Classifying Decks

How do we classify decks into strategies? Most people are comfortable with the concept of an archetype but it is not a well defined concept. As a community, we could change that. The issue is mainly that there are many dimensions to the game of magic and each deck will act on a subset of the different dimensions of the game in order to advance a specific strategy. The first step in classifying decks and strategies of playing magic is to define what the dimensions of interest are. Below is a short list of some dimmensions of strategy that I use to classify decks:

Reactivity: The tendency to interact with opposing strategies in a manner which is specifically reactive. Measured by the quantity of cards which could be used to react to an opponent's plays during a game. This is distinct from interactivity in that the decision to react is made after knowing the action that is being reacted to (ex: Force of Will, Mental Misstep).

Proactivity: The tendency to be the primary actor during a game. Measured by the quantity of cards which could be used proactively during a game. This could be defined as the opposite of Reactive; however, it also has a relationship to interactivity. There is often overlap between interactivity and reactivity but it is less common to have overlap between reactivity and proactivity. Similarly it is possible to be proactively interactive (ex: Duress, Arcbound Ravager).

Velocity: The tendency to cause the game state to change rapidly. This is important because it will correlate strongly with the number of cards in a deck which will be seen by the pilot of a given strategy during a game. Measured by the number of cards included which cause cards to change zones or to be moved from one region of the library to another (edit: such that new information is gained) (ex: Golgari Grave-Troll, Preordain). The term velocity has been used a few times before to describe different aspects of a strategy and I'm open to suggestions if someone can think of a better word to describe this dimension of strategy.

Mana Production/Dependence: Measured by the average casting cost paid for spells and abilities exluding cost reduction provided by other cards. This could be separated out into two categories but it may not be useful to do so since they should be highly correlated. Production could be defined as the total amount of mana the deck could produce with all of its mana sources in play. This becomes less meaningful when including cards such as academy which will not produce the same amount of mana consistently. Instead it would be better to calculate the expected mana the Tolarian Academy would produce by using the velocity of the deck to calculate the maximum number of artifacts that would be expected to be on the battlefield during the fundamental turn. Since that is a more involved calculation I suggest we only model strategies by measuring the mana dependence in those cases since they should correlate strongly with mana production. I would expect that the two extremes of the mana dependence/production dimension are Dredge at the very low end of the spectrum and Mud at the very high end of the spectrum; however, I have not taken the measurements yet.

Synergy: The tendency for a strategy to rely on the strength of interactions between its cards or to use cards in combination to achieve a key outcome. Measured by the number of cards which can only be used with a specific subset of cards within the same deck excluding the addition of mana except where special conditions apply (ex: Mishra’s Workshop, Voltaic Key).

Card Advantage: The tendency for a strategy to generate card advantage during game. Measured by the number of cards included which can generate card advantage.

Virtual Card Advantage: The tendency for a strategy to create game states that are comparable to ones where card advantage would occur. Virtual card advantage can be gained in mayn ways. One example would be by having mana sources that produce multiple mana continually thus reducing the number of cards needed to achieve the same quantity of available mana. Measured by the number of cards included which can generate virtual card advantage. Card Advantage and Virtual Card Advantage should be measured along the same axis but also distinguished as separate values.

Edit: (thought of this one after initial post)
Win Conditions: The tendency to proactively close out the game. Measured by the number of cards that can win the game.

Three examples of other dimensions that have historically been used to classify deck type would be the fundamental turn and fundamental cards. I think that they are useful but I also think that their usefulness as measurement device is hindered due to subjectivity. I would define them as follows:

Fundamental Cards(Pillars): Cards which the strategy requires in order to execute the designed sequence of its fundamental turn (see below).

Fundamental Turn: The turn upon which the deck is designed to execute the proactive component of its game plan in a goldfish scenario. Measured by gold fishing the deck and taking an average based on which turn the fundamental cards are played.

Edit: (moved from above due to subjectivity)
Interactivity: The tendency to interact with opposing strategies. Measured by the quantity of cards which are included to force interaction with opponents (ex: Sphere of Resistance, Force of Will). The inverse of interactivity, non-interactivity, is also an interesting metric and may be even more important to metagame analysys.

Once we have a set of dimensions that are communally deemed to be the most relevant to game play the next step is to draw out boundaries in that n-dimensional space to define where each archetype resides. If we used the above classification system to profile some of the decks that people consider to be control, aggro, combo and everything in between then we could use that as a basis for defining archetypes that is less subjective.

Can anyone think of any other dimensions that they use to classify decks and strategies of playing Magic?

last edited by Aaron Patten

@Aaron-Patten said:

Can anyone think of any other dimensions that they use to classify decks and strategies of playing Magic?

I tend to classify decks with a type (control, agro, combo) and a sub-type (gush, shops, dredge)

There can be cross-over decks - combo/control for example. The old tps or the perfect storm was a good example of this. As opposed to a dps deck now that is not running (m)any control elements.

I firmly believe in the classic 3 of combo, control, agro. Any label beyond that is just clarification of the tools being used. Its important clarification but clarification nonetheless. I also believe this method of classification allows for the most descriptive labels possible with the least confusion or cross over.

@Khahan Where would you draw those boundaries according to the dimensions listed so far? Is there a preferred data source you can provide that will allow me to take an average of each of the decks in the three tiered archetype system so that I can construct a set of boundaries that would encapsulate the archetypes in the manner you describe? There is always crossover but defining boundaries would mean that we would objectively know where that crossover starts and where it ends.

last edited by Aaron Patten

There was a long post on the previous version of The Mana Drain on classification schemes for Magic decks. It would be nice to link you to those, but my position begins with the notion that we have to dispense with the idea that there are perfect taxonomies.

Not even in the natural sciences or genomics do we have good classification systems.

In fact, many of the most long-standing classification systems from the natural sciences (going back to Linneaus, if not earlier) were shown to be inaccurate with recent advances in decoding genomes. This recent publication, for example, is pretty amazing, but very different then earlier systems.

People once thought, for example, that Whales were fish (read Melville's Moby Dick). We figured out that wasn't true even before the genomic revolution. People also once thought Panda's were Bears, and now we know that they are closer to Raccoons.

In fact, science hasn't even been able to determine exactly what is meant by "life."

So, with the big fat caveat that all classification systems are flawed or imperfect, here are my two cents:

I think you can organize Magic decks into many classification schemes, but the best are those that are flexible and multi-dimensional.

First of all: Strategic Orientation.

I think this is a very helpful way of organizing decks. In my Gush book, I have a whole chapter on this. Typical strategic orientation categories include:

Combo
Control
Combo
Aggro-Control

Also used are:
Combo-Control
Prison
Midrange
And even Aggro-Control-Combo

I think those are helpful categories, but certainly imperfect.

Another way of classification decks is by "Engine." An engine is any resource that used to power a deck, and can be mana, cards, or otherwise.

Draw engines include: Thirst, Gush, Fact, Standstill, or broader engines like Gushbond generate mana and cards.

Mana engines include Dark Ritual and Mishra's Workshop.

When trying to understand metagame structures, I think it's helpful to categorize decks in both ways. I also think it's helpful to track frequencies of card appearances, even when they aren't used as engines, but as what I call "strategic objectives."

Classifying decks according to cards in them is objective as you can be, and is always a good starting point, but I have no objection to overlaying other classification categories, like strategic orientation.

For the record, I think terms like Interactive, Reactive, etc. are inherently flawed, and I try to avoid them, although I don't always succeed.

last edited by Smmenen

@Smmenen I like the taxonomy analogy. I think it goes fairly deep. If the individual cards are the individual genes then the complete deck list is the genome. A taxonomy based purely on the presence of certain genetic sequences would be akin to a pillar deck classification system. If it contains gene A it must be type X , if B then type Y, etc. What I'm proposing is measuring the frequency of cards sharing common game actions and using that as a basis for comparison between decks. This would be similar to counting the different types of proteins the genes encode. Once established archetypes are coded in this way it's possible to define regions for each archetype similar to how you would draw a map. If we were to only model three game actions by measuring their frequencies in described model we would be able to define regions of that three dimensional space and classify decks by where they land among those regions. Adding more than three dimensions may be difficult to visualize but basic geometry holds in n-dimensional space and so it would be similarly possible to define n-dimensional regions for decks to be placed. The truth is that these regions have already been largely agreed upon by groups of players but haven't been measured empirically. We can all agree on a deck list we've played and decide it's control or aggro or combo but measuring it empirically allows for a whole new level of analysis which sidesteps issues of personal bias or differences in nomenclature.

I'm using Interactive and Reactive as titles for empirical sets of the model. They seem fairly easy to define and agree on for the purpose of counting and classifying individual cards. Why have you found them to be inherently flawed?

As an aside, I suspect that with enough of the right dimensions of game play included in the model there's a way to determine the function of each deck in a given match by counting these dimensions. Ultimately you'd be able to answer the question "Who's the Beatdown" before playing the match. This doesn't account for in game actions that a machine might not be apt to take such as bluffing but I would expect it to come close enough that it would be useful.

last edited by Aaron Patten

Can someone explain to me what the purpose of classifying even is? What is to be gained?

@Topical_Island I don't know if we can have an absolute answer to that question, but classifying decks allows for a (relatively) quickly identifiable codex of decks in a meta. If we didn't classify decks, we'd all be discussing "decks that do things."

I'm being (slightly) hyperbolic, but my point is that classification systems allow for efficient and reliable categorization of a meta game. Data is easier to analyze using blocking (categories), rather than totally randomized data.

As an aside, using blocking methods (deck categories) reduces variability when analyzing metagame data.

last edited by Horologium

@Topical_Island said:

Can someone explain to me what the purpose of classifying even is? What is to be gained?

By categorizing decks, we can make better decisions with less information.

@Aaron-Patten said:

As an aside, I suspect that with enough of the right dimensions of game play included in the model there's a way to determine the function of each deck in a given match by counting these dimensions. Ultimately you'd be able to answer the question "Who's the Beatdown" before playing the match. This doesn't account for in game actions that a machine might not be apt to take such as bluffing but I would expect it to come close enough that it would be useful.

For example: When you've pre-categorized a decks, you can make your in-game decisions a bit better. After a turn or two, you might not know exactly what deck your opponent is on but you can likely guess the category. This can steer your decisions while you try to decipher the more intricate details of your opponent's deck.

@ApolloGod Ok... so again. I really think that much of the difficulty around this issue stems not from the nature of cards and decks, but from our own poor tending of the terms and goals of the discussion as a community. There are two reasons people categorize decks (generally). To index them, and to analyze them.
These don't have to be the same system. I think we should have two (at least).

I don't agree with @Smmenen that Linean Taxonomy isn't "good". It functions just fine for what it does, create a rough index of living things based on likeness of form and then later, genetics. Has it needed some patches? Well, it used to classify minerals as living things... also humans of different races were considered species... so that was bad. But systemically if works just fine as an index. It's purpose is not to analyze the ecosystemic behavior of organisms. It's just to order them into categories and sub-categories in a logical way.

Take another system, this one also having to do with games: When I look at my encyclopedia of Chess openings, they're ordered by the opening sequence. King Pawn openings, Queen Pawn openings, Knight Openings, etc... They aren't ordered by other characteristics. I can't turn to the section on aggressive openings, or safe openings, or deep openings, or anything like that. It's just taxonomy to order them. That is all. Is the Orangutang opening more like the King's Gambit, since they both tend toward open games and involve a pawn sacrifice? Maybe in function... but the very stable French Defense is in the same section of the book as the craziest King's Gambit games, because they both begin with E-5. It's just taxonomy.

If I wanted to look at openings along a different vector, I could start talking about which openings lead to the most aggressive games. We could do data analysis and see which opening sequences lead to the most checkmates, to the most draws, to the longest or shortest games. We could do the same thing in biology to see what the relationship between different sub-categories of organisms might be, to certain ecosystemic niches. I'm willing to bet that Fungi tend to for symbiotic relationships much more often than Animals, for example...

But with these cards, the current state of this discussion is hopelessly nebulous. We have some people trying to create taxonomies, other people trying to create ad-hoc categories for analysis purposes (which is a fine thing to do, but really confounds the conversations on taxonomy) And categorization along different vectors is flying around everywhere like a snowstorm. We are essentially having to deal with the magic equivalents of arguments like "A remora should be classified as a fungus, since its function is to be symbiotic to larger fish." We can learn a lot from looking at other taxonomic systems. The first thing we should learn is that doing taxonomic classification based on ecosystemic function (or position in the metagame, as it's known in this case) just creates a mess.

I propose we come up with a purely taxonomic system first, based only around what cards are and aren't included. This will be rough of course, but that's going to be true of any taxonomy that gets imposed over an organic system. There are going to be oddities. (Leopards and Jaguars can interbreed and produce a viable offspring... sometimes, but they exist on different continents, so for facility's sake) But doing this for Vintage decks would be useful, if for no other reason than tidiness of threads on this website. To the extent that its made clear that that system is purely taxonomic, we can put to bed the fruitless discussions that conflate taxonomy and function and do nothing but chase their own tails.

@Aaron-Patten said:

@Khahan Where would you draw those boundaries according to the dimensions listed so far?

Who says I would draw those boundaries with the dimensions you listed? Don't take that the wrong way. I'm not suggesting there is anything wrong with the ideas you presented. Its just that you have your ideas, I have my ideas, topical_lsand has his own ideas and every single one is valid for that person.

@Aaron-Patten said:

As a community, we could change that.

This is a good goal. It would be nice to look at a thread on this site or any other site and be on the same page as the OP with basic common terminology. I'm going to do something that I hope will direct this thread a bit better. Right now we have different people arguing over their own personal methodology in an abstract form. I'm going to link to a few decklists and ask those of us who are discussing their own systems - how would you classify each deck in your own system?

1. Fenton Oath list in the OP of this thread:

2. UWR deck piloted to 2nd at a 20 man tourney last month:

3. Gush Mentor decklist found here:
http://www.mtgstocks.com/decks/114568

4. Pitch Dredge:
http://tappedout.net/mtg-decks/pitch-dredge-vintage/

I just pulled the few decklists I either found or that popped into my head. Again, the purpose is just simply to demonstrate how your own personal categorization system works.
I would use the following descriptors:

So first, my categories:
a) control - wins the long, slow game. Multiple control elements to deny your opponents gameplan. This can be done thru countermagic (spell denial), mana denial (waste/strip/crucible/chalice/null rod) or any other aspects (discard, creature removal etc). It generally slows your opponent's gameplan down until you whittle them away

b) agro - wants to win by quickly overwhelming your opponent with creatures or spells. It sets an aggressive tempo to the game that tries to get your opponent to 0 before they can execute their gameplan

c) combo - Relies on interactions from cards within your own deck to create a game winning situation.

These are the 3 main categories I use to define a deck and my goal in defining decks is to help me develop deck building, sideboarding and in-game play strategies. Its an overly simplistic method that needs further refinement which I give with sub-categories. While these 3 main categories have held true since day 1 of Magic, the sub-categories shift and change but help define how a specific deck wants to accomplish its goal. Its based on the card(s) central to that deck.
So for the above examples:

1. Landstill Control. Landstill at its heart is a control deck. It controls multiple aspects of the opponents gameplay thru mana denial, counter magic and card advantage. I don't need further descriptors beyond landstill because it generally uses wastes and counter magic to slow down the opponent, standstill for card advantage and man-lands for a win con. Generally speaking we all know this last part and recognize a landstill list just from calling it landstill. This is a deck where its namesake is sufficient but beyond that I'd call it control.

2. Fenton Oath - combo/control. Like landstill the deck name gives us a lot of information we need to know how to play against the deck as Fenton Oath is a specific style of oath. It wins thru a combo of forbidden orchar/oath of of druids on the board or key/vault combo. It also employs heavy control elements to make sure it gets its combo off

3. UWR - UWR is really not a good descriptor as it just states colors and doesn't give us much information. But that particular one I would classify as a control deck. If somebody told me I was playing against a UWR control deck I would have a pretty good idea of most of the cards they are playing. I would expect to see FOW and some other counter magic. Lightning bolt and/or sudden shock, swords to plowshares. And I'd expect to see a small number of creatures to execute a win. could probably guess at dack and/or jace as well. I may not know the exact number of each card in the deck and I may not know if I'm seeing bolt or swords but I can guess I'm probably seeing a 1CC removal spell.

4. Gush mentor - this is why I wanted to link a decklist because gush/mentor encompasses so many different styles. I would call this list a Gush/Mentor Control deck (probably the most common). Calling it gush/mentor control tells me I don't have to worry about a common back up tendrils kill and I would need to really pick/choose how to use my own counter magic. Knowing its gush/mentor means I'm going to see mentor and a lot of cheap and/or alternate casting cost draw spells that will cycle thru the deck a keep my opponents hand full. Calling it the control version tells me to expect more counter magic than my own build for example.

5. Dredge - some may disagree with me but I would call classic dredge an agro deck. Its trying to set up a horde of creatures to beat my face in and its trying to do it so quickly that I don't have a chance to respond and there is little interaction for me - either proactive or reactive interaction (at least from the perspective of the dredge player trying to execute his game plan). Dredge could arguably be called a 4th category - a rule breaker. It takes a fundamental rule of how to play magic and wins by exploiting that rule and turning it on its head. But comfortably fits into agro for me. I could also conceivably define it as combo. But either way calling it dredge gives me the important info

6. Pitch dredge - agro control. Same agro gameplan but it sacrifices a bit of redundancy for control components. I know I'll be seeing free/alternate casting cost counter magic to back up its gameplan of getting creatures from the graveyard

The combo/control/agro monikers hold true across the full lifespan of MtG. The use of these 3 broad categories along with other descriptors like deck name or primary cards in a deck tells us a lot of what we need to know to achieve my goal of categorizing decks.

@Aaron-Patten said:

I'm using Interactive and Reactive as titles for empirical sets of the model. They seem fairly easy to define and agree on for the purpose of counting and classifying individual cards. Why have you found them to be inherently flawed?

As I said, there was a HUGE thread in the old forums on this subject started by MTGFan, but the gist is that there is no objective and empirical way to distinguish, at the tactical level, between a card that is being interactive and a card that is being non-interactive.

Consider, for example, that a Force of Will stopping Necropotence appears interactive, while a Force of Will defending Necropotence appears non-interactive. Yet, they are both the same tactic. Only when we widen the lens to the strategic context, does one appear interactive and the other not. That context is very difficult to define in a way that is universalizable, consistent, and objective, let alone empirical.

Or a Sphere of Resistance to lock a player out seems non-interactive, while doing so to slow down an Academy combo deck (Mike Flores example in his article on interactivity) appears interactive. Trinisphere is both the epitome of non-interactivity and ultimately interactivity.

I'm not interested in getting into a debate on this point, but I don't believe that interactivity, as distinguished between non-interactivity, is logically defensible. I think all magic cards are inherently interactive.

Reactivity suffers the same problem, but worse. Is the Force of Will stopping Necro reactive? If so, why isn't the Force of Will countering the Force on your Necro? Sure, you can give me the common sense answer, but I can reply ad infinitum challenging that answer as lacking an objective basis. We can create crazy hypotheticals that can disprove any broad rule that can be developed.

For example, even something that appears to be aggressive can be framed as reactive or controlling, like attacking a player who controls Necropotence, not primarily to reduce their life total to win the game, but to reduce their capacity to draw cards.

last edited by Smmenen

@Topical_Island said:

I propose we come up with a purely taxonomic system first, based only around what cards are and aren't included. This will be rough of course, but that's going to be true of any taxonomy that gets imposed over an organic system. There are going to be oddities. (Leopards and Jaguars can interbreed and produce a viable offspring... sometimes, but they exist on different continents, so for facility's sake) But doing this for Vintage decks would be useful, if for no other reason than tidiness of threads on this website. To the extent that its made clear that that system is purely taxonomic, we can put to bed the fruitless discussions that conflate taxonomy and function and do nothing but chase their own tails.

This is no-doubt useful. It also has little difference to the pillar system of classification that the original mana drain site used for navigation. I too would like to put fruitless discussions to bed which is why I would like to construct a common means to empirically measure strategic classification. Both are useful and both are empirically measurable so they'll both help put fruitless discussions to bed. So I agree, let's use both.

@Khahan said:
Thanks for the reply. In answer, I would have to measure the count of each card type based on the dimensions described. Currently my goal is not so much to change the classification system that has stood for so long but instead to measure it imperially so that unbiased comparisons can be drawn from it. So, what I would do is count each card in the main deck according to the multi-dimensional classification system and then use your own classifications as labels to mark the regions of n-dimensional strategy space that those decks fall into. The more decks that are classified based on historical accounts the better defined each region becomes. Part of the source of the disagreement is that people have different opinions about where those boundaries should lie for the purpose of labeling a deck as control, aggro, or combo. This won’t really solve that by its self but I hope it can act as a common basis for comparison when analyzing strategies.

Reactive/Proactive 9:
1 Ancestral Recall V C
1 Brainstorm V vc
1 Decree of Justice C W
1 Dig Through Time V C/vc
1 Strip Mine
4 Wasteland

Reactive 16 (only affecting things that have already occurred):
1 Disenchant
1 Flusterstorm C
4 Force of Will
2 Mana Drain
2 Spell Pierce
3 Swords to Plowshares
3 Mental Misstep

Proactive 17 (only affecting the future):
2 Crucible of Worlds C S
2 Jace, the Mind Sculptor V C/vc W
3 Mishra's Factory vc W
2 Monastery Mentor C W
1 Ponder V vc
1 Preordain V vc
1 Sensei's Divining Top V vc
4 Standstill V C
1 Time Walk V

Purely Mana Production 17:
1 Black Lotus
2 Island
1 Misty Rainforest V
1 Mox Emerald
1 Mox Jet
1 Mox Pearl
1 Mox Ruby
1 Mox Sapphire
2 Polluted Delta V
2 Scalding Tarn V
4 Tundra

Total Reactive: 25 (9+16)
Total Proactive: 23 (9+17)
Total Velocity: 18 (attained by counting V)
Mana Dependance: 56/42=1.333 (average minimum mana cost)
Synergy: 2 (Crucibles are the only cards that do nothing alone)
Virtual Card Advantage: 7 (counting vc)
Win Conditions: 8 (counting W)
I've excluded the dimensions that I deemed subjective and thus not strictly empirical such as interactivity (as Steve pointed out), fundamental turn, and fundamental cards. Virtual card advantage seems the most debatable metric in this model currently. I may decide to move Wasteland and Strip Mine to strictly reactive since it's only proactive action is as mana and the other mana sources were not included in the list of proactive cards.

1. Fenton Oath list in the OP of this thread:

2. UWR deck piloted to 2nd at a 20 man tourney last month:

3. Gush Mentor decklist found here:
http://www.mtgstocks.com/decks/114568

4. Pitch Dredge:
http://tappedout.net/mtg-decks/pitch-dredge-vintage/

I just pulled the few decklists I either found or that popped into my head. Again, the purpose is just simply to demonstrate how your own personal categorization system works.

I’ll model the others some time in the near future (perhaps as an edit) unless someone else would like to give it a try. I largely agree with your descriptors and the way they are used in tandem and I suspect others do as well. We can use them as a basis for labeling.

@Smmenen said:

@Aaron-Patten said:

I'm using Interactive and Reactive as titles for empirical sets of the model. They seem fairly easy to define and agree on for the purpose of counting and classifying individual cards. Why have you found them to be inherently flawed?

As I said, there was a HUGE thread in the old forums on this subject started by MTGFan, but the gist is that there is no objective and empirical way to distinguish, at the tactical level, between a card that is being interactive and a card that is being non-interactive.

Consider, for example, that a Force of Will stopping Necropotence appears interactive, while a Force of Will defending Necropotence appears non-interactive. Yet, they are both the same tactic. Only when we widen the lens to the strategic context, does one appear interactive and the other not.

I see both of the examples you've presented as being reactive since they are both "acting in response to a situation rather than creating or controlling it". I don't think that my previous post disagrees with any of the statements you've made here but I didn't strictly define "non-interactive" except as the inverse of interactive which I see now is difficult to define. I can agree that all game actions are interactive by virtue of the nature of the game being interactive but if we apply a filter by only considering the changes in state within a single turn there is still useful information to be gained by that consideration.

Force stopping Necropotence is interactive because it involves the decisions of both players within that turn. It is reactive because it is a decisions made after the initial action of casting the Necropotence that turn. It is not proactive since it is not the action being performed first in the turn.

Force stopping the Force stopping the Necropotence is interactive because it involves the decisions of both players within that turn. It is reactive because it is a decisions made after the initial action of casting the Initial Force of Will targeting Necropotence that turn. It is not proactive since it is not the action being performed first in the turn.

In both cases the result is the same, It is a reactive card in either case. It's the Necropotence which is proactive. An example that I would consider to be only proactive would be attacking a player with a Ball Lightning when there are no blockers. It's a rare event but the only interaction is with a player's life total instead of other cards. By ending the game a player is ending interaction. It's necessary but not sufficient to the game. The game collapses to have no more strategic consideration than war (the card game) if all plays are proactive. The goal of the model is to classify cards and use them to classify decks so individual game actions are not as important as the set of possible game actions involving individual cards. A creature with no abilities other than to generate or change power and toughness on the board could be counted as proactive without being considered in the model as interactive or reactive. At this point though I'll concede that defining "interactive" may not be as sound of a resource for empirical measurement. It may not even be necessary.

The simpler, and perhaps better, approach could be to only consider reactivity and proactivity of cards. reactive, and proactive have overlap but are well defined concepts. Their dictionary definition appears readily applicable to actions taken in games of Magic such that they can be counted/measured.
Proactive: Creating or controlling a situation by causing something to happen rather than responding to it after it has happened.
Reactive: Acting in response to a situation rather than creating or controlling it.
Assault and battery would be an example of a card that could be counted as both proactive and reactive since it can be played both in a way that is "acting in response to a situation rather than creating or controlling it" but also in a way that is "creating or controlling a situation by causing something to happen rather than responding to it after it has happened".

Or a Sphere of Resistance to lock a player out seems non-interactive, while doing so to slow down an Academy combo deck (Mike Flores example in his article on interactivity) appears interactive. Trinisphere is both the epitome of non-interactivity and ultimately interactivity.

Agreed, it is interactive but the action taken is proactive since it is "creating or controlling a situation by causing something to happen rather than responding to it after it has happened". The Sphere of Resistance happens first and the actions it interacts with happen later. A counter example would be Swords to Plowshares since the spell can only be played once a creature is on the battlefield. It requires a game action to happen first before it can happen. Of course playing land and adding mana have to happen before anything else but the mana dimension can be left out of determining whether a card is to be counted as reactive and/or proactive. It is, after all, accounted for along another dimension of the model.

I'm not interested in getting into a debate on this point, but I don't believe that interactivity, as distinguished between non-interactivity, is logically defensible. I think all magic cards are inherently interactive.

Agreed, all magic cards are interactive on some level since the game involves an interaction between two people at its simplest. So, perhaps it is best to not use this as a metric since it is not as well defined in a game that is inherently interactive to begin with.

Reactivity suffers the same problem, but worse. Is the Force of Will stopping Necro reactive? If so, why isn't the Force of Will countering the Force on your Necro? Sure, you can give me the common sense answer, but I can reply ad infinitum challenging that answer as lacking an objective basis. We can create crazy hypotheticals that can disprove any broad rule that can be developed.

I don't see how that is the case. The definitions for proactive actions and reactive actions can be defined by which of them happens first, relative to the point of decision, for the purpose of this model. People may disagree with it; however, it would still be meaningful and consistent. If we use the dictionary definition of reactive and proactive for the model then the claim that it is lacking objective basis would be difficult to defend without compromising the defenses' definition of objective. To say that it is lacking objective basis would be implying that the dictionary is not objective which, while true, is not necessarily pertinent. It would be similar to argue that a person is wrong simply because perception is un-provable and thus there is no such thing as knowledge. Ultimately I have the utmost faith that swaths of magic players will inevitably find a way to wholeheartedly disagree with the dictionary; however, this force of nature will not dissuade me from using it as an objective basis for a modeling tool.

The first thing that comes to mind as something that someone would present as an argument against using these definitions is to say that Sphere of Resistance and Force of Will are both intended to prevent actions of the opponent. I would agree with that assessment; however, there is still a distinction between the two that I find relevant. Similar to the distinction between classic control and prison. "Preventativness" may be difficult to measure empirically though so fitting it in an empirical model may not be possible. Channel Fireball for 20 on turn one also prevents a lot of game actions the opponent may have been planning

This reply went a little longer than I thought it would and I’ve noticed that can sometimes be taken as adversarial so allow me to qualify this post by saying that I appreciate all of your input and that I hold your opinions in high regard.

last edited by Aaron Patten

@Aaron-Patten said:

@Smmenen said:

@Aaron-Patten said:

I'm using Interactive and Reactive as titles for empirical sets of the model. They seem fairly easy to define and agree on for the purpose of counting and classifying individual cards. Why have you found them to be inherently flawed?

As I said, there was a HUGE thread in the old forums on this subject started by MTGFan, but the gist is that there is no objective and empirical way to distinguish, at the tactical level, between a card that is being interactive and a card that is being non-interactive.

Consider, for example, that a Force of Will stopping Necropotence appears interactive, while a Force of Will defending Necropotence appears non-interactive. Yet, they are both the same tactic. Only when we widen the lens to the strategic context, does one appear interactive and the other not.

I see both of the examples you've presented as being reactive since they are both "acting in response to a situation rather than creating or controlling it". I don't think that my previous post disagrees with any of the statements you've made here but I didn't strictly define "non-interactive" except as the inverse of interactive which I see now is difficult to define.

Now that you conceded part of my argument, let me just say that I am not interested in drilling down too far in the interests of helping you refine your model here, because I think it's a fundamentally flawed project, but I will poke a few more holes.

Force stopping Necropotence is interactive because it involves the decisions of both players within that turn. It is reactive because it is a decisions made after the initial action of casting the Necropotence that turn. It is not proactive since it is not the action being performed first in the turn.

Force stopping the Force stopping the Necropotence is interactive because it involves the decisions of both players within that turn. It is reactive because it is a decisions made after the initial action of casting the Initial Force of Will targeting Necropotence that turn. It is not proactive since it is not the action being performed first in the turn.

In both cases the result is the same, It is a reactive card in either case. It's the Necropotence which is proactive. An example that I would consider to be only proactive would be attacking a player with a Ball Lightning when there are no blockers. It's a rare event but the only interaction is with a player's life total instead of other cards. By ending the game a player is ending interaction. It's necessary but not sufficient to the game. The game collapses to have no more strategic consideration than war (the card game) if all plays are proactive. The goal of the model is to classify cards and use them to classify decks so individual game actions are not as important as the set of possible game actions involving individual cards. A creature with no abilities other than to generate or change power and toughness on the board could be counted as proactive without being considered in the model as interactive or reactive. At this point though I'll concede that defining "interactive" may not be as sound of a resource for empirical measurement. It may not even be necessary.

The simpler, and perhaps better, approach could be to only consider reactivity and proactivity of cards. reactive, and proactive have overlap but are well defined concepts. Their dictionary definition appears readily applicable to actions taken in games of Magic such that they can be counted/measured.
Proactive: Creating or controlling a situation by causing something to happen rather than responding to it after it has happened.
Reactive: Acting in response to a situation rather than creating or controlling it.
Assault and battery would be an example of a card that could be counted as both proactive and reactive since it can be played both in a way that is "acting in response to a situation rather than creating or controlling it" but also in a way that is "creating or controlling a situation by causing something to happen rather than responding to it after it has happened".

Or a Sphere of Resistance to lock a player out seems non-interactive, while doing so to slow down an Academy combo deck (Mike Flores example in his article on interactivity) appears interactive. Trinisphere is both the epitome of non-interactivity and ultimately interactivity.

Agreed, it is interactive but the action taken is proactive since it is "creating or controlling a situation by causing something to happen rather than responding to it after it has happened". The Sphere of Resistance happens first and the actions it interacts with happen later. A counter example would be Swords to Plowshares since the spell can only be played once a creature is on the battlefield. It requires a game action to happen first before it can happen. Of course playing land and adding mana have to happen before anything else but the mana dimension can be left out of determining whether a card is to be counted as reactive and/or proactive. It is, after all, accounted for along another dimension of the model.

I'm not interested in getting into a debate on this point, but I don't believe that interactivity, as distinguished between non-interactivity, is logically defensible. I think all magic cards are inherently interactive.

Agreed, all magic cards are interactive on some level since the game involves an interaction between two people at its simplest. So, perhaps it is best to not use this as a metric since it is not as well defined in a game that is inherently interactive to begin with.

Reactivity suffers the same problem, but worse. Is the Force of Will stopping Necro reactive? If so, why isn't the Force of Will countering the Force on your Necro? Sure, you can give me the common sense answer, but I can reply ad infinitum challenging that answer as lacking an objective basis. We can create crazy hypotheticals that can disprove any broad rule that can be developed.

I don't see how that is the case. The definitions for proactive actions and reactive actions can be defined by which of them happens first, relative to the point of decision, for the purpose of this model. People may disagree with it; however, it would still be meaningful and consistent. If we use the dictionary definition of reactive and proactive for the model then the claim that it is lacking objective basis would be difficult to defend without compromising the defenses' definition of objective. To say that it is lacking objective basis would be implying that the dictionary is not objective which, while true, is not necessarily pertinent. It would be similar to argue that a person is wrong simply because perception is un-provable and thus there is no such thing as knowledge. Ultimately I have the utmost faith that swaths of magic players will inevitably find a way to wholeheartedly disagree with the dictionary; however, this force of nature will not dissuade me from using it as an objective basis for a modeling tool.

The first thing that comes to mind as something that someone would present as an argument against using these definitions is to say that Sphere of Resistance and Force of Will are both intended to prevent actions of the opponent. I would agree with that assessment; however, there is still a distinction between the two that I find relevant. Similar to the distinction between classic control and prison. "Preventativness" may be difficult to measure empirically though so fitting it in an empirical model may not be possible. Channel Fireball for 20 on turn one also prevents a lot of game actions the opponent may have been planning

This reply went a little longer than I thought it would and I’ve noticed that can sometimes be taken as adversarial so allow me to qualify this post by saying that I appreciate all of your input and that I hold your opinions in high regard.

Ok, so the key part are your definitions:

Proactive: Creating or controlling a situation by causing something to happen rather than responding to it after it has happened.
Reactive: Acting in response to a situation rather than creating or controlling it.

So, what is meant by "response"? And how do you define "a situation"?

If you play a turn one Birds of Paradise, and, ten turns later, I cast Wrath of God, destroying it, is that reactive? By your definition of reactive, it is.

What if I Swords to Plowshares a Wall of Fire, even though it never interferes with my game plan, since I have all flyers? Is that reactive?

Does a "situation" have to be temporarily proximate - that is, close in time? Does it have to happen on the same stack? In the same phase? In the same turn?

Your terms here are large enough to drive trucks through.

There are all kinds of cards that can really screw up your definitions.

What if you play a spell, say Ancestral Recall, and I respond with Snapcaster Mage to replay a Lightning Bolt on your Delver? Is that "reactive to the situation"? It's not reactive to you playing Ancestral.

Or, what if we put Vedalken Orrery/Leyline of Anticipation into play? What if you cast Ancestral Recall, and I respond by casting Necropotence? Is that responding? I can't use the Necro to draw cards to stop your Ancestral? But maybe I want to maximize it's chances of resolving?

Also, what if the situation is just the game? If you play a land, don't I have to play lands to develop my board? Isn't play land then reactive to your development?

According to your definition, I believe that playing lands after the opponent has played a land is reactive.

• 13
Posts
• 7379
Views