ClusterFuck

[Matt]

The idea of "who is the beatdown" is a good concept, but in the game, resources are randomly distributed and only through playtesting will you understand what role your deck plays. In this sense theory is flawed because it provides imperfect data... It is not a predictive engine so much as a statement of direction.

Well at least now I see where your distaste for theory comes from. If I hadn't been able to predict things using theory, I would almost certainly be calling it flawed too. But I have, and in fact, WTB? theory has been hugely accurate at predicting how a matchup plays out for me, even before playtesting begins. So I really can't agree with you on that fundamental point.

@Carlos: I've been working on a rough draft of how to combine tempo and CA theory into one, larger theory, and it feels like it covers almost everything...but not quite. There's definitely a missing component, and your board advantage theory might be it. But currently, like Ric_Flair said, it needs some work. If you could write a longer-style post or article, explaining the nitty gritty of your theory (providing examples along the way), that would be tremendously helpful in advancing the improvement of general magic theory.

So given that this does occur, the question is why NOT play the best deck?

Since I know you're so much one to look to the pros for the best way to play the game, I might add that being familiar with a deck can make a huge difference, and is largely touted by the pros as being a good way to win more often. When people say, "play a deck that suits your style of play," that's really just a sub-part of this - if you've been playing Keeper for five years, you're probably better off playing a control deck than switching to aggro.
Ease-of-play is another factor when choosing decks that isn't considered often enough.
Both of these factors can be minimized by enough playtesting, but in the real world, you may not have the luxury of enough hours to devote to testing, and that's where these step in.

The Fish vs. Academy example isn't very good. That final match was a very close thing. Also, you've got your facts wrong. Hovi won the whole thing, it was Nicholas Labarre who played the fish deck.

Reports from both sides of the table may be found here:
http://web.archive.org/web/19991007065536/thedojo.com/t984/ptro.981125nla.txt
and here:
http://web.archive.org/web/20000521111958/www.thedojo.com/t984/ptro.981208tho.txt

The correct call may seem to be let it go, and try to find StP, but if you lost the game it obviously wasn't the correct call.

This is the exact problem encountered over on the SCG boards. I'll quote myself:
"Some Magic players define the correct play as "the play that wins the game." And under this definition, they are correct, there is no way to know what the correct play is. But the opposing side has a different definition of correct play: the play that maximizes the chance of winning. Using that definition, it is more than possible to find the modernist-correct play"

Playtesting can never blind you.

I'm going to call bullshit on that one. Bad playtesting blinds you in the same way bad theory does. If you playtest against poor players, or against poor builds of a deck, or while tired/distracted, you'll make incorrect conclusions just as surely as ignoring tempo theory will give you incorrect conclusions about the viability of a card.

[Ishi]

Great thread!

1st off, it looks like point #3 is quite solid and non-debatable.

Point #2 has gotten us off into some really weird territory, but the debate has been quite interesting.  I think that Matt's points on the topic are really good.  Card advantage and tempo are important aspects of the game and we need to take them into account.  Of course, putting too much emphasis on them is going to lead to dumb ideas, but that doesn't mean we should thrown them out all together.

Card advantage theory probably did cause people to look past cards like LED, but also it was quite usefully in causing us to look past tons of other cards.  Spy network, Tahngarth's Glare, Rolling Stones, the "lucky charms", etc. etc.

People have been talking comparing magic theories to scientific ones, and I think there are good parallels there.  Newtonian physics wasn't "correct" but it was pretty usefull.  Sure it caused scientists to be blind to the ideas of relativity, quantum mechanics etc.  But it quite rightly caused scientists to reject astrology, numerology and Ishi's theory that if you throw an orange at exactly 20mph it will never stop.

Scientists would never make any progress if they had to rigorisly test every possible screwball idea, so they just ignore certain ones because they trust their theories to a certain extent.  Similary magic players would never get anywhere if we spent our time playtesting 100 matches of crystal rod-Test of Endurance.dec vs thallids.dec and then seeing if the winning deck could take on Griffin Canyon-Khamal Fists of Krosa-Artificial evolution.dec.

So playtesting is to magic what experimentation is to science.  Playtesting is useless without theory and theory is useless without playtesting.

Rob Dougherty's experiment is an excellent example.  The suggested "improvements" based on their theory alone were really poor.  But Rob's original idea was strongly based on theory.  He didn't just take 60 random cards and ask people to see what they could do with it, he tried to make a strong tempo deck and chose cards accordingly.  What should have happened (and what did after a while) is that people should have chosen cards that worked with the tempo theme and then playtested them to see if they were any good.  Theory and experimentation have to go hand-in-hand.

Also, we have to look to ways to improve theories or discard theories if the become obsolete.  The bazaar example was excellent.  It shows that card advantage can't be counted anymore by simple the number of cards in your hand + cards in play.  With flashback, reanimation, chalice etc. thinking in terms of which cards are active, versus which cards are dead is much more usefull.

[Ric_Flair]

I guess the best way to sum up my opinions on the matter are this:  Like when Newtonian physics gave way to quantum physics Magic theory is so much in the crapper right now that we need an equivalent shift.  Magic theorists just cannot keep up with the game and predict which cards are going to be good any more accurately than the average player can.  Aside from Matt, most of us are not smart enough to rely primarily on theory to make good decks  

We need some smart people to think up new and useful theories that keep pace with the innovative cards that R and D is putting out.

[Matt]

We need some smart people to think up new and useful theories that keep pace with the innovative cards that R and D is putting out.

Yay! Something we can agree on, sort of. I'm not so sure that entirely new theories are required, but I can agree that at the least the old ones need revision.

[Dr. Sylvan]

How about this:

The reason Steve is perceiving tempo to be supreme over card advantage is that the game is designed to make it so. If you started out with seven lands in play and no cards in hand, the natural focus of all of your efforts would be on drawing more cards, and you would run more powerful, more expensive spells, because the early-game mana shortage would be gone, and you would need to maximize the effect of each draw. In the hypothetical tempo-irrelevant world, Force of Will would be useless because you already have enough mana to run better cards--I'm sure Dismiss would be considered one of the premier counters of such a universe, for example.

Back to actual Magic, the abundant resource is cards-in-hand, so you're struggling to create mana resources at the maximum possible rate, allowing you to unload your oversupply. Card advantage is still important, though, because the game isn't so short that a tempo advantage can achieve victory before the pendulum swings back. This is why people were so freaked out over Long: it had the power to break this rule. I think this is the perspective that informs Steve's perception of CA as a meaningless theory, even though he happens to have become intimately familiar with the only known deck which makes this near-true.

This ties in directly with "Who's the Beatdown?" (maybe it's just a rephrasing; I'm not sure). So my theory breaks the game into two stages:

Stage 1: Tempo - Early in the game, when it's possible to use smaller, more efficient cards and nonrenewable acceleration to destroy the opponent before his countermeasures are even castable.

Stage 2: Card Advantage - As the game progresses and the efficiency of the opponent's cards becomes irrelevant, it's better to have bigger effects on each card which make use of the abundant resource in Stage 2: mana.

The best decks are those which are well-suited to both portions of the game.

[zero]

I'm glad that skeletal scrying was brought up by dr. sylvan, because it so clearly illustrates the differences between CA and tempo. They both provide the same amount of CA : lose a card from the hand, draw x cards. And yet the switch was made for tempo reasons, even vs. decks like sligh where it may seem counterintuitive to lose life. CA theory as I've seen it does not explain this.
@Matt : My point about winning was to help illustrate the difference between theory and reality. (Oh, and the point of the game is to win). Sure, if I flip a coin 1000 times, I expect 500 heads and 500 tails. If I flip it 3 times, I expect to get 3 straight heads 1/2^3 times, or 12.5%. But can anyone tell me exactly when this will happen? AFAIK, no. I just know it will happen once in every 8 three flip attempts. That's your "expected value", to use a financial term. This is not a solution. Knowing this is useful, but not the be-all and end-all of magic playing. Putting yourself in a position to win, which seems to be more related to tempo theses days (see skeletal scrying above) than CA, would be where I'd want a magic theory to be pointing me (reiterating your point).  I guess to me a solution is a "sure thing", and I'm skeptical there is a grand unified theory of magic that will enable this. We'll just have to use expected value for now...
Your point about playtesting is interesting also. As an example, I'd look to meandeck's latest spoils mask and slavery lists, which seem metagamed to do best vs. a high powered field, hence the disagreements on maindeck blood moons, etc.  Sure, you can argue that you're only worried about the people in the  upper brackets, but you still have to get there. Crappy playtesting does follow the garbage-in, garbage-out maxim (I'm not saying that crappy playtesting was done by the team).