does not bother me.

[hsiale]

depending on where you set the bar for "experienced" player is going to change whether a set is A, B, or C, but as a rough guess, I mentally set the bar at around a level 15 on isotropic.

IMO that's too low to talk about "experienced" player. I am level 21 and still make many horrible mistakes when it comes to seeing viable strategies on the board (especially in noticing strong engine possibilities).

[onigame]

I know some people might disagree with me here, but I do think people should choose what sets get chosen for the randomizer. As it is, someone can buy Prosperity and have just Prosperity and base come up. In real life, many people only play with two sets at a time. So, I think that the option for full random or choosing which expansions should exist.

Absolutely.  It's why the expansion sets are at the top of my list in my ugly-looking HTML interface.  Almost certainly it's going to be the most requested feature in any set generator Goko uses.  The tricky bit with anything that has to face a casual player, though, is to not overwhelm them with options they don't understand yet.