1 in 4

Recently Balatro’s developer, localthunk, gave two interviews about his game. One with Dan Gheesling and another with the Eggplant Show. In both interviews localthunk mentions negativity bias, which is the tendency people have to focus on negative events more than positive ones, in the context of how people perceive chances in his game:

Dan: Right then, the last two questions. This one from Tosho says: will there be balance changes to Wheel of Fortune? I can never get it to not say nope”.

localthunk: OK, uh, no, there won’t be. I hear this like literally all the time, there are conspiracy theories about this stupid card. Um, I think people need to just look up negativity bias because you’re predisposed to remember negative outcomes of a random event more than the positive ones, so it will feel worse than one in four but I promise you it is actually one in four chance.

And then on the Eggplant Show at 26 minutes:

Interviewer: This is a thing that comes up a lot when I talk to people about randomness in games. They’re always assuming, mostly wrongly, that the designer has like an ace up their sleeve and is constantly messing with them and there’s some super complicated algorithm analyzing the current game state and then deliberately throwing a wrench in it…

And I think in a lot of games that’s not the case, and my strong assumption is that that’s also not the case in Balatro. But maybe you wanna correct me or maybe you don’t wanna disclose this, I just assume that those boss blinds are coming up randomly, I’m not assuming that there’s anything in the back end being like Hah! He has a flush deck, I’m gonna throw the boss that hard counters that!”. You’re not doing that, right?

localthunk: So I briefly worked as a data scientist before I started doing this full-time, and I respect statistics way too much to be doing that to my game. I have, like, an anecdote about that with this game because it really does bother me, especially because you can see the source code too. So, there’s a card in the game called Wheel of Fortune that has a one in four chance of upgrading a joker card. And then there’s a card in the game called Gros Michel which is a banana and it has a one in four chance of going extinct every round.

And it’s a wonderful illustration of negativity bias where the negative result of a random event, you remember it a lot more strongly than the positive results. So, you would not believe the amount of comments that I’ve gotten about, Are you really sure it’s a one in four for the Wheel of Fortune? I’ve done 19 in a row and I haven’t gotten one.”, You need to check the code, there must be a bug.” An absurd amount of comments like that, and I’ve heard nothing about the banana in the same way. The code is identical, it checks to see if this value is less than one in four and if so then it’s true and if not’s false. Like, it’s psychology, it’s all psychology, people have a fundamentally hard time intuiting statistics and probabilities.

And this is all true. People think they want true chance” in their games, but when a game comes out and does it they complain a lot, which is why devs prefer to mess with randomness in their games so it’s more consistent. I have no strong opinions on if devs should or shouldn’t do that, but it’s a topic that surely generates a lot of discussion.

However, this is not why I’m writing this post. Games that are about randomness are ultimately environments of RNG control practice, and the abilities involved in that can roughly be divided in two: risk mitigation and risk taking. localthunk has mentioned in both interviews how his game is about risk mitigation, and that’s fine, but I would argue that as an RNG control practice artifact, it is lacking in modelling the risk taking aspect of randomness correctly.

I will use Artifact to illustrate this point. In Artifact there’s a hero called Ogre Magi that has the following passive ability:

MULTICAST: After you play a blue spell, there is a 25% chance to put a base copy of that card into your hand.

This is an ability that also uses a 1 in 4 chance, and it makes Ogre Magi my favorite hero in Artifact. It’s a very powerful ability because you can copy any spell at all, and blue spells can be extremely good. For instance, consider this one:

Bolt of Damocles: Deal 20 damage to the enemy tower.

Towers in Artifact, what you have to destroy to win the game, have 40 health points. This spell takes ten mana to play, so can only be played in the late game, but it deals half the damage needed to win a lane. If Ogre Magi is in that lane, then when this spell is played, there’s a 25% chance you’ll get a copy of it in your hand. And if you have a way to refresh mana for that lane, which blue has in the form of:

Then you get to play Bolt of Damocles twice, and win that lane, and likely the game, since by the time of ten mana you likely have already taken another tower down. This is a very unfair and lucky way of winning a game, but by trying to do it you’re taking a risk. The risk is that Ogre’s passive doesn’t trigger, and now you used your finisher card, which could have taken your enemy by surprise had you managed to get this tower down to 20 health or below, and so you don’t have that play available anymore.

Now, how is this relevant to Balatro? I believe that Artifact is the best game in the RNG control genre, and it’s the best because it gets the balance between risk mitigation and risk taking right. In real life, in high variance environments, you have to mitigate against unlucky events, but you also have to take risks, because if the risks pay off the returns tend to be outsized. Someone who only mitigates against unlucky events will generally fail, someone who only takes high risks will generally fail. The one who succeeds is the one who gets the balance right and who knows when to defend and when to attack.

RNG control games are practice environments for this aspect of real life, and so the best games in the genre model this correctly and thus allow for the expression of both of these tendencies that people have evolved to be attuned to. People enjoy these games because they intuitively understand this to be true, they intuitively understand that threading this needle is what’s interesting and what they want to engage more with.

Balatro, unfortunately, in my opinion, gets this balance wrong. It is too much about risk mitigation and not enough about risk taking. Most of the times when something good happens to you it wasn’t really because you took a specific risk that that good thing could happen, it’s just something that happens that you had a small part in architecting. The tarot pack just happened to spawn a soul card and that’s it, there’s nothing comparable to the Bolt of Damocles + Ogre Magi combo where the risk you take is completely in your hands and if you get it right you simply win the game. In this sense, Balatro is a too safe game, and also not as much about RNG control since you have way less points of control available.

Does this mean that I think Balatro should fix this problem, or that it should be a different game? Absolutely not. There’s a place in the market for all kinds of games, including games focused more on risk mitigation than anything else. Clearly there are lots of people in the world who enjoy this, and the developer also seems to enjoy it, so it’s really not on me to suggest that things should be different.

But take this post more as a brief steelman of the Wheel of Fortune complainers’ position. I’m a big believer in intuition and a skeptic of reason. Generally, when lots of people are complaining about something, there’s a good intuitive reason behind it, even if they can’t express it correctly themselves. In this case, it is true that the Wheel of Fortune complainers are engaging in negativity bias, but just because that’s true it doesn’t mean that they’re wrong (you could read this as them engaging in negativity bias being objectively true, but pragmatically false, see Luck isn’t real post). People’s brains evolved in the way it did for reasons that are beyond our understanding, it’s a good idea as a game developer to trust that process and its results, even as they manifest to you in seemingly illogical complaints, instead of defaulting to dismissal.

As for my personal experience, I went into Balatro blind (didn’t play any of the demos before release and didn’t watch many streams on it) with the expectation of a good RNG control game, but found out it was something different, and that was slightly disappointing. However, after playing it for about 20-30 hours I figured out why exactly I was disappointed with it (the contents of this post), and that allowed me to then enjoy the game for what it was rather than what I wished it to be. I now have around 80 hours in it, I’m taking a break to play the new Path of Exile league, but I’ll probably keep playing it more afterwards. It is a really good game that has new design tech that I don’t think many game designers have taken notice of yet.

In his interviews, localthunk also mentions this:

localthunk: So I started working on that, and then at the time Northernlion was playing Luck be a Landlord on his YouTube channel, so I was watching his videos and I got inspired, I really liked that there was no fantasy theme, you’re not facing an enemy, it’s like a score attack style game, I really connected with those parts of the game. So I thought, I’m going to transition this project into something in that direction and at that point I kind of cut myself off from roguelikes. I wanted to try and explore that idea myself, make some mistakes and not play any other deckbuilders so I could kind of naively explore that idea space myself. So this ended up being the first deckbuilder I ever played because of that and I’m glad I did it that way.

And then on the Eggplant Show at 7m15s:

localthunk: I made a decision to stop playing all roguelike games at that point. Balatro is the first deckbuilder I’ve ever played, and I really wanted to avoid playing any game that was similar because the whole reason I make games is because I love the process of making games. It just felt like I would rob myself of the opportunity to explore a design space naively if I went into it with all these preconceptions about the best practices in the genre instead of making my own mistakes, and trying to reinvent the wheel and do all these sorts of things. I’m not sure that answered all the questions there, but for the design methodoly of Balatro I think that was an important step on why it turned out the way that it did.

Two years ago I mentioned on my indiedev creativity post that games that win on creativity, like Balatro, generally come from amateurs because the amateurs have nothing to lose and thus will be more likely to take chances of silly and wild ideas. This usually happens by accident, so it was very perceptive and wise of localthunk to intentionally remain a fool so he could explore the deckbuilder genre with fresh eyes, thus giving himself an inherent creativity boost.

The fool can become anyone and anything because he is unburdened by what is, and is focused only on what could be. The fool is infinite potential, and it is from him that all truly new things come from. Congratulations to localthunk for creating something truly new. We should all aim to be fools like him.

2024-04-01 01:04