Wednesday, December 12, 2018

Tactics Versus Strategy

Many people think that strategy is merely tactics on a larger scale. It is immensely frustrating to discuss because no matter how many times you describe the qualitative differences they recognize only quantitative differences. People conflate tactics with large numbers with strategy and ignore the different kinds of thinking at work. It's almost like people think that strategy is the smart word for tactics, similar to people who always use "whom" for "who" because it sounds smarter. Let's agree the world needs both tacticians and strategists, but conflating them is counterproductive.

This is particularly relevant to game design because much of what we call strategy is really tactics. We rarely ask whether a game models a process or the thinking one uses when engaging with a process. Therefore any game involving armies is called strategic even if they are nothing but hit point containers. Let's take this to the most absurd limit. Maybe you've designed a game of immense scale: galaxies fighting galaxies. Each galaxy has a number of civilizations in it, and every time you deal damage to an enemy galaxy it loses a few civilizations, but every few turns (called eons) each galaxy regenerates a few civilizations. Furthermore, within certain constraints you can control how your galaxy expands, contracts, or revolves - how it moves - through the universe, so you have some choice which neighboring galaxies you can attack. Maybe the board on which the combat takes place expands for the first half of the game and contracts for the second half, and whoever controls the last remaining galaxy before the next Big Bang wins. We could even add some push/pull mechanics so attacks affect the expansion / contraction of the universe.

Great, but is anyone going to argue that this 4th edition D&D miniatures game (but I repeat myself) is strategic? You laugh, but if I included a little more detail (each civilization contains a number of societies, and a civilization is destroyed when its society points reach zero!) people would claim this is a strategy game. Maybe more detail, like a rock-paper-scissors game? When a galaxy fights, it uses tech, culture, or politics. Politics beats tech, tech beats culture, and culture beats politics. Wow, so deep, so complex, so strategic! - and yet all we've done is mask the underlying mechanics of a basic tactics game.

On the other hand, maybe in real life you have a basic policy of being honest, and not pretending to be someone you're not. Because you have a few quirks, you rub some people the wrong way, but over time you surround yourself with honest people who appreciate you. In any given situation you're not too concerned if someone likes you, so much as being yourself. You're certainly not calibrating your persona to match the environment like some kind of calculating, social chameleon. You've seen that backfire more times than you can count.

Can we call this strategic thinking? (Could it be that nerds are worse at strategy than cheerleaders?) You're not concerned with maximizing each individual encounter. You looked at a situation and picked a basic course of action with a tendency to move you in the right direction.

Scale alone isn't the key factor. You can have a tactical galaxy game and a strategic interpersonal situation.

So here's my attempt at working out the difference. I doubt my definition is complete. I hope the comment section improves it - so long as people don't conflate the Galaxy Game, tactics with big numbers, with strategy.

Tactical thinking is calculating odds, risk, reward, and taking measures that maximize gain / loss. Strategic thinking sidesteps the issue of odds entirely and develops outcome independent systems of analysis / action. Those systems are designed to produce situations mappable within themselves and (ideally) not mappable within your opponent's, sort of like a conceptual OODA loop. Tactical thinking is here's how my guy beats up your guy. Strategic thinking is let's focus on controlling the camera. (When my guy wins he's strong and your guy deserved it. When my guy loses, he's the underdog and your guy is a bully.)

I think it's the outcome independent aspect that defines strategy. The mappable part is simply saying you've thought far enough ahead that every situation is a win-win, as opposed to pushing circumstances into an area you're not prepared to analyze or take action in. Mappable also means, all things being equal, you prefer action that gives you more information about your environment / opponent, again, so you can engineer win-win situations. For instance, if you're a freelancer, you learn more about the market if you charge too high than too low. Or, you set boundaries with people to reduce conflict.

How does this help game design? It will help you ask better questions. With accurate definitions you can see if your game offers competing strategies, or if the fundamental level offers only one proper way to play.

What do you think?

Personally I'm not completely sold on the definition. I'll need time to mull it over. There are some things to look for, however. A proper distinction between strategy and tactics needs to take into account:
  • You can lose on a lower level and win on a higher level
  • Strategy is such that although you create the conditions for victory, you rarely predict exact circumstances
  • Tactics has more to do with discreet cause and effect
  • At the strategic level cause-and-effect is harder to sort out; a situation can't be understood without taking action; consequently, certain types of action preclude certain types of knowledge; alternatively, there is less of a distinction between observation and action
  • Tactics tend to shift from situation to situation, as circumstances arise
  • Often a strategy will recommend one basic action, often one that creates its own luck


  1. Thanks for this definition that gets at what Vox talks about with this distinction. I feel like I understand what he's getting at a lot better having read this.

    I can see why he doesn't go into detail about it very often; it's tricky to explain what strategy is in a meaningful way.

    In a game context, is it necessary for the goals the player has to be set by him in order for strategy to be relevant? Do you need a wide scope of win conditions? Or can you have meaningful strategy when the goal is "score the most points"? In your example, the cheerleader sets her own goal for how to handle her interactions.

    By this definition, would an NFL coach's approach to one game be tactical, and potentially his approach to the season could be (but wouldn't necessarily be) strategic?

    For example, if Belichick pulls back in most games in order to both conserve his best players and their health, and to create uncertainty about how good his team actually is, is that strategic?

  2. I have no idea what Vox thinks about this. I was reading Creveld's history of military strategy last week and even he didn't offer a definition. So I'm prepared to be wrobg.

    I think you can have strategy when there's only one wincon. There just has to be multiple ways to achieve it or, more importantly, enough complexity that you can't map out chains of cause and effect from start to finish. Instead you have to recognize shapes, tendencies, abstractions, efficiencies - and be able to map those out like a big flowchart. (The difference between this knowing discreet cause and effect versus situation mapping is that in the ladder you don't know much much time it takes to move from flow chart box to the other. Instead you focus on accurately determining which box you're in and what actions you should take to shit off movement to unfavorable boxes.)

    Also, any game more complex than tic tac toe allows for strategic thinking. The question is if a game allows for competing strategic paradigms, or if there is only one proper way to think about a game's long term solution.

    Belichick risking games by conserving player health is certainly short turn loss for long term gain. But its missing that element of creating its own luck. It's still tradeoff thinking.

    Belichick's habit of trading away expensive super stars for underrated value players (with the exception of Brady) has that create its own luck quality, however. Not only do you get good players cheap, but you train them from the get go in your own system, so despite the numbers any individual achieves he's actually less valuable to another team - effectively trapping them in the Patriots roster. Also, because Belichick has been doung this for a long time no one tries to bargain for too much money. You get too expensive, you're out, and Belichick hires 3 guys for the price you wanted, and one of them ends up being better than you. This is how you get guys like Bruschi, Wilfork, Gronk, etc., that were great Patriots players, year after year, but never scooped up by another team. Now even more reinforcing loops. From that kind of a core you build a team where people want to stay because theres some actual comraderie.

  3. Now that I think about it, the best summary of my proposed difference would be this:

    Strategic thinking is making true statements about variables even if their values are unknown.

    Tacticial thinking is determining the value of a variable.

    1. Revised: Tactical thinking is determining the value (or most likely value) of a variable.

    2. Revised, again: ...or most likely set of values...

  4. I like the definition that comes from the chess world. In chess, tactics are the moves you make as you play the game. Strategies are the long-term goal you hope to achieve with those tactical moves. Strategies are built out of tactical moves. You can have tactics without strategy, but can you really have strategy without tactics?

    > Belichick risking games by conserving player health is certainly short turn loss for long term gain. But its missing that element of creating its own luck. It's still tradeoff thinking.

    I agree with that. The goal is to win the most games possible in a season. The tactic is to conserve a key player's health on the games they're going to win anyway, similar to sacrificing a queen to draw out a king for check-mate. But, in both cases, the tactic is part of the strategy. Different overall strategies can have specific tactics in common with each other, but the overall goal can be different.

    1. That's a good definition when the contest has discreet moves. There's still a difference between a move and a tactic though. Moving your bishop to H1 is a move. Pinning the knight on F6 is a tactic. Maximizing your own options while minimizing your opponent's is a strategy.

      We'd have to call the tactic the reason for the move. But I can see how in everyday speech that distinction gets blurred.

      You can play with a poorly thought out / unrigorous strategy: i.e., capture as many pieces as I can, so for a player to have tactics and no strategy at all is probably rare. But solid tactics and poor strategy is a common pairing.

      Also, I misrepresented the above commment. Another reason Belichick pulls players is to create uncertainty. I would still call this a tactic, maybe a subset that I would define as a strategem: a tactic that becomes less effective woth use.

  5. For a game if you offer multiple goals to choose from for victory it is easy to say you have a strategic layer. You have to pick which goal to chase which will define what you do. And that means if your game does not have enough room for you to define goals underneath the single goal of say, defeating the opponent, then it has no strategy.

    But doesn't that make any game have a strategic layer implicit to the tactical layer? If I'm playing a squad based game we consider that tactics. But in the context of the game, I have strategy for perhaps choosing forces and deciding which hill or woods to occupy to win.

    Those choices are only tactical if there is a higher level above them to be strategic at. We think of them as tactical because we implicitly understand they operate underneath the army.

    I think that strategy would be deciding on your path to victory, and tactics is making that path efficient.

    1. >I think that strategy would be deciding on your path to victory, and tactics is making that path efficient.

      That's true, at least the first half, but it doesn't address the issue of losing at one level and winning on the other. Moreover, the definition presupposes victory and omits interaction with your opponent.

      "Path" implies a certainty of information not always available at the strategic level. Its not wrong, but it doesn't invite the best questions.

      The principle unknown information is what is your opponent going to do? A great strategic mind is going to have a plan that incorporates that unknown information.

      A strategy by its very nature makes itself effecient, like Belichick's salary scheme as I described it. The tactics work because the strategy works, if its a good strategy. It's more than efficiency though. In your analogy, the ideal strategy would be one that lengthens your opponent's path without him knowing, renders its goal irrelevant without him knowing, or changes the location of his goal without him knowing.

      Path... effeciency... it's just not that simple.

    2. Perhaps. I was thinking from my perspective as a programmer.

      When you have a problem space, you pick an algorithm to solve it with. You want one that gives you the solution no matter what specific instance of the problem shows up from that space, you've solved the whole set.

      If you choose a bad algorithm, your answer may be wrong or slow. But even when you choose the right algorithm, now you have to implement it. That's where the efficiencies come in that I was referring to.

      The algorithm will naturally give you various features like speed or correctness. But within the bounds of that design, you can do things like loop unrolling to improve it.

      I don't think it's a simple concept for certain!

    3. Well thats a lot more to chew on. I don't know enough about the kinds of problems you're addressing to know how it compares to interacting with a hostile will.

    4. If you're interested, here's a short paper by Dijkstra. About 2/3 of the way through, he describes a problem of a chessboard overlaid with dominoes. The thinking he describes for how to find the solution for not only the specific instance, but every possible permutation as well, is what I'm getting at.

  6. >If your game does not have enough room for you to define goals underneath the single goal of say, defeating the opponent, then it has no strategy.

    Or, no strategic depth. As I noted in another comment, it's hard to make an enjoyable game with no strategy whatsoever. It's also hard to make a game with competing strategies as well. Its easy to make a game with one optimal strategy.

    >But doesn't that make any game have a strategic layer implicit to the tactical layer?

    Yes, but that doesn't mean the strategic layer is interesting.

    >Those choices are only tactical if there is a higher level above them to be strategic at. We think of them as tactical because we implicitly understand they operate underneath the army.

    True, but thats a low bar because practically any game with a goal has at least 1 viable strategy.

    >I think that strategy would be deciding on your path to victory, and tactics is making that path efficient.

    I need time to consider how that undersranding differs from my variable idea.

  7. I generally use the definition that tactics is immediate; strategy is future-oriented. The goal of tactics is to make efficient use of one's resources at hand; the goal of strategic planning is to ensure you have an advantage in future tactical encounters.

    Take your galaxy game. The tactical decisions in the game you've described might be very simple: make sure the galaxies are directed to specifically target the type that is weak to their type. You haven't described much strategic depth either, but it does have the basic strategic decisions in the sort of rock-paper-scissors wrinkle:
    1) You could make an even mix for balance,
    2) OR you could try to crank one one type in hopes of overwhelming one-third of the enemy's troops very quickly,
    3) if you're going for (1) you should get intel to see if the enemy is doing (2) so you can prepare the counter in time.

    Starcraft is more complex; tactical decisions are things like the movement of troops to get surrounds, the timing and placement of spellcasting, focus-fire on key targets, and so on. Strategic decisions are the decisions of what type of troops to make, how much to spend on troops and how much to spend on economic expansion, when to feint and when to commit to an attack, and so on - in the hopes that, when the lines collide and the players switch to micromanaging their troops, you will already have set yourself up to get a better outcome.

    Chess, as mentioned, has tactical maneuvers like pins and forks and material trades; the openings, though, are almost entirely strategic, because the goal of both players is to try to set up conditions for future opportunities. There are heuristics one may study - material points, the strength of the pawn structures, how many pieces have been developed off the back row, which player has more control of the center squares - which do not guarantee any tactical advantage. They are strategic advantages, because they make it more likely that you will have better tactical opportunities in the future.

    Also, since I just listened to the audio sample, consider the opening battle from "A Throne of Bones." Tactical decisions would be things like, when to switch the front lines, when to fire a volley, when to advance the cavalry, when and where to fire the artillery. The decisions made before the battle was even joined were strategic decisions - where to set up the lines, which cohorts to place in which order, which hill and the specific layout to place the artillery. The strategic plan was to wear the goblins down until they broke and fled; so, the cavalry made the tactical decision to attack from the side instead of attacking from the rear and trapping them.