site banner

Culture War Roundup for the week of March 20, 2023

This weekly roundup thread is intended for all culture war posts. 'Culture war' is vaguely defined, but it basically means controversial issues that fall along set tribal lines. Arguments over culture war issues generate a lot of heat and little light, and few deeply entrenched people ever change their minds. This thread is for voicing opinions and analyzing the state of the discussion while trying to optimize for light over heat.

Optimistically, we think that engaging with people you disagree with is worth your time, and so is being nice! Pessimistically, there are many dynamics that can lead discussions on Culture War topics to become unproductive. There's a human tendency to divide along tribal lines, praising your ingroup and vilifying your outgroup - and if you think you find it easy to criticize your ingroup, then it may be that your outgroup is not who you think it is. Extremists with opposing positions can feed off each other, highlighting each other's worst points to justify their own angry rhetoric, which becomes in turn a new example of bad behavior for the other side to highlight.

We would like to avoid these negative dynamics. Accordingly, we ask that you do not use this thread for waging the Culture War. Examples of waging the Culture War:

  • Shaming.

  • Attempting to 'build consensus' or enforce ideological conformity.

  • Making sweeping generalizations to vilify a group you dislike.

  • Recruiting for a cause.

  • Posting links that could be summarized as 'Boo outgroup!' Basically, if your content is 'Can you believe what Those People did this week?' then you should either refrain from posting, or do some very patient work to contextualize and/or steel-man the relevant viewpoint.

In general, you should argue to understand, not to win. This thread is not territory to be claimed by one group or another; indeed, the aim is to have many different viewpoints represented here. Thus, we also ask that you follow some guidelines:

  • Speak plainly. Avoid sarcasm and mockery. When disagreeing with someone, state your objections explicitly.

  • Be as precise and charitable as you can. Don't paraphrase unflatteringly.

  • Don't imply that someone said something they did not say, even if you think it follows from what they said.

  • Write like everyone is reading and you want them to be included in the discussion.

On an ad hoc basis, the mods will try to compile a list of the best posts/comments from the previous week, posted in Quality Contribution threads and archived at /r/TheThread. You may nominate a comment for this list by clicking on 'report' at the bottom of the post and typing 'Actually a quality contribution' as the report reason.

13
Jump in the discussion.

No email address required.

I don't think it has to be an either or. I think AI can solve a lot of problems that currently exist in human spaces with the result being that humans are more drawn in to those spaces.

Take an ELO matchmaking algorithm as an example. In a 'pure' setting there is a pool of players looking for a match and the algorithm matches the players to their closest ELO available. But what happens if you are having a bad day? Or the players around your ELO happen to just be better than you? If the algorithm is 'pure' it wont care, because technically the ELO will balance itself out, so it wont account for the fact that you just lost 3 matches in a row and are probably tilted to the point where you will stop playing if you lose again. But if the algorithm isn't 'pure' and is instead designed with the goal of keeping players playing as long as possible, it can pick up on the fact you are losing to much and send you to play a lower ELO player so you don't burn out. The problem there being that a lower ELO player has to take a loss.

Now the algorithm has a lot of 'power'. It essentially dictates for 80% of players whether they win or lose. The only way to make the field 'fair' is to segment the playing population until the vast majority of players trend towards a 50% winrate. Having a good day? Face higher ELO players, lose, go back to your own ELO. Got tilted? Playing bad? We happen to have a player that's significantly lower ELO, who still has one loss to go before we have to give him a 'win' game, cheer up.

The problem with a 50% winrate is that it isn't satisfying. The problem with ELO is that you can see it go up and down and it might demoralize you. The problem with hidden ELO is that you start feeling the algorithm working behind the scenes. A 50% winrate feels like a slog. It burns people out and they stop playing.

So what happens if we inject the player population with bots? Bots that just lose. Or if need be, bots that win. We can use the bots to break up the predictability of the algorithm. Just throw in random bot games. Give players an extra win because winning feels good. Don't worry about feeling lonely, the vast majority of players are human. We can even make the bots emulate a bad player. Have it make obvious rookie mistakes so that instead of suspecting it of being a bot, you just feel sorry for it. No one is worse off here. Matchmaker has happier players playing for longer.

As an example for the motte, I am sure the AI can figure out what kind of a post will garner the most replies. Why would it be bad for the motte to have an AI that constantly fuels discussions that keep people glued to their screen? If we are completely honest, what else is this place good for?

AI isn't bad for humans from a hedonistic perspective. If we have some higher goals for humanity than wasting time playing chess and arguing online then, sure, AI is probably bad. But for the internet? So long as you know that there are real people watching, like twitter recently started showing, the interaction is real. It doesn't even have to be typed by human hands. A new age of Robot Wars. Watch an AI expertly rattle off all the arguments of 'your side' against the 'opposition'. And if we are being honest, how different is that from the type of representative politics we already settle for? Be it in parliament or in media or online.

The problem with a 50% winrate is that it isn't satisfying. The problem with ELO is that you can see it go up and down and it might demoralize you. The problem with hidden ELO is that you start feeling the algorithm working behind the scenes. A 50% winrate feels like a slog. It burns people out and they stop playing.

I made be a weird outlier, but I find a 50% winrate perfectly acceptable and satisfying as long as the matches feel fair and the competition is close.

I neither want to feel like I'm effortlessly cruising to victory nor like I'm struggling just to keep pace. Okay, there are times I'd like to go on a power trip and just crush everyone, but that's not the same kind of satisfaction as a hard-fought victory.

I think the 'problem,' then, with most matchmaking algos is that they aren't so good at optimizing for close wins except at the very highest levels, and only in games like Chess where random factors effecting outcomes are minimized.

Basically, if you get a couple 'lucky' wins you get paired with people who will absolutely stomp you and that is demoralizing. If you get stomped badly enough you're paired with people who are probably not so good and you win handily, which increases your morale but isn't as satisfying as eking out a hard-fought win. Win too hard and you get launched up back to the big leagues to be smashed.

Maybe 1/4 of the matches you play, if that, are genuinely close to your actual skill level. Thus, the 'quality' of every match, in terms of its enjoyability, varies immensely even if your win rate is consistent.

This yo-yo effect is what I find frustrating. I'd like to play against people whom I feel challenge me when I'm playing at my general 'best' without exerting myself to try to keep pace.

If AI can optimize for that better I'd say "AWESOME." If that means I end up playing against AIs that are optimized to give me that experience, I'd be rather annoyed.

The problem, insofar as there is one, is that Elo is a metric, not a target. And like all metrics measuring things people want, it immediately gets treated like a target. Which doesn't really make it lose its value as a metric because it's very hard to fake, but it does make people miserable.

Past the first handful of games in a Trueskill style system (where your rating has both an expected value and a variance, and when the system has little knowledge of your skill the variance is huge), the yo-yo effect isn't very real. If the matchmaker is actually matching people close in ranking (a big issue in many games is prioritizing queue times over match quality), it should take extremely improbable streaks to get matched significantly outside your skill range. Realistically, the gap between you at your peak and you on an off day is much bigger than random Elo fluctuations. It's just that when you're treating ranking as a target, not a metric, any random upswings feel like long deserved gains and random downswings feel like the matchmaker is out to get you.

I'd like to play against people whom I feel challenge me when I'm playing at my general 'best' without exerting myself to try to keep pace.

Your level of exertion can't help but be factored into you ranking. If you do ever exert yourself to keep pace, you'll be ranked higher than your general 'best' without exertion. So you're explicitly asking to be matched against people worse than you.

I mean, I'm explicitly asking to be matched against people who will give me a challenge without forcing me to pull out ALL the stops in order to compete, and will fight fair in any event.

It's like a boxer cutting weight to go down a weight class. I still want a real fight, but I don't want to have to focus on optimizing my performance along every possible metric to stand a chance of winning.

The problem, insofar as there is one, is that Elo is a metric, not a target. And like all metrics measuring things people want, it immediately gets treated like a target. Which doesn't really make it lose its value as a metric because it's very hard to fake, but it does make people miserable.

Yeah. But we can surely design algorithms that consider ELO but also consider, I guess, the fact that ELO doesn't capture all the factors that might go into the outcome. In some games, weird random factors can impact who wins, or certain particularly cheesy strategies work really well unless you specifically counter them.

Chess doesn't have the problem, mind.

Maybe design the game to ask "Was the previous match fun for you?" and takes your feedback to figure out what level of competitiveness you actually enjoy.

I dunno.

My point is, if you prefer that level of challenge, but will pull out all the stops if challenged more, you're explicitly saying that you'd prefer to not be matched against yourself, but someone worse. The way to get that level of challenge more consistently then is to not tryhard when challenged more, but take the loss and drop in ranking.

The beauty of going off pure Elo is that it doesn't care why you're performing at your level. Whether you're talented but goofing around, or terrible but trying your best, all it does is match you against someone performing at the level you've recently been performing at.

Yeah. But we can surely design algorithms that consider ELO but also consider, I guess, the fact that ELO doesn't capture all the factors that might go into the outcome. In some games, weird random factors can impact who wins, or certain particularly cheesy strategies work really well unless you specifically counter them.

Elo is that algorithm. Trueskill if you want to add in the factor of uncertainty. As a metric, it doesn't matter if your rank jumps around +-100 due to random factors, they'll even out in the long run and are not a precise enough measure that these jumps matter that much. Any consistent change will still only come with an actual change in player skill. It is only as a target where these random jumps leading to losing 200 from where people feel the 'deserve' to be (which is nearly always their peak, though of course the peak is also a random jump ahead) are a problem.

Cheesy strategies are a separate design issue, a match can be perfectly balanced 50/50 but be a boring blowout either way depending on random rock-paper-scissors.

My point is, if you prefer that level of challenge, but will pull out all the stops if challenged more, you're explicitly saying that you'd prefer to not be matched against yourself, but someone worse.

Lets say, for arguments sake, that I can expend my full, sweaty, maximum try-hard effort for about one (1) match before I'm exhausted and return to baseline performance, then I have to rest up before I can expect to pull out that kind of performance again.

In that case, if I go full, all-out aggressive to try and win at all costs, I may pull out a victory, at which point I'm exhausted and I either stop playing for the night, or I go ahead despite knowing my performance will degrade, and guess what, I'm matched to a person with a higher skill level and I can expect to lose pretty handily unless I dig deep for that serious effort again.

It's not clear to me why a ranking algorithm should assume that I play full-tilt at all times, or that I would want to play at full-tilt at all times.

I think this is where my assumptions about competition sort of diverge from the usual. I'm generally not playing to 'win' and improve my ranking. If there's no money or prestige on the line, what would be the point? So I'm playing to have fun and not end up more stressed and angry than when I started.

Which means I want to match with people who will challenge me at the level I'm most comfortable at, in most cases, and not be 'forced' to go all out. Because my goal is having fun and not 'winning,' I don't have to nor want to switch into full on effort-mode if I start losing. I just want it to be fair.

So I don't WANT a ranking algo that expects me to be trying as hard as I can at all times.

I'd like to be able to say, for instance "computer, match me with someone who is a about equal to my my baseline skill level," or "computer, match me with someone who will push me to my limits" or "computer, I don't really care who I match with today."

Again, so I can optimize for having a fun and enjoyable experience, even if I don't win.

it doesn't matter if your rank jumps around +-100 due to random factors

It matters from the player experience side, is my point. If my ELO can jump up or down +/- 100 due to random factors, then I can end up matched with players who can, respectively, stomp me easily or are a relative cakewalk, in quick succession, and neither of those experiences is particularly enjoyable.

The yo-yo effect is my issue, and based on my experience with most online competitive multiplayer games (looking at you, COD) the algos tend to yo-yo you around mercilessly. Although that is quite possibly due to the algos they employ being designed to keep you addicted and push microtransactions on you, and thus are not optimized for a smooth experience.

At that point, just let me have a server browser so I can CHOOSE who I play with.

I'd like to be able to say, for instance "computer, match me with someone who is a about equal to my my baseline skill level," or "computer, match me with someone who will push me to my limits" or "computer, I don't really care who I match with today."

This can achieved by having multiple accounts and switching based on your level of intensity. It is often frowned upon because it can be easily abused, of course. The computer can't really tell whether you're honest or just asking for the first one but actually planning to go all out and stomp people worse than you.

This can achieved by having multiple accounts and switching based on your level of intensity.

Sure, there's plenty I can do 'manually' to try to fix the issues I'm speaking of. The big one is I just play games with people I already know.

Genuinely, I think I'd be satisfied if all games just included a 'casual' and 'ranked' mode by default, so I can hop into ranked if I ever feel like going all out and seeing how good I can REALLY be.

But I'm still going to point out my issue with the current state of game design.

And if I'm being fully open, my core problem with multiplayer online games these days is rampant cheating, since I can't even feel like the hard-fought matches were fair.