Germany NRW
Sprechen Sie Talk?

Is it allowed to mix cards? What I mean is does it overpower some cards or make others weaker? Say mixing the Minor Improvements from basic, complex and interactive decks together and then drawing the 10 for each player and then reduce down to 7 to play? Can this also be done with Occupations?

Eddy Richards
Scotland Allanton Duns

Yes to everything!
Experience will tell you if you find some cards overpowered, underpowered or boring  just leave these out if you want to. But if you are newish to the game, just try any combination you like.
We usually use a mix of two decks (E and I, for example) for both Minors and Occupations, typically choosing 7 from 10. But other groups use different ways of mixing things up. There is no right and wrong!

Tibs
United States Amherst Massachusetts

They should all work okay together but I recommend removing Minor Improvements 083, 089, 094 (the roads) because their mechanic is based on other people playing them. With more than just the I deck, they become a bit too rare for this mechanic to be effective.
In this case, it translates to the weaker roads being overpowered (specifically: why not play a 1wood road for two points when there's a reduced chance that another player will play a trump?)
Certainly this is not viable at all with more than E, I, and K.



Yes, mix it all together. Draw 7 occupations and 7 minor improvements at the start. Pick one of each and pass the cards on to your neighbor. Repeat this until everybody has 7 occupations and 7 minor improvements. This balances out a lucky draw and offers great opportunity for strategy and combo's. That way any strong combination of cards is mostly due to great picking skill. If all works out of course... :)

Germany NRW
Sprechen Sie Talk?

Thanks for all the tips.

Lawcomic
United States Tennessee
The stuff that dreams are made of...

What "mechanic" do you use if you mix all the cards together for dealing with player limits on the occupations?
We tend to just discard and draw new cards until we have a complying hand/draft pool.
Anyone do it differently?

Todd Parker
United States Denver Colorado

I recommend buying 3 copies of the game, and keep 3 different decks, one for each player count. Anything to speed up the game setup so you can get more games in.

Dave Brown
Canada Toronto Ontario

kungfro wrote:
In this case, it translates to the weaker roads being overpowered (specifically: why not play a 1wood road for two points when there's a reduced chance that another player will play a trump?)
In practice this isn't true.
1 wood for 2 pts is good, but it is certainly not overpowered.
Here are stats on the Wooden Path when played with EFrIKWm decks mixed together (4er):
115 Iwoodenpath  4.49  Plays (316 / 440 = 71.8%)  Wins (101 / 316 = 32%)  PWR = 3.3 [3.7,2.9] [3.8,2.8]
It is ranked as 115 out of 537.
Interestingly it is drafted quite late (4.5 pick on average) so it usually ends up in good player's hands which is probably why it wins above average (32% compared to the average win of 27%)
The Clay Path, which you have surmised would be not as powerful in this format is actually better than the Wooden Path:
92 Iclaypath  3.59  Plays (325 / 433 = 75.1%)  Wins (106 / 324 = 32.7%)  PWR = 3.5 [3.9,3.1] [4.0,3.0]
This is likely because clay is easier to get than wood.
And finally here is the Paved Road:
327 Ipavedroad  5.08  Plays (160 / 444 = 36%)  Wins (55 / 160 = 34.4%)  PWR = 1.8 [2.2,1.4] [2.3,1.3]
Not very good as there is usually a better way to spend your stone.

Tibs
United States Amherst Massachusetts

But how do those stats compare to Ionly decks? To different player count? The probability that the paths are in the game is proportional to the number of players, after all, and my claim is that the number of decks in use is extremely relevant to these cards.
Certainly, the trump mechanic weakens with more decks. Using every published deck is going to make the trump mechanic nonexistent, which takes away a lot of the purpose of these cards.

Dave Brown
Canada Toronto Ontario

kungfro wrote: But how do those stats compare to Ionly decks? To different player count? The probability that the paths are in the game is proportional to the number of players, after all, and my claim is that the number of decks in use is extremely relevant to these cards.
Certainly, the trump mechanic weakens with more decks. Using every published deck is going to make the trump mechanic nonexistent, which takes away a lot of the purpose of these cards.
Sure, though even with 4 other decks I still see trump happening. It's a bigger deal when it does.
You originally said that they would be overpowered. That simply isn't the case. Are they less interesting? Maybe. If we are removing cards for not being interesting there are at least 100 I would put ahead of the roads.
I don't have the data, but I would guess that wooden path is quite weak in an I only game. I don't think that is a good thing.

Tibs
United States Amherst Massachusetts

I'm not convinced my claim of overpowered isn't the case, because I have no idea what the values you posted actually mean; and they aren't presented alongside competing data (i.e. only I deck).
Where can I see the raw data?

Dave Brown
Canada Toronto Ontario

Here you go:
http://playagricola.com/forums/index.php?topic=2677.195
Here is a thread explaining that thread:
http://playagricola.com/forums/index.php?topic=4165.0
The data from I deck only wouldn't make a difference to whether or not the roads are over powered when mixed in with more decks.
All it would tell you is whether they are more or less powerful in that format. My guess would be that the wooden path is less powerful in I deck only than in mixed. That does not make it over powered in mixed.
Also, you don't really need to go to stats to see that some cards will be better than others.
Take the ponds for instance. I will take a point or two and some food for free over a wood for 2 pts, even if those pts were guaranteed.

Tibs
United States Amherst Massachusetts

I mean, where are the raw data? This person is posting data that has been analyzed. I'd like to know how many times each card was dealt and drafted and won, in a more machinereadable format.
And having access to the I deck alone absolutely would tell me what I need to know. I would compare the "PWR" between the Ionly paths to the multideck paths while controlling for player number.
Also I'm not sure if PWR is the relevant stat here. It's proportional to the number of times the card was played and wound up winning divided by the number of games where it was available as a draft. I'd like to know how the number of times the card was played and won versus played and lost (w/x in that formula) changes with the number of decks and players.
Also I keep seeing discrepencies: 23 FRBoatswain  2.09  Plays (322 / 375 = 85.9%)  Wins (117 / 321 = 36.4%)  PWR = 4.5 [4.9,4.1] [5.1,3.9]
This stat says the card was played 322 times out of 375 games it was dealt, and win 117 times out of the 321 times it was played. Why is there a difference between 322 and 321? Is that one game a tie, or is something being incorrectly tallied?
Lastly, I don't even think the description of PWR is correct:
Quote: PWR is proportional to the percentage of games that C will appear in the winners' played cards. The larger the PWR, the more percentage of games you will see C in the winners' played cards. And when you play specifically draft 9 with 500 cards, PWR of Braggart = 5.5 literally means that you will see Braggart in winners' played cards in 5.5% of games.
If we imagine a card (called "Toy" for this example) that wins 100% of the times it's dealt, then we get: PWR = 100 * 1/7 * (played 100% it's dealt) * (wins 100% of the time it's played) = 14.2. The claim is that if you examine each winning player's card tableau, 14% of the times you look you will see Toy in the tableau.
But this can't be correct. The percentage of times this card appears in a game, and thus in the winner's played cards, is dependent on the number of cards in the deck. Toy will show up far more often when dealt from a 50 card deck than, say, a 5000 card deck.
So PWR cannot be equal to the percent of won games that include it, because the size of the deck of cards it's dealt from is not included in its calculation. PWR is proportional to (but not equal to) the number of games it's included in that it's won... but we know that value is 100% in the case of Toy, so we've demonstrated that the 1/7 constant is arbitrary [1].
[1] I suspect the author may think that 1/7 normalizes for the fact that players have 7 Occs and 7 MIs, but that's not reason enough to use 1/7. In truth each player results in 18 cards being drawn from the available deck (regardless of type) because of the 9and9 draft, so if any normalization constant should appear anywhere, it should involve 18.

J
United States Alexandria Virginia

Quote: [1] I suspect the author may think that 1/7 normalizes for the fact that players have 7 Occs and 7 MIs, but that's not reason enough to use 1/7. In truth each player results in 18 cards being drawn from the available deck (regardless of type) because of the 9and9 draft, so if any normalization constant should appear anywhere, it should involve 18.
Based on reading his comments I think he used 1/7th because in the games where this data is tallied there are 504 cards in the pool to be selected and 72 cards are initially in the draft (4*18).
When you play 4ER draft 9 with 500 cards (which is approximately decks EFrIKWm11), card C is dealt into (1/7) of games since each games has 72 cards dealt. Therefore PWR is the percentage of games that C will appear in the winners' played cards when using a deck of 500 cards. If you use a deck of M cards then you should expect to see card C in the winners' played cards in (500/M)*PWR percent of games.
(The factor 1/7 does the normalizing. Each 4ER draft 9 game has 9*2*4=72 cards dealt. The number 72 is 1/7 of 504. If we didn't include 1/7 in the PWR formula, then NL General would have a PWR of 43.4. That would mean that if NL General was dealt into EVERY game, you would see it in winners' played cards in 43.4% of games. When you use a pool of 500 cards, NL General will be dealt into 1/7 of all games, so you will only see it 6.2%)
kungfro wrote: If we imagine a card (called "Toy" for this example) that wins 100% of the times it's dealt, then we get: PWR = 100 * 1/7 * (played 100% it's dealt) * (wins 100% of the time it's played) = 14.2. The claim is that if you examine each winning player's card tableau, 14% of the times you look you will see Toy in the tableau.
But this can't be correct. The percentage of times this card appears in a game, and thus in the winner's played cards, is dependent on the number of cards in the deck. Toy will show up far more often when dealt from a 50 card deck than, say, a 5000 card deck.
So PWR cannot be equal to the percent of won games that include it, because the size of the deck of cards it's dealt from is not included in its calculation. PWR is proportional to (but not equal to) the number of games it's included in that it's won... but we know that value is 100% in the case of Toy, so we've demonstrated that the 1/7 constant is arbitrary [1].
Given that 100*72/504 = 14.2 (1/7) his description of PWR and your calculations for your Toy card are spot on given the conditions he's using for his data (Pool of 504, 4 player9 card draft. For Toy every game it's in one of those 72 cards it wins hence it win's 1/7th of the time or 14.2.

Dave Brown
Canada Toronto Ontario

Quote: This stat says the card was played 322 times out of 375 games it was dealt, and win 117 times out of the 321 times it was played. Why is there a difference between 322 and 321? Is that one game a tie, or is something being incorrectly tallied?
It's explained in how the stats are compiled. It's basically errors when the game was played (like if the card disappeared at the end from their board, it's not supposed to but apparently it does 1 in 322 times). Chris knows that his data collection method is flawed, but it's close enough.
Quote: [1] I suspect the author may think that 1/7 normalizes for the fact that players have 7 Occs and 7 MIs, but that's not reason enough to use 1/7. In truth each player results in 18 cards being drawn from the available deck (regardless of type) because of the 9and9 draft, so if any normalization constant should appear anywhere, it should involve 18.
PWR made sense at the time 7 or so years ago when it was conceived. There have been other ways put forward since then that will show other things better.
The key here though is that PWR is just a comparison. 5.5 or 3.1 or whatever doesn't matter in a vacuum. What matters is what it means in the context of the game and other cards.
If a card is played 100% of the time and wins 50% of the time, that card is too good.
I think it is preposterous that you can think that a minor that costs 1 wood for 2 points is overpowered and yet also think that a minor that is free and gives 2 points and 5 food is perfectly fine, or a point and 6 food, or 2 points and 3 stone every time you take DL, or a point, a place to sow veggies and a free sow action. And so on.
There are a ton of minors that are simply better than Wooden Path, even if it were a guaranteed 2 points.
The stats show that, and just looking at the cards will show that too.
You can take Wooden Path in its supposed brokenness. I will take Duck Pond. Or for that matter, Reed Hut, if we're going to talk about actual broken cards.

J
United States Alexandria Virginia

I was going to comment but I was 90% sure that this conversation had happened already so I went and dug up the thread:
https://www.boardgamegeek.com/thread/753324/mixingdifferent...
My general view on the issue is that the roads as they are are slightly underpowered anyway and only are par with most other cards when they win the bonus points so there's nothing really wrong with including them. I understand Tibb's point about them being "more reliable" but I think classifying them as "overpowered" is an overstatement.
Without a draft there were instances where multiple roads did appear in the game so it's not like it's an entirely safe play and anyway if you are playing with a draft the likelihood that a road is drafted in the first 4 picks I'd think is pretty low so players should probably have an idea anyway which road would win.

Tibs
United States Amherst Massachusetts

allstar64 wrote: Based on reading his comments I think he used 1/7th because in the games where this data is tallied there are 504 cards in the pool to be selected and 72 cards are initially in the draft (4*18).
Okay, but which 504? If it's a random 504 out of 600 possible, then the culling is artificial and the pool is actually 600, so 1/7 is no longer valid.
And what about games with different player counts and different card pool sizes? Instead of (1/7) shouldn't it just say (number dealt/number in pool) or something? That way it is always dynamically adjustable, and PWR could be compared between disparate player counts and deck sizes.
ad_hoc wrote: I think it is preposterous that you can think that a minor that costs 1 wood for 2 points is overpowered and yet also think that a minor that is free and gives 2 points and 5 food is perfectly fine.
Is there such a card? Certainly it's not "free." Certainly there are limitations on its playability: round, player count, prerequisites, etc. Wooden Path has none of these things.
allstar64 wrote: My general view on the issue is that the roads as they are are slightly underpowered anyway and only are par with most other cards when they win the bonus points so there's nothing really wrong with including them. I understand Tibb's point about them being "more reliable" but I think classifying them as "overpowered" is an overstatement.
That is something I had not considered: the paths by default are underpowered. But I still maintain that the risk/trump mechanic irreparably suffers with increased card pool size, so regardless of power the cards suffer from a fun depletion.
I'll have a refreshed look at this problem a little later.


