News for hackers. Stuff that matters.
Re:EA is permanently banning Linux players on Battlefield V by naniwaduni on Friday January 03, 2020 @07:29PM (#21951833)
> Second, I think these systems ignore player incentives. I think that global competitive ranking encourages the worst of playerbases, that skill is an arbitrary mechanism to optimize on instead of something like how close each match was, or player-reported satisfaction levels, or variety of play-styles, or any other of a dozen other metrics.

I think you're totally missing the point. Computing a global competitive rating isn't intended to optimize for skill. Rating is a proxy for skill, sure, but generally it's being used in matchmaking because (a) rating gap is a not-completely-terrible proxy for closeness of matches (because skill gap is a decent proxy for closeness of matches), which contributes to player satisfaction, and (b) you can derive it from easy-to-measure data and while there are fiddly details, everyone is roughly on the same page as to what you're trying to approximate.

It happens that, once you compute that, it's really tempting to sort and report. Perhaps that's a mistake and invites the worst of people's instincts to make numbers go more and bigger.

This still has the constraint that all the players are playing the same game, but as long as you can somehow avoid ascribing value to rank placement then many forms of cheating just stop mattering; the cheaters self-sort themselves into a tier where they don't interact with anyone who can't deal with them.

Re:EA is permanently banning Linux players on Battlefield V by danShumway on Friday January 03, 2020 @08:32PM (#21952133)
I think my argument there would be that despite common assumptions, rating is actually a very poor proxy for skill, and an even worse proxy for a good match-up. Similarly skilled players won't always like to play with each other. Nor do I think that win-percentages are the only metric we have to work with -- most multiplayer games have reams of much better data to sort through, particularly in games with high variability of play-styles, player types, and team compositions and setups (games like League/Hearthstone/Overwatch, etc...)

That being said, I would be more amiable to a system that used player skill as a matchmaking tool if it didn't report those numbers to players.

However, I'm not sure I would say that sorting and reporting is usually a side effect or minor mistake on top of those systems, given that many of these games have seasons, rewards, an entire meta-game built on top of their rankings. The rankings aren't just serving the purpose of getting players matched together -- they're serving as player motivators. The responses that players have can't be seen as a purely accidental side-effect when designers are going out of their way to present this information to the player and to encourage the player to pay attention to it.

Hearthstone is a good example here. Hearthstone has brilliant design, but its ladder system is horrible and dumb. It explicitly encourages net-decking and grinding instead of experimentation. Fast decks are objectively better than slow ones with the same win percentage, because ranking is largely determined by win percentage combined with the number of games you can fit into a season.

Season rewards are then tied to rank, so the best way to get new cards if you want to experiment is to build a boring deck and grind. This is despite the fact that for most players, knowing within the first few turns whether or not they're going to win is not fun.

So of course people bot Hearthstone, because before you can build creative fun decks and play against interesting people with interesting strategies, you have to pay the "fun tax" of grinding on the ladder. Everything brilliant about Hearthstone is bogged down by being tied to this kind of toxic experience.

There's no way to say this about such a popular game without sounding naive, and I'll take the criticism, but I do not believe for one second it would be difficult to build a better public ranking system for Hearthstone, especially given the ludicrous amount of player data they have. Even something as simple as balancing rank with deck variety so players got fewer rewards the more times they played the same cards would go a huge way towards getting rid of net-decking.

These systems do increase engagement among very invested players, so maybe that's the only intention. But to heck with engagement -- is laddering at the higher ranks of Hearthstone fun? People do it, people say they enjoy it. Again, I'll take the criticism a statement like this warrants, but I don't believe them. I don't believe the experience of grinding a single 55% win-rate deck for hours and hours is an experience that is worth having or producing.

And when I think about why players do that, the only cause that comes to my mind is, "the game tells them to."

Re:EA is permanently banning Linux players on Battlefield V by Mirioron on Saturday January 04, 2020 @02:37AM (#21953522)
>That being said, I would be more amiable to a system that used player skill as a matchmaking tool if it didn't report those numbers to players.

Almost all of these games have an "unranked" mode in matchmaking that doesn't publicly report your rank.

>Hearthstone is a good example here. Hearthstone has brilliant design, but its ladder system is horrible and dumb.

Hearthstone is probably the worst collectible card game I've ever played. The actual gameplay of it is poor and heavily RNG reliant. It's not bad if you put in an inordinate amount of hours into it, because then you'll get a feel for what the opponent likely has and how likely you are to succeed, but this is generally not something you can just easily reason about without an absolute metric ton of experience. That's why net decking is so common - it's too difficult for most players to figure out near-optimal combinations, because they don't have as full of an overview of the game as the experienced players. The search space is just too large.

Re:EA is permanently banning Linux players on Battlefield V by danShumway on Saturday January 04, 2020 @11:23AM (#21955712)
> Hearthstone is probably the worst collectible card game I've ever played.

:shrug:

I'm not going to argue about what games are good. Substitute out another online card game you like, I promise 90% of the time it will have the same problem. My experience with MTG Arena has taught me that net decking there is just as common, and I consider MTG to be a much more skill-based game than Hearthstone.

Like Hearthstone, MTG Arena has "unranked" modes, but like Hearthstone, the majority of rewards and design is meant to force you into ranked modes.

> it's too difficult for most players to figure out near-optimal combinations

Take a step back. Why do most players feel that they need to find near-optimal combinations? We have a ton of research on card-game archetypes going back over a decade. It is generally understood by every designer in the field that competitive optimizers are one player archetype of many, and often not even the largest one. But the design of our ranking systems explicitly encourages players to become competitive optimizers to the exclusion of every other archetype.

Of course, everyone wants to build good decks, but the experience of playing paper MTG with people even in semi-competitive settings is often very different than the experience of playing MTG Arena -- the "need" to optimize is on a whole new level. It's the same game, but the incentive structures of the metagame that surrounds it changes how the players act.

That's not to say that competitive players don't exist, but it is to say that in every area of card-game design, we recognize that the design space needs to be wider than that. And in our ranking systems, we don't. It's also not to say that stuff like tournaments are bad. It is to say that every online card game I have ever played suffers from these problems, and every one of them uses the same exact ranking system, and maybe the problems are related to that common element.

Re:EA is permanently banning Linux players on Battlefield V by Mirioron on Saturday January 04, 2020 @06:44PM (#21958488)
>Like Hearthstone, MTG Arena has "unranked" modes, but like Hearthstone, the majority of rewards and design is meant to force you into ranked modes.

Is it that the game designers sat down and deliberately designed a system to get you to play more ranked or did the players demand rewards for ranked (or the designers anticipated such a reward)? I've seen the latter happen in an online game. The players in World of Warships demanded more rewards after seeing the initial competitive game mode (team battles). Eventually the developers gave it to them.

Re:EA is permanently banning Linux players on Battlefield V by danShumway on Saturday January 04, 2020 @08:52PM (#21959057)
That's very possible -- a lot of game features come out of that process.

But, I would say the responsibility of a designer is to be able to parse player feedback and disregard bad suggestions. If your players knew how to design a game, they would be game designers.

One of the big reasons design is so hard is because you're trying to understand another person's motivations better than they themselves do. Feedback shouldn't be ignored, but neither should it be implemented verbatim. Players just aren't good at sussing out their own motivations -- they'll tell you, "we want more rewards for ranked" instead of "we don't want to feel obligated to dip into unrelated game modes", or "we want to feel more like we're progressing towards something."

Players also aren't good at giving feedback about what feels wrong, their feedback instead takes the form of half-thought ideas of how to fix what feels wrong. Responding to that feedback means digging under it to find out the actual problem and addressing that specifically.

Note that this isn't just me talking. Rosewater himself has talked about this phenomenon a lot in his design blog on MTG, and he has a good quote on this[0]:

> The key is whether or not what the players want will in the end make them happy. Players ask for things all the time that if we actually gave them would make the game less enjoyable for them. We do spend a lot of time trying to understand what they players want, though, because when we are able to give players things they want, we do.

The point being, I have no doubt that if MTG Arena announced tomorrow that it was getting rid of ranked rewards and seasons, the playerbase would be livid. I still suspect an alternate system for rewards would be better in the long run.

[0]: https://magic.wizards.com/en/articles/archive/making-magic/s...

Re:EA is permanently banning Linux players on Battlefield V by celticmusic on Saturday January 04, 2020 @02:29AM (#21953491)
> I think my argument there would be that despite common assumptions, rating is actually a very poor proxy for skill

you can't know that without knowing the specifics of how that rating is generated.

I promise you ELO for chess is a hell of a lot more accurate than ELO for league.

Re:EA is permanently banning Linux players on Battlefield V by danShumway on Saturday January 04, 2020 @11:27AM (#21955734)
I do know how the specifics of how these rankings are calculated because they're publicly visible[0]. I'm not arguing that behind the scenes Hearthstone's matchmaking isn't sophisticated enough (although it probably isn't), I'm arguing that publicly, the public mechanisms for determining rank are really bad and encourage bad play habits.

Hearthstone ranking is not a good proxy for skill, it's a good proxy for the amount of time you have to play and the speed of your deck. Objectively, the best strategy to get good ranks at Hearthstone is to build a fast, >50% win-rate deck and then to grind. That's not a good proxy for skill.

[0]: https://hearthstone.gamepedia.com/Ranked

Re:EA is permanently banning Linux players on Battlefield V by celticmusic on Sunday January 05, 2020 @04:52AM (#21960545)
I'd love to see the details on League of Legends ranking algorithm.
Re:EA is permanently banning Linux players on Battlefield V by danShumway on Sunday January 05, 2020 @11:54AM (#21962156)
Sure!

https://dotesports.com/league-of-legends/news/league-of-lege...

> As players win games in ranked, they gain League Points, also more commonly known as LP. They also lose LP when they are defeated. Upon reaching 100 total LP in a given division, players enter a promotion series to try and get to the next available division.

> Players must win a majority of their games in order to advance to the next division. Promotion series within certain ranks, like Bronze III to II, are a best-of-three series. From one rank to the next, such as Silver I to Gold V, promotion series are best-of-five instead.

> If a player doesn’t win that amount of games, then they’ll need to get back to 100 LP to try again.

----

Yes, the exact details of how your MMR is calculated is kept secret, but it doesn't matter if you still know exactly what you need to optimize for: winning sequential games, often in chunks, and going out of your way to avoid losing streaks which can result in demotions.

Arguing that hiding the exact MMR formula means that players won't alter their behaviors to fit its approximate contours is like arguing that because Google's page-rank algorithm isn't public that SEO is not a thing.

(Some of) the consequences of this setup include:

A) losing matches becomes much more costly, which exasperates the problems League has with feeders and trolls, and lends itself to a more toxic environment.

B) Experimentation and learning new heroes is best done off of the ladder until you feel like you're competitive with that hero. Experimenting on the ladder is not only bad for your rank, it's also likely to get you reported by other teammates as a feeder (see the toxicity mentioned above).

Players are effectively optimizers. If you give them a system with a clear success metric (ie, a public rank going up), they will optimize their play-style to fit that system. Designers often assume players will stop doing this at some point. They won't grind in an RPG because that's boring. They won't bot a competitive game because that's not fun for other players.

Most players don't think in those terms during gameplay (myself included). Thinking that way is hard. As a designer, if you want a player to optimize for creating a fun experience for themselves and others, you have to explicitly build your publicly-facing systems to encourage that behavior.

Re:EA is permanently banning Linux players on Battlefield V by celticmusic on Monday January 06, 2020 @03:10PM (#21972725)
That's an overview, not specifics and not details (verbiage that first you and then I used). It's a shitty sleight of hand, so I'm going bring this back around and then I'm done with you.

You stated that rank is a poor proxy for skill. https://news.ycombinator.com/item?id=21952133

I stated you can't know that without details. Knowing that winning gives you more LP and losing takes it away is not detail nor specifics.

You can't know, but you're going to hold onto that branch rather than be a little introspective and consider that you may be wrong.