Print

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

[Accatitippi]

...not to mention validity. 

I compiled these charts for fun a few months ago. Yesterday I was cleaning my desktop and found it and they were an interesting read so I updated them and thought maybe they'll be interesting to you too.
I believe I started working on this as a response to Dingan's excellent post: http://forum.dominionstrategy.com/index.php?topic=15011.msg582479#msg582479. The idea is the same (describe card distribution over time), but I'm analyzing it cumulatively instead of expansion-by-expansion.
I got somewhat carried over and I described a few quite hard-to-define cathegories (eg: remodelers, virtual Buy, Gainers...), so the data carry a big dose of subjectivity. But we're here for fun, so I hope you don't mind. I'll try to post the file so that you can have your own subjective version too.
Split piles are considered singularily, Prizes are ignored, upgraded cards (travellers, Urchin and so on) are counted as one card.
All graphs are cumulative, i.e. they describe how the total base of cards has changed over time. With the second edition, the 12 removed cards are thrown out of the analysis.
On the x axis are expansions, and promos released near them.

So, let me show you how I waste my time!

Here is the obvious one. Note that Actions are out of scale, I had to divide them by ten to make you see the interesting stuff happening.
You can clearly see alt-treasure and alt-vp being geared up in their respective expansions, and then Victory declining slightly as time goes by (but Landmarks and vp token cards are not considered), while treasures staying constant after Prosperity.
Donald has said several times that he had to lower the percentage of attacks in each set to cater to his fans, and we see the effects here. I didn't do a "nonAttack interaction" line but it would be interesting.


Here you can see the cards divided by cost. The "Other" cathegory is essentially made up of Potions, Debt, and Peddler. I translated them to coins to have a cleaner graph:

We can see how ers cards are steadily taking over the once much more common ers.
The other three lines have been very constant since Prosperity, after some turbulence due to Seaside and Intrigue's cheapness that was then balanced by Prosperity's spendiness. (in this chart Alchemy also boosts -costers at the expense of -costers due to how I translated potion)

Now, let's delve into the subjective charts:

Thinners. I divided them into mass thinners (can do 2 or more cards at once), single thinners, and remodels (can't usually trash without gaining). I put Trader as remodel and sentry as single. Events are counted as things. Total is the sum of all three groups.
Anyway, Total has gone somewhat up, and we're glad for that because thinners are cool.
After an early disappearance, Remodelers went back to being a common card paradigm, wich is nice.


This is probably my favorite chart.
I divided cards into terminals and villages (the uncounted cathegory being nonterminal actions, treasures, and victory). Some cards are only sometimes terminal (like Conspirator, Ironworks, or Nobles), and some are villages with significant twists (like Cotr, Throne Room, or Nobles). These make up the dotted lines.
Anyway, wow. 68% to 45% is a huge fall.
And on the subject of terminality:

Reliable Lab variants are cards that you can use to draw in most games. Unreliable Lab variants include Menagerie (*dodges rotten tomatoes*), Herald, Magpie, Vagrant, you get the general idea.
Menagerie should probably be reliable, and maybe Page/Warrior too, but that's what I did 6 months ago.
Anyway, look at all that terminal drawness drowning in nonterminalness. It seems that drawing one's deck has become much more about adapting to the perks of the Lab of the day than about lining up a Village and a Smithy.

A flexible gainer is something that lets you choose what you gain. Limited gainers are stuff that goes "gain a Gold", "gain a Magpie", "gain one of the trashed cards", you know who I'm referring to.
True +Buy says +Buy, virtual +Buy gives you cards according to how much money you can field, like Haggler, or it gives you extra turns to buy more cards using the money you (or someone else) can field, like Outpost. Black market and Ball also ended up here.
This one's quite messy.


Attacks.
I went with:
Discard attacks hurt the opponent's economy next turn, like Militia, Minion and Bridge troll (I bet you didn't expect that)
Topdecking attacks hurt the opponent's turn after the next, or their ability to draw useful cards next turn, like Rabble, Relic, and Spy.
Junking attacks and Trashing attacks are exactly what you expect.
Stuff like Torturer, Sea Hag and Ghost Ship got several classes, and also the Combined class.
Junkers got a big boost in the first expansions, and then kept their position as the most common attacks out there. Topdeckers and Trashers seem to be in decline, the latter of which is a shame if you ask me. Sab and Spy leaving unreplaced made a significant dent in the respective populations.


And lastly, for something a bit different, I wondered about power level in expansions.

I chose the quartile because I felt that that's where the set-definers started for most cost levels.
Unfortunately, I didn't think about finding over-/underrepresentation. I'll just say that Cornucopia's share is pretty impressive. Base seems to be suffering a bit, and Adventure and Dark Ages might be just benefitting from having more content than the others. Overall, the split is much more even than I expected.
Alchemy is not present because most of its cards are ranked separately.


And here is the bottom decile, which is where I felt laid the pity boundary.
Honorable mention to the absent: Dark Ages, Guilds, and Adventures, which interestingly are the most recent expansions.
Base takes a quarter of the pie, the other early expansions split the rest quite evenly. Which is surprising, I was expecting to see Intrigue dominate a bit more.
Removed cards make up for about a third of the cards, which means that we can look forward to a bottom decile smaller by 33%, how great is that! 

So, I hope this was interesting. Thanks for your attention.

No, I wasn't serious about the decile.

Edit: trying to attach the file. If you have suggestions for stuff to plot, I can try and make it happen if it's reasonably simple.

Edit2: New file with expansions in the correct order!

[LastFootnote]

Cool! Nice to see the evolution of Dominion over time.

[JW]

Great charts!

Thinners. I divided them into mass thinners (can do 2 or more cards at once), single thinners, and remodels (can't usually trash without gaining). I put Trader as remodel and sentry as single. Events are counted as things. Total is the sum of all three groups.
Anyway, Total has gone somewhat up, and we're glad for that because thinners are cool.
After an early disappearance, Remodelers went back to being a common card paradigm, which is nice.

What does "events are counted as things" mean? Having Bonfire, Plan, and Donate increases the odds that there will be a way to thin on a given board. But there's no spike in "total thinners" for Adventures even though it had Bonfire, Plan, and Trade (two of which thin, plus four Kingdom trashers, three of which thin).

Are you counting all of the events as cards, regardless of whether they thin or not? Because events are added in addition to the normal Kingdom cards, even if only 1 Event out of 20 thinned (lower than the ratio among all cards), this increases the odds that a Kingdom would have a thinner, compared to pre-Adventures (when there were no events).

A very rough approximation is to add events that trash to the numerator and denominator (of Cards That Trash / Total Cards) and to ignore events that don't trash altogether.

An improvement in accuracy is that events that don't thin reduce the chance that the events that do thin will show up because you may have already hit the 2 event/landmark limit.  As of Empires  the total number of Landmarks/Events has increased a lot so the recommended 2 event/landmark limit is often met and so adding about 0.7 * events that trash to the numerator and denominator would be more appropriate. More details here

[J Reggie]

Why did you switch Cornucopia and Hinterlands?

[Dingan]

Someone has free time on their hands.. 

This is awesome!  Always love me some data.  Way prettier and more insightful than my article, I think.

Wow, hard terminals are going way down.  I wonder if that's because they're in any way "less fun", and hence DX makes less of them, or something.  Idk.  You would think there would at least be some reason though.

It's pretty cool that Qvist rankings are more-or-less evenly distributed across all sets (and they look like they're even more even if you normalize the pie slice sizes by expansion size).  A true testament to DX's ability to balance cards, even from the very beginning (outliers like Chapel aside).

[Accatitippi]

Why did you switch Cornucopia and Hinterlands?

Haha, I just noticed that they are switched, it was probably an error when I set up the order constants. My bad. Goes to show, don't trust these charts.
Anyway, I fixed it, I'll update the file so that you'll be able to see the new charts. (generating, uploading, and linking images is a lot of work, and the difference is minor 

Here are a couple of bonus thingies: what proportion of each expansion made the top quartile/bottom decile? The orange line is the expected value (it's actually a bit off from the average because I had to make one rounding per cost category. The real mean values are 26% and 11%.

Cornucopia here is the big outlier, but check that, Adventures is as bad as Base!


Alchemy aces this one with a whooping 1/2 terrible cards. Anyway, with two huge sets at 0, all the other are unsurprisingly above the average.

JW, thanks for the suggestions! Yes, I counted all card-shaped-things as equal for those totals. I'll look into doing something more precise, but it would require a big overhaul of the formulas, and I'm afraid I've spent too much time already on this project for the time being.

Edit: I posted the same thing twice, whoopsie

[sorawotobu]

You counted Relic as a topdeck attack but Minion as a discard attack? How are they not the same? I think counting them as topdeck attacks makes sense.

[Accatitippi]

You counted Relic as a topdeck attack but Minion as a discard attack? How are they not the same? I think counting them as topdeck attacks makes sense.

When I posted this I knew that my category choices couldn't be shared by everybody, and that's why I added all the disclaimers about subjectivity. But as a matter of fact, I'm quite confident about that one choice.
In absence of draw, Minion hurts your next hand, rather than the one after it, like Relic does.
If you have draw, Relic hurts your next draw, just like Fortune Teller and Sea Hag. Minion does nothing like that, even though the 4-random-cards hand hurts your likelihood of starting your engine more than say Militia.

Anyway, I realized that I ignored Warrior and Soldier in that graph, so here's an update:

[Witherweaver]

[sorawotobu]

In absence of draw, Minion hurts your next hand, rather than the one after it, like Relic does.
If you have draw, Relic hurts your next draw, just like Fortune Teller and Sea Hag. Minion does nothing like that, even though the 4-random-cards hand hurts your likelihood of starting your engine more than say Militia.

So the deciding factor is not the way in which they hurt you but rather the timing of it? Because a hand that has been hit by a Minion attack is going to be exactly as good on average as a hand that has been hit by Relic, it's just four random cards (as opposed to the three out of five with Militia for example).

[Chris is me]

In absence of draw, Minion hurts your next hand, rather than the one after it, like Relic does.
If you have draw, Relic hurts your next draw, just like Fortune Teller and Sea Hag. Minion does nothing like that, even though the 4-random-cards hand hurts your likelihood of starting your engine more than say Militia.

So the deciding factor is not the way in which they hurt you but rather the timing of it? Because a hand that has been hit by a Minion attack is going to be exactly as good on average as a hand that has been hit by Relic, it's just four random cards (as opposed to the three out of five with Militia for example).

Relic is like a Bureaucrat that always hits. Relic hurts your draw, which is only a handsize reduction if you don't draw that turn.

Minion is like an Urchin without the choice, basically. Minion is always*** a handsize reduction.

Seems clear to me?

[Beyond Awesome]

I see the logic in this, but I would still consider Relic a hand-size reducer.

[Mic Qsenoch]

For me at least, Relic is an attack that gives my opponent their -1 card token.

[DG]

If you want another graph, how about cards with player interaction that are not attacks?

[markusin]

Both Relic and Minion reduce the number of cards you get to hold in your hand. The difference between those cards and topdeck attacks like Rabble or Fortune Teller becomes apparent when you play Cellar/Warehouse/Oasis/etc. Bureaucrat and Ghost Ship fit into both these categories.

[Accatitippi]

Both Relic and Minion reduce the number of cards you get to hold in your hand. The difference between those cards and topdeck attacks like Rabble or Fortune Teller becomes apparent when you play Cellar/Warehouse/Oasis/etc. Bureaucrat and Ghost Ship fit into both these categories.

Good point, but sifters are out only in a minority of games, and for all the other games Relic plays much more like Fortune teller than like Minion.
Ghost ship got both categories, Bureaucrat didn't because in a vacuum (no sifters, dual types, and Baron) it's doesn't really hurt your current hand.
I know it's not perfect, but they had to go somewhere and that's where I felt they fit best. You're free to imagine the discard line a tad lower and the topdeck line a tad higher, or even a new curve labeled -1 card token.
(this change is actually trivial to apply to the spreadsheet, I encourage you to try that - I'm sure you'd find many other debatable classifications which would benefit from debate)

[theory]

This post is awesome!