Against randomizing character selection

#1
@keithburgun has argued that if a game contains multiple characters, the player should not be expected to make a choice between those characters at the start of the match. Instead, a character should be chosen for the player randomly by the game software. I think this idea is very wrong, and in this thread I'll explain my perspective on the issue.

(Note: I am ONLY talking about single player games here. Dealing with multiple players who each pick their own characters complicates things, and anyways Keith and I both agree that "the default number of players should be 1", so I won't be talking at all about multiplayer games.)

Keith spelled his argument out in detail in this article, but the basic argument can be summarized as this:

1) The player chooses their character before there is any information available about the gamestate, so it is a very shallow strategic choice.
2) Most of the time, the player does not make the choice from a purely strategic perspective, but also by thinking "which would be most fun to play?".
3) The choice is thus an instance of "playing the designer", which should be avoided at all costs.
4) Randomizing character selection means the player doesn't have to make this choice, and is therefore a good idea.

I agree with 1) and 2), but 3) is where I think Keith makes a big mistake. A choice can be "playing designer" only if it is made within the match. The choice of which character to play is more analogous to the choice of "which game in your steam library do you want to play" than it is to a strategic choice. Keith is making a mistake in the way that he thinks about character choice: it isn't that the match starts and then the player has to make a choice about which character to play, but instead that the player is presented with the choice between several different games (which are identical in all but the player's character), and the match starts after they've chosen which game to play.

Let me use Slay the Spire as an example, since it's the new book club game. When you run Slay_The_Spire.exe, the game hasn't started yet. Slay_The_Spire.exe is not really a game, it's a piece of software which actually contains three different games, one for each character. Choices made in pieces of software that aren't games can't be instances of "playing the designer", that term only makes sense once the player is actually playing. In my mind, suggesting that Slay_The_Spire.exe should pick a random character for you is like suggesting that Steam should pick a random game for you instead of letting you pick which one you want to play.

This distinction is not a mere technicality either; I think the model I'm proposing is much more psychologically realistic than Keith's. The "playing designer" model implies tension, that there are two forces, strategic value and fun value, that are competing to influence your decision. But there is no tension; I don't think that strategic value generally has any influence when people choose which character they want to play. I don't think that I've ever picked a character because I thought it gave me the best chance of winning (remember, I'm only talking about single-player games here!). The only thing that influences my choice is which character I feel like playing, which is exactly the same thing that influences my choice when I decide whether I want to play Slay the Spire or another game.
 
#2
You make a good point here, and I agree with you. Can you apply this to all pre-match decisions and give designers an excuse to have super complicated game set-ups?

An addition. If the match length is sufficiently short, I think it's good for the game to choose a character (and other game start configurations) for you on your first play, but then let you choose thereafter. All of the resistance I have to letting a player choose their character stems from how annoying that choice is the first time I play, when I do not have enough information to make an informed decision about which character I may like because I don't know what the game's like.
 

vivafringe

Moderator
Staff member
#3
A choice can be "playing designer" only if it is made within the match.
This is just a definitional argument, I think? Whether or not “playing the designer” choices can happen within a match or before the match, the player experiences the game the same way. The truth value of “are we playing the designer?” doesn’t lead to any revelations of the actual experience because once we start quibbling over its definition, it stops mapping to the real world.

As evizaer points out, in Civilization you are given lots of toggles to fiddle with before you start a match. This very much feels like you’re “playing designer”; it’s an annoying amount of cognitive load on the player that the designer arguably should have handled. Some players like having these options but for a lot of the potential player base, being given these choices is burdensome. I think it’s a pretty clear counter example for this argument.

That being said, I think Keith makes a definitional mistake as well because he lumps the egregious amount of menu toggling in Civ with the relatively simple choice of character selection. Civ introduces a huge amount of cognitive load and looks like lazy design, while character selection could imply several well-tuned experiences that the dev spent a lot of time on (e.g. for Slay the Spire).

I think ultimately all of this comes down to cognitive load. Bear with me, I’m about to make a funky analogy. Let’s say you want to ask someone out on a date. A lot of people mistakenly give TOO many choices to the person they’re asking. They’ll ask something vague like, “are you free sometime next week to do something?” This is analogous to the million menu toggles in Civ.

At the same time, it’s unclear that saying “Do you want to meet with me at 7:15pm at the Landmark Cinema to go see Thor 2?” is great either. This is offering them ONE choice. Probably the person would like to be given a bit more wiggle room than that - either in the showtime, the location or the movie.

This is why I tend to like choosing characters in games. It gives the player a fairly simple choice that doesn’t burden them with a lot of cognitive load (“playing designer”), but gives them a little bit of wiggle room over their play experience.
 
#4
I think we can reject "playing designer" as a concept and discuss the matter more concretely in terms of cognitive load, how much variation the different settings actually introduce, and the cost of the game choosing settings the player doesn't like.

For instance, Civ has a super long campaign, so maybe the designer shouldn't want the player stuck playing an islands map if they didn't opt into it somehow. Since Civ is a game with so many moving parts, a player may reasonably be disinclined towards some of those parts and wish to minimize contact with them. They may choose to not play instead of having to deal with those parts at the expense of parts they like. The cost the player has to pay for getting a game optimized for the wrong kind of experience is too high--this indicates it may be better to let the player have some say in where the emphasis of their particular campaign may be.

As strategy games get more complex, you will more often end up in a situation like Civ's, where there are parameters of the procedural generation that can skew the game in non-degenerate directions that some players will like much more or less than others. You can only do so much as a designer to even out how "fun" each of these different skews may be to play through. There will always be the matter of the player's taste/mood. This is the second edge of the sword of complexity: the game appeals to more kinds of players, but it also has the capacity to vary a lot per-match in terms of who it appeals to or repels.
 
#5
I agree that you could apply my same argument to a huge screen with a bunch of parameters, such that each unique set of choices for the parameters is considered it's own game. I don't have too much of a problem with that in principle, but I think that generally there would be better solutions than just "here's a huge screen of parameters, set them to whatever you want!" Instead, the designer should select several sets of parameters that they find most interesting and display those sets as special "gamemodes" that the player can choose, instead of just expecting the player to guess around in the parameter list until they find something interesting. You can still include the "set them to whatever you want" option as well, and that could definitely provide some fun to players who have played the default mode and all the provided gamemodes. But giving the option to set all the parameters without any guidance as to which areas in the possibility space the designer thinks are interesting seems like a very poor choice.

Also, I don't think the term "playing designer" should be discarded entirely, it just has no place in the conversation about character selection (or other pre-game decisions). The stuff you guys are saying about cognitive load certainly seems useful. But "playing designer" is still useful for describing something that happens within matches, when there genuinely is tension between strategic value and fun value. Take Spelunky as an example: say you beat Olmec, and you now have the option to either end the game, or go to hell. If you end the game immediately you get a gold bonus. So, if you were at 1 health and had no bombs or ropes left, you might think "Well, I almost definitely won't survive hell, so the right thing to do is just end the game here to get the extra bonus. But I played this whole match trying to get to hell, am I really just going to end it here? Hell is a pretty fun area." That thought process is exactly the type of situation where "playing designer" is a useful phrase; that thought process shows tension that needs to be addressed.

@vivafringe I wasn't just making a definitional argument. I wasn't saying "Choosing a character isn't playing designer because I have arbitrarily chosen to define 'playing designer' as something that happens during the match", I was saying that playing the designer is something that happens when there is a tension between strategic value and fun value (and I didn't choose this definition, Keith did). The fact that it can't happen during the match follows from that, because the player doesn't care about strategic value yet. In a player certainly could care about strategy before the match starts, they could select the character they think is best, but I don't really think that really happens very often in single-player games (I acknowledge that it definitely happens a lot in competitive multiplayer games).
 
#6
I like all the arguments given here. Let me give it a slightly different spin. One of these pre-game decisions that is very common is the difficulty slider. Sometimes this is even explicitly part of character selection, where on the select screen the game clearly tells you that character/race X is for experienced players and more difficult.

Mark Brown made the best video on difficulty levels. I think it basically solves the difficult-slider problem. I think this also partially resolves the conflict here.


I think there are three kinds of choices:

Customisation
If the option you introduce would benefit from a tooltip that says: "This is not the intended way to play the game to have an optimal experience". Then we make the player do design work. This is okay for difficulty slider, colour blind mode, or variant rules that break the game but might be interesting for some players (like god-mode, or sandbox mode, or developer mode).

Variants
If the tooltip should read: "This choice results in a variation of the game rules that is also optimal albeit different." Then this is the designer giving the player an option to get more value from a similar game. This would be character choice or variant rule-sets like a practise mode, or win conditions for competitive games (CTF vs TDM vs KotH vs Payload).

Strategy (listed for completeness sake, we all know it exists)
If the tooltip reads: "This choice is meant to either reward or punish you by winning or losing." Then that is an in-game strategic choice. This would be talent trees or item selection.

My conclusion: All three of these options can be correct, but they should be clearly labelled as such (if necessary) and the designer needs to think about which choices does what. For example a choice of the second kind should not change the difficulty by too much, and a choice of the first kind should not be asked of the player without them knowing about the game or what it entails.

Failure examples:

* A talent that is completely unbalanced which breaks the game. This is a strategy choice which turns out to be a difficulty slider.
* Civ check-boxes that pretend to be variant rules, but have too much influence on the game difficulty, essentially asking the player to balance the game.
 
Last edited:

keithburgun

Administrator
Staff member
#7
Great thread. I am not convinced, though.

Yes, it is true that when a game offers different characters in the typical way, you could think of each of those as its own game. That makes sense! However, it also kind of makes sense to consider every difficulty level its own game, and every map generation its own game. If you have other options like "turn off Blood Curses" or something, that also is now its own game.

The point is that we can get pretty granular with that "it's its own game" thing, if we want to. Specific example:

Match 1 of Auro. In this match, I got Jump.
Match 2 of Auro. In this match, I did not get Jump.

We could say that those are each their own game. Getting really technical, sure. But it also kinda makes sense to think of those as two iterations, two manifestations of the same game, the way that "different map generation" is probably not best-thought-of as a new game.

So yes, if you want to think of characters as being their own sort of "variants", I don't oppose that. That's fine. I think for some systems that would make sense. But for me I think it's healthier to think of all your characters as being randomized starting conditions. You start with randomized tools to interact with the world.

But anyway here is my much stronger counter-point:

If you are really are selling me 3, 4, 10, 50 "variants" of your game, if that's really what they are, can you just finish doing your job as game designer and tell me which one is best? Prescribe to me the best ruleset? That's literally what I'm paying you for.

It sucks when I go into a game's options screen and it has all these options like "TURN OFF BARBARIANS" and "NO BLOOD CURSES" or whatever. My reaction is, I don't know man, this is what I'm paying YOU to do for me. Is it better with barbarians?

So my answer is that the game is best with ALL of the characters. It's best when you must play all of these characters and think of it more like random Auro skills. The whole idea of having a big old pack of variants to choose from just sucks.
 
#8
If you are really are selling me 3, 4, 10, 50 "variants" of your game, if that's really what they are, can you just finish doing your job as game designer and tell me which one is best? Prescribe to me the best ruleset? That's literally what I'm paying you for.
That's what I said: Have one setting be the default, then make it clear that choosing something else is not recommended. Darkest Dungeon does exactly that. It lets players choose if they want, but it clearly states what the designers think is idea.
 

keithburgun

Administrator
Staff member
#9
The "playing designer" model implies tension, that there are two forces, strategic value and fun value, that are competing to influence your decision. But there is no tension; I don't think that strategic value generally has any influence when people choose which character they want to play.
This is true in single player games where each character really has its own rank and all of that, like they're totally treated as separate games. But not in multiplayer.

Also I still think players have to make bad choices about like "which character am I better with/I know some combos with/I am comfortable with" vs. "which character would actually lead to the best gameplay".

That's what I said: Have one setting be the default, then make it clear that choosing something else is not recommended. Darkest Dungeon does exactly that. It lets players choose if they want, but it clearly states what the designers think is idea.
Yeah, I'm okay with this. As long as "THE REAL GAME" is clearly marked as such, sure, let players do whatever.
 

Juli

New member
#10
I don't think pre-game decisions can really be compared to selecting a different game if different pre-game decisions lead to outcomes that are valorized under the same category.

In Monster Hunter there's a competitive thing people do called Time Attacks, which is sort of like a speedrun, but for a single encounter. You start a hunt, go kill the monster, then record your time and submit it. The submissions are then categorized according to the weapon you are using, and which hunt you completed. So for example, if I decide I'm going to attempt a time attack using the lance against Nergigante, then I am only compared against other people using the lance against Nergigante. If somebody else kills Nergigante in less time, but used a greatsword, I am not placed lower than that person, because I am not compared to them at all. So in this case, choosing which weapon I want to use, and which monster I want to hunt, is comparable to selecting a game from my steam library. I picked lance and decided to fight Nergigante, so I am playing the "lance vs. Nergigante time attack" game.

If I enter a Street Fighter tournament and pick Cammy, I am not just being compared against other Cammy players. If I come in 2nd place, and the 1st place player was using Ryu, then I'm actually still 2nd place. There is no prize for being the best at the Cammy vs. Bison matchup. Thus, 'playing designer' in this case is incongruous to selecting a game to play. If I boot up Battlerite, I am not punished for choosing to play it instead of Far Cry. If I pick Cammy into a bad matchup, I *am* punished for not picking another character.

ok I missed the part where you said you're only talking about singleplayer games here. This still all applies to singleplayer games if you *do* care about strategic outcomes, but these examples aren't about that.

So here's another example. Monster Hunter again. Let's say that I get an investigation for a Black Diablos with 5 reward boxes. Here's the important stuff to know: An investigation is a type of encounter with a monster, and Black Diablos is the type of monster I will be encountering and hunting. 5 reward boxes are a lot, I care about getting them, and I have a limited number of attempts at completing this investigation. If I fail the investigation (by dying too many times, or not completing it in the time limit), I don't get the rewards, and it uses up an attempt. So I want to fail as few times as possible, hopefully 0. I also want to have fun. Do I A) use Lance, a weapon I am experienced at, and has a great match-up against Black Diablos, arguably the best in the game? OR B) use Switch Axe, another weapon I am roughly equally experienced at, that has perhaps the worst match-up in the game against Black Diablos? The answer is probably B, because I think Switch Axe vs. B. Diablos, despite being a terrible match-up, is one of the most fun match-ups in the game. It's wrong to say that I am choosing a different game here, though, because the outcomes are equally valorized regardless of whether I use the very-good lance or the terrible, awful, no-good switch axe.

I don't think this should be avoided at all costs, but I do think it comes at a significant cost that goes largely unaccounted for, or at least underestimated, within discussion of games.


-

Another cost of heavy customization features in games is that it muddies communication about those games. In a game with little customization, there is a shared experience. In a game with a high degree of customization or a great variety of game modes, it's possible there will be no shared experience at all. Off the top of my head, I can remember having this problem Audiosurf. I discussed the game with a lot of people, and almost none of us had a shared experience, because that game's opening menu is a selection of like 15 different modes. The only person I could really talk to about the game was an IRL friend of mine who I played Audiosurf with in meatspace, sharing songs on a USB stick and playing the same modes. Meanwhile in something like Dark Souls, there's a common shared experience among all players. Not identical, but shared enough that if you and another person both played and enjoyed it, you can probably talk to and relate with each other over it.
 
Last edited:
#12
If Super Smash Bros. 4 with 55 characters is 1,485 games in one, why are all characters allowed at tournaments?
Well, this is why I mentioned I was only talking about single-player games. In a competitive multiplayer game (especially a tournament), character selection is definitely a strategic choice, so what I'm saying doesn't apply. You don't select your character arbitrarily based on which "game" you feel like playing, you pick the one you think you are best with.