Dominion Strategy Forum
Dominion => Dominion General Discussion => Topic started by: GenericKen on May 13, 2012, 04:48:22 am
-
It seems bizarre that the annotated games on the site are such a pain to read and write. Why haven't we developed a standard notation system for dominion games? The isotropic logs are nice and (somewhat) comprehensive, but they're cumbersome. Most games don't last longer than 22 turns - we shouldn't have to scroll to review a game.
Here's my proposal. I've put maybe 40 minutes of thought into this, so I'll try to keep it short.
-Most of the information available in a dominion game is unnecessary. +buys are only important in some games. Cursers and trashing are only important in some games. The most important thing about a dominion notation system will be extensability.
-It almost never matters if you draw 5 coppers or if you draw 3 coppers, a silver, and an estate. It also almost never matters if you end the turn with extra + actions or extra + buys. For treasures, VPs, terminal silvers and villages, the only things that matter are what you buy, what you could have bought, and what's left in your deck.
-It almost never matters what order your action cards are played in. A lab chain is a lab chain. While the decisions made during the turn may be important to prolonging the engine's run that turn, they generally will not affect the next turn outside of what actions are played and what actions are discarded, which tells us what's left in the deck. If the order is important
Rough notation format:
<turn#>. $<coins>(-$<costReduction>), +[<cardsGained>(<top>)], -[<cardsTrashed>], @+[<cardsGivenToOpponent>], @-[<cardsTrashedByOpponent>], <#played><*onShuffle><KC played>(mode, reaction(mode)) #<#discarded><KC discarded> <*onShuffle>
Details:
-A * goes at the end of turn if we shuffle drawing our 5 cards for the next turn, and a * prefixes an action if it causes a shuffle in turn.
-When multiple gains/trashes are made, the most expensive should listed first.
-CSGP, EDPCol are copper, silver... and Estate, Dutchy... adb$ are +action, draw, buy, and coin.
Here's a standard doublejack:
http://councilroom.com/game?game_id=game-20120507-234742-e0c1d27c.html
Vineyard; PearlD; Look(Lookout); Vil; Potion, Joat; HoP, Gov, Lab, Tres(Treasury); GM
cassandre ; GenericKen
1. $4 +Joat ; $4 +Joat
2. $3 +Look * ; $3 +S *
3. $3 +S ; $4 +[Joat, S], -E, 1Joat
4. $4 +[S,S], -C, 1*Look, 1Joat ; $5 +Lab *
5. $3 +S, -C, 1Look ; $6 +G, 1Lab
6. $7 +[G,S], 1*Joat ; $1 +S, 1Joat, #1Joat *
7. $6 +GM ; $8 +[P,S], 1Joat
8. $7 +G, -C, 1Look, 1GM ; $5 +Lab, 1Joat *
9. $8 +[P,S], -E, 1Joat * ; $7 +G
10. $10 +[P,S], 1Joat ; $9 +[P,S], 2Lab, 1Joat *
11. $5 +T, -S, 1Look * ; $8 +[P,S], 1Joat
12. $7 +[D,E], 1GM ; $7 +D
13. $6 +[G,S], 1Joat ; $5 +[D,S], 1Joat *
14. $9 +P * ; $11 +P, 2Lab
15. $9 +P
Points: 29 ; 30
Here it is again w/ some unnecessary info stripped out ($ amounts when not overspent, 1s on single actions). I'm not sure if this makes it easier or harder to read:
cassandre ; GenericKen
1. +Joat ; +Joat
2. +Look * ; +S *
3. +S ; +[Joat, S], -E, Joat
4. $4 +[S,S], -C, *Look, 1Joat ; +Lab *
5. +S, -C, Look ; +G, 1Lab
6. $7 +[G,S], *Joat ; $1 +S, Joat, #Joat *
7. +GM ; +[P,S], Joat
8. $7 +G, -C, Look, 1GM ; +Lab, Joat *
9. +[P,S], -E, Joat * ; $7 +G
10. $10 +[P,S], Joat ; $9 +[P,S], 2Lab, Joat *
11. +Tres, -S, Look * ; +[P,S], Joat
12. $7 +[D,E], GM ; $7 +D
13. +[G,S], Joat ; +[D,S], Joat *
14. $9 +P * ; $11 +P, 2Lab
15. $9 +P
Points: 29 ; 30
And again with no played actions (shuffles should tell you the rate of draw):
cassandre ; GenericKen
1. +Joat ; +Joat
2. +Look * ; +S *
3. +S ; +[Joat, S], -E
4. $4 +[S,S], -C, *Look ; +Lab *
5. +S, -C ; +G
6. $7 +[G,S] *Look ; $1 +S, #Joat *
7. +GM ; +[P,S]
8. $7 +G, -C ; +Lab *
9. +[P,S], -E * ; $7 +G
10. $10 +[P,S] ; $9 +[P,S], *
11. +Tres, -S * ; +[P,S]
12. $7 +[D,E] ; $7 +D
13. +[G,S] ; +[D,S] *
14. $9 +P * ; $11 +P
15. $9 +P
Points: 29 ; 30
Another random one:
http://councilroom.com/game?game_id=game-20120406-224153-8d57b904.html
Haven, Pawn; Tunl(Tunnel); Envoy; Inn, Mine, Trib(Tribute), RS (Royal Seal); GM; Forge
gorney ; GenericKen
1. +Pawn ; +S
2. +Inn(shuf 2) * ; +Envoy *
3. +Envoy, Pawn(a,d), Inn * ; +RS, #S *
4. $4 +Tunn ; +Envoy
5. $4 +Tunn, #C * ; $7 +G, #RS *
6. $4 +Tunn, Inn ; +RS, #RS
7. +Haven, #P, #C ; $9 +GM(top), #RS
8. $4 +S * ; $9 +P, RS, #G, #Envoy
9. +S, Inn, Haven, #C ; +RS(top) *
10. +Inn(shuf 3), Pawn(a,$) * ; $9 +GM(top), #RS
11. $6 +Inn(shuf 1), 2Inn, Haven, #S ; $7 +G *
12. +[Tunn, 2G], *Inn, Pawn(d,$) ; $9 +P, #GM
13. +P, Haven, #Inn ; $7 +G(top)
14. +[Inn(shuf 4), G], Inn ; +[P, Trib(top)], #G
15. +[P, 4G], Inn, *Envoy, 2Inn, Pawn(d,$), #G ; $9 +P, Trib(2d,$2(G))
16. $7 +D * ; $9 +P, #Envoy, #RS
17. $6 +[D, 2G], Inn ; $6 +D
18. $7 +D ; +Tunn
19. $6 +D ; +[P,E], Trib(C,Haven)
Points: 35 ; 45
Unfortunately, Isotropic doesn't log what Inn discarded, so it's hard to know whether gorney was discarding coppers or green cards, and hard to know how much value was left in his deck.
This is just a rough idea. It would be nice to strike a balance between fidelity, ease of reading, and ease of writing (it's still a pain to transcribe, but if the rules are formal enough, we could write a parser).
Please contribute with any suggestions or amendments to my notation format.
-
Have you seen the notation sometimes used on the main blog (http://dominionstrategy.com/2012/01/20/guest-article-annotated-game-9/)? That presents a similar amount of information, but the tabular spacing and unabbreviated names makes it much easier to read.
One factor working against developing a good notation is that whilst Dominion is a fun game to play, it isn't so fascinating going over games played by others. It's less useful than for a game like Chess, as for Dominion the set-up changes all the time and knowing exactly how someone played a particular Kingdom once isn't going to do all that much to help you in the future.
On the occasions when you do want to go over a game, the log does a reasonable job. Apart from adding the missing information (what was discarded? what was passed with Masquerade?), for me the biggest improvement over a log would be a log "player", that allows you to step through the game turn by turn and action by action, displaying the information you get during a game plus extra details of the game state such as deck/discard pile contents for each player (along the lines of that currently provided for the logs on Council Room).
-
A log player would be a very nice thing to have.
I have the same detached feeling when I read Poker hand logs. You read it from top to bottom and don't feel any of the excitement that the players must have felt.
A log player somewhat helps with this. You can stop and ask yourself: "What will he do here?" ...click.... "Oh my, a reraise!"
Such a replayer where you can't immediately see the end result and can focus much more on how the game evolves and the choices the players are faced with would be nice.
-
I already mentioned this in another thread and tried to find a notation which clearly can log all cards without having a different notation for each card.
The goal was to be able to log games played IRL and to post solutions to puzzles here in the forum. I won't post all the notation here, but this is what I came for:
First I assigned a two-letter abbreviation to each card (and single-letter abbreviations for standard cards). Then I tried to figure out what all cards have basically in common, and that is manipulation of the location of all cards.
All is about drawing in the hand, discarding, putting into play, revealing and so on.
I assigned different brackets to each location, so I defined {draw pile} (hand) <discard pile> and |play area| and the opening bracket can define where a card comes from and the second where it goes to, examples:
TN{E) means I play a Tournament and draw an Estate (from draw pile into hand)
when notations refer to a specific card played I put that in these [brackets]
CE[(EX1>; {32D)] means I play a Cellar and discard (from hand to discard pile) an Estate, a Curse and a Copper to draw a Gold, a Silver and a Duchy.
For revealing I simply add °, e.g.
AD[{°AD,E,X>; {°31)] means an Adventurer revealed another Adventurer, Estate and Curse from the draw pile and discarded them and the Gold and Copper were revealed and go to my hand (note that the order drawn is not displayed because it's not that relevant).
Cards gained are displayed with an +, and trashed with an -, and other players are referenced with p1, p2 etc
+IG[p1+X>]> means I get an Ill-Gotten Gains (from supply to discard) and player one gets a Curse (from supply to discard).
The goal I tried to achieve (condensed logs without ambiguos meanings) were achieved, but I know that this is not a very good readable format. But if anyone still want all info about my logging system, I will post a full explanation here. Here what a log could look like:
5(1111E)0 |1111| +TN> (111EE)
0(111EE)6 |111| +2> (&2111E)
7(2111E)0 |2111| +3> (TN,111E)
2(TN,111E)6 |TN{E)111| +TN> (1&3211)
9(13211)0 |32111| +P> (111EE)
4(111EE)6 |111| +2> (TN,TN,1E&P)
11(TN,TN,1E&P)0 |TN{°P>[+FW)](FW},TN{1),FW{11)+E>,1111| +FL[{E-+TN>]>
-
The problem with a lot of these systems is that it's also not intuitive. Chess suffers from this problem, but if you're into chess you've already demonstrated a willingness to invest that kind of time to learn a notation system. Dominion, on the other hand, has a much broader audience to account for.
Also, Chess has only 8 pieces that you need to know about. Dominion has over a hundred.
I think the old blog system (where you selectively choose which Actions and game bits to annotate) is probably best, but it's a pain to write up.
-
Why excise all the actions? Sure, a Lab chain is a Lab chain (and that could probably be abbreviated), but I'm totally fine with Iso logs because you see exactly what the players do. Sure, it's the end result of each turn that matters, but it's the means to that result that a) is the whole point of playing the game b) shows who the better player is c) can be very satisfying to play and to see played. True, unless you're playing a Bank, you don't really need to know the exact makeup of the Coppers and Silvers thrown on the table, so I agree that certain things should be excised or abbreviated in notation. But to get rid of all the Actions just makes it all rather bland.
-
The problem with a lot of these systems is that it's also not intuitive. Chess suffers from this problem, but if you're into chess you've already demonstrated a willingness to invest that kind of time to learn a notation system. Dominion, on the other hand, has a much broader audience to account for.
Also, Chess has only 8 pieces that you need to know about. Dominion has over a hundred.
I think the old blog system (where you selectively choose which Actions and game bits to annotate) is probably best, but it's a pain to write up.
Gotta agree here. A while back I tried my hand at creating a notation system like one of the ones described here... and even from the simplest perspective possible, it was too complex to be human-readable. And I disagree with theory about chess's notation not being intuitive; every single line in a chess log consists of a move from one coordinate to another coordinate, and these coordinates are easily understood by anyone who took algebra in 9th grade.
The other notations in a chess log--captures and checks--aren't actually necessary to the system, but they add something for people who aren't going to set up a board (or can't follow in their heads).
Meanwhile, Dominion has at least a dozen possible "moves": take an action, buy a card, draw, trash, top-deck, gain, reveal, discard, shuffle, look, pass, react, I've likely missed some. Often you'll do several of these on the same turn. So each line is now a mishmash of multiple actions with multiple cards. Not to mention that things can happen on other players' turns; notating a 3P game becomes ridiculous if the players are spamming Militia. And how do you plan to log a 30-action turn full of KC, Goons, GM, Bazaar, Bishop?
I don't think you can do better than a machine-readable notation here. But in an era where the transfer of 1 MB of data takes only a few seconds even on very slow connections, is there really a big functional difference between a log 1000 characters long and one 50k characters long?
-
The problem with a lot of these systems is that it's also not intuitive. Chess suffers from this problem, but if you're into chess you've already demonstrated a willingness to invest that kind of time to learn a notation system. Dominion, on the other hand, has a much broader audience to account for.
Also, Chess has only 8 pieces that you need to know about. Dominion has over a hundred.
I think the old blog system (where you selectively choose which Actions and game bits to annotate) is probably best, but it's a pain to write up.
Chess notation is incredibly intuitive...
Letter of Piece you Move) (Square it Moves to) with X for capture, + for check, ++ for checkmate, O-O, O-O-O, and E.P. for the three special moves..
I explained it in little over a line :P. Unless you still use one of the obnoxious archaic notations...
Edit: OK.. also =Q,R,N,or B for pawn promotion... and you don't need to write P for pawn.. and you need to define the original square when two of the same piece can move to a square.
I resign.
-
http://web.mit.edu/puzzle/www/12/sheila_sunshine/sovereignty/
-
http://web.mit.edu/puzzle/www/12/sheila_sunshine/sovereignty/
Exactly, that was what I was refering to and it was the inspiration for my notation.
-
I thought about how to programatically remove unimportant information from the logs. One possibly reasonable way to do it is to try to predict the next line from the previous one, as in language modelling (http://en.wikipedia.org/wiki/Language_model). If the next line is predictable, drop it.
EG, most of the time,
"Foo plays a warehouse"
is followed by
"Foo draws 3 cards and discards 3 cards" (or whatever, don't want to look it up the exact message)
That probably happens >95% of the time someone plays a warehouse, and so it's basically contentless.
But if they draw < 3 cards because they've drawn most of their deck, that is useful/interesting/unexpected, and should be kept in the log.
This approach gracefully handles what to display empirically and it's a nice familiar format that anyone can understand.
-
I bet it would be possible to define a notation like GenericKen's that also takes that kind of expected effect into account.
Suppose you expand the format a bit to show what happens when you play a card. Maybe "c" for drawing a card, "d" for discarding, + for gaining/getting a card in some other awy, - for trashing/returning/passing, $ for +$1, b for +1 buy, a for +1 action, etc, possibly with a multiplier. You would put it in brackets after the card. But you could also have defaults for each card, like [2c a] is assumed for a Laboratory so you can just say "Lab" instead of "Lab 2c a", or Baron always comes with +1 buy so you leave that out.
So for example, playing a Baron to get an Estate is Bar[+e]. Playing it to get $4 is Bar[d $4]. A Spice Merchant for cards and actions is Spi[-C 2c a], and for coins and buys it's Spi[-C $2 b]. A Cutpurse that connects is Cut[$2; d], indicating that the next player discards, or just Cut[; d] because the $2 is always there. A Village that draws nothing because your deck is empty is Vil[0c].
With the expanded format, it's probably a bad idea to try to put everyone's turn 1 on the same line. I'd probably also stick to 1 to 3 letters in the card abbreviations. This also leaves room to show the hands drawn, though I didn't do that below. Here's a transcript of http://councilroom.com/game?game_id=game-20120512-004426-b37dc249.html, and yes my shuffle-fu is weak in this game:
a = rij, b = rspeer. Relevant cards are Smu=Smugglers, Alc=Alchemist, Pwn=Pawn, Mkt=Market, TM=Treasure Map, Ped=Peddler, P=Potion, S=Silver, G=Gold, lowercase cards are victory cards.
1a. $4 -> P
1b. $4 -> P
2a. $3 -> Smu *
2b. $3 -> S *
3a. Smu[+S] $3 -> S
3b. $3P -> Alc
4a. $2P -> Pwn *
4b. $4 -> TM *
5a. Smu[+TM] $5 -> Mkt
5b. $4 -> TM
6a. Pwn[c$] $3 -> S *
6b. Alc $3P -> Alc; >Alc * (that's returning an Alchemist to the top of the deck)
7a. $6P -> Alc
7b. Alc $4 -> P
8a. Mkt Smu[+P] $5 -> Mkt
8b. Alc $3P -> Alc; >Alc *
9a. Pwn[c$] $4 -> TM *
9b. Alc $3P -> Alc; >Alc
10a. Mkt Alc Smu[] $3P -> Alc; >Alc
10b. Alc $3P -> Alc
11a. Alc Pwn[c$] $7 -> Mkt
11b. Alc* Alc Alc Alc Alc $6P -> Mkt; >5Alc
12a. Mkt TM[-2TM, +4G] $5 -> Mkt
12b. Alc Alc Alc* Alc Alc Mkt $8P -> pro, Ped; >5Alc
13a. $12P -> pro *
13b. Alc Alc Alc* Alc Alc Ped $6PP -> 3; >5Alc
14a. Mkt Alc Pwn[c$] $9 -> pro
14b. Alc Alc Alc Alc* Alc Mkt $10P -> pro, Ped; >5Alc
15a. Mkt $9 -> pro
15b. Alc Alc Alc Alc Alc* Ped $6PP -> 3; >5Alc
16a. Alc Mkt $10PP -> pro, Pwn; >Alc
16b. Alc Alc Alc Alc Ped Mkt Alc* Ped $11 -> pro, e
17a. Mkt Alc* Alc Pwn[a$] Smu[+e] $5 -> duc, Ped
17b. $4P -> e *
18a. $7P -> duc
18b. $4P -> e *
19a. Mkt Mkt $5 -> duc
19b. Mkt Alc $5 -> duc
20a. Mkt Pwn[c$] $10P -> pro, e *
rij 44 - rspeer 27
-
Chess notation is also extensible. If it's ambiguous which knight or rook is advancing to a space, the notation must expand to specify which is moving, but the vast majority of the time, it's explicit what happened. I feel the same about dominion.
I think I didn't communicate my goal well enough.
The point of a notation system isn't to replace the isotropic logs, but to summarize them into a format that's readable WITHOUT scrolling. We don't need to manually & losslessly compress the data - we have gzip for that. I want to compress the data in the physical space it takes on the page. As much information as there is in a game of dominion, the vast majority does not affect the decisions you make as a player. We can compromise a bit on fidelity of information for the sake of readability.
If the first two buys were S and S, and if turns 3 and 4 resulted in $5 and $4, it doesn't matter if they were 5C or 3CSE or C2S2E for the $5 or some other combination for the $4. The last 2 cards of the deck contain $2. Noting turns 3 and 4 as simply $5 and $4 does introduce ambiguity, but it's an ambiguity that doesn't actually affect your decision making in game. If a kingdom card in your hand (say, Upgrade), does make that ambiguity matter, the ambiguity can be clarified by the notes on the action.
This ambiguity can extend to engine chains. As difficult as it can be to keep an engine running, most engines do run themselves. That is to say, a best-decision tends to exist. Even the most complex King's Court chain can be easily defined by the end-state - what cards were played and what cards weren't - who gained what and how many times you shuffled. Whether you draw your deck or fizzle rarely has anything to do with your decision making while running the engine - if you do accidentally kill your own engine, the fizzle is fully described by the cards you drew but did not (could not) play.
This may be a bit overambitious, as the isotropic logs do not show what got discarded. If a player bought warehouse-militia, it's hard to know whether the milita is getting discarded to warehouse or to the opponent's militias.
I've tried cleaning up my example w/ monospacing & tabs here:Vine(Vineyard); Perl (PearlDiver); Look(Lookout); Vil; Pot (Potion), Joat (JackOfAllTrades); HoP (HornOfPlenty), Gov, Lab, Tres(Treasury); GM (GrandMarket)
cassandre ; GenericKen
1. +Joat ; +Joat
2. +Look * ; +S *
3. +S ; +[Joat,S], -E
4. $4 +[S,S], -C, *Look ; +Lab *
5. +S, -C ; +G
6. $7 +[G,S] *Look ; $1 +S, #Joat *
7. +GM ; +[P,S]
8. $7 +G, -C ; +Lab *
9. +[P,S], -E * ; $7 +G
10. $10 +[P,S] ; $9 +[P,S], *
11. +Tres, -S * ; +[P,S]
12. $7 +[D,E] ; $7 +D
13. +[G,S] ; +[D,S] *
14. $9 +P * ; $11 +P
15. $9 +P
Points: 29 ; 30
It's better, but it can still stand improvement. Adoption is important, and most people understand 4/3 vs 2/5 openings. I had great hopes for listing the total $ each turn, as people can extrapolate from turns 1 and 2, but all those '$'s were very distracting. Joat might not be the best example of a simple game log, as the card does like 4 things.
It's also still hard to read when the turns are so uneven - some turns have nothing but treasure, and some turns draw the deck several times.
-
I don't think anything with columns is going to catch on. It's already hard to do on a forum, and if you have a big turn then you have to try to do things with wrapping that most text editors are not equipped to do.
Seeing the money amounts is far from distracting; you can glance at them over the course of the game and get an idea of the flow of the game. You can't do that if you're trying to mentally substitute how much each card is worth.
-
Still I think the best way to describe Dominion turns is human processing combined with skipping/compressing non interesting moves.
For instance:
6. PlayerA plays his 5 Alchemists, buys a Province and puts all Alchemists back
From this line you can infer almost everything:
- He has 5 Alchemists
- He had at least $8
- He had drawn his Potion
Often, there's no need to be explicit.
If someone has $9 and 2 Buys, is it important to know he didn't buy anything besides a Province or that he had the extra buy?
I like the old notation from the annotated games.
-
How about something like this (annotated game from the blog (http://dominionstrategy.com/2012/02/17/annotated-game-10-geronimoo-vs-wanderingwinder/)):
Turn 1:
WW: Masquerade with $4
Geronimoo: Masquerade
Turn 2:
WW: Silver Reshuffles
Geronimoo: Silver with $4 Reshuffles
Turn 3:
WW: trashes Estate with Masquerade Venture (-1VP)
Geronimoo: Young Witch
Turn 4:
WW: Silver with $4 Reshuffles
Geronimoo: trashes Estate with Masquerade Festival Reshuffles (0VP)
Turn 5:
WW: trashes Estate with Masquerade Venture (-1VP)
Geronimoo: Young Witch with $5
Turn 6:
WW: Gold Reshuffles
Geronimoo: trashes Estate with Masquerade Crossroads with $3 Reshuffles (0VP)
Turn 7:
WW: trashes Copper with Masquerade Province (+6VP)
Geronimoo: trashes Estate with Masquerade Festival with $6 (-7VP)
Turn 8:
WW: Venture Reshuffles
Geronimoo: Reshuffles Festival + 2 Young Witches deal out 2 Curses Walled Village (-5VP)
Turn 9:
WW: Gold
Geronimoo: Watchtower + Native Village with $6 Reshuffles
Turn 10:
WW: Province Reshuffles (+11VP)
Geronimoo: deals out a Curse with Young Witch Festival (-10VP)
Turn 11:
WW: Gold
Geronimoo: Festival + Watchtower Reshuffles
Turn 12:
WW: Pawn
Geronimoo: trashes Copper with Masquerade, deals out 2 Curses with Young Witches Watchtower + Native Village with $6 (-8VP)
Turn 13:
WW: Province (+14VP)
Geronimoo: Reshuffles Festival/Watchtower chain, deals out 2 Curses with Young Witches Festival + Watchtower Reshuffles (-12VP)
Turn 14:
WW: trashes Curse with Masquerade Province (+19VP)
Geronimoo: Festival/Watchtower chain, trashes Curse with Masquerade, deals out Curse with Young Witch Highway + Festival + Watchtower Reshuffles (-19VP)
Turn 15:
WW: Masquerade with $4
Geronimoo: deals out 2 Curses with Young Witches DISCONNECT (-17VP)
The idea is to quickly see the buys because they are underlined. The remarkable plays/buys are in bold. The interesting actions and reshuffles in italic and finally the difference in VP-points in brackets.
-
I think if this:
cassandre ; GenericKen
1. +Joat ; +Joat
2. +Look * ; +S *
style catchs on, you will lose a lot of casual interest in the board and discussion over logs. Its just a pain to look through unless you know what they all mean, and only the proper hard core people will want to do that.
Geronimoo's looks much nicer to read (but assume it takes more time)
-
I really don't like the abbreviations. People aren't computers who can easily map the abbreviation to the complete name.
-
I really don't like the abbreviations. People aren't computers who can easily map the abbreviation to the complete name.
+ in contrast to dead trees, there should be enough space on the internet to write the full names...
-
I heard the internet was nearly full up though...
-
Here's another example where I transcribe the Annotated game #2 between MMM and theory (http://dominionstrategy.com/2011/02/14/annotated-game-2/)
(I used the * for reshuffle)
Turn 1:
MMM: Silver with $4
theory: Silver
Turn 2:
MMM: Silver *
theory: Baron *
Turn 3:
MMM: Silver
theory: Silver with $4
Turn 4:
MMM: Apprentice with $7 *
theory: Gold with $7 *
Turn 5:
MMM: Silver with $4
theory: Gold
Turn 6:
MMM: Apprentice with $7 *
theory: Gold *
Turn 7:
MMM: Apprentice an Estate Gold with $7 (-1VP)
theory: Mine with $7
Turn 8:
MMM: Apprentice an Estate Apprentice * (-2VP)
theory: Platinum
Turn 9:
MMM: Gold
theory: Embargo with $3
Turn 10:
MMM: Apprentice an Estate and a Copper Gold with $7 * (-3VP)
theory: Embargoes the Platinums Mine with $10!?
Turn 11:
MMM: Apprentice a Silver and a Copper Gold with $7
theory: Apprentice with $8
Turn 12:
MMM: Apprentice a Silver Colony * (+7VP)
theory: Mine Gold into Platinum Apprentice with $9 *
Turn 13:
MMM: Apprentice a Silver and a Silver Colony * (+17VP)
theory: Apprentice an Estate, Mine Copper into Silver Embargo with $4 (-18VP)
Turn 14:
MMM: Apprentice a Gold Colony (+28VP)
theory: Mine Silver into Gold Tactician?
Turn 15:
MMM: Apprentice an Apprentice and a Gold Colony * (+38VP)
theory: Apprentice an Estate Colony with $16 (-29VP)
Turn 16:
MMM: Apprentice a Colony Colony * (+29VP)
theory: Province with $10 (-23VP)
Turn 17:
MMM: Apprentice a Colony Colony * (+23VP)
theory: Embargoes the Colonies Province with $11 (-17VP)
Turn 18:
MMM: Apprentice a Colony Colony (Curse gained) * (+16VP)
This took me 20 minutes to transcribe (to get an idea)
-
I like Geronimoo's/Davio's prose style. I'll try that out next time.
-
I like having the score differential posted at the end of each turn. It allows me to get a feel for what stage the game is at without having to keep track in my head or waiting for an update.
-
I like having the score differential posted at the end of each turn. It allows me to get a feel for what stage the game is at without having to keep track in my head or waiting for an update.
I think I'd actually prefer the score written rather than the score differential.
-
I agree with Kirian. Once I realized what it meant, the differential was fine, but I the players total would make more sense to me.
-
The opening 2 Turns can easily be condensed as we're so used to seeimg them in the form of Card1/Card2.
From this we can infer the opening hands as well.
Buying a Silver first with your $4 and the $3 you wanted on turn 2 is so common it doesn't deserve its own entry.
Examples:
Opening:
Bob: Masq/Silver
Lucy: Silver/Baron
I like the shorthand for reshuffles, but I would prefer an alternative character to the asterisk. I use it for so many different things (mainly searching and regexes) it confuses me. Maybe pound (#)?
Example
Opening:
MMM: Silver/Silver
theory: Silver/Baron
Turn 3:
MMM: Silver
theory: Silver
Turn 4:
MMM: Apprentice with $7 #
theory: Gold #
Turn 5:
MMM: Silver
theory: Gold
Turn 6:
MMM: Apprentice with $7 #
theory: Gold #
----
Here I've left out almost all but the essentials.
The key is to convey the feel of the game and the strategies both players are pursuing.
Any notation should do that as good as possible.
Leaving out some of the ... with $4 makes it a bit cleaner.
I've left in the "Apprentice with $7" entries, because they signify a very important choice by that player to skip Gold.
I don't think passing up a Baron for a Silver is such an important choice, Silver is kind of the default move, much like Gold is with $6.
-
Leaving out some of the ... with $4 makes it a bit cleaner.
I've left in the "Apprentice with $7" entries, because they signify a very important choice by that player to skip Gold.
I don't think passing up a Baron for a Silver is such an important choice, Silver is kind of the default move, much like Gold is with $6.
I'm not sure this is a good idea. The problem is, the question of whether buying a cheaper card is an important choice is subjective. I'm not saying you are wrong here, but it creates an inconsistency. as you say "I don't think passing up a Baron for a Silver is such an important choice..." but perhaps sometimes it is, and there will be situations where players will disagree on what is an important choice.
I would actually prefer to see the buying power every turn. Buying a Duchy with $7 in the end game is pretty automatic, but seeing that a player got $7 4 turns in a row is information that says something about the game.
I also think that, while it is pretty uninteresting to those of us regularly posting here, showing things like opening Sliver with $4 is potentially interesting for newer players who don't regularly contemplate Dominion strategy. I still occasionally space out and grab the shinny 3 I want to get when I could have delayed revealing my strategy by opening Silver.
-
I agree with you now that columns and abreviations are too cumbersome. I was trying to mono-space the format, but it looks like that's a lost cause given the very nature of the game.
A couple of points:
1. Nitpicky: "deals out a curse to" is functionally the same as "deals a curse to".
2. The VP differentials can be confusing because the context of +VP vs -VP changes depending on which line it's on. I think posting both VP totals is less ambiguous.
3. I think "with $4" is important for consistency. While buying a $4 card is usually terrible, it's still an option that the player declined.
4. I think we need to resolve how to denote shuffles that occur as the result of actions. The nice thing about using a symbol like '*' (or '#', whatever, but I find '#' more visually distracting, and it tends to denote checkmate or script comments rather than something that happens regularly, but I digress) is that it visually separates when bought cards become available in the draw deck.
If you buy a card, the '*' nicely tells you that it becomes immediately available for draw on the next turn:
Silver *
HOWEVER, if you shuffle as a result of actions, where should the * go?
*Smithy draw, Silver
Smithy draw*, Silver
...imply different things. If a draw action causes a reshuffle, should the * go before the action (because previously gained cards were not available when choosing to play that action), or after the action (because those previously gained cards can now be drawn by the action)?
I'm not sure what the right answer here is.
I still believe strongly that there's much profit to condensing the format to be readable without scroll. This is still possible without mono-spacing for a 22 turn game. Tweaking Geronimoo's log:
(I used the * for reshuffle)
MMM - Apprentice Chain
theory - Mine Platinum, into panic
1. Silver with $4
Silver
2. Silver *
Baron *
3. Silver
Silver with $4
4. Apprentice with $7 *
Gold with $7 *
5. Silver with $4
Gold
6. Apprentice with $7 *
Gold *
7. Apprentice an Estate Gold with $7 (2-3VP)
Mine with $7
8. Apprentice an Estate Apprentice * (1-3VP)
Platinum
9. Gold
Embargo with $3
10. Apprentice an Estate and a Copper Gold with $7 * (0-3VP)
Embargoes the Platinums Mine with $10!?
11. Apprentice a Silver and a Copper Gold with $7
Apprentice with $8
12. Apprentice a Silver Colony * (10-3VP)
Mine Gold into Platinum Apprentice with $9 *
13. Apprentice 2 Silver Colony * (20-3VP)
Apprentice an Estate, Mine Copper into Silver Embargo with $4 (20-2VP)
14. Apprentice a Gold Colony (30-2VP)
Mine Silver into Gold Tactician?
15. Apprentice an Apprentice and a Gold Colony * (40-2VP)
Apprentice an Estate Colony with $16 (40-12VP)
16. Apprentice a Colony Colony *
Province with $10 (40-20VP)
17. Apprentice a Colony Colony *
Embargoes the Colonies Province with $11 (40-28VP)
18. Apprentice a Colony Colony (Curse gained) * (39-28VP)
-
Like Davio, I wonder if the underlines are actually needed; in many turns, it's the main thing that has happened, so half the log or more is underlined!
I'm going to past one of my own logs, which is longer and has more going on with it, in a style hybridized between Geronimoo's, Davio's, and GenericKen's, along with further abbreviation that I don't think harms anything. The game is here:
http://councilroom.com/game?game_id=game-20120427-192224-4b43b626.html
Greystripe77: Horse Traders / Silver
Witch-BM with Philosopher's Stone support
Kirian: Silver / Silver
Witch-BM with Bishop support
3. Silver
Witch
4. HT ($6) Witch + Copper #
Bishop #
5. Silver
Gold
6. HT Potion + Estate? [4]
Silver #
7. Witch (Hit) # : ($7) Gold [3]
Witch (Hit) : ($7) Gold [2]
8. Silver
Gold
9. ($4P) Philosopher's Stone
Venture
10. HT : Potion + Estate?? [3]
Witch (Hit) : ($8) Gold
11. Gold
Bishop (Curse) : Gold # [4]
12. Stone ($4) : ($9) Province? [9]
Platinum
13. Witch (Hit) : ($5P) Philosopher's Stone #
Gold [3]
14. Witch (Hit) : ($7) Gold [8]
Witch (Hit), Venture : Colony # [12]
15. Stone ($5) : ($9) Province [14]
($12) Colony [22]
16. HT : Philosopher's Stone, Estate [14]
Witch (Hit) : ($9) Province [28]
17. ($2P) Estate [15]
Bishop (Copper) : Province [35]
18. Stone ($6) : ($12) Colony # [25]
($7) Gold #
19. Stone ($6) : ($9) Province [31]
Estate [36]
20. Witch (Hit) : ($1P) Copper
Colony [45]
21. ($9) Province [37]
($12) Colony [55]
22. Duchy [41]
Bishop (Curse, *Curse) : Bishop [57]
23. Stone ($7) : ($10) Province [47]
Witch (Hit), Venture : Colony # [67]
24. ($3P) Estate # [47]
Witch (Hit, *HT) : ($12) Colony [77]
25. HT, Stone ($7) : Colony, Estate [57]
Greystripe77 ends the game on Colonies rather than hoping to get nearly all the remaining VPs
-
All these abbreviations still miss the details and the details are important. Sometimes the difference between a gold and copper in hand compared to two silvers can be significant. A beginner might never realise that an expert player has discarded actions cards with a cellar just so they can be all drawn again with a scrying pool. This leaves me wondering who these abbreviated logs are for. If these are for the casual player then they need to be longer and easier to read. If they are for the expert player then they need as much detail as possible. The fact that it is abbreviated offers no advantage in itself and becomes a disadvantage when you can't see how decisions are made during play.
-
I think we collectively have skipped an important step in coming up with a notation system, and so we're trying to do 2 things at once. The goal of a notation system is to display relevant information and filter out unneeded information. But we jumped straight into discussions of syntax, without first talking about which information is relevant or not! Thankfully Geronimoo's system, which closely resembles plaintext, seems to have focused the discussion back to this important topic. I suggest we explicitly focus our efforts on that topic before we deal with anything about style.
I think we've all implicitly agreed about one thing that needs to be included in the gamelog: Whenever the player makes a non-trivial decision. However, in order to include this I think a gamelog needs to include not just what actions were taken, but what alternatives were available, as well as significant data gained. This will sometimes require going beyond what actions were taken, and explicitly call out actions that were not taken.
For example, I want to know if terminals collided. This is a big deal when it happens: the player has to make a decision of what to play, they gain knowledge about what actions are still available in their deck, and in general it has a big impact on their mental state. Same thing with, e.g., wasted (targetless) King's Courts, or electing not to play a trashing Action card due to lack of a good target, or not playing 3 copper to buy a Grand Market when Platinums are available. These are player actions available that were not taken, which are as significant as actions that were taken.
I think the hardest thing to include is deck state, and how players manage their reshuffling. Top-level players will be making decisions to play or not play drawing cards based on how many cards are in their deck, and what they know those cards are. It's hard to reconstruct this information without card-by-card breakdowns of all hands since the last reshuffle, or including deck state annotations where it might be relevant. I honestly have no idea what to do about this.
-
PSGarak> I agree that showing alternatives is important for a notation system - that's why I feel strongly about showing "with $4" on silver buys. Additionally, my original notation had a section for discarded cards (#C, #Joat) precisely to capture clashing terminals, but there're two serious problems with it.
1. It's painfully time consuming to figure out what the two clashing terminals were. You generally have to walk backwards through a turn to reconstruct what was played and what wasn't.
2. Warehouse. It's not actually possible to know what was drawn and what was discarded with a large number of cards - Isotropic simply doesn't log them. If I discard a clashing terminal to warehouse, there's just no way to know from the logs.
Herein lies the beauty of using '*' to delimit shuffles. You can't precisely know what cards were available for a turn or what cards are left in the deck from the logs, but by tracking gains and trashes, you can know what cards are in the deck between two shuffles. If a player buys three swindlers and only plays two between two shuffles, that should tell you what you need to know most of the time.
In particular cases where a subtle trick cannot be captured by notation (say, not triggering a shuffle with a draw card while buying a gold, or re-drawing discarded estates for Baron), I think it's best to concede that it cannot be captured by notation. Just briefly describe the trick in parentheses: "(discard Smithy)". There's a blog-full of neat tricks in the game, and I don't think there's enough commonality between them to warrant encoding them.
-
I would like to comment that developing a notation system should also not be limited by the information currently available in isotropic logs. You shouldn't exclude potentially useful information from your system just because it isn't currently available. These are the type of things that would be nice to get added to isotropic logs, but they won't be because it is going away... which is another good reason not to design your notation around its limitations.
-
On the occasions when you do want to go over a game, the log does a reasonable job. Apart from adding the missing information (what was discarded? what was passed with Masquerade?), for me the biggest improvement over a log would be a log "player", that allows you to step through the game turn by turn and action by action, displaying the information you get during a game plus extra details of the game state such as deck/discard pile contents for each player (along the lines of that currently provided for the logs on Council Room).
A "log player" is an excellent idea. Unfortunately, it's impossible with the present version of isotropic. The game state cannot be reconstructed from the logs -- a limitation I hope will be absent from the official app.
-
You could still have a "buys" log player. You don't get to make the action choices, but you do get to make the buys decisions.