So you've just come up with a cool new idea for a deck and slapped it together, now how do you know if it works? My method of testing to this point has been erratic. I build a deck then run the gauntlet, playing it against a number of random decks until it gets beat, then going back and tweak it. This testing technique has a number of problems. On average does the deck lose because it doesn't function as intended? Did it lose because of a bad draw, or an opponent's good draw? Did it lose because it is vulnerable to a particular strategy? It's hard to do more than make gut decisions without appropriately diversified and documented testing.

What I am working on is a logical testing sequence to subject a deck to. My biggest challenge is decks being vulnerable to certain strategies. Often these are strategies that I don't often employ myself—like a creatureless burn deck. The goal with this first attempt at formalized deck testing is to make sure that a new deck is not vulnerable to a particular deck type.

But this shit is boring as hell! I know, I know, that's why I'm only tackling one problem at a time, and making sure to build in a large variety of opponents. In fact, if this works as I intend you will actually end up playing against a greater variety of decks.

To test a new deck I propose to play it against a series of decks in a variety of categories and keep track of wins and loses. Also, because losing against the computer AI is rare, unless the decks are particularly mismatched or you get a really bad draw, I would also keep track of games that were "close" because had a good human player been running the opposing deck it would most likely have been a loss.

There can't be too many categories so you can run a deck against a random opponent in each without getting bored playing the same deck. But they must also be specific enough to prepare a deck for a good range of strategies. Of course in reality better decks will usually deploy a variety of strategies, but you've gotta prepare for them all, right?

The following are the categories that I'm trying out, with sample decks (mostly my own and Mark's).

Counterspell/Control

• Control
• Type One Counterspell
• The Deck
• Kim
• Four Color Counterspell

Land Destruction

• Type One Land Destruction
• Edge
• Edge of What?
• Land Destruction (type 1.5)
• Landkill T1

Fast Creature (Large)

• Sadomasochistic Djinn
• 5 is good
• Leif Type One
• First-Turn T1
• Juggernaut (type 1.5)

Fast Creature (Small)

• Armies of Light
• Explosion
• Plague Rat
• Lightning Sprites
• Nils' Type One

Burn

• Atog 1
• War Mage
• Crag Hydra
• Dante
• The Flame Thrower (from Master Magic)

Environment/Lock/Combo

• Barbed Apes
• Blue Skies
• Nevinyrall's Wrath
• naf's psychic orb
• Island of Pain (coming soon)

So that's about it. I play a new deck against a random deck in each category. Then I evaluate where the deck's weaknesses are. If it lost to Land Destruction or Fast Creature (Small) then it might not be fast enough. If it lost to Fast Creature (Large) or Burn then it might not have enough staying power. If it got locked down by a nasty environmental enchantment or artifact then it might not have enough non-creature threat removal. Every deck will have it's own means of dealing with these problems, but the first step is finding the weaknesses.

posted by Nils T. Devine  # Sunday, June 05, 2005