Random Games and Skill Caps

Discussion in 'Game Design' started by keithburgun, Aug 2, 2013.

  1. keithburgun

    keithburgun Administrator, Lead Designer Staff Member

    "I felt like I pretty much knew as much as them despite having barely played the game"

    When games are significantly random, the skill cap is almost always god-damn low, but there's this "phantom skill area" that people spend years and years pursuing. I mean it's what I've talked about before with my Game Placebo article and my other articles on randomness, but I don't know, I feel like I haven't been strong enough on this problem.

    Or more specifically, I've been too lenient on games like Poker which have these giant crowds of apologists who will fight all day long to defend the position that it has a ton of skill to it. And I can't disagree because I myself haven't spent years pursuing it of course....

    What it appears to me is that there's a lot of games where people really solve them after just a few plays, but then spend years in placebo-skill land where they think they're learning lessons and stuff but really just random information is hitting them over and over.

    I think what I've said in the past is, well randomness sucks because it severs the tie between player agency and feedback and therefore understanding, which is the whole point of games.

    But what I'm saying now is slightly different - in addition to that, random games necessarily have a super low skill cap. Is that always true?
  2. EnDevero

    EnDevero Well-Known Member

    No, it's not true that random games automatically have a super low skill-cap, but, as I said in your last FS thread, every bit of randomness you add does slightly lower the skill-cap as far as I'm concerned.
  3. Nachtfischer

    Nachtfischer Well-Known Member

    I just talked to a guy about why I think Summoner Wars is a really bad (serious) tournament game due to the dice rolling all over the place. And also how I feel kind of "pranked" by such games heavy on randomness. Especially in SW, it's like "Oh, sorry, you simply don't matter this time!" - "Well, nevermind then, game. I'm sorry I even made a decision to begin with."

    Therefore I'm not even interested in squeezing out every bit of skill there could possibly be somewhere in this noisy system (that naturally draws win percentages towards 50 %, at least to some extent). You know, it kind of feels like having to POKE THROUGH ALL THE CRAP to finally find some (dirty) gold nuggets here and there, whereas perfectly fine nuggets are lying around openly (in the form of Puerto Rico, Through The Desert etc.).

    His reaction was the usual one: "But there IS strategy!"
    And it's true that it's hard to argue with that. I mean, I never claimed there was NO strategy (that would be Candyland). But it also cannot be declined that randomness is a HUGE factor in these systems and certainly waters down the competitiveness to varying extents. It's just that my tolerance for these systems keeps getting lower and lower.

    Oh, and then my favorite:
    Yeah... I obviously had to tell him about some actually good games that aren't Chess. Funny thing, he thought Carcassonne would surely be one of my favorites, too... well, NO.

    Anyways, I don't think you can say that any random game has an OVERALL low skill cap. I think they can have higher skill caps than non-random games and therefore probably in some of them you can validly learn more for quite a while (although I myself wouldn't want to do that due to the above reasons). But it definitely has a LOWER skill cap than it could, if it actually had a real game mechanic in place of the randomness. I think blox has an interesting point with that! And obviously randomness just feels like the cheap and easy glue you can pour over ANY system, whether it has some decent mechanics or not.
  4. Dasick

    Dasick Well-Known Member

    Well, by that logic Auro is a game with a low skill cap because of all of that random levels, candy and monster placement. But that's not true. So it's not a direct correlation.

    Though if someone is giving you grief, pull up win-rates of newbs vs pros. Obviously, even in Poker pros tend to be able to beat newbies, especially in the long run, but the win-rate is probably something like 60 or 70 percent.

    The mental model I use is the shifting bottleneck. Challenge is when trying to achieve a goal, but there is something in the way. This something is the 'bottleneck', and over the course of mastery it is constantly shifting. When the player first starts playing, the bottleneck is knowing the rules, the possible inputs, and learning to execute them. Then there's learning basic strategy, more in-depth tactics, grander strategy, counter-picking, valuation, memorizing patterns and opening and whatever, etc etc. Whatever stands between you and a higher win-rate. At a certain point, the bottleneck is unsurpassable. Perhaps there is nothing left to learn, and you've solved that game. But if you have sufficient randomness, usually it becomes that unsurpassable bottleneck, so you're changing one ceiling for another. I personally have no problem with either one, so long as it's a high ceiling.

    Also, the placebo effect is when medical treatment works because you think it's working, as in, there's an actual, observable, measurable effect. Obviously, if I think that I can control dice, I still can't control how they roll (I doubt anyone studied this, for obvious reasons).
  5. Nachtfischer

    Nachtfischer Well-Known Member

    I guess Keith meant games where the randomness is regularly a decisive factor ("significantly random"), which would exclude Auro.

    Not sure I get this. I mean, an unsurpassable point is always problematic in a system you want to learn something about... do you mean if there's enough to learn BESIDES the randomness then it doesn't matter to you if the FINAL bottleneck is a solved game or the fact that there is just randomness "left"?
  6. keithburgun

    keithburgun Administrator, Lead Designer Staff Member

    Ascension and Cosmic Encounter? Those are two EXTREMELY random games, especially CE. CE is practically as random as Candyland.

    Yeah specifically I was talking about output randomness (AURO has zero) and unfair input randomness (AURO has a level of that's below a threshold I think).
  7. Nachtfischer

    Nachtfischer Well-Known Member

    Right, that's just what I thought. But I guess there are things to exploit in these games whereas you can't exploit Candyland's dice.

    I think what the top Ascension players do is basically some spreadsheet-like analysis of what's best to buy when, and then they memorize it (better than all the other players) and beyond that it's pretty much all hoping. Yes, the top players will usually end up on top of tournaments (because of the analysis memorization), but between these players it's just decided by whoever has the most luck.

    And in Cosmic Encounter (haven't played it much to be honest), I imagine you could probably exploit the "psychology" of other players by tricking them into constellations favorable to you?
  8. vivafringe

    vivafringe Well-Known Member Staff Member

    This is incorrect. Imagine a game called "Go-Dice." Here's how it works:
    1. Players play a game of go.
    2. After the game, both roll a die, with the winner of the go game adding 1 to his score.
    3. The winner of the roll wins the game of Go-Dice.

    Is this game shitty? Yes. But it proves you can add an arbitrary amount of randomness and keep the skill cap the same. Adding variance to games basically just means it takes a larger sample size to find the better player. It has nothing to do with skill cap.
    Redless and Zirk like this.
  9. Nachtfischer

    Nachtfischer Well-Known Member

    Yeah, I don't know. It's a matter of how you define "significantly random". Maybe your Go-Dice isn't?

    What about another version where the winner of the die roll gets 1000 points? This would surely be significantly random and the skill cap is zero. And then what about 5, 10, 20 points?

    It's really a matter of shifting this "significance limit" around. But I guess we all agree that randomness does not necessarily cause a (globally) low skill cap.
  10. Dasick

    Dasick Well-Known Member

    Card draws are usually input randomness. Since we were talking about poker, I assumed we're discussing randomness in general.

    Precisely that. Design is optimization. You take on some negative 'weight' that can support/manufacture your positive 'uplift'. So long as the uplift is higher than weight, that's a design that will fly with me (ahaha, geddit?) - it's good design. Randomness is a huge weight because it fuzzes the feedback loop; but it effectively removes the line after which you know you're not getting better, and can be used to great effectiveness if you can keep the fuzzy area clear enough for a long enough time.
  11. Nachtfischer

    Nachtfischer Well-Known Member

    I think we are. Just the "bad" kind though (i.e. output randomness and "unfair" input randomness).

    I see your point. But doesn't it still feel like "gold nuggets inside a pile of crap"? Well, I guess it really depends. Maybe it's more like a pile gold with some crap sprinkled over it if it's an actually interesting system. :D
    But I assume the randomness would then have to be not be that significant/decisive...
  12. Dasick

    Dasick Well-Known Member

    I don't even think that you can have 'fair' randomness. At a certain level, when the skill level difference is really close, RNG will fuzz it up. So, the less is the skill difference, the more annoying RNG gets, and I think we all know that the better you are at a game, the more diminished the 'learning' return is, so at a high level of play, it becomes more of an issue. The way to mitigate it though is by giving players decisions that mitigate randomness.

    I wouldn't call it 'crap sprinkling my gold'... unless you want to get alchemical the metaphor I would use is lemons vs lemonade. There is a gradient between sweet and sour, and the big difference is composition. Obviously if you add more lemons it will be more sour, but it also creates a bigger base for more lemonade. So the more lemons (RNG) you add, the more sugar (decisions/options) you have to add to compensate.
  13. Nachtfischer

    Nachtfischer Well-Known Member

    But isn't the Quick Match mode of Auro totally fair randomness? (Trials is not 100 % fair, but surely not as obviously unfair as many other games. It's a matter of defining the borderline again...)
  14. Dasick

    Dasick Well-Known Member

    To me, fair-er randomness is when you can make a decision/play/pay a price to get a certain result, and when both players have to deal with the same random thing. In Auro, you can mitigate random monster movement through the flame trap skill, by freezing monsters, and by using monsters on cooldown. In deck builders you get more certain results by trashing cards and through the 'search for' or cycling cards. etc etc
  15. Nachtfischer

    Nachtfischer Well-Known Member

    Good point.

    But doesn't randomness always hamper the ability of learning, of getting good at making decisions in a specific system? Siginificant randomness even moreso obviously.

    Think about Summoner Wars again: There IS strategy to the system. But after a match the loser will oftentimes attribute his loss to bad rolls, and the winner to his superior strategy, even if it was the other way around (in both cases). And the thing is, it's mostly pretty hard to know what exactly about the outcome was caused by you and what was caused by randomness/luck. The feedback is basically disturbed by random noise...

    So, if you (like me) get the most fun/value from a system by learning, i.e. improving, then that's kind of a huge problem.
  16. Dasick

    Dasick Well-Known Member

    The big reason to use randomness is learning to deal with unpredictable turns of events. So long as you have a sufficiently deep core mechanic to do so, and the randomness is 'fair' to a certain point, I have no problem with randomness in games.

    I don't know much about summoner wars. My go to example of deck-building is Puzzle Strike, as little as I've played it (that's a problem I really should fix), but I do think that it gives it's players plenty of opportunity to deal with random draws ('bag' system, trashing cards, cycling & arrows). At least as a complete noob I don't see that fuzzy ceiling.

    Here is another thing - randomness tends to average out over a period of time, especially so if you use weighted distribution techniques or a 'bag' system. In my opinion, a good random system will have enough cycles to actually get that 'average' value before the game ends. Combined with the ability to control the randomness, these are my guidelines for minimizing the negative effects of randomness.
  17. keithburgun

    keithburgun Administrator, Lead Designer Staff Member

    One thing that we haven't talked about here, which is this concept about a game having a certain amount of "information space". Basically, a feeling that I've had that's been increasing over time is that you only have so much space in a system, and randomness is one of the things that can take up space. I don't have this all hammered out yet verbally, but basically, deeper games generally have more "practically understandable information space". So go has a ton of that.

    But randomness can also take up information space. Card games are "so simple to play!" because randomness generally takes up a huge % of their information space. You have this huge spigot of noise that makes an otherwise super flat system seem interesting to play. And randomness, of course, is not understandable, so a giant % of the information space of most card games is basically just... wasted, at least for people who want to build skill and understand a system etc.

    So that's why I have this feeling that there really can't be a highly random game that also has a ton of skill to it. Certainly they can have some skill, but yeah, I believe that with almost all card games, all of the pros might as well be rolling dice against each other, because the non-noise part of it they've all already solved. I'd love to be proven wrong on this though.
  18. Dasick

    Dasick Well-Known Member

    I hereby suggest that from now on we call "Go" "Weiqi" or "Baduk" or "The Encircling Game", because 'go' is a common enough english word that it makes parsing the text difficult enough.

    By 'information space' you mean a quantity of information that the user can process at the same time? Like, how many rules and interactions a player can remember at one time?

    I'm not sure your statement regarding cards is correct, but how would we go about proving it one way or another?
    keithburgun likes this.
  19. keithburgun

    keithburgun Administrator, Lead Designer Staff Member

    Actually by information space I think I mean more like, the information that will make up the workings of the game. Maybe Mechanics Space is a better term. So you only have so much mechanics space, and randomness can take a lot of that up.

    As to how we'd prove it, it's hard because it's really difficult to quantify depth / skill depth. It might be something that we can agree in a general way is true but not mathematically prove or something, dunno.
  20. Nachtfischer

    Nachtfischer Well-Known Member

    I don't think so. Even if some outcomes are more likely than others, when they're just random noise it's still bad. In Summoner Wars you e.g. hit on 1-4 and "only" miss on 5 and 6. So you could basically see that die as a coin, that falls on the "hit" side in 66,6 % of all tries. That's not uniform, but still noisy and to me it certainly feels bad.

    Yeah, it does. But the thing is, even if you make close to perfect decisions in a significantly random game, you might still lose. It "fuzzies" up the whole decision-contest. Again, in Summoner Wars attacking is somewhat encouraged by it being "more likely" to succeed than not. But still a player could just roll 1-4 for almost the whole game (happened to me, also happened to my opponents) and he has no chance, no matter how great his decisions were.

    What exactly do you mean? The fact that you exclude strategies in a TCG before actually playing (by constructing your deck in a certain way). But then you could see the deck construction itself as a strategic decision (and by the way, a design-wise pretty problematic one: it's disconnected from the actual game and usually based on asymmetry in owned cards to begin with.)

Share This Page