No Man's Sky

[Khaldun]

AI that has a range of personalities and playstyles and is thus actually kind of hard to interpret--that's the Holy Grail. The problem with even a "good AI" is that once you dope out what it's been programmed to do, it becomes easy to beat, in relative terms, except in games like chess. Think Civ V: you want an AI that occasionally does impulsive, weird, or surprising things, not just one that methodically uses the game's ruleset to maximum advantage.

[Sky]

Right. You have to model it to do a ton of things. I wouldn't worry about people who play at the hardcore end, they'll always abuse AI, just the nature of the sperg.

I would think there are such huge volumes of data being generated by multiplayer games across multiple genres, just farm that and use it to feed the AI. It would most likely be slanted to people already drawn to multiplayer, of course. Maybe pull data from single player games?

[Yegolev]

Skynet isn't that smart yet.

[Malakili]

Plus you don't actually want, say, your Skyrim dungeon baddies acting as smart as a PvPer in online RPG because those games have you, or you and a small party, basically able to kill any threat. This is not realistic. A "smart" dungeon full of guys would just get 35 people together and all attack at once, in which case the enemies need to be trivially weak, or you need to have an equally powerful group on your side.

Oh man, did I just imply people in online games actually use teammwork. Maybe I have this all wrong. Maybe realistic AI will just result in everyone running around like an idiot and rage quitting.

[HaemishM]

At least half of my games in League of Legends makes me wish my teammates were bad AI as opposed to presumably (but not provably) thinking human beings.

[Khaldun]

Well, think of your average Skyrim dungeon. Say a crypt with those ancient Norse zombies and so on in it.

There's in-world reasons why they don't methodically cooperate to cut you down. They're tied to their burial site, so they arise as you violate the immediate area you're in.

There was an old D&D module series that's now called "Against the Giants". One of the things I really loved about it when I was young and being the game master for my group of friends was that the first one advised the GM, "If players leave the hill giant fortress to heal up, resupply and be in better shape for the next attack, be sure to move some of the guards from their current posts and establish new well-defended positions, put in some new traps in the entry hall, and so on: the giants should be ready for the players to renew their attack."  If you had agent-based AI and you specified that in some cases, the AI would actively adapt to player activity, then you'd get experiences that feel "right" for the kind of enemy you're facing. Intelligent enemies that work together should intelligently work together; brute monsters or supernatural foes might not actually adapt or be intelligent in that way.

[Sky]

My players burnt the hill giant fort down. Accidentally.

And again, good AI doesn't mean all npcs are as smart as an online pvper. More like Khaldun says. They'll have morale checks, secret agendas, different levels of training, etc.

Then again, I want npcs that react more realistically. "Oh my god no, I have orclings at home, why"

[Khaldun]

That too! I want NPCs who run like mad if they know they're in a hopeless situation. The Gunners in Fallout 4, for example, should be smart enough to know that you're overrunning their defenses. It would be interesting even if they'd send a delegation to you suing for a truce or something, whereas Raiders never would because they're mad dogs. That's really what I'd love to see--agent-based AI that had different motivation hierarchies for different NPCs where they'd change behavior dynamically over the course of a game.

[Venkman]

But why spend all that time creating human facsimile game opponents when instead we could just keep amping up connectivity with human opponents?

[Khaldun]

I think partly because in something like a role-playing game, human players actually don't fit the immersive fiction well at all. In fact, in an odd way, human beings tend to converge into a single mass more than AI characters do--they focus on the competitive and/or griefing aspects of the game they're playing. If you want to be immersed in another world, you don't actually want other people in it unless they completely share that objective.

[Kail]

But why spend all that time creating human facsimile game opponents when instead we could just keep amping up connectivity with human opponents?

Because you could then tailor make player "opponents" to act however you want.  Enjoy League of Legends but don't want to be called a gayfag for thirty minutes because you missed a skillshot once?  Spawn a bot that will play the game without the grade school dick waving.

Because multiplayer only games with lower player bases would then still be playable, even in ten, twenty years when nobody remembers them but a tiny niche of old people.

Because nobody wants to play Brahmin Farmer number 103 but you might still want some Brahmin Farmers in your Fallout game which can do more than stand around and say "what do you want" when you click on them.

Because some people still exist who do not have reliable high speed internet and some of them might buy your game if they had some way to actually play it.

Because you might want to establish some kind of setting or atmosphere for your game, and no human player is going to go around spouting exposition about the Crimson Veil or whatever your backstory is when they could be spamming "LF2M 4 DM run 1 healer and 1 DPS pls" in the trade channel instead.

Because some people feel nervous or anxious around other people (even anonymous internet people) and jumping in to a very competitive multiplayer only game with garbage tutorials is very intimidating if you stand a good chance of humiliating yourself in front of another actual person, rather than a bot who doesn't care if you know what you're doing or not.

Because you might want some level of agency for some enemies that's higher than you get with a modern AI script, but also not at "can become the most powerful thing in the world" level that every human player exists at.

Do you want some more?

[Rendakor]

What Kail said. I want better single player games; I don't want all my single player games to turn into multiplayer games. The biggest reason for me is that, with a family, I need a pause button if I'm playing when anyone else is around.

[Venkman]

Ok this makes sense.

But some of what you're saying doesn't justify the investment when so much has already been made from broadband allowing the players being the content. I mean, LoL is not a jillion dollar business because players want to play a fun game. It's that because it's super competitive and for a lot less cost than developing scores of believably performing AI.

For RPGs though, I'm right with you. Games we're you're the hero get screwed up by everyone else proving you're not.

Games designed to be competitive by nature seem to me better off as multiplayer. It's the more economical approach to producing matches with ever changing conditions. More universally buff tutorials would be nice. I remember wishing strongly for that in Planetside 1, and when I had to make critical class decisions in EQ2.

In my opinion, Overwatch would be perfect if they'd add a sub-class of match type called "no chat, no voice, just lemme join people looking for a good time". Lots of matches I've been in have zero communication anyway. The practice mode against AI is believable for newbies and gives you that positive feedback of a high likelihood of winning. It's the kind of game you don't need to give a shit about but enough there for those who do (and which will pay all the money).

But would a significant base of players playing singleplayer buy accroutrements for their Overwatch characters enough to pay the investment in AI? I doubt that doesn't fit their strategy (eSport!) or maybe it would if they thought it feasible.

[Rendakor]

I would love that "no chat, no voice" mode in basically every online game ever. Unless I'm queuing with my friends I rarely want to interact with the mooks on either team.

[Kail]

Ok this makes sense.

But some of what you're saying doesn't justify the investment when so much has already been made from broadband allowing the players being the content. I mean, LoL is not a jillion dollar business because players want to play a fun game. It's that because it's super competitive and for a lot less cost than developing scores of believably performing AI.

Oh, sure.  I'm not trying to argue that it's always the best idea financially.  But as a consumer, it's very worrying to me that a company would depend on something as unreliable as farming random people on the internet for the BASIC FUNCTIONALITY of the game.  There are a lot of indie games on Steam working on the exact same thing you're saying: multiplayer only games are way easier, we don't need to write AI since we can rely on other players, we don't need to add single player stuff like stories or progression, we only need a few maps and people will play that over and over forever.  Then they launch and the instant the player count dips, wait times rise.  And as wait times rise, people stop playing, and wait times get higher and soon the game just dies.  Nobody can play it, no matter how much they want to, because there aren't enough people online to start a match.  People who bought the game are out of luck.  If the game does something interesting or impressive, nobody will ever be able to see it.  People who made the game and want to use it for their demo reel have to hack test footage out of the client or something.

It's not always an issue for the publisher, because the company just throws the game on the market and dumps it the instant it gets unprofitable, no regrets.  But as someone who wants to PLAY the game, it's a significant long term problem.  People in fifty years will still probably be able to watch every episode of Knight Rider, but they won't be able to play Team Fortress 2 to any degree more complex than wandering around an empty 2fort map.

[Sky]

But why spend all that time creating human facsimile game opponents when instead we could just keep amping up connectivity with human opponents?

I deal with the public all day at work. I don't want to deal with them at home while I'm relaxing. I don't play multiplayer games, I'm over them (as someone who played BF1942 competitively).

Do you buy the cheapest food possible at walmart, too? Why spend the time going to the butcher and getting a hand-cut slice of cow when you can get a frozen burger for the microwave, right? Could we maybe aspire to more than 'hey, this is cheap and easy'?

And hey, nobody is saying stop making competitive (or even cooperative) multiplayer games. Just saying maybe there could be, you know, better single player games. And we're talking about No Man's Sky, specifically, which is not a multiplayer game.

[01101010]

But why spend all that time creating human facsimile game opponents when instead we could just keep amping up connectivity with human opponents?

I deal with the public all day at work. I don't want to deal with them at home while I'm relaxing. I don't play multiplayer games, I'm over them (as someone who played BF1942 competitively).

Do you buy the cheapest food possible at walmart, too? Why spend the time going to the butcher and getting a hand-cut slice of cow when you can get a frozen burger for the microwave, right? Could we maybe aspire to more than 'hey, this is cheap and easy'?

And hey, nobody is saying stop making competitive (or even cooperative) multiplayer games. Just saying maybe there could be, you know, better single player games. And we're talking about No Man's Sky, specifically, which is not a multiplayer game.

Sadly, no one wants to dump money into developing fabulous single player games anymore. Not with the market chasing the COD/Battlefield dragon. This is the age of the bubble... get in on it before it bursts or get left in the dust.

Maybe a company will rekindle the single player genre, but it won't be for awhile... at least not a good offering.

[Sir T]

Warframe has been successful with a "Teams versus AI" model.

[HaemishM]

Multiplayer games that can continue to sell new cosmetic content a la skins is going to last longer than single-player game with an ending, even with a significant DLC pipeline (though games like Crusader Kings II might disagree with that).

[Sky]

Sadly, no one wants to dump money into developing fabulous single player games anymore. Not with the market chasing the COD/Battlefield dragon. This is the age of the bubble... get in on it before it bursts or get left in the dust.

Maybe a company will rekindle the single player genre, but it won't be for awhile... at least not a good offering.

Meanwhile Witcher 3 is the best CRPG ever released.

[01101010]

Sadly, no one wants to dump money into developing fabulous single player games anymore. Not with the market chasing the COD/Battlefield dragon. This is the age of the bubble... get in on it before it bursts or get left in the dust.

Maybe a company will rekindle the single player genre, but it won't be for awhile... at least not a good offering.

Meanwhile Witcher 3 is the best CRPG ever released.

Fair enough, but does it hit on all the notes? I have no reference as I, 1) hate coming into a series in the middle, 2) never played it or heard anyone outside of here talk about it. I'll gladly take your word for it though.

[HaemishM]

/squint

Not sure if serious.

[Khaldun]

I barely played the first two and gave up on both because of not liking the combat. You don't really to have played the previous to get the story and characters. Here and there it improves the game some to know the background.

[Sky]

Anyway.

Still oddly tempted to want to play this stupid game, despite knowing it's probably shallow and repetitive. 

[Khaldun]

I will play it. I am an optimist on this stuff. There's always something to learn even when it doesn't work right.

I think the only procedural-content game I skipped because it was obvious bullshit by a team that just didn't bother to actually develop the game was the SimCity debacle that was supposed to be agent-based AI, procedural content.

[Malakili]

This is the kind of thing I've been known to say I won't buy and then impulse buy it on release because I'm bored that day. I don't really believe this thing has legs, but if I get to wistfully explore the galaxy for a week before it gets old it might still be worth it. Eh, my birthday is between now and release anyway, so why not treat myself. See, I'm talking myself into it even as I'm typing this.

[KallDrexx]

Fight trailer up

[apocrypha]

Hmm, looking at the UI's on show there it seems like there's very, very little information on them. Weapon with heat/ammo, distance to (something), shield... and that's about it. On the FPS UI there's... ammo count. Nothing else. So I'm inferring from that that the combat and space flight are extremely simplified and arcadey.

Whether that's a good thing or a bad thing remains to be seen, it's just an observation.

[Reg]

If the fighting requires actual hand-eye coordination then I'm out. I just can't do that stuff at all anymore.

[jth]

They may have some legal troubles over their procedural generation, which it turns out isn't exactly their own work:

http://www.eurogamer.net/articles/2016-07-21-why-no-mans-sky-fans-are-worried-about-a-patented-superformula

No Man's Sky uses fancy maths to generate its infinite, procedurally-generated galaxy. But is the maths Hello Games used to build the game its own creation?

In 2003, a Belgian plant geneticist named Johan Gielis published an equation that can describe a large number of natural forms, such as starfish, spiderwebs, snowflakes and even crystals. He called this equation the superformula. (...) Telegraaf quotes Jeroen Sparrow, of Genicap, as saying the company did not consent to the use of the formula for No Man's Sky. This formula needs to be licensed, Sparrow said, because it's protected by a patent.

[Sky]

My stepbrother is a patent lawyer. And a giant scumbag douche.

[HaemishM]

My stepbrother is a patent lawyer. And a giant scumbag douche.

I think you're being redundant.

[Yegolev]

I happen to know a patent lawyer who is a pretty cool dude, but I also know a rich artist and a hillbilly that loves Nikki Minaj, so ....  nevermind.

[Sky]

Just wanted to be clear  

He takes douchebag to such a legendary status, I feel bad for my father. We don't ask about him, but secretly relish in the stories of his moronic exploits (last one is his dodging student debt by going to do contract work in Saudi Arabia for three years, which he was fired from almost immediately and hid out in Belize to avoid arrest on tax fraud, I think he's finally back in the states now). Best part is, we hear the optimistic filtered version from his mom, so you just know it's way worse than the version we get.

As far as I can tell, he tries to patent other people's unpatented research so he can profit from licensing.

[Yegolev]

Belize is a pretty good place to hide, so he's not a complete moron.