Today I’d like to share a few aspects of my Pro Tour: San Juan experience that I think illustrate important things about the game.
When to splash: a case study
In my first draft of Pro Tour: San Juan, I made the most of a bad situation. I went 2-1 despite the person to my left playing the same colors as me, and despite the packs not feeling tremendously deep in those same colors. The ability to emerge from a difficult situation with a “decent” or “acceptable” outcome is something you have to master if you intend to win.
I knew from a brief discussion the day before the Pro Tour even began that Martin Juza valued the card Nest Invader very highly. He likes drafting green, and he likes Nest Invader in his green decks. Martin Juza was seated to my left (I would be passing to him in packs 1 and 3, and he to me in pack 2) in this first draft. I opened up Consume the Meek, Nest Invader, and some cards I knew Martin would value less than Nest Invader, all else being equal. There were cards in the pack he would take over the Nest Invader, but only if his first pick went very well with one of those cards and not very well with Nest Invader. I took the Consume the Meek. What I later learned was that Martin had first picked a black card, and he decided to pick another black card second instead of Nest Invader when he received my pack. As I began taking red and black cards, so did Martin behind me. This isn’t as big a disaster for me as it is for Martin, since he will be feeding me only 1 pack instead of two.
Because you don’t get to see what the person to your left is passing in pack 1, making it harder to determine what colors they are playing, and because they only pass to you in 1 out of the 3 packs, I don’t tend to be all that concerned with what colors the person behind me (to my left) is drafting. Other people disagree with me, and like to expend great energy trying to figure out what the person to their left is likely drafting. I’ve observed that people aren’t very good at figuring it out, and don’t really have much to do with this information when they are able to figure it out.
Getting back to my draft, pack 2 is when things got interesting. Around 4th pick, Martin passed me a [card]Narcolepsy[/card]. This signaled to me that Martin likely wasn’t blue, and the person to his left likely wasn’t blue either, so the rest of pack 2 might have a few more blue playables. At the time I had several playables in both red and black, and none in blue. Splashing the Narcolepsy is a reasonable option, since the card is good on any turn of the game and its cost contains only one colored mana symbol. The card I would be taking if I passed along the Narcolepsy was something decent for my deck, but nothing special. I decided to pass the Narcolepsy along despite its superiority to the red card I took because I was drafting an aggressive deck and I prefer consistency over “raw power” in my aggressive decks. When you need to “curve out” to win, and you must play your threats quickly before they become obsolete, splashing a third color is dangerous. I don’t expect my red/black aggro deck to win by having more powerful spells than my opponent, I expect it to win by being effective enough in the earlier turns of the game to make the later turns of the game about my opponent “trying to survive” long enough to get his strategy online. You don’t see a slow deck win with “worse” cards nearly as often as you see an aggressive deck do so. For this reason, I am unlikely to splash just one Narcolepsy, and since I didn’t see any blue in pack 1, chances are I won’t in pack 3 either. Picks 5-9 or so in pack 2 might contain another worthy splash in blue, or they might not. It will depend on how the packs break, i.e. whether the packs happened to be deep in powerful blue spells. Even though those to my immediate left are not blue, someone must be. Those someones will be taking the really good blue cards early, unless, again, the packs break deep. Remember that I’m not interested in splashing a “good” blue card like Halimar Wavewatch, I’ll need a specific type of blue card to come around, the type worth splashing. My criteria for splashability are whether the card is easy to cast and whether it has potential to swing the game in the mid-late game (as opposed to only being very effective in the early game).
The next pack I was passed contained another Narcolepsy. Had I taken the first Narcolepsy, I would gladly take a second one, and splash them both. However, given that I did not take the first Narcolepsy, my decision here is essentially the same decision I’ve already made (again the pack contained an average card for my red black deck), with the added benefit of being one pick deeper in the pack, making the probability of getting splashable blue cards even lower. I passed the second Narcolepsy and certainly wasn’t letting myself get caught up in any “what if” daydreaming.
The next pack contained a Drake Umbra. At this point, Martin probably thinks I am doing backflips, and Martin’s other neighbor probably thinks Martin is doing backflips, when in reality, it was the player to my right who ended up with the 2 Narcolepsy. Two things made me pick the Drake Umbra. First, the alternative black or red card was playable, but not as “decent” as my last 2 picks, meaning it was easier to take the potentially splashable card with this pick. Second, Drake Umbra is a different kind of card than Narcolepsy. A fast deck loves an effect that can provide a big burst of evasive or direct damage, and it loves a card that the opponent must “answer” or surely lose to. Drake Umbra provides both of these things while Narcolepsy provides neither (it CAN tap a blocker, after a turn is allowed to pass, but hopefully you can see how this differs from making one of your creatures a “dragon” right away). Narcolepsy is a fine splash, since it meets my important criteria of low color requirement and good in the mid-late game, but it isn’t nearly as enticing as a Drake Umbra. The only other splashable blue card I was able to pick up the rest of the draft was a Mnemonic Wall, a card with highly variable value depending on what else you’ve got, and I had plenty for it. I played both blue cards in my final deck, along with 3 islands. The Drake Umbra won me the match in game 3 against Martin Juza, when I topdecked it the turn after he dropped a board-dominating Ulamog’s Crusher.
Hopefully you can see how knowing what my deck’s plan was (aggressive victory) shaped my decision of whether or not to take splashable cards, a decision I see players struggle with all the time.
Choosing a constructed deck: trusting results over reasons
I prepared for PT San Juan with 5 other players: Ben Rubin, Patrick Chapin, Paul Rietzl, Brian Kibler, and Ben Seck. We had a “gauntlet” of decks to test our ideas against that included primarily (meaning most frequently played) mono red, big Eldrazi monsters ramp, and white/blue control. These decks were our best guess at some likely popular decks, as well as some “extremes” of the format that would be good to test against. We all took turns playing the gauntlet decks against each other and against our “brews.” 4 out of the 6 of us ended up playing one of our “brews,” a UGb Oracle of Mul Daya Ob Nixilis deck. Paul played mono-green Eldrazi ramp because he likes to play Affinity and anything that feels like affinity. Eldrazi Temple felt enough like Seat of the Synod to win him over. I played “stock” white-blue control. Nothing unknown, nothing controversial, just Jaces, Gideons, and the supporting cast.
I’m aware of the value of playing an unknown deck. In fact I hate playing a deck that I know the other good players are expecting to show up in force. Also, I went 0-5 at PT San Diego with white-blue control, so it wasn’t like I was reaching to find another opportunity to pick up the deck. However, in testing, I lost very few games with the deck and even fewer matches. Sure, some of the wins came against unproven “brews,” and the PT will to some extent select against brews that can’t beat UW just by virtue of the fact that they lose to UW control. That said, some people will play decks they know to be underdogs to UW, and still more players will thinkthey are favored against UW, but will not actually be favored when they face it at the PT. My version could be different from their expected version, or their testing could be flawed (these things matter more than you might think). The combination of known and unknown underdogs could be large enough to justify a decision to play UW. The fact that different people were getting different results with the same UW deck in our testing indicated to me that the deck was capable of different lines of play that skew other people’s testing. Additionally, the fact that I personally was winning a lot with the deck meant that I naturally was tending to find the correct lines of play. I didn’t need to isolate the differences to suspect they were there, I had observed enough games to suspect a difference in play skill and the direction of that difference.
It takes a special set of circumstances to make me want to play a “stock” deck like UW control, but this was one of those cases. At the PT, as I played the UW deck to a 4-0 start and 7-2 finish (plus an intentional draw), I became more and more satisfied with my decision. I was 2-0 vs. big Eldrazi ramp decks, which most people believe is a bad matchup, but my testing suggested was very winnable. I was beating opponents who showed up with “brews” that I had never played against before, just like in testing, by balancing a proactive planeswalker gameplan with a reactive countermagic and creature kill gameplan. When to be aggressive with tapping out for ‘walkers or attacking with Celestial Colonnade, having an intuitive sense of what elements of the opponents deck need to be countered or killed, knowing how to mulligan, etc. were all things I had developed during playtesting by accident, as I tried to determine which of the other decks could beat UW control.
Why didn’t my 5 other teammates play UW control too? Two main reasons. First, their personal results with the deck in testing weren’t quite as strong, which means they must at least account for the possibility that they hadn’t developed the right feel for the deck yet, and didn’t have time to do so (we settled on decks the night before the tournament). Second, they are all expert players who are aware of the dangers of playing a known deck. They may have been less willing than I was to look past this. Third, and most interestingly, there were several ways the UW deck felt “underpowered” to us. If the mana ramp decks got their mana engines online, sometimes all you could do was watch them cast several spells (like Eldrazi or Avenger of Zendikar) that were all difficult to answer game-winning threats. Furthermore, cards like Vengevine and Luminarch Ascension seemed almost designed specifically to defeat a deck like UW control. These are valid reasons not to want to play UW control, and we all knew them.
When reasons conflict with results, sometimes you have to trust results. In science, this happens all the time. In fact, this is almost a restatement of the principles justifying scientific methodology. One can create “reasons” (or “theories” in a loose sense of that word) justifying anything, but not everyone can obtain observable results that justify their reasons. When you have reasons or a theory that is conflicting with your observed results, you have an opportunity to learn. Either your observations are incorrect or non-representative, or your reasons/theories are flawed. Often you can’t see how the observations are incorrect or non-representative, and you can’t spot a flaw in your reasons/theories. In my case, the fact that I was beating players like Ben Rubin and Patrick Chapin gave me pause about the results being “fool’s gold,” i.e. non-representative (in a harmful way) in terms of opposing skill, but they could still be “fool’s gold” if I was playing against a non-representative sampling of opposing decks. Had I just been winning a local Friday Night Magic or Magic-League event against mediocre players, I would have been much more skeptical. A large sample size was also helpful; I played many games with the UW deck. I also felt like the “reasons” why my deck was supposed to lose to ramp decks, for example, was a result of a biased way of thinking about the deck. Every time the ramp deck got their engine going, you felt helpless. This makes these games a more salient event than the games where the opponent couldn’t get their engine going. I made sure the deck I tested with played 4 Tectonic Edge the entire time I tested with it. Partly because of this, I found that stopping the opponent’s attempt to “ramp” up to a ton of mana was happening more than I thought it would. I also thought that Luminarch Ascension was a powerful enough sideboard card against ramp decks that it would help me win games even when my opponent’s draw was otherwise unstoppable. The results couldn’t be ignored. I had to play the deck.
I offer this next example without mentioning names because the names aren’t necessary to understand the lesson of the story. Someone I know at PT San Juan made a very different deck selection decision than I did. Like me, they weren’t wowed by any of their team’s decks heading into the eve of the tournament. However, unlike me, on the eve of the tournament they did not select a “stock” deck that they felt was performing well. Instead, they played a rogue deck that was under-tested, but sure to catch people by surprise. In another context, this might have worked, but at PT San Juan it did not. My point with this story is that “going rogue,” as Sarah Palin might say, has advantages, but they need to be weighed against the advantages the known decks offer. Just as there is collective knowledge about how to beat the known decks, there is also collective knowledge about how to win with the known decks. There is also a large amount of data suggesting the known good decks are doing powerful things without many obvious weaknesses.
When you’re faced with a decision to go rogue or go known, think about the particular context in which you developed and had initial success with your rogue deck. If the context is sufficiently likely to produce a good deck (representative opponents, representative decks), by all means, trust that context and play your unknown deck. You might enjoy the benefit of surprise without having to endure the frustration of an ineffective deck. If, however, you don’t have a deck that meets these criteria, don’t feel bad playing the deck that you think is the “best known deck.” Your opponents won’t be very surprised, but that isn’t the only thing that determines who wins. Besides the issue of rogue vs. known decks, be cognizant of whether you find yourself continually using “reasons” to explain away results. Your results mean something. They may not be representative or definitive proof of your expected outcome, but they are evidence of something. Your job is to answer this question “are my results with this deck evidence of something important, like my comfort level with the deck and the deck’s ability to interact with a wide range of opposing decks (this was my answer in San Juan), or merely evidence that my opponents have not been using ‘real’ decks.” Answering this question will force you to examine the representativeness of your testing, and force you to trust your testing should you conclude it is sufficiently representative and proceed with caution otherwise. If you can’t complete this task with regularity, you might as well not be testing at all.