Mastery vs. Hoping – Games and Gambling Machines

If you have any output randomness at all in your game, or any of a certain kind of input randomness, then it’s quite likely that a compulsive gambling mentality will be the thing driving players to play. This is interesting to me because I have always said that single player games require some randomization to maintain their status as games, but there’s an interesting pitfall with regards to this that I think a lot of people aren’t totally aware of.

At a basic level, you could say that there are two totally different desires – the “game desire” of creative mastery (I want to see what I can do), and the “gambling desire” of hoping (I want to see what the system will do). I should also mention that both of these desires are referring to subconscious desires, not conscious ones – I’ll explain that further later on.

First, I need to delineate classical gambling machines and how they work.  This section is a bit obvious to most people, but I just need to be clear on it before I move on to my real point.


Gambling Machines

RouletteI’ve talked before about how “gambling machines” are really quite different than games.  The classical, obvious examples of gambling machines are things like Blackjack, slots, roulette, or a lottery.  While the best gambling machines give players some amount of choice to give them some illusion of agency, this choice doesn’t usually matter very much and is highly solvable.

While this solv-ability would be a terrible thing in a competitive game, it’s actually something of a boon for gambling machines.  This is because it allows players to have the following thought pattern:

“I’m playing SUPER well.  The only reason I haven’t won yet is just a series of flukes.  If I keep playing, I’ll definitely win.”

And eventually, they will of course win, so it’s not that that thinking is wrong.  They are likely playing optimally (or close to it), and eventually the dice will tumble in their favor.  So, this is roughly how this kind of compulsion loop thing works (more info on that here).  It tells you that “you’re doing everything right, it’s only a fluke that you didn’t win, so if you play again, you’re likely to win”.

The lottery is a great example.  You’re absolutely playing optimally by picking any numbers at all.  You’re doing the best possible move, so if you don’t win, that’s just some fluke.  Therefore, you should probably try again.  Even if the odds aren’t good, it’s only a matter of some random numbers going your way…


Random Videogames

What players – and possibly many developers – don’t understand is that if you have a videogame that has even some of a certain kind of randomness, that game may be driven by the same kind of compulsion that causes someone to continue playing the lottery.

Most people already know that this is how games like Diablo get people to keep playing.  What item is that monster going to drop this time?  But I think few people realize that many games which seem totally skill based are actually just as driven by compulsion as Diablo.

For instance – my friends and I spent about 2 hours playing Super Crate Box on the OUYA (by the way: there will be an article forthcoming about how awesome the OUYA is and how you need to get one, stay tuned).  I had played this before on PC, but with the whole console atmosphere, it was much more playable.

Super Crate Box

Super Crate Box

Super Crate Box, at first, seems like a completely skill-based competitive game.  Most people would certainly not think of this, upon first sight, as a “random” game, and certainly not as a gambling machine.  You run around, jump over monsters, and shoot, all of which requires lots of twitch dexterity.  There’s even a decently interesting strategy metagame going on, in terms of “when do I slow down and kill a bunch of monsters to clear out the screen” vs “when do I just rush for crates” (which are the objective).

However, none of that is what made us play this game for two hours straight.

We were handing the controller around and playing round after round, and as my friends were playing, I was thinking – what is the reason that this friend of mine, a super critical hardcore strategy game player, is so motivated to play dozens upon dozens of matches of this game, which isn’t particularly dynamic or exciting to play?

I realized it’s because Super Crate Box is most certainly, at its heart, a gambling game.

You see, at the top of the screen, monsters are streaming into the game.  They stream in at seemingly random rates, of random types, and heading in a random direction.  Now I’m sure there are some constraints on this;  I don’t think you’ll ever see a stream of 6+ large-head monsters come out of the thing in a row, but with that said, the constraints are still loose enough that sometimes, you get a harder game than other times.  This is particularly true since the three possible monster types aren’t at all equal in their difficulty.

This is compounded by the fact that the goal of the game – collecting crates – is also super-random.  The crate spawns in what seems to be a truly random position.  Sometimes it spawns very close to you, sometimes it spawns on the other side of the map.

Oh, not only that, but the weapons are also given out randomly, and they are certainly not balanced, especially if you consider the context of the current game (certain weapons are better versus certain spawns).

So, I think that when we play this game over and over again, we’re subconsciously compulsively HOPING for that one game where the monster spawns are favorable/easy and the crates all spawn really close to each other.  That, in a nutshell, is what gets people to play lots of Super Crate Box, and I don’t think the developers realize that.

I also don’t think most players realize it either.  Consciously, they’re thinking “yeah I gotta perform better”, but the thing that’s really drawing them to play over and over again is not mastery at all, but hoping.


Unfair Randomness

A game that has Unfair Randomness is a game that has some significant level of difficulty variance due to its randomness.  To avoid having Unfair Randomness, then, every play of a game must be equally (or very close to equally) difficult.

By “Unfair Randomness”, I don’t mean randomness that makes a player die right away and has the player going “What?!  That’s completely unfair!”  Unfair Randomness is usually very subtle and many players don’t even consciously notice it (but their brain does!).

The most common form of Unfair Randomness is simple “to-hit %” found in most RPGs.  While people will counter-argue with this by saying “oh well over many many hits, it averages out”, in reality, this isn’t the case, because not all hits take place in the same situation.  Sometimes, you’ll get a critical hit when you REALLY need one.  Sometimes, a monster will miss you when you REALLY need him to, and sometimes… that monster will hit you in those moments.

So taken over the course of one match, sure, maybe things always head towards 50% in general, but that doesn’t mean that you weren’t very fortunate with when the extremes struck, or very unfortunate when they did.

Another note:  Unfair Randomness doesn’t actually have to be “random” (well, actually nothing we deal with in any games is truly random, but you know what I mean).  For example, perhaps people don’t consider the way a ball bounces down in Pachinko or Peggle to be “random”, but it’s certainly effectively random, in that no one can possibly predict how it will land.  It’s quite similar to a real life dice roll in this way.


100 Rogues

Let’s take another example – 100 Rogues, my own first commercial game.  I designed this game back in 2008-2009, and I’ve learned a lot about interactive systems design since then.  Largely, 100 Rogues was based off of the existing “Roguelike” template, and so it has tons of input and output randomness.

I’ve been spending some time with the new OUYA version of the game (by the way, it’s the best version of the game ever, upcoming article on that, too) and I’m realizing that the most exciting part of the game is “finding out what item was in the treasure chest”.  This is a thing which the player has absolutely zero agency to affect, and yet it’s the most engaging thing about the game.

The Crusader's "Faith" ability - one of the best abilities in the game, and also the most random.

The Crusader’s “Faith” ability – one of the best abilities in the game, and also the most random.

Now that I understand this about 100 Rogues, I can make it a much better gambling machine.  The abilities need to be more random.  There need to be fewer item restrictions on the classes – for instance, the Fairy is probably less fun to play than the Crusader almost entirely because she has less chance of drawing great random items she can use.

So here’s the litmus test for whether or not your game has unfair randomness, and is therefore possibly more of a gambling machine than a game:

Is the most exciting thing that happens in your game something that the player has absolutely no control over?

For 100 Rogues, and many many other videogames, I’d say the answer is yes.  This doesn’t mean at all that I’m saying 100 Rogues is bad, mind you.  Instead what it does is inform me that I need to take it in a more gambley direction.  I didn’t have any idea of this “unfair randomness” concept back then, so I tried hard to take Roguelikes and make them more strategy/tactics gamey.  Much later, I realized that to do this I had to start from scratch, which resulted in AURO, which has essentially tossed out the entire Roguelike template and has very controlled randomness specifically to avoid the unfair kind, because I really wanted AURO to be a competitive strategy game.

We’ve worked hard to make sure that in AURO, the random arrangement of tiles and the placement of monsters is always fair, and that each game should be pretty damn close to the same in terms of difficulty (for Trials mode – the highest-score single-player mode).  In the future, I’ll be keeping an eye on this to make sure that it remains the case.  If I find that it’s still too swingy, I’ll look for solutions.


Other Examples of videogames that are accidentally gambling machines:

Civilization – “Hoping” to start near friendly/weak civilization, or large amounts of valuable resources, or to be the only one on your continent, etc.

Dominion – “Hoping” that the synergistic cards I bought will get drawn together

Arkanoid – “Hoping” that when the ball gets stuck up behind the wall, it will bounce around up there a billion times and basically clear the level for you.  Again, this “isn’t truly random”, but in the same way that a dice roll isn’t truly random.

Match-3 abstract games – This one’s a bit obvious, but, basically these games are all about “hoping” that gems fall in a favorable pattern.  I think probably stuff like Candy Crush and Bejeweled embrace this, which is part of why they’re more popular than stuff like Puzzle Fighter or Dr. Mario.


The point here is not even to make a value judgment about “strategy games VS. gambling machines”.  I, of course, greatly prefer strategy games, but as my two-hour stint of Super Crate Box shows, I can get sucked into a skinner box just as quickly as the next guy.

The point is that we should be aware of what it is that we’re doing.  There’s only so much room in a given system for sources of interest, and if your game has significant Unfair Randomness, that’s going to be the primary draw of your system.  Of course, there are other ramifications that Unfair Randomness has on a strategy game, which I’ve talked about at length.  But beyond that, you can strive to make a very good gambling machine, if you figure out that that’s what you’ve got.

  • Ido

    For a lot of games you’re point stands, but I think that Super Crate Box sessions pretty quickly (faster than 2 hours of play for players already skilled in twitch games) fall into the law of large numbers ( which alleviates the gambling effect.

  • Ido
  • Rickard Elimää

    This article seems a little schizophrenic. When talking about crates, all I can hear is you talking about reward systems. When talking about lottery, all I can hear is using “investment” to create engagement. (An investment can be putting time, money, relations or and other stuff at stake.) When talking about Arkanoid, all I can hear is uncertainty created by effort rather than something the player can’t have any control over.

    When talking about randomness, it will make me think of two distinctions. If it’s going into the system or out of the system. The latter is when you do decisions and then the random result tells you if you succeed. Craps works like this, and D&D too. If it’s going into the system, you get the result and then may do the best you can with that. The boardgame Kingsburg works like this and so do Settlers of Catan.

    It’s good to be aware of the amount of randomness in a game. I do agree with your points, especially the effect of randomness in strategy games, but randomness isn’t the only way of creating uncertainty. I would rather hear about how to create uncertainty other than randomness instead of hearing how bad randomness is. Marc LeBlanc and Thomas M Malaby may be useful guides in this subject.

  • Nachtfischer

    Great article! Specifically because I noticed the issue multiple times during the last months. Two iOS examples:

    SHIFTS is a strategy game where you control a space ship crew. Essentially your goal is to find four colonizable planets as fast (i.e. by using as few resources) as possible. The problem: Let’s say you play pretty well in two seperate games and explore 20 planets in both. You might find zero “goal-planets” in one game, three in the other due to random placement. And in another third game you might win on the third turn because you were pretty much surrounded by colinizable planets from the beginning (you then get a really high score, but were YOU better?). It’s a shame and I quickly lost interest in the game even though I like the mechanics (deciding which abilities of your crew to use at which point etc.). But it’s true that this is totally gambling on the unfair input randomness to be favorable.

    The other example is ORB COMBAT SIMULATOR, a game about collecting positive and negative orbs that interact in some clever ways. Again, just like Shifts this seems like a game I should enjoy: turn-based, no dexterity wall, deterministic outcomes of your actions (so no OUTPUT randmoness) and some decently put together new mechanics. But again, it’s largely gambling and I noticed this because I played quite a few games, trying to beat my highscore. And one time I surely just reached this very high number of points and it felt good naturally. But in the following games I didn’t even come close. I think I was caught in the machine there for like two days of playing hardly anything else (hoping for the right random orbs coming at the right time). I finally realized the wild swings of (unfair) randomness and again lost interest soon.

    One thing about (non) value judgements: It’s obviously fine IF you know what you’re doing to gamble for some time, to enjoy Skinner’s box essentially. Just like drinking SOME alcohol SOMEtimes is. But beyond that I think there SHOULD totally be some value judgement.
    Assume a person is obsessively playing and learning a really deep strategy game (e.g. Go). This person will probably constantly get better at the game, but on top of that even advance on a personal level (games teach learning, rational thinking etc.). Now if this person was instead obsessed with a slot machine or Diablo, what reason is there to be
    obsessed with these systems? I’d say it’s only the compulsive Skinner-box induced addiction. What do THESE systems teach? Is it fine to devote MANY hours of your life to them or rather harmful? I’d say at that point there definitely is a huge gap in terms of the ACTUAL quality (though maybe not with the perceived one, which is exactly the danger)
    between (strategy) games and gambling.

    And actually this value judgement is already well-established: Obsessive gamblers or alcoholics are seen as ill. What’s not nearly as well-established is applying the gambling-lense to videogames even though some of them make heavy use of the exact same methods.

  • Nachtfischer

    How exactly is it schizophrenic? In Crate Box you put time and even money (though not regularly as with the lottery) at stake just like with the lottery or Arkanoid. And all of them have elements you can’t control that determine how well you do. In the case of Crate/Arkanoid you obviously have SOME control, but the point is that there is still an UNFAIR amount of randomness, that is oftentimes decisive (also see my other two examples from my longer post). That’s something to watch out for!

    Keith has written about the difference between input and output randomness extensively and I’m sure you, him and me all agree that OUTPUT randomness in games (e.g. I decide to attack you and then have to roll a die to see if I hit or miss) is pretty much always problematic and if at all it should be used extremely carefully in your system. Because it takes away impact from your decisions (which is what games are all about). This kind of randomness is probably even more closely related to gambling. And btw, I think this (output) randomness is used heavily in Settlers (you might DECIDE to build next to a wood tile but luck decides if you really GET the resources), whereas the input randomness in this game is just the setup of the board tiles.
    What I think the article was about though was that INPUT randomness can if used in UNFAIR swingy ways also make a game break down into a gambling machine. You could argue that the dice rolls in Kingsburg are this kind of randomness, but the outcomes are definitely not all equally valid. So what you do is you HOPE for the input randomness to produce the right numbers essentially. Just like you hope for the right cards at the right time in Dominion.

    And yeah, randomness is not the only way to create ambiguity. In fact, it’s probably just the cheapest way to go with. And it’s one of the core skills of a game designer to come up with solutions for the problem of keeping ambiguity WITHOUT rather cheap things like huge amounts of dice rolls and card draws. Still, you CAN use these things in your game, but you should be aware of the issues and use them very carefully.

  • Nachtfischer

    I think Keith addressed that to an extent with the following:

    “While people will counter-argue with this by saying “oh well over
    many many hits, it averages out”, in reality, this isn’t the case,
    because not all hits take place in the same situation. Sometimes,
    you’ll get a critical hit when you REALLY need one. Sometimes, a
    monster will miss you when you REALLY need him to, and sometimes… that
    monster will hit you in those moments.

    So taken over the course of one match, sure, maybe things always head towards 50% in general, but that doesn’t mean that you weren’t very fortunate with when the extremes struck, or very unfortunate when they did.”

    Really, the question is: If you look at Crate Box as a GAME, wouldn’t it then be better if you didn’t have to play a LARGE NUMBER (however many that means) of games to make the gambling effect go away? Wouldn’t it be better if there was no gambling from the start? It’s basically just the same with Poker (you’ll see who is better over a large number of games, but not really in 1, 10 or maybe 20, or who knows how many games it takes).
    Or even look at a slot machine. I mean, if you play it a bazillion times you’ll end up with the average “expected winning percentage”. Does that mean it’s not gambling?

  • Keith Burgun

    Interesting that you have that opinion about ORB COMBAT SIMULATOR. I know the designer, he’s a super cool guy and I’m almost positive that he didn’t intend for the game to work that way, so that would definitely be another point in the favor of my argument I think. Especially because he’s a really good game designer.

    Yeah, you are right that at a certain point, a value judgment about gambling machines is completely useful. And I agree that it’s unfortunate that people who are addicted to video-gambling-machines such as Diablo aren’t seen as just as ill as someone addicted to casino gambling or a drug, because many of them are.

  • Nachtfischer

    Yeah, I am pretty sure Orb Combat was NOT designed with gambling in mind and in fact pretty much along your own guidelines. It is by accident though, that it breaks down. Something has to be done about these wild swings of input randomness (just think about how hard it is to make sure all games are decently comparable in Auro’s Trias mode). But as it is described as beta (or even alpha?) in the App Store, maybe the developer will do something about it if he feels the same way. You should probably redirect him to this article and these comments.

  • Rickard Elimää

    I found this article talking about several things but label it as something else (randomness), that’s why I said it was schitzo. I gave three examples how it did that in my first post in the first paragraph.

    We could argue until we’re blue about Settlers, but at least I play it as changing strategy depending on what resources I get. Can I build roads instead? Do I have to exchange something for the missing resources? Should I buy a card instead? But you got your point, there is a fair amount of randomness i Settlers too, like in Kingsburg. Dominion got both in- an output randomness and I think it’s important to see both. Input supports decision making while output destroys it.

    Otherwise, I fully agree with what you said.

  • AJ

    “For example, perhaps people don’t consider the way a ball bounces down in Pachinko or Peggle to be “random”, but it’s certainly effectively random, in that no one can possibly predict how it will land. It’s quite similar to a real life dice roll in this way.”

    You are looking for the right word for this in a few places in the article. It’s probably “chaotic” — deterministic but in a computationally intractable way.

  • Greg

    A very thought provoking article. I got here through cardboard edison so approached it thinking about board games, which I think is a slightly different perspective on how things go.

    I wrote up my thoughts on the thing here ( I think that we agree on some things and disagree on others, but that’s pretty normal. Thanks for the interesting article.

  • jimbob

    “The most common form of Unfair Randomness is simple “to-hit %” found in most RPGs. While people will counter-argue with this by saying “oh well over many many hits, it averages out”, in reality, this isn’t the case, because not all hits take place in the same situation. Sometimes, you’ll get a critical hit when you REALLY need one. Sometimes, a monster will miss you when you REALLY need him to, and sometimes… that monster will hit you in those moments.”

    These hypothetical people are presumably made of straw, because you can regulate these chances even in a small sample size by using pseudo random number distribution, which will appear in all competently designed competitive video games.

    Also, just because these random chances will be characteristically random at inappropriate times does not make them inherently unfair, so long as it’s possible to play around the randomness. If you are attacking and have no back-up plan for the attack missing, or are entirely treating it as a gamble, then you are playing badly. This is also why your Dominion example is bad, and this is why your FS statement that non-random options are always better is wrong.

  • Nachtfischer

    The thing is, even if you’re playing optimally (e.g. in Dominion) the randomness can just screw you.

    Input randomness is a good thing, but you need to make sure to control it. That means in a single-player game all playthroughs need to be decently comparable to all others. And in a multiplayer game the input randomness shouldn’t affect different players to randomly (wildly) differing degrees. So, if e.g. a random catastrophe happens in a game (e.g. by rolling a die) it would be unfair just to punish players to randomly differing extents. Sure, some players could have taken counter-measures to prevent getting hurt by the possible catastrophe and then that’s their well-playing, their STRATEGY that helps them. But all things being equal, you shouldn’t just say “Oh, this player loses 20 resources and this other player just loses 5.” (That would just be gambling… mere hoping to not get damaged.) But that is essentially exactly what happens in some multiplayer games and some highscore games with large and decisive swings in the input randomness.

  • Bob Shelline

    I’ve been thinking about this for days now. I’ve come to the conclusion that every game (by your definition, including ambiguous decisions) include input randomness. If it is a competitive head-to-head game, the human mind of your opponent supplies the input randomness, otherwise the game system itself must supply the random element.

    So, to some degree, every game has a gambling mechanism where you act in a certain way hoping that either your opponent or the game system will move into a state that is optimal for the strategy you are developing. If it doesn’t have this element, it becomes solvable and drops into the puzzle state.

    I’m trying to think of an example game that would disprove this, but haven’t thought of one yet. Can you think of any examples?

  • Rob Seater

    As you allude, one thing ‘unfair randomness’ does is create memorable play experiences. Often, the games I hear players talking about after the game is over are the ones where something dramatic and unexpected happened, not where it was a tense showdown of skill. This effect seems to be most pronounced in cooperative games, where players are more likely to commiserate in comradery (even if they would be frustrated in the same situation as a solo player).

    It doesn’t necessarily improve the game itself, but I think it’s another example of how tense strategic games lose out in modern markets. I don’t hear anybody talking about Go games they played, because it’s hard to grasp and describe the core of the excitement in a close game. Whereas, if you get a wildly unbalanced scenario in Betrayal at House on the Hill, everybody talks about it.

  • Keith Burgun

    A good way to think about “random surprise” is that it’s cheap. It’s like if you were watching some romantic comedy and then suddenly at the end aliens came down and abducted someone, or something. That would indeed be very surprising, but it doesn’t make sense. Random surprises are incoherent surprises that don’t *mean* anything. It’s much more valuable to have “surprising yet inevitable” events happen in the game – things which were indeed surprising, yet still coherent.

    Games with fair randomness or even purely deterministic games can be wildly elastic and surprising if they’re designed well.

    I think there are *other* reasons you don’t hear people talking about Go games they played or other “tense strategic games”. I simply don’t think there has really been a well-designed, tense, deterministic strategy game that is also accessible, yet. It’s not *because* they’re not random.

  • Keith Burgun

    Well I don’t consider “the other player’s mind” a source of input randomness. You’re not wrong for doing so, but I think that the term “input randomness” loses basically all of its utility when you do that. I think it’s just more useful to actually have input randomness refer to game-rules only.