Gamasutra: The Art & Business of Making Gamesspacer
Minmaxing - Is turn-based fun anymore?
Printer-Friendly VersionPrinter-Friendly Version
View All     RSS
April 18, 2014
arrowPress Releases
April 18, 2014
PR Newswire
View All
View All     Submit Event





If you enjoy reading this site, you might also want to check out these UBM TechWeb sites:


 
Minmaxing - Is turn-based fun anymore?
by Wesley Paugh on 02/06/12 04:14:00 pm   Expert Blogs   Featured Blogs

The following blog post, unless otherwise noted, was written by a member of Gamasutra’s community.
The thoughts and opinions expressed are those of the writer and not Gamasutra or its parent company.

 

Why do turn-based games exist? Will they remain interesting to us forever given advances in computing power?

I ask because I have very little interest in playing chess, or learning Go, because I know I could just write a program to do a better job of either than I ever could manage on my own. Any turn-based game, and any real-time game whose only continuous parameter is time, can be min-maxed to find a single ideal move each turn. If both players make those moves, then both are playing optimally and a balanced game will end in a stalemate of some sort. Unbalanced games and games that leverage randomness cannot be said to be games of skill (at least, not between players that know how to play optimally) which, in my mind, makes them emphasize more the act of playing than the outcome (whether that outcome is a score or a win / loss state).

If players willingly choose to make different moves than the optimal ones, then the game being played is no longer the same; players are now playing the original game, and another game whose goal is not to win at, say, Chess, but to deceive their opponent into making much less optimal moves than he might, otherwise. That social game, I'll call it 'Deception',  can be factored out of Chess, Counter-Strike or any other game or even instance of two humans interacting. It should not be omitted from any critique of a game's quality (i.e., don't call Poker a great game because its rules are exceptional, call it great because its rules are compatible with so many metagames).

Any efforts to design games within turn-based, discrete-space limits and have the game be balanced and challenging are efforts to make the utility of game states less knowable, and the state space less traversable through thought. Becoming a better player in those games involves seeing through those obfuscation efforts and learning to better estimate which moves yield higher-utility states. Games become, very much so, solving a puzzle in the sense that there is a single, highest-utility sequence of moves you must figure out that guarantee stalemate in balanced games (or, if a player wins, the game may as well have been a single dice roll, as randomness could have been the only deciding factor). If you play against an opponent whose moves are deterministic (and a player that always plays optimal moves would be deterministic, for all intents and purposes), the game becomes indistinguishable from a sliding-block puzzle, for example.

I've never played Civilization games, but, to hear them described, they have added so many and so complex a set of legal moves that establishing utility becomes difficult. In the end, though, it's still just the same min-max puzzle as Chess or Tic Tac Toe, yes? Figure out the best move and play it, and assume your opponent will play his best move. If the move he plays isn't optimal, then one of two things has happened.
 
One option is that your opponent was not aware that his move wasn't optimal, because his means of assessing utility wasn't very good. If that's the case, continuing to make optimal moves guarantees victory, rather than stalemate, in balanced games.
The other possibility is that your opponent is playing both the original and a different social game like Deception. Even though Tic Tac Toe is a game humans can solve easily, an extremely clever player could conceivably win the game more often than others by playing social games like 'Deception' on top of it, and not all of the moves in 'Deception' would appear to interfere with Tic Tac Toe.

I should clarify, I'm not saying I have any problem with Civilization or Chess; there are surely certain people that enjoy studying how series of successive game states aggregate into 'strategies'. Nothing about playing the game to win is fun in and of itself, though; the 'decisions' we make are meaningless guesses at finding an optimal moves, and the fun is in studying and recalling information about the game, not playing it. Playing those games for fun means either accepting that you are solving a puzzle by crunching numbers on utility and game states (which I can't imagine anyone finding fun), or accepting that your goal in playing is to amuse yourself rather than to win. I suppose you could do a little of both, but the importance of the latter is where I truly believe new sources of fun will be found in games. Well, that, and the ability to include social games like 'Deception', which I have a hard time envisioning ever being min-maxable by human or computer.

This post is largely thinking out loud, but it's helped me put into words some of my feelings towards what games are really supposed to be doing, and what game design ought to entail. Rules, states, and systems are great, as they give context for play. But it's sounding more and more reasonable that, if all a game designer has as a goal is to devise a state-space generator and utility-estimator whose complexity is beyond the human mind's ability to traverse exhaustively, that designer will not necessarily make a fun game.

The problem with my thinking like this is that it defines games out of existence. Continuous-space games and real-time games can be examined as generalizations of turn-based, discrete games that only further complicate, rather than somehow magically eliminate, the ability to min-max game states to solve games as you would puzzles.

Now, I don't know of any designers that design by envisioning a state-space and transitions, and planning states with the goal of slowing down efforts to estimate utility. But! This is the sort of meaning I take from people (and there are a fair number of them) that talk of games as being 'interesting-decision generators' and calling for games to move away from visceral or aesthetic appeal and back towards board games and other more abstract games like Chess.

The trick of finding the right balance with other disciplines of game development is of much greater use. Examine games you design as though they were state machines, certainly, to assess and tweak your game's complexity, but it is important not to give too much development priority in creating the emergent depth of games like Go at the cost of enhancing other gameplay quality attributes (e.g. usability / UI design).


( For context, this post is coming after 6 months doing encounter design for a turn-based iOS game called 100 Trials )


Related Jobs

Nexon America, Inc.
Nexon America, Inc. — El Segundo , California, United States
[04.17.14]

Web Designer - Temporary - 3 month
Darkside Game Studios
Darkside Game Studios — Sunrise, Florida, United States
[04.17.14]

Mid-Senior Graphics Programmer
Digital Extremes
Digital Extremes — LONDON, Ontario, Canada
[04.17.14]

UI ARTIST/DESIGNER
Digital Extremes
Digital Extremes — LONDON, Ontario, Canada
[04.17.14]

UI ARTIST/DESIGNER






Comments


Greg Holsclaw
profile image
A bit a stretch to use Chess and Go as your examples of min-max optimal moves. It is true that known openings, and the end game can be very deterministic in chess, but the true life of the game is in the middle game, where choice progressions can last 5-10 moves. Find those optimal moves is very difficult.



Placing Tic-Tac-Toe and Chess in the same min-max category is not a thoughtful comparison. TTT only has a few hundred possible game in total, there are hundreds of chess openings and viations. No matter the skill of a player, if you know the trick, you will never lose at TTT. Never. Nor in checkers.



Anyway, aren't many games played in the sense that the first to make a 'mistake' in calculating the optimal move is defeated by the other. If you are playing a novice, you might be able to make several sub-optimal moves due to their propensity to also make such sub-optimal moves.



Chess and other higher level abstract games, are not fun in their mechanics alone, but because of the opponent interaction, bragging rights for the victor, or the sense of accomplishment.

Wesley Paugh
profile image
> Chess and other higher level abstract games, are not fun in their mechanics alone, but because of the opponent interaction, bragging rights for the victor, or the sense of accomplishment.



Yep, that's entirely my point here. We enjoy competition in any form; the actual rules of chess aren't that important, and I suspect it's social role is sufficiently filled by it that another similar game hasn't caught on.



It wasn't my attempt to directly compare Chess to TTT; chess has a vastly deeper state space and much richer strategies, as you indicated. My goal with the comparison was just to show that the thought processes to becoming better at Chess are very similar to the ones we used when we first realized TTT was not a difficult game. It might not be everyone, but I became fairly disinterested in chess when I realized that (except the occasional game with my dad which, again, has little to do with the mechanics of the game being great).

Keith Burgun
profile image
>or learning Go, because I know I could just write a program to do a better job of either than I ever could manage on my own



Not true about Go. Computers can't get past Kyu. You could get to 1 Dan in a year or two of semi-serious play, and probably 3-5 Dan in your lifetime. Computers notoriously suck at Go.



>Any turn-based game, and any real-time game whose only continuous parameter is time, can be min-maxed to find a single ideal move each turn



Not true, again. What you're thinking of is perfect and imperfect information games. In a perfect information game, there theoretically exists a totally optimal move. However, in a game like Poker or Yomi, there is always some amount of risk management, luck, or guessing. So it has nothing to do with real time or turn based.



Players don't know all of the optimal moves of any game that hasn't been solved (like tac tac toe) and therefore discarded as a legitimate way for an adult to spend their time. The best Chess players in the world don't know all of the optimal moves in Chess.



So your whole article seems to be based on this idea that just because a game CAN be solved, it IS solved. It's not true.



And also, games that have some randomness - let's take Poker which has a good amount of it - still are games of skill. Since when do games have to have no randomness at all to be games of skill? What does "game of skill" even mean?



>Games become, very much so, solving a puzzle in the sense that there is a single, highest-utility sequence of moves you must figure out that guarantee stalemate in balanced games



That is the case. But in a good game, you never figure it out.



>Figure out the best move and play it, and assume your opponent will play his best move.



Deception is a core mechanism of just about *any* game. I can't think of a game that doesn't have deception.



Just for the record: I think aesthetics are valuable in games. But in the exact same way that a body is important for a car. If the car doesn't work, it doesn't do much good to look pretty.



And finally - the fact that you think that playing a game is a tiresome number-crunching chore confirms to me something I've thought about you for a long time: that you actually don't really like games all that much. There's something about the modern video game you do like but I don't think it's the game part. Turn based games probably make you face the fact that what you are playing is indeed a game too much, whereas real time games allow you to futz around and do something to amuse yourself.



For me, playing a good game is an exciting exercise in imagination, and not at all how you described it. A great game is an exploration of a reality that never was - and I mean that mechanically.



UI design.

Wesley Paugh
profile image
> Not true about Go.



I strongly disagree. Whether I could write an algorithm that could perform better than *any human*, certainly not. But myself? I would suck at Go, and the way I'd improve at it would be to learn strategies, figure out some patterns to avoid and try to reproduce, think ahead a few moves, etc. I could easily write an AI faster that would try the exact same things and do a better job *than me*, and I am completely uninterested in putting that level of effort forth to learn how to better navigate its state space because of that fact.



> However, in a game like Poker or Yomi, there is always some amount of risk management, luck, or guessing. So it has nothing to do with real time or turn based



The risk management part can be number-crunched and minmaxed, first of all. The luck and guessing part have nothing to do with skill, then, which also doesn't interest me with regards to gameplay.



> So your whole article seems to be based on this idea that just because a game CAN be solved, it IS solved. It's not true



Maybe not. What I've been getting at, though, is that games ought to be more than just getting a little closer to solving a puzzle each time.



> Deception is a core mechanism of just about *any* game. I can't think of a game that doesn't have deception.



Me neither, but it's present in some games waaaaay more than others, and in something like poker it accounts for most of the reason people enjoy the game, not the actual rules of the game. This goes for a lot of turn-based games.



> And finally - the fact that you think that playing a game is a tiresome number-crunching chore confirms to me something I've thought about you for a long time: that you actually don't really like games all that much



Yes, I'm well aware of your definition for Game, and how it ignores the importance of Deception in most of the games you enjoy, and how it excludes a ton of things that we've accepted as games for as long as we've had them. That definition is of little value to me for the reasons I've stated; if all you care about in playing a game is solving that puzzle a little further each time, then I see no reason why you shouldn't design games that leverage such deep states.



And if you think that means I don't like games, by your definition, then yes, you're probably right. I'll still enjoy soccer or a game of chess with friends, but it has nothing to do with them being good games with lots of opportunity for skill-building, as you seem to believe is mandatory for good games, and everything to do the fact that games are fun to play with people.

Keith Burgun
profile image
>>I'll still enjoy soccer or a game of chess with friends, but it has nothing to do with them being good games with lots of opportunity for skill-building, as you seem to believe is mandatory for good games, and everything to do the fact that games are fun to play with people.



"Mister automobile mechanic, I don't know why you are trying to fix my SPARK PLUGS. The problem isn't spark plugs, the problem is that my car won't start!!!"

Wesley Paugh
profile image
> "Mister automobile mechanic, I don't know why you are trying to fix my SPARK PLUGS. The problem isn't spark plugs, the problem is that my car won't start!!!"



Oh, Keith, that's what I love about you: a plainly-stated argument is good, but why settle you can mock someone at the same time?



Let me put it this way: Go completely fills the requirement for any state-space-solving game we will ever need (until we solve it, at least). Any other games that are made with similar levels of complexity won't be any more fun, or really introduce any new or meaningful interactions above 'generate as many states deep from current as possible, estimate how close to winning I get'. So why bother making more games?



We bother because, for one, we can find new and interesting ways to rely on Deception and other similar games. That was part of what I'm stating in the blog; understanding and leveraging that is a better way to begin designing than designing by planning what you can do in a turn that will most affect the state machine.



We also bother because there are other games that have nothing to do with balancing game states and more to do with finding acts of play like running, jumping, shooting, etc., that are fun and structuring a game around those. There might be some skill development that goes along with them, but it only works because the play was fun to begin with, and nothing more. Sure, 'fun' means a lot of things, but we have games built around it alone, and those games are worth continuing to make and just as valid as Go.

Keith Burgun
profile image
Not mocking you. Using an analogy to explain the flaw in your logic.



>>Sure, 'fun' means a lot of things, but we have games built around it alone, and those games are worth continuing to make and just as valid as Go.



"Sure, a functioning vehicle that has an engine that burns gasoline using internal combustion to turn wheels is one thing, but we also have cars that are just based on DRIVING, and those are just as valid."

Keith Burgun
profile image
I will leave one last comment:



Did you really mean to ask if something as broad as "turn based" is "fun anymore"?

Wesley Paugh
profile image
To use your analogy, the first car we ever built was (appropriately) called the Go. It runs smoothly, it's elegant, gets to top speed immediately, never needs maintenance, runs forever and is always fun to drive. Forever. What point is there in trying to do better than that? Sure, there's an argument to be made that sometimes you just want to drive something different, even if it is slightly larger, or eventually dies, and I fully support that. Someday we might find something better than the Go.



What I'm saying is we could just try and find some fun ways to drive cars at all, or drive with other people, or make jet-skiis, or motorcycles, or jet-packs to get around to the same places. A lot of them might not be fun or practical for more than a tiny bit of time, but hell yeah we ought to embrace those as vehicles.



And your argument against that seems to be: "only vehicles that are awesome cars like the Go are actually vehicles. If you like jetpacks you should just read the book". I'm bad at analogies.

Wesley Paugh
profile image
And I responded to that question in your other comment thread.

Keith Burgun
profile image
Go is not the perfect game. In fact, I've recently kinda pulled back on my go-playing, because there are some problems.



Namely, the amount of investment you have to put in up front - the learning curve - is DECEPTIVELY huge. This is just to be able to like, really play. That's a problem! Also the endgame. Really, you have to be a bad-ass in order to be able to even judge when the game is over! So for noobs, the game can really drag. And that's a problem.



I think you maybe just assumed Go was perfect without playing it yourself? You shouldn't do that. You should also play Civilization once in your life.



Go isn't perfect. But furthermore, even if it were, having one perfect game does not mean we shouldn't try to create other great things. Beethoven's 9th symphony is pretty much a perfect melody, but that doesn't mean other lesser melodies can't be super awesome too!

Wesley Paugh
profile image
Learning curve was not within the scope of my post; there ought to be pretty clear arguments for and against a steep learning curve on a game you can literally play until the day you die.



That aside, I think we've made our points (though if I was unclear in my analogy about why nothing I've said precludes making imperfect cars, let me know). Yes, I should play Go, but from what I've read it is the least 'puzzley' game I know of because it is so difficult to just min/max your way through. I will likely get to Civ first, since I've had some friends recommending it for a while. Once I settle on IV or V. (probably IV, from the sounds of things).

Keith Burgun
profile image
And what does "is turn based fun 'anymore'" mean?



Are you implying that turn based games were once fun? If so, what happened? Your article doesn't address the title-question of your article.

Wesley Paugh
profile image
True, I didn't address that strongly in the article. The illusion of chess being anything but puzzle was lifted for me as soon as I considered how state min-maxing would be used to create a Chess A.I.. Until that point, I hadn't really thought about the game as being solvable, so yes, it was more fun for me before the introduction of that concept. I was more posing the question for readers to ask themselves for their personal answer to it than I was trying to answer for the community with certainty.



Although, there's also the case to be made that, the closer we get to an algorithm solving Chess, it stands to reason that we will learn more from computers, and studying algorithms we write, and learn to play more perfectly, getting closer to a point where the game is uninteresting to us (although yes, we may never get there).

Keith Burgun
profile image
LIFE can be min-maxed. Good luck doing it, though.

Wesley Paugh
profile image
I would have no clue how to min-max life. I know exactly how to min/max a turn-based game, though. Anybody that beats me is just somebody that can imagine more states in his head at a time, and that's not a skill I would really pride myself on improving, and I would care even less about doing it within the contexts of multiple games, unless I have some other driving force like friends or aesthetics to set it apart.

Keith Burgun
profile image
>Anybody that beats me is just somebody that can imagine more states in his head at a time,



That's true of any game and of reality. Welcome to "planning"?



You can have friends and aesthetics without that annoying "game" part, you know.

Wesley Paugh
profile image
Yes, but not play. Play can exist without the ability to minmax in any way I can conceive, through games like Deception. I have made this post because I am a greater advocate for making games built around that than I am of making games that resemble chess.

Keith Burgun
profile image
But chess has deception. And you can have play without games.

Wesley Paugh
profile image
Chess does not have Deception. Chess' rules consist of the game you would play over a network, without any chat: just the pieces and how they can move. Any attempt to use strategies to make your opponent not see what you are trying to do, for example, is an attempt to play a game using the rules and actions of human psychology, rather than just those of chess.



And you are aware that I do not believe you can have play without a game. The entire reason you and I are unable to discuss game design is because of that difference of opinion on what a game really is.

Darren Tomlyn
profile image
@wesley



All my posts are based upon the contents of my blog - (which I recommend everyone to read) - here (please use this link instead of clicking my name, since I've updated/replaced the first posts): http://www.gamasutra.com/blogs/DarrenTomlyn/20111208/9066/Part_1A
_The_Problem_With_The_Word_Game_v3__Relaying_The_Foundations_Of_O
ur_Understanding_Of_our_Language.php ).



I'd like to post a complete reply to this particular subject, but I'm planning on getting round to discussing more of this in detail in later posts - mainly when I examine solitaire/patience, probably.



But I can post a small reply here, in relation to the further reply you wrote above:



Game != play (noun).



How do we know this for certain? Because games have always been played (verb) for work (noun). (Usually for training/selection purposes, and long before we had the word game, (and even this language), to represent such an activity.) (I give you the original Olympic games as immediate evidence.)



Game, art, puzzle, competition, work and play are all the same type of word, representing a similar concept, (application of behaviour), but individually different from each other, based on how they are used.



Whether or not a particular activity is recognised and understood to be a game, puzzle or competition, (or even a work of art), can sometimes be nothing more than the subjective perception of those taking part or perceiving such an activity taking place.



Asking whether or not turn-based activities are fun, isn't quite asking the right question, and therefore isn't really helping you understand what is happening and why.



What you're wondering about, (whether you recognise it or not), is the nature of the behaviour different activities (and pieces of software etc.) enable for those taking part, and how that relates to their perception of it.



What you are really wondering about, therefore, is:



If an activity etc. enables an application of behaviour that would be more consistent with a puzzle or a competition instead of a game, will people still find it fun if they intended and wanted to play a game instead of a doing a puzzle or taking part in such a competition?



(Notice that the phrase turn-based isn't mentioned).



But you can only ask this question, (and understand the answer), if you fully understand what a game, a puzzle and a competition actually are, and this is where you're having problems...



So, I suggest you read my blog - it might help...

Simon Ludgate
profile image
I'd point out that solving a Sudoku puzzle is fun for many people, even though computer programs can solve them far more quickly and efficiently. Perhaps single-player turn-based games should be thought of as stimulating problem-solving areas of the brain, rather than being examined from the perspective of a competitive challenge. AI "opponents" should be designed as sufficiently predictable challenges for players to overcome, rather than mathematically superior opponents which can only be defeated with a strictly dominating strategy.



For multi-player competitive turn-based games, randomization can alleviate the design stagnation: see turn-based games with die rolling or card drawing mechanics, such as D&D or Magic: The Gathering.

Wesley Paugh
profile image
Excellent points! I sincerely appreciate that you sort of describe games as having some degree of entertainment value for its own sake, rather than needing to be a skill challenge against an opponent that wants to be the very best. Even if winning strategies are known, a single-player game can have good replay value if the same problem-solving technique is even subtly randomized. The goal of agent / obstacle design ought to be to match the estimated player skills, rather than play as optimally as possible (this goes doubly in realtime).



How much a designer ought to plan for player skill improvement, or whether that planning ought to include learning algorithms is a related can of worms I'd like to open at some point.

Robert Boyd
profile image
Your question seems to be more of a case of "Is Min/Maxing fun?" rather than "Is turn-based fun?" And for some people, the answer is yes. For others, the answer is no, but they may still enjoy turn-based games.



Even in cases where there's a clear optimal choice, some people may choose to make less than optimal choices for fun. Maybe, they choose to add a character to their RPG party because they like their personality even though another character might be more powerful statistically. Maybe they choose a less popular choice because they want to see if they can make it work. Maybe they don't care much about min/maxing and just want to see how the game's story progress. There are any number of reasons why someone might not care too much about min/maxing and yet still enjoy turn-based games.

Wesley Paugh
profile image
I zeroed in on turn-based games because they are the easiest examples to demonstrate as min/maxable. Real-time and continuous-space games are, in spirit, although it isn't as obvious and much less gameable, especially by humans.



I am 100% a proponent of players making sub-optimal moves for the fun of it, or to watch a story play out, or because a separate game like Deception is being played alongside the first game. Those kinds of meaningful decisions are great, but they would exist in tandem with turn-based gameplay, which was my focus here. It would be especially great if story roles interacted directly with the turn-based gameplay (e.g. the love a Tank's character has for a Healer makes him more effective at protecting her, that kind of thing).



But yes, I am definitely with you on encouraging sub-optimal choices as a way to make games more interesting than min/max.

Luis Guimaraes
profile image
Rock-Paper-Scissors...

Wesley Paugh
profile image
Sure! We can call that turn-based for these purposes. A lot of turn-based games incorporate randomness (and imperfect information, several have informed me) to mitigate min/max strategies.



This is a perfect example of how complementary games like Deception ought to work alongside games that can be min/maxed; knowing and perceiving how your opponent thinks, and taking moves outside the games moveset like body language, can give you an advantage in RPS.



Otherwise, RPS becomes a game of complete randomness. Playing against a random-number generator (which would be a strict instantiation of the game) is no fun, but playing against an AI that has studied patterns in human play? *THAT* is a fun opponent to play against.

Luis Guimaraes
profile image
Exactly. Thinking about the fact the you mentioned Counter-Strike (also Quake came to mind: http://www.youtube.com/watch?v=XdkDjsBiO58), RPS came up as the over-simplified exemple.



Limited-information games have to be the most interesting games possible for human opponent gameplay. You can guess, you can adapt, and try to trick, mislead or control the opponent. The more sandbox-like the mechanics, the more interesting a match can be, with emerging strategies on the fly.



I'm talking a lot about real-time here, but the same is true for a match of Frozen Synapse, CS Manager or NFL Street, which are all very close to the RPS simultaneous turn system.



By the way, I have to write it again, the semantics of word "game" sounds so complete you can feel that are many sub-games going on: "limited-information games".

Keith Burgun
profile image
In great games, attempting to make the most optimal move IS ALSO the fun thing. That's what good game design is - luring the player into doing something which is fun. Not that you'd know it from looking at modern digital games...

k s
profile image
I'm a big fan of turn based strategies but I don't play to win so much as I play for the experience.



My favorite game to date is still Alpha Centuari, it never offered the same "match" twice. AC offered not only challenge but also uncertainty as to what I might find on Planet.



I'm sure I too could write a program that would play better then me but I wouldn't bother cause I enjoy seeing where the game takes me.

Chris Hendricks
profile image
This is probably my favorite response on this particular topic so far (and not just because I'm also an Alpha Centauri fan). "I enjoy seeing where the game takes me" is a great sentiment, one that should apply to all great games for me. In Scrabble, for instance, I get great enjoyment out of not only seeing what I can do with my letters, but reacting to the unexpected turn that my opponent gives.

Chris Hendricks
profile image
Double post.

Jeremy Reaban
profile image
More fun than "Whack-a-Mole" which is basically what all real time games devolve to.

Luis Guimaraes
profile image
You forgot "single player" in your sentence.

Roger Tober
profile image
That's kind of what people tend to do. Take the other type of game (not their favorite) into it's lowest denominator, and explain how senseless it is. Turn based games work better in a social context as far as I'm concerned. They are also more mentally stimulating and at the same time more relaxing because the time pressure is taken off.

Kristen Erwin
profile image
Food for thought:



"The number of legal chess positions is 10^40, the number of different possible games, 10^120. Authors have attempted various ways to convey this immensity, usually based on one of the few fields to regularly employ such exponents, astronomy. In his book Chess Metaphors, Diego Rasskin-Gutman points out that a player looking eight moves ahead is already presented with as many possible games as there are stars in the galaxy. Another staple, a variation of which is also used by Rasskin-Gutman, is to say there are more possible chess games than the number of atoms in the universe. All of these comparisons impress upon the casual observer why brute-force computer calculation can't solve this ancient board game. They are also handy, and I am not above doing this myself, for impressing people with how complicated chess is, if only in a largely irrelevant mathematical way."

David Holmin
profile image
"Why do turn-based games exist?"



Because it allows for a design around making choices instead of manual dexterity.



"Will they remain interesting to us forever given advances in computing power?"



Yes. People still play chess, don't they?



"I ask because I have very little interest in playing chess, or learning Go,"



Oh... Well, people still play Civ, Heroes, Magic and Pokémon, don't they?



"...because I know I could just write a program to do a better job of either than I ever could manage on my own."



And I could design an FPS where every enemy hits your head every time they shoot, and shoot within 1 ms of seeing you. But I don't.

Wesley Paugh
profile image
All very good points!



> Because it allows for a design around making choices instead of manual dexterity.



There are optimal paths through chess, and so the choices you make come down to optimal/suboptimal ones, solving the puzzle or not. My goal here was to emphasize the importance of the games played through social interaction, like Deception, and generally de-emphasize the importance of states and state transitions and try to show that trying to win by playing perfectly isn't the reason most people play games.



> people still play Civ, Heroes, Magic and Pokémon, don't they?



Of course, I only mean to say that the interestingness of the puzzle being solved when you play those isn't the reason the majority of people play them. Designing them for perfect balance doesn't need to be, maybe even shouldn't be, the highest priority.



> And I could design an FPS where every enemy hits your head every time they shoot, and shoot within 1 ms of seeing you. But I don't.



Also right on. This goes hand-in-hand with the clarity I gained that prompted this post, that playing perfectly isn't the point of playing. Still, turn-based games have a harder time maintaining the illusion that there aren't perfect moves you ought to be taking.

Joe Cooper
profile image
Minmaxing is not enough to beat many of the games in question, though as you explain it may well be enough to beat -you-.



I regularly beat the computer in Civilization and when I started playing against a human it became far more interesting; you could come up with all kinds of long term plots. At one point I sold a bunch of my cities to another human player to fund espionage and find his missile shield. When it finally came time to war with him, I crushed him in a ballistic missile war.



And the fun part is that, unlike the computer, it will never work again with the same person.

Joe Cooper
profile image
I'd also like to add, basically the only game I've had any fun with at all on my iPad has been Civilization, and I've been playing it over and over again for about a year now.



Turn based games don't jive with simplistic stereotypes of "fun" where we envision people chasing balls and flashing colors but that doesn't mean a hell of a lot of people don't enjoy it.

Kenneth Blaney
profile image
You say "turn based" but in fact I think you mean "games that can be represented by a tree". A game like Calvinball or other nomic games can easily be turn based but still defy solution this way where as a number of games you wouldn't consider turn based actually can. (Imagine each individual game state as a node in the tree and the controller state as the player's input. Its how tool assisted speed runs are created.)



That said, even though you could create an AI that could beat you, that doesn't mean that everyone else could make an AI that beats hisself or herself. So a more accurate thesis is why you don't find these games fun as opposed to the more objective sounding "these games aren't fun."

Wesley Paugh
profile image
Fair points, although I think it is often enough of a case in turn-based games that you only ever have a discrete number of possible moves each turn to generalize them as being trees. As I mentioned, real-time games suffer from the problem just as much, but their nature masks the puzzley-ness of the gameplay to a greater degree.



> ... that doesn't mean that everyone else could make an AI that beats hisself or herself.



True, but they might still recognize the approach they beat Tic Tac Toe with as making all tree-games puzzles. I can't speak for everyone, and just meant to pose the question for people to ask themselves, 'is it fun to play chess to win, when you know that optimal moves exist each turn?' I think the answer is yes, chess is still fun for many and widely played because the illusion of something more than the puzzle it is holds up in human brains, and because the social games played alongside it are very fun. But anything that calls the illusion to someone's attention reduces the likelihood that Chess would still be fun.

David Holmin
profile image
@Wesley



You seem to be assuming that every turn-based game is like Chess, where everything about the game state is known to the players, or players. In Civ, for example, there's a lot about the game you don't know at all times, such as the nature of the map at the start, and not least the moves of your opponent(s). It comes down to interpreting what the game presents you with, and what you learn of the game throughout playing it, to make the best choices available, long term or short term, etc. Furthermore, turn-based gameplay can be a component of a game, and not the game itself, as in Fallout (1). There, a lot of the fun comes from the synergy between combat and the choice making/exploration of the rest of the game. Much of the fun is pulling stuff in the game world and see what that gives you, and how what is given is incorporable in the combat aspect of the game. I'm typing this tired just before going to sleep, but hopefully my point comes across.

Michael Parker
profile image
It's interesting that as designers, we still aren't using a common vocabulary to describe the types of mechanics involve in game design.



Looking ahead at possible moves is one thing.



Manual dexterity is another.



Social interaction, or psychology / bluffing / prediction is another (maybe more than 1 field here)



Ability to recognise patterns or situations is another.



And so on. Different games use these different concepts in different amounts. Personally, I agree with you - my favourite games involve some form of social interaction (I wrote a blog post on why this is important in game design). This is why I much prefer poker to chess.

Gary Dahl
profile image
I've enjoyed your article (and the controversial title), and agree that it is very important to consider how players strategize about the games we design. However I don't understand why only turn-based games can be generalized into potentially enormous decision trees. Can't the same kind of min-max generalization be applied to real-time games (or life itself)? You just have one more resource to decide how to manage as you play: time.



However I don't think it follows that no games are fun anymore. Games include many different kinds of fun experiences. The trick seems to be to focus more on what fun experiences you'd like your design to evoke; rather than focusing so heavily on the structure of the underlying game.

Wesley Paugh
profile image
'Turn-based, discrete' probably would have more accurately identified the games I'm talking about, for just that reason, but it wouldn't have been as catchy a title. Real-time and continuous space also have the problem of being solvable through exhaustion, although turn-based wears that fact more prominently.



You have very elegantly restated my realization: we aren't playing to win so much as we're playing to play. Designing for difficulty or designing games that have inexhaustibly deep or wide state spaces aren't as important goals as making sure the interactions are meaningful in other ways, like interacting in a social game with a human or accomplishing tasks aside from winning ( e.g., create an society that interests you in an RTS, despite its inability to compete in the game's overall goals ).



I am very glad to hear you enjoyed the article, thanks!

Eric Schwarz
profile image
All games have optimal modes of play, whether turn-based or real-time. You can just as easily make those ideal decisions in a real-time setting, and in fact it's often simpler, at least if you cut out the smallest of steps, due to the fact that most real-time games gain their difficulty from execution and not planning. For example, it's relatively simple to program aimbots for first-person shooters and I'm sure with a bit more work you could build one that plays a game start to finish for the player.



With that in mind, we're getting onto the question of "why do players find games fun?" I know your question about automation is a little rhetorical, but even so, it's worth mentioning that it's precisely because we are imperfect beings that games are interesting. In single-player games, the unpredictable nature of the game's scripting, level design, etc. is what keeps us engaged. Once we've fully learned that allowing for optimal play, we can still turn to multiplayer, where the presence of other players gives effectively unlimited permutations of unpredictability to respond and adapt to.



But then, what happens when you end up with total and complete mastery over a game? We see it happen - there are master chess players who know the ideal move to make in every single situation, for instance. Why is the game still fun for these people? Well, then the enjoyment transfers not from necessarily winning, but to puzzling out what the opponent's strategy is, and in correctly selecting a counter-strategy. We can't know the future, so even someone who has utterly mastered a competitive game still has something to look forward to.



Anyway, good read, although I guess I feel the questions aren't necessarily worth mulling too hard over. I think the proof is in the pudding - clearly people find turn-based games fun, and we already have a pretty good idea why, so why question it too much? It's a decent thought exercise but you can apply the same type of thinking to just about any game style, to no practical outcome.

Bart Stewart
profile image
A word that's come up frequently in defense of the original thesis is "perfect," as in "players min-max turn-based games to win by finding and playing the perfect sequence of choices." But it's not necessarily the case that there is a single perfect choice-sequence -- in fact, I would say that part of the art of good design for games with large state spaces is that there are multiple paths to victory.



Turn-based structure does come into this discussion because, as noted above, that structure rewards careful thought over manual dexterity. That's why many turn-based games are strategic games of (as Luis pointed out) information management. Winning these games is fun because that validates the player's perception and effective implementation of a superior strategy. Not a "perfect" strategy -- just one that was better than those used by the other players.



This is also part of the reason why many TBS games (unlike linear "play it once and never again" games) are designed to be highly replayable: even experienced players can discover new and better strategies. That process of improving intellectual competence is the "play" part that's fun about TBS games.



[Edit: I want to add that the comment that turn-based games (in their strategic forms, I'd say) are essentially puzzle games is a very astute observation. This is the "figure things out" Explorer playstyle revealing itself, I think.]



And finally, it's sort of morbidly fascinating to see so much thoughtful support for turn-based games in the comments here. If so many people are so good at seeing the fun of these games, why are there so few of them these days?

Wesley Paugh
profile image
> But it's not necessarily the case that there is a single perfect choice-sequence -- in fact, I would say that part of the art of good design for games with large state spaces is that there are multiple paths to victory



By 'perfect', I meant 'uses only those moves that maximize the possibility of winning'. In a game with multiple paths to victory, that just means there are multiple ways to play perfectly. Being able to identify a game as a state-generator removes the meaningful choices from the game, because it becomes calculable how to play perfectly. I would be insane to claim this has made Chess not fun, but I did want to take the opportunity to portray social games like Deception as avoiding the problem altogether, rather than just obfuscating it.



Given infinite time between turns, your best strategy in a turn-based, discrete game is to sit and plot out every conceivable turn you or your opponent might take in the future, and pick the one that yields the most and most likely victory possibilities. That is a terribly dull, yet guaranteed-best, strategy. Chess was brought closer to real-time with the addition of clocks (still not fully real-time, of course) to address this.



I do feel a little bad criticizing these games when there are much, much more problematic trends in mainstream games that ought to be chastised. Still, it has been on my mind a lot (partly because of my recent work) and I felt it was as good a time as any to throw my support behind the more socially-focused games like poker, where it's generally understood that you shouldn't try to play perfectly as often as you should try to play interestingly.

Eric Schwarz
profile image
We see so few turn-based games because game developers, in the AAA space especially, don't decide the direction of their own games; rather it's the publishing side, and marketing especially, that determines whether a game gets the green light or not.



Consider the sorts of questions most publishers ask when looking at a game pitch:



1) What target market does this game appeal to?

2) What other games does this game compete with?

3) What does this game do to differentiate itself from said competition?

4) How can this game be marketed? What are its selling points?

5) How does this game fit into our product lineup?

6) Does this game conflict or compete with any of our existing products?

7) Can this game be made with existing technology and assets we already have licenses for?

8) Does this game have the potential for sequels, potentially on a yearly basis?

9) Similarly, is there potential for downloadable content, microtransactions, etc.?

10) Does this game have potential for cross-media promotion and franchising, i.e. comic books, films, action figures, etc.?



I could go on, but the point is, publishers want games that are safe bets, they want properties that can be built upon for the long term, they want products that can be targeted and marketed clearly, and so on. Due to the evolution of the current games industry over the last two console generations, this equation does not typically include turn-based games.



Publishers like action games, and especially shooters, because they are generally guaranteed to sell given the state of the market, and the bar to reach is clear as far as competitors are concerned. Marketing any game is a big challenge, but if you have a standard to follow then it is made much easier.



Big publishers are slow. They want to do things by their books, and right now they are geared towards making games in very particular ways. If Call of Duty and Gears of War had been turn-based card-battling games and met the same success, you'd find the gaming landscape a very, very different place than it is today. With the name of the game being risk management, safe bets rather than innovation makes sense, and I certainly don't blame publishers for sticking with what works.



Moreover, their entire business model has been built around the idea that players want action, excitement, cinema-quality production values, multiplayer and so on. Anything that doesn't fit into that equation in the way they're used to is going to be an unnecessary risk, an anomaly.



All that said, there are a lot of turn-based games, more of them than there were ten years ago in fact. Smaller and indie developers have been putting out tons of awesome games thanks to the rise of digital distribution and cheaper development tools, and there are actually quite a fair number of such games on portable platforms (especially smartphones, but also the DS and PSP). Big publishers, however, simply are not interested in them, and that won't change until one has a turn-based game become a smash success... and even then, chances are it'd be chalked up as a fluke and ignored. Given the state of the market, unless those publishers start thinking smaller, that will pretty much never happen unless a major shake-up occurs.

Bart Stewart
profile image
Wesley, in all fairness that seems like a pretty tortured definition of "perfect." I think of perfection as I think of uniqueness -- do you also say that something can be "very unique?"



This is why the concept of information management is crucial to the design of large-state-space games. It is practically impossible to have a "perfect" strategy (or sequence of play actions) in such a game because no player is omniscient -- you know enough to form what could be an effective strategy, and no more. The fun of these games is testing and improving these strategies in the absence of perfect information.



In other words, maximizing is not the same thing as perfecting. It's the fact that there is no perfect strategy in a well-designed game, only a number of strategies that might be good enough, that makes thoughtful turn-based games fun.



And Eric, my question was mostly rhetorical. ;) But I'm sorry to say I think you're right on nearly all counts.



Other than some JRPGs, I can't think of many TBS games I've seen lately. Most indies seem to be platformers or seizure-inducing clickfests. But if some good TBS games are starting to show up, that is absolutely a Good Thing -- I am convinced that there's a market for true strategy games that's not being fully tapped right now.



Somebody's going to make some money, I think!

Jason Wilson
profile image
I would agree that a game in which I always know the best (or optimal) move to make in every situation, would quickly become boring. However, I think for many games, including the ones you've listed, knowing the best move is very difficult, or at the very least, would take many years of practice and experience to learn. Therefore, the majority of players do not always know the best move to make, so they are essentially gambling that the move they have chosen will be the best one. And like in gambling, the player uses whatever information they have at their disposal to come to their best guess at the best move. The excitement and fun comes from finding out the result of their choice.



I would also like to say that real time games are no different in this regard. Playing a game of Street Fighter is a constant guessing game of what is the best move to play at any given time. Star Craft is a constant set of decisions that the player is faced with as to what is the next best move. One could even argue that the real time games are in fact a series of "turns" that quickly pass the player. An inexperienced player with slow reaction time is merely choosing to do nothing on many of their turns. A very simple real time game could have the same problems of mastery leading to boredom, and I imagine an AI programmer could easily create an AI that could excel at a simple game.

Roger Tober
profile image
I agree. It's all about acquiring a skill. Games are just models of our lives where acquiring skill doesn't require such a high price in time or circumstances.


none
 
Comment: