Game Placebo - Understanding the illusion of agency

Discussion in 'Game Design' started by keithburgun, Dec 24, 2012.

  1. evizaer

    evizaer Well-Known Member

    The concept of randomness being a placebo for agency still just doesn't make sense to me.

    Output randomness diminishes your impact on what happens to the system. Does that mean it's a placebo for agency? No. It just means that you have less agency in the game's outcome when you have output randomness.

    The word "placebo" makes it sound like you're being cheated out of agency you're supposed to have. That's simply not the case. There is no expectation of perfect agency when people play colloquial games. It's totally unlike the expectations people have when they are given a pill that they are told should cure or help against some illness. It's more like taking a pill someone gave to you randomly and then being upset that it doesn't heal your broken arm.
    friiik and Remy77077 like this.
  2. Logo

    Logo Well-Known Member

    I'd take it from a neutral standpoint, not as a criticism. Some types of randomness are placebos for agency: they reduce your agency over parts of the system, but still gives you a perception of agency (the Placebo effect part of it).

    Whether that's as good as real agency is up for debate. Keith argues its not because as players understand the placebo effect the effect disappears. Other people would disagree over that.

    Either way I think Placebo describes it quite nicely. Some types of output randomness can give the feeling of agency beyond what the game system actually allows. A good example would be XCOM's false choice of taking shots in the 40-50% hit range. If you remove randomness you'd drop these shots down to 0% (XCOM is about lining up good shots with flanking and tactics) and the players would feel like they have less agency. In reality having these shots at 50% is adding very little agency (since it's almost always a false choice), but the player feels a lot more agency from the additional choice with trade-offs (hence Placebo effect).
    Remy77077 and ratxt1 like this.
  3. evizaer

    evizaer Well-Known Member

    But isn't the point not that false choices lead to agency placebo affect, but that randomness causes it to happen?

    I agree that false choices can have a placebo-like effect that causes people to believe they have more agency than they actually do.
  4. Logo

    Logo Well-Known Member

    Well in that particular case it's both right? Without randomness the false choice is just a false choice instead of "a really bad decision". With randomness the choice is probably technically not a false choice, just a really bad one; you could probably get pretty far in XCOM with 50% shots, but it'd be bloody and highly sub-optimal.
  5. keithburgun

    keithburgun Banned

    Hey everyone, just got back from a vacation. Will respond to some of the discussion here, but can't hit it all.

    Not all actions result in any meaningful effects. Agency is the sum of your input on the game.

    Let's not confuse "too much agency" and "too many options". If you mean "too many options", or "too much power/control", then I agree. Obviously limitations are what make a game a game, but within those limitations, players should basically have maximal agency, as far as I can tell.

    Regarding input/output randomness, I took them from the Ludology podcast:

    I have reservations about them too, but they work in a rough way.

    Like Evizaer mentioned, output randomness may be useful in a simulator. If your goal is to simulate something, rather than necessarily be the best game you possibly can, then the pluses of randomness may outweigh the minuses.

    Not really. They partially do but they are partially fantasy simulation driven - all of them. To their detriment, in my opinion.

    It doesn't fail for Outwitters, or Go, or Through the Desert. Why is this question limited to videogames, btw? Videogames happen to have a history rich in attempts at fantasy simulation, is the answer.

    Nailed it, except you forgot to point out where anything was "circular".

    Gambling does, but my answer to that is that "gambling games" are completely their own animal that work in a fundamentally different way than games of strategy do.

    I agree, but there are better ways to acheive "you can't quite control everything". The things that are out of your control should still follow some kind of logic that can theoretically be understood.
  6. Logo

    Logo Well-Known Member

    I don't think I am. Well I agree with what you said in response, but for multiplayer games it seems like 'too much' agency can cause stage fright. The example being solo games (SC2, Outwitters, etc.) vs team games (DotA, Lol, etc.). A single person's overall agency on the system is less in the latter in the sense that their teams factor in to the outcome a great deal. Team games also have much less stage/ladder fright. I don't think the two are unrelated.

    The way I'd reconcile the idea of maximizing agency with what I just said would be that generally you'd look more at the agency of a team game by the combine input of all players (or all players on one side). That and not all players get ladder anxiety.

    You could argue single player games don't have the same property, but maybe they do? Most singleplayer games have low consequences, but those that have high consequences & are relatively popular (like roguelikes with permadeath) often mitigate a fair bit of agency (in this case through randomness).
  7. infernovia

    infernovia Well-Known Member

    Right, so you have described how it works well for some of the games but not for others. I am not saying you are wrong for all games ever, I am saying according to your model, we would have to fundamentally change a lot of games that have been historically praised, when it's not clear how you would change it from the current functioning design to a very different one.
  8. keithburgun

    keithburgun Banned

    This is true. I think we can do much better than most of the things that have been historically praised in the super short history of videogames.
  9. infernovia

    infernovia Well-Known Member

    So... again, how would you improve Civilization?
  10. evizaer

    evizaer Well-Known Member

    Please don't derail this thread, infernovia. I don't want this to turn into an off-topic shouting match and/or become a replay of the "I just whatevered a book on game design".
    Targie and link6616 like this.
  11. infernovia

    infernovia Well-Known Member

    Alright, I won't argue anymore.
  12. major_shiznick

    major_shiznick Well-Known Member

    Ah okay, I think I see where you're coming from here. To a person who understands games, you're absolutely right. If the player is at all game-savvy, they should be able to see through a game that uses randomness to shove aside agency. However, it often acts as a placebo for gamers who are not aware: people who are more susceptible to the psychological patterns Keith mentions in the beginning of the article. It is not just children who are fooled into sensing false (or at least exaggerated) agency; it happens in most every "standard" card game, Bridge being huge a offender for example, despite being taken very, very seriously.

    Not sure whether or not I agree with that characterization; it seems overblown. I see it more like the player is a patient who has to take a medication every day. Some days their medicine is real (they actually earn their satisfaction), and some days their medicine is a placebo (where their satisfaction is the result of a "cheap" psychological trick). Sometimes the placebo works, and sometimes it doesn't, but if the patient catches on and feels short-changed, they should of course seek a more satisfying, "legitimate" medical regiment.
  13. keithburgun

    keithburgun Banned

    Great example, and easy to answer. Civilization is far, far too long and gets far too large in scope mid/late game because it's trying to simulate the path that a civilization would somewhat realistically take. Everyone agrees that the early game is far more interesting and less tedious than the late game. There are a number of answers to this problem but none of them have been tried in five iterations of the game so far, because all of the good answers would require dropping the fantasy simulation to some extent.

    The tech tree also could be randomized a bit more to stop players from just following an optimal path every game (actually Master of Magic does this), if you weren't as concerned about "the real way that these techs unfolded in history".

    The major point is that when you drop any semblance of history/fantasy simulation, you're free to make the best gameplay decisions possible.

    Civ is full to the brim of bullshit; if you play through a full game, you can expect to hit Skip Turn without doing anything many times, and you can expect to choose some building out of a list of 100 almost at random on city after city. Late game Civ is hell, to the point where I find it curious that you'd even ask how Civ could be improved.

    My point is that this placebo kind of affects us all. I find myself sucked in by it sometimes, and internally conflicted, knowing that while I enjoy this now, it's an illusion that will break soon.
  14. evizaer

    evizaer Well-Known Member

    Let's talk about how you could randomize a tech tree to make it more interesting in play.

    I see two approaches here:
    1. Randomize the whole tree and let the player see the whole thing at the start. In this case, I think the player looks at the tree for like 10 minutes and can see the new optimal paths without much difficulty. Or the tree is a jumbled mess and you don't care THAT much where you go on it because it's so disorganized due to being random that specialization is too difficult or untenable all-together.
    2. What techs are available to you NEXT are randomized, but this seems like it just leads to a reduction of agency because you can't plan ahead much.

    Then there's the problem of techs generally facilitating growth of your empire, which leads to the scaling problems that plague 4X games. That's a yet BIGGER design problem which I don't know if you can address through reworking the tech tree--unless you just make it very short and render the game unrecognizable to what it is today. At that point we're just designing an entirely different game.

    Perhaps the flaws are just endemic to the design goals and traditions of the genre.

    Agency is not the guiding light of people's interest in colloquial games, though, and I think that your article operates on the assumption that it may be. Some degree of agency will be required depending on the player's taste, but that amount of agency isn't fixed and there are lots of complicating factors. For instance, I know that Auro gives me much more agency than many of the game I enjoy, but at the same time I find it to be too sterile and restrictive an experience for me to get much enjoyment out of the agency.

    Of course Burgun-games are all about agency, but that's by definition and isn't particularly worth discussion.
  15. Bucky

    Bucky Well-Known Member

    Then the obvious improvement is to remove stuff from the game and shorten it a lot. Prune the tech tree. Make building production faster in general, and have fewer of them. Shrink the board. Incentivize worker-raiding. Make the computer players act more aggressive.
    keithburgun likes this.
  16. Lofobal

    Lofobal Well-Known Member

    Ah, I see you've met Mr. Burgun.
    Scarbo, CWheezy, Targie and 1 other person like this.
  17. Logo

    Logo Well-Known Member

    The point of 1 isn't that players wouldn't take an optimal path, it's that the optimal path(s) would be different. One game you may end up with a super early navy because of the tech tree the other a lot of economy. yes it has the drawbacks you mentioned, but the purpose of it wouldn't hinge of players being unable to figure out the optimal tech tree path. I also think he meant it as an incremental improvement/example, not a solution to all of the problems of a civ like game.
  18. evizaer

    evizaer Well-Known Member

    Isn't the point to make it organically more difficult to figure out the optimal tech path given your situation without diminishing agency too much? Changing the playing field every game is a kind of cheap way of accomplishing this that doesn't get all the way to improving depth. The procedurally-generated tech tree will definitely have balance issues game-to-game.

    There are two kinds of games:
    1. Games where the starting state is always the same, but the game rules allow interactions between players that promptly cause any two instances of the game to diverge. Your search for optimal solutions is mapping out different extremities of the decision tree of the game each time. Such a game can be said to have a lot of depth in the sense that from the same conditions you can arrive at wildly variable intermediate and final situations.
    2. Games where the starting state varies. These games eschew the kind of deep search one can do in a game with fixed starting state. The more elements of the starting state vary, the more each instance of the game produces a different decision tree that requires new exploration that won't carry over the future games.

    So your choice in game design is between a kind of organic depth, or varying the start conditions such that you can never explore deeply but you must explore anew every game (to a varying degree which depends on the number of variables in the start state that are manipulated.)

    Doing (2) is cheaper than doing (1), and I actually tend to prefer it. But I think that better and more efficient game design is to create a game that does (1) well. Go and Chess fall into the (1) category.
    Targie likes this.
  19. major_shiznick

    major_shiznick Well-Known Member

    What about the part of (2) where randomization of initial states can test player adaptability to starting conditions that they may not be comfortable with? This is especially relevant to games where those starting conditions are hidden and there is an intellectual battle to deduce your opponent's strengths or weaknesses alongside the strengths and weaknesses of their initial state.

    Of course, if the initial conditions are procedurally generated instead of randomly selected from an authored set, then I would more closely agree with how you describe (2). Maybe it would instead be really interesting to have a tech tree where each level of it is selected from a bank specific to each level of the tech tree (or whatever organization makes sense, depending on how crazy you want it).

    Tracing back, stating that all games are (1) or (2) makes for some weird classification decisions. Where do Yomi or Android: Netrunner go? What about Fischer Chess? What about Marvel vs. Capcom: Clash of Super Heroes (where assist characters are randomly selected)? I think it's much more accurate to say that any one element of a game's beginning phase falls into (1) or (2) and that instead some games are made to be entirely one or the other.
  20. evizaer

    evizaer Well-Known Member

    All games have some combination of random and non-randomly determined start conditions. I drew the line there in the easiest way, which is to say "some random elements" vs. "no random elements." That classification isn't really the payoff, the important part is that the manipulation of start conditions procedurally can provide a facsimile of depth, but it's fundamentally distinct from depth.

    And also the point at which a game "starts" is an issue of dispute here, as your examples show. I was thinking that a game starts when the players start interacting with the system. So in Arimaa, the game starts with a blank board, for instance, so the start conditions are non-random.

    The tech tree in Civilization is actually a set of game rules. By randomizing the tech tree, you're changing the rules of the game in a sense. You can see game start as the point at which the players begin operating within the rules of the game.
  21. keithburgun

    keithburgun Banned

    To get things back on topic a little bit, I just made a sequel post. This lays out my 14 points leading me to the conclusion that randomness should not be a part of ideal game design.
  22. 2000 IQ Killjoy Gamer link6616

    2000 IQ Killjoy Gamer link6616 Well-Known Member

    Interesting points, although I had been hoping the follow up was going to head into looking at the placebo effect or agency. Within your own definitions I can't disagree with them really other than the poker points listed by Dave in the comments already. Outside your definitions I think it's incredibly limiting since games have many purposes but that's covered in your clarification points to an extent I guess.
  23. Caphriel

    Caphriel Well-Known Member

    The following is harsh, but it exemplifies why people say you are bad at arguing:

    I disagree with point 2. Or rather, you seem to be claiming that the value in point 2 is the only value in games, which I disagree with. Or maybe this is a games vs. games* issue, in which case I might say you're begging the question, because you have defined games* such that they have this value and only this value.

    Point 4 implicitly assumes that all games involving randomness (output randomness, specifically) are gambling applications that "inducing compulsive play by randomizing reward schedules." You've created here a false dichotomy, between the "good" value of endless learning and the "evil" value of skinner boxes. You don't state that outright, but it's implicit in the way your argument progresses, and is important in order for point 2 to stand, c.f. clarification 2.

    In fact, as you progress through these points, you continue to quietly exclude games that don't meet your ideal criteria or the criteria of heavily-random skinner box addiction machines. That continues in points 5 and 6. You say " Many modern games that we know today are actually some mix of what I’ve described as “games” and what I’ve described as “gambling applications”."in point 5, then in 6, continue to address that subset of games. You ignore modern games that are not a mix of games* ("what I've described as "games"") and "gambling applications."

    Point 7 continues the above exclusions, and adds a new one: " It is then reasonable to assume that there could be games that we enjoy even as adults which are illusory in a similar way." If it is reasonable to assume that, then it is also reasonable to assume that there could be games we enjoy as adults that are not illusory in a similar way. You ignore that category of games, too, and continue as though all games we enjoy as adults are illusory in a similar fashion as the ones we enjoyed as children.

    Point 9 sets up another fallacy. "I’ve never said that highly random games have no strategy involved, but merely that at a certain point, the strategy gets largely solved and it breaks down into sheer gambling." You correlate randomness and strategy-solving, imply a causitive relationship, then associate strategy-solving with gambling for negative association.

    C.f. point 2, where you said, "If a game... has a very long, seemingly endless set of lessons to teach (depth / difficult-to-master), then that is a game that has great value to humans." Games, with or without randomness, eventually have the strategy become largely solved. Games without randomness just become boring and unplayable (Tic-Tac-Toe, Checkers for some), while games with randomness "break down into sheer gambling."

    Point 14 now adds another problem. Chess is not a "popular, well-liked game" by any reasonable statistical definition. The US Chess Federation currently has about 80,000 members. There are currently 170,000 people playing DOTA 2 on Steam at the time of this writing. If a list of "ideal guidelines" for game design would include "no randomness" or even "no output randomness" then games would be restricted to a much diminished subset of the population. Let me repeat that emphasis:

    Excluding all randomness as a factor in ideal game design reduces gamers to a fraction of their current population.

    It's fine if you want to say "I only care about nonrandom games with a very long learning curve." But to say, "Randomness should not be a part of ideal game design." means that you think that guideline should apply to all games. Unless you meant games*, in which case your article is unclear, because in some places you specify "games as I have defined them" and in other places you use "games" to clearly mean the common usage.

    As a final note, I disagree with clarification 4. As I have argued here in this thread, and elsewhere, input randomness can be as bad as or worse than output randomness. All randomness (whether input or output) is like any other tool in a designer's toolbox: used well, it can improve a game, and misused, it can ruin one.
    link6616 likes this.
  24. Caphriel

    Caphriel Well-Known Member

    For clarification of the the initial statement:
    You engaged in several logical/rhetorical fallacies (begging the question, false dichotomy, association fallacy, red herring/straw man), your conclusions don't follow in good logical form from your premises, and you make factually untrue claims, like Chess being popular. Don't get me wrong, I love Chess, but it's not popular by any stretch of the imagination.
  25. ratxt1

    ratxt1 Well-Known Member

    Basically here is the main problem. Burgan's arguments are misleading as he redefines what a game is than makes arguments about said new game definition where it is unclear whether he is talking about Burgan games or regular games. This whole misleading method of argument isn't even that usefull as it only helps us in making or analyzing Burgan games, when most games don't fall into this category. It's kind of like saying that removing betting from gambling games makes them bad (which is pretty obvious they'd be boring as hell), and than phrasing your argument in such a way so it seems like your saying that games that don't involve betting are bad games.

    I'd liked your first article on Placebo and agency much more than this one, as agency applies to all games. The fact that you control what's happening is one of the main things that sets games apart from other media.
  26. keithburgun

    keithburgun Banned

    I have a prescriptive definition for the word games. Feel free to swap in a different word if you prefer. The points I am making only apply to that kind of thing.

    I believe that to be a true dichotomy, actually. If you want to explain how it isn't, I'd be happy to hear that out.

    There are many problems with this statement. Firstly, 80,000 people "registered to the US Chess Federation" doesn't mean there are 80,000 chess players in the world. There are probably millions, or at least hundreds of thousands. Furthermore, even if your numbers were useful, a game doesn't have to be "more popular than DOTA" to be "popular".

    I should also mention that DOTA has almost no randomness (only critical hits, if I'm recalling correctly).
  27. Shiri

    Shiri Well-Known Member

    DotA has plenty of randomness. Last time I checked even the basic autoattacks did random damage which varied quite by like 8% of the initial value in some characters. Then the gold you get from creeps is random, there are characters with entire random moves like Ogre Magi, random stun procs on items (the most obvious and worst offender)...random creeps in the jungle spawns which makes a huge difference if you're playing a character like Chen...might be more stuff I'm missing.
  28. Logo

    Logo Well-Known Member

    DotA2 has:
    -Jungle spawn type
    -Rune spawn type
    -Rune spawn side
    -Ogre Magi & Chaos Knight, PotM offensive abilities
    -Attack damage spread (physical attacks)
    -Pseudo-random: Stun chance, crit chance, dodge chance, proc chance (faceless void's rewind for example)
    -random miss chance to high ground
    -random travel of chained abilities (arc lightning, omnislash, chain frost thing on lich)
    -random/single draft (one mode only)
    -Starting side (asymmetrical map, only random in matchmaking though)

    DotA 2 is reasonably deterministic despite all that, but yeah there's a fair # of randomized things. Rune spawn type & side might be the thing that has the most impact, though jungle spawn type can be huge as well (mostly because of the magic immune spawn and that one really tough spawn). The others can generally be pseudo predicted and have less unexpected variance.
  29. Caphriel

    Caphriel Well-Known Member

    Okay, then you are begging the question. Your (prescriptive) definition of games defines "games" such that randomness is bad/should not be an element of those games. Taking that definition of games, you use as a premise of your argument "by my definition of games, the only value in games is endless learning." Then you use that premise to conclude that randomness is bad. Your conclusion that randomness is should not be a part of ideal game design relies on the premise that randomness is not a part of game design by the definition of game.

    I don't mean this to be rude, but do you understand what a false dichotomy is? When you say that it's a true dichotomy, what you are claiming is that those are the only two options. Either a game has endless learning or it is a skinner box. What about games that don't have endless learning and don't have randomness? Checkers, for instance.

    What about games that only have a little bit of randomness? Does that make it a skinner box? That's a rhetorical question. They are not skinner boxes. This is a false dichotomy because you have claimed there are only two options when there are actually more.

    ratxt1 pointed out that there were way too many statistical and methodological flaws in my comparison of Chess to DOTA2, so I've deleted it, and I retract my claim that Chess is not popular.
    zem likes this.
  30. ratxt1

    ratxt1 Well-Known Member

    Wow is that argument to why Chess isn't popular full of flaws. And it is in the same post where you are criticizing someone else for using logical fallacies. First off you are totally biasing the results when you are choosing Chess' versus DOTA2's target market. I could just as easily say that DOTA2's target market is people who have good internet access as I could say it's people on Steam. And in your Chess' target market you are including many people who can't even play chess like people under the age of 5, or people who are mentally handicapped. Chess is defintely a popular game, not as popular as DOTA2 sure, but DOTA2 is huge. But what really makes Chess CHESS is it's influence and legacy, there is way more information on Chess than there is on DOTA2, there has been way more money earned from playing Chess than from playing DOTA2, there have been way more tournaments for Chess than for DOTA2, there have been way more devotees to Chess than to DOTA2 etc.
  31. Caphriel

    Caphriel Well-Known Member

    You're right. I'll delete that so it doesn't continue to draw responses and retract my statement about chess's popularity.
  32. evizaer

    evizaer Well-Known Member

    Here's my detailed response to the argument presented in the article.

    If that article is just about burgun-games, then it's circular. Burgun-games are defined such that any limitation of agency in play is bad, therefore it is true by definition that output randomness is bad because it can act to negate player agency.

    If the article is about colloquial games, then I have a number of additional points and criticisms, which I've written out below.

    Point 1: People fundamentally care about coming to understand things that are relevant to their interests. Often this means that games may not interest people even though there's a huge possible depth of understanding available simply because the content of the game doesn't relate to a person's interests. Playing games itself is often not an interest of people. So there's no inherent value in understanding.

    This invalidates point 2. I'll argue as if it doesn't later because otherwise this is nowhere near as interesting/fun.

    Point 4: You are defining the intentions of a class of games that you just invented. Because some games have a feature doesn't imply in any way that there should be a class of game that focuses purely on that feature. Don't confuse "I think that these games retain players through reward scheduling trickery" with "these games are intentionally designed with the objective of presenting tantalizing reward schedules and that's their only merit." This argument doesn't even work, though, because well-designed non-random games are appealing because of their reward schedules, as well. If the feedback a game provides is routinely interesting to a player, reward scheduling has worked. Random reward schedules can be recontextualized as in Poker, with its added layer of betting, to provide the basis for a game that hosts a fair bit of skill and allows plenty of headroom for understanding.

    So your problem is randomized reward schedules (I'll just refer to it as RRS). These are reward schedules that specifically have some element of randomness involved.

    I'd restate your argument as:
    Games that have RRS are bad because RRS provides noise to the process of understanding a game. Since games are all about iteratively improving understanding, adding noise to that process makes it less effective...

    But it doesn't. Adding some additional variance complicates an in-game situation up to a certain point, beyond which it begins to obfuscate. Note that there's a continuum here, not a dichotomous relationship. Complications are required to have the kind of indeterminate situation that players will love to spend their time pondering in a game of strategy. Randomness isn't the only way to achieve variance, but it's one tool. Randomness can provide variability and create a more dynamic game situation-to-situation. When applied poorly randomness can rob the player of enough agency to make him question the value of his efforts. But it's not advisable to outright dismiss randomness as always being a poor tool based on this fact.

    Point 6: I would suggest that a more likely explanation of why children play candy land is to see what happens instead of to express agency and compete in a contest of skill. I think you're disregarding the role of play in childhood. From my experience and reading, the role of play seems to be to prepare the child for the kinds of tasks and roles he may face or become as an adult. Playing games as a child is a struggle of learning to subject yourself to the rules of others and their consequences always and by definition. Play is an exploratory tool first and a means of expression much later from my experience with the way that children play and the reading I've done. Children don't just pick up how to take turns and follow rules the first time perfectly and then start coming up with strategies and ways to express their skills. It takes years for a child to get to the point where he can start strategizing at any level. It seems highly unlikely that a child is deceived into believing they have agency when they play any game because they may not have the capacity to believe they have agency and that capacity is probably one of the furthest things from their minds.

    There are also probably a significant number of better reasons why adults don't play candy land. Candy land is, at a base level, too simple to entertain an adult for long, independent of randomness. It's also themed for children, which makes an adult much less likely to believe it's befitting of them to try to play it and take it seriously.

    Point 7: When we're adults, we're entirely capable of being deceived about agency. But not everyone has the same base level of expectation of agency, let alone expects all games to manifest the same amount of player agency. People's goals when playing games aren't just to build decision-making skills. As such, I feel it's dishonest and incorrect to position agency as if it were central to everyone's desires and any substitute for agency is deception.

    Point 8: There's no distinction between your example and a chess player knowing that in a given situation a certain move is optimal. The poker player has learned through investigation that the probability of him winning a certain hand is 90%. He can take those odds or not. Both him and the chess player express their skill and understanding through their play with this information in mind. Note that neither of them has anything to gain in terms of understanding from playing the move. The chess player who choses the optimal move understands the move's optimality just as much as the Poker player understands the he's maximized his chances of getting a good payout by betting on the 90% probability of success and neither of them in your example have any understanding missing. The fact that the chess player wins the match after playing a series of memorized optimal moves doesn't impact his understanding any more than how draw luck in a few poker effects the poker player's knowledge. The poker player has to pick up on new tells from each of his opponents when he meets them, read their betting strategies, and pick his own to suit the situation. This allows quite a bit of an expression of skill. Notice how high the skill ceiling is on reading people vs. how comparatively low the skill ceiling is on, say, chess end-games or openings. The chess player is hard-capped at knowing optimal moves--and he can just memorize them for a non-trivial number of situations that actually come up in games. I'm not seeing the chess player necessarily as having less potentially for increased understanding in the context of playing the game.

    Given your case relies on Point 1's truth, I think I've just shown that Poker is not necessarily a worse game for gaining understanding over time than Chess is. By that, at least your example is bad.

    There also is no opponent-independent optimal strategy in poker, so Point 9 is factually incorrect. There are opponent-independent optimal strategies in Chess.

    Point 10: In most games with randomness, you can tell, with some non-trivial amount of understanding of the game systems, that you have chosen the optimal path, though the results may not be readily apparently optimal. Then you have to strategize from contingencies if your optimal path doesn't end up giving you optimal results, which adds more potential points for a player to express his skill. If you vaguely look at the game from its end state and wonder how a player won, perhaps randomness would wash it out, but if you actually reviewed a match with any degree of carefulness and attention to detail, you can identify situations in the game where an optimal move was missed and where opportunities were not suitably exploited and gain understanding. This is even possible in games with a significant amount of randomness--you just don't get as much benefit from doing it.

    Point 11: Same could be said for chess but with play of optimal moves instead of gambling--and gambling actually leads to variance while optimal play is just boring resolution. So I think games with randomness actually win on this point because they'll present the player with more ambiguous decisions.

    Point 12: Irrelevant. You use practical examples and you try to design real games.

    Point 13: Irrelevant. Games can exist with no actual randomness but still have enough moving parts to adequately confuse players and render their decisions seemingly irrelevant. This would indicate that non-randomness is no indication of increased capacity for players to gain understanding. Randomness is a way to add variance (which is called "noise" if you want to give it a negative connotation, it seems), but there are plenty of ways to approximate the same effect with non-random game features that simply outstrip the player's ability to memorize and keep in mind while playing.

    Considering that even when I accept your premises your conclusions still don't follow (in fact, the opposite of your conclusion may follow), I don't think your argument is sound.
    zem and link6616 like this.
  33. infernovia

    infernovia Well-Known Member

    Solid post Evizaer, pretty much hit all the points that I wanted to hit and then some.

    I would consider this to be the major argument against the model Keith is proposing.

    Interesting point, and not something I had thought about. I would say that the feeling of creating outcome in the game is enough for a child, because his mind can only handle positives and negatives in the game then. When the mind develops beyond it you desire for something more interesting than just postive and negative feedback.

    This was a solid example.
    link6616 likes this.
  34. keithburgun

    keithburgun Banned

    Well, firstly the rules of course are "limitations on agency" and they aren't bad. But further, agency can be limited otherwise, as long as it's done in a coherent way that makes some kind of logical sense. This is important, so that it can be learned. So, a game could have a totally good rule that forces you to skip a turn - a huge limitation on agency - just so long as this happening was based on something that can be learned, not noise.

    That's the big point of the article.

    Terrible counter-argument. There is inherent value in understanding - the human brain feeds on understanding, and this is well understood by modern science. Secondly, people not being interested in games isn't even related to what I'm talking about here, and you know that. I'm arguing that cucumbers have vitamins and you're counter-arguing that some people don't eat cucumbers.

    Right. This has been the status quo position on randomness for as long as there have been people talking about it. I'm positing that this view may be somewhat wrong - that even with a very small amount of randomness, we actually may be obfuscating the actual feedback loop of gameplay, but many of us just aren't picking up on it.

    Will respond to more later.
  35. ratxt1

    ratxt1 Well-Known Member

    I will agree that randomness does disrupt the feedback loop, which I will than again agree is a bad quality for most games, however, there is always give and take. Maybe I am ok disrupting the feedback loop a little so that I can get more variance in my gameplay, or so my game can actually work how it's I intend for it too (some games need randomness). I mean I can just as easily say that the less random a game is the easier it is for the person to work out the decision tree of the game as there are less possible options for outcomes of a given move, this is a bad property of a game, so no game should be completely deterministic. The problem with your logic is that the feedback loop is not the only thing that matters when making a good game.
  36. evizaer

    evizaer Well-Known Member

    I was showing how understanding alone isn't sufficient for people to enjoy games, so only analyzing games as if they are only about understanding is not adequate and justified.

    If you admit that people's interest in games is more than just about understanding, then you can see how games with randomness can work.

    Even if we only and solely admit increased understanding as the valid reason to play games, we are still left with a bunch of the issues I point out later. Particularly that some additional variance provided by randomness can provide the player with more chances to test his or her understanding; and that odds can be understood just as certainties can be and people seem to be able to pick up on the fact that choosing odds for one outcome over another is an agency-conferring action meaningful to the player and game.

    "The human brain feeds on understanding" and "Games are entirely and only about understanding" are not the same statement. You are making the latter statement and not the former. I am only going to discuss the latter, because it's the entire foundation of your argument as stated in point 1. (Your rhetoric here is an example of what others have called "moving the goal posts.")

    Your positing that the view is entirely wrong. You're saying that feedback loops must be unsullied by randomness because it "clouds" the result.

    Clearly you aren't against clouding results categorically, because games are just the artificial clouding of the path to agreed-upon goals. Specifically, it seems you are against any fixed uncertainty in a game system that can't be conquered by understanding. This only stands up logically if the only role of games is to provide an understanding scratching post, which I maintain that games require more than just some understanding headroom to interest players. I think there's more to the value of games than understanding.
  37. evizaer

    evizaer Well-Known Member

    So in the interest of not getting woefully off-topic and derailed, I'll restate the Burgun argument as I see it:

    1. Players play games solely because understanding has inherent value to humans.
    2. Therefore the only way to judge the value of a game is by assessing its ability to sustain incremental increases in player understanding for as long as possible.
    3. Therefore games should pursue being limitless sources of additional understanding.

    4. Randomness provides a hard, fixed point of uncertainty that cannot be penetrated by understanding.
    5. Because understanding is the sole and fundamental action of playing and valuing games, any mechanic that cannot be penetrated by understanding conflicts with the core value of games.

    Conclusion: Randomness fundamentally and always is negative and should be avoided in game design.

    Please correct this restatement if it misses the point or is missing something. I think we can use this as the canonical statement of your argument without the additional verbiage that seems to be confusing people. You seem to serially pick poor examples (cf. the poker discussion), so we can dispense with them all together and just discuss the argument on its own merits.

    The reason I disagree with your conclusion is because I think premise (1) is incorrect. I agrue that (a) people don't value understanding by itself, they only value relevant understanding, and (b) people seem to value spectating things that interest them and an inevitable part of every game experience is spectation (seeing what happens).

    (a) gives the Burgun argument problems because relevance to the player's interests must be established first before the player will engage in understanding improvement. Throughout the process, if the game ceases to be relevant to the player he won't care about gaining more understanding, so a high understanding-ceiling does not necessitate the player valuing the game highly. Games must be designed understanding this need for relevance to the player's interests, which is a complex and difficult process. We usually discuss this in the style of "identifying a core experience you'd like to represent or create."

    (b) means that the understanding feedback loop is not fundamental qua agency, so randomness that leads to interesting outcomes is valuable.
  38. keithburgun

    keithburgun Banned

    That's the issue you're getting hung up on. Of course if we use the word game generally, people play them for all KINDS of reasons. What I've meant is that the unique value of games - the value that can only be gotten from games - is the understanding through decision-making in a contest thing.
  39. evizaer

    evizaer Well-Known Member

    "the unique value of games - the value that can only be gotten from games - is the understanding through decision-making in a contest thing" is a totally different statement in all dimensions than what you say in Point 1 and 2:

    I argue that this statement is false because people don't find understanding anything and everything valuable. This is why you can be bored by someone explaining something new to you whereas someone else may be riveted by the same explanation. If gaining understanding were valuable independent of our personal interests, we'd be enriched and entertained by an explanation of anything. This clearly isn't true in reality.

    "It can be both enriching and engaging..." would be a more accurate statement. At this point you'd have to admit the other contingencies that would cause it to be enriching and engaging, and then address those conditions in the rest of your argument. It's the only fact about how people value games that you included in your argument.

    This is a very vague statement.

    A minimal interpretation would be "People value games only if we can make sense of them at some level." This says that the ability to understand a game is a necessary condition--but not a sufficient condition--for people to value games.

    When I read your argument, I get the impression that you mean something more like "People value games solely because they reward increased understanding." I have this interpretation because you argue later as if that is the only fact about games relevant to how people value them, and thus the only fact relevant to whether or not we should include output randomness in games.

    Can you clarify your wording there?

    Regardless, "Feature A is unique to a thing" doesn't imply "that thing should focus on Feature A to the exclusion of any other feature."
    zem and major_shiznick like this.
  40. vivafringe

    vivafringe Moderator Staff Member

    Yeah, evizaer seems mostly on target here. Here is a counterexample to the idea that "unique" means "focus on it." Skinner box methods are far more effective in games than in any other (known) medium. Games are uniquely well-equipped to deliver randomized rewards for our lizard brains. Indeed, that is why skinner box games are being released so damn much; they are far better at it than old standbys like lotto tickets.

    It seems like rather than trying to talk about properties of games, Keith should be talking about things people enjoy doing in the long term that run counter to short term impulses. Then, make games that maximize enjoyment.
    link6616 likes this.
  41. Caphriel

    Caphriel Well-Known Member

    This. If all I want to do is come to understand something, I'll go learn something with practical value, like another programming language or another design paradigm.
  42. ratxt1

    ratxt1 Well-Known Member

    Question, if understanding through decision-making in a contest thing, is the unique value of games, and the ONLY thing that should be desired in a game. How come we aren't all "playing" geography tests? these seem to be a perfect fit for your criteria. No randomness, only thing it does is test your understanding, and a great feedback loop (it even tells you what you did wrong), once you get 100% on one of the quizes you can just switch to another one, seems like this site has hours of "enjoyment" that we are all missing (though to be fair I have found taking geography tests on that site pretty fun when preparing for geography tests in the past).
  43. major_shiznick

    major_shiznick Well-Known Member

    I'm tempted to comment in detail on "the purpose of games", but I think evizaer basically has it. I don't think I can contribute more than what he's laid out, because it's incredibly complete and deserves detailed response at some point.

    I was able to like the placebo article because it addressed an actual problem in "serious" games: the use of randomness to cover up a game's lack of depth or decision-making potential. I think it is sufficient for a designer to ask themselves, "will randomizing X mechanic/rule/whatever endanger overruling the importance of the player's decision-making?" The conclusion of the placebo article suggested that this question should not be asked, though an informed reader was left enough intellectual space to challenge this.

    This follow-up feels more like the reader is told that output randomness should be categorically rejected. Then goes on to massage the reader into agreeing, hoping that they don't catch on to the fishy initial statements. I felt like I gained nothing from reading it, unlike most of Keith's articles where I can disagree with individual points or conclusions but in doing so refine and shape my own perspective on game design. To put it a bit too harshly, the 14 points felt like I was being given a rhetorical placebo to replace actual understanding of the role randomness can/should play in a game.

    ratxt1, I don't think Keith or anybody would say that geography tests fit his criteria (or most people's criteria) for "game". Keith would call it a "puzzle", since it is only posing a problem to solve; no decisions are actually made. I get your point, but I don't think it's quite in the same ballpark.
    Remy77077 and link6616 like this.
  44. ratxt1

    ratxt1 Well-Known Member

    Well decisions are being made (you have many choices of what to pick), but I guess they arn't ambiguous decisions, but than if the decisions are ambiguous how are you supposed to make perfect feedback loop, these seem at odds with each other (which they are and I would say that ambiguous decision making is a bigger hinderance to the feedback loop than randomness most of the time). My real point was that if the only purpose of games is to get a better understanding of a subject matter than why don't you just read a book (or take a test), also to link to cool site (Just got 100% on Asia test!!).
  45. evizaer

    evizaer Well-Known Member

    Keith would say that a geography quiz is a crappy game because you only have a binary tell of understanding: you either know the answer or you don't. There's no fuzziness to the process for "high skill" players, and the skill ceiling is low. There's no calculation based on prediction and untangling of complex valuation criteria, just the regurgitation of memorized facts. Memorization contests are distinct from Burgun-games, regardless.
  46. ratxt1

    ratxt1 Well-Known Member

    It is a crappy game, that's not really the point. All this fuzziness and complex valuation criteria, seems at direct odds to a feedback loop just like randomness. I mean have you ever played an abstract board game (turn based, no randomness, no double blind, emphasis on complex valuation and strategy and tactics), and felt no agency like you were just moving pieces and you didn't know what effect each of your moves had? This happens all the time in abstracts in my experience, humans are not like computers, we don't process stuff well through straight up looking x moves ahead, we need to use abstract concepts like tempo, and space, and positioning and material, and initiative to make sense of it all. Once you have wrapped your mind around all these concepts and understood the system better you can feel a huge sense of agency as you now have great understanding and know all the minute nuances and subtle give and takes that come with each move. The same thing that gives the game high agency at high levels of play, gives it low agency at low levels of play. I guess you could say we only care about high level play, but these agency barriers have swayed me from more than one acclaimed abstracts (which is a genre I quite like in general).
  47. evizaer

    evizaer Well-Known Member

    I agree with you and made such a point earlier in my first post responding to the point-by-point argument.


  48. keithburgun

    keithburgun Banned

    Well, I'd argue that the reason we get bored by things is because we don't have the proper context required to actually understand the thing. Listing facts for someone isn't the same thing as helping someone understand something. So, I maintain that human beings are indeed enriched and entertained by understanding, but getting people to truly understand something is difficult.

    To clarify my other statement: the more games allow people to understand them (again, not a simple process, requiring both accessibility and depth), the more value they have to humans.

    I never argued that games should have "no other features".

    You're confusing "knowledge" with "understanding". Right now, all of us have a complete understanding of the workings of the geography test system. The only thing keeping us from being masters of these systems is a list of hard facts, i.e. Nigeria Is Here. So geography tests kind of fail on both the accessibility AND depth side of the equation. And yes, it's a puzzle not a game.

    With that said, I think that if you package a geography test (basically, a trivia game) properly, you could find everyone playing it on iOS or something. (Who'd have thought 10 years ago that everyone would be playing Scrabble now?)

    Uh, actually the reason that I put it as numbered arguments was precisely so that people WOULD catch on something and be able to refer to a specific number that doesn't work.

    If the chess player is making a TRULY optimal move, then the game has broken down at that level, which is bad, but also which is bound to eventually happen. So it's kind of unrelated that a game can also be broken in a different way.

    The major point is, let's say we have a chess player who makes a GREAT move, and yet it still falls apart and he loses. He can break down exactly what happened and see the logical course of events which caused him to lose. In Poker, if you make a great move and lose, it could just be because "random, sorry".
  49. infernovia

    infernovia Well-Known Member

    What's wrong with that again?
  50. evizaer

    evizaer Well-Known Member

    Ah, so now we have to differentiate "understanding" from "actual understanding." More language problems.

    Regardless, a game about WW2, even if it has immense depth and is accessible, is not going to appeal to someone who thematically dislikes WW2 games (perhaps they find that the kind of strategizing they promote is not of interest to them, and they only like medieval-style warfare); or even a person who doesn't like games that force them to consciously strategize. No amount of "proper context" is going to change people's tastes and preferences. So clearly depth is not sufficient for people to value games.

    Another example: No amount of explaining the complexity of flavor in a certain wine is going to get me to like that wine. If I don't like the taste of red wine, it doesn't matter how much you hector me about how many flavors it cleverly presents to my tongue, I'm just not going to want to drink it.

    This is starting to sound like "you aren't getting someone to understand something unless they are engaged and enriched by it."

    You can keep saying "it wasn't a good enough explanation!" forever. The fact that the character of an explanation needs to be different to appeal to different people reveals that building understanding itself isn't sufficient for people to be enriched and engaged. There have to be other parts to the experience in place. You have to address how those other parts (which you have to characterize) are impacted by output randomness in order for you to have a sound argument.

    The more value they have to humans who are interested in playing them to begin with.

    Understanding-rewarding (we need a reliable term for this) is a very broad and vague feature.

    Your argument says that the value of games is as systems we can gradually come to understand. No other ways to value games are presented or treated as relevant. If you believed there were other ways we could value games, you'd have to include them in your argument and show why output randomness is irrelevant to them in order for your argument for the elimination of output randomness to be sound.

    Presenting a series of points that we can contest individually probably saves us a bit of time in interpretation. It'd be best if you could write conversational articles where we could actually reliably interpret your argument, though.

    What's the difference between said chess player's "GREAT move" and a 5% underestimation of probability on the part of the poker player? It's still quite difficult to suss out the difference in valuation effectiveness in either case, and if either player comes to understand his valuation failure, they could still improve their chances of victory.

    These are just two different kinds of optimality.

    People can deal with fuzziness in feedback and do so naturally. Having some randomness-caused fuzziness doesn't break the feedback cycle, it just changes the mode of optimality. Now if you have too much randomness and the game hides it from you, there can definitely be enough fuzziness to decouple the player's actions from the feedback--that's really bad, but that doesn't happen with just any randomness. I think coming to understand where this decoupling happens and the nature of the decoupling is important to a thorough understanding of game design.

    The variance caused by randomness would seem to increase the perfect understanding threshold (by making results intrinsically somewhat fuzzy; the degree of fuzziness can be damaging if excessive) and make the game playable for longer than a deterministic game where you know the results of your play in the long term due to your expertise. At least in a partially-random game there's some chance that you will actually have to start making difficult decisions again, whereas in a deterministic game it seems like you'd just fall into a groove and be kind of bored if you were nearing optimality.

    This topic is sufficiently vague that you can just argue endlessly, though. Maybe it's not worth talking about without the discussion being scoped to issues in a specific design.

    Regardless, arguing that any randomness is bad seems to me to be missing the forest for the trees.

Share This Page