Print

Please login or register.
1 Hour 1 Day 1 Week 1 Month Forever
Login with username, password and session length

[Biag]

Black and White

And the Creatures series (1, 2, and 3/DS)

Expanding on Black & White: you play as a god and you have a giant animal pet/servant/buddy called a "Creature" that has some reasonably advanced AI. It learns by association. So it might try to eat a villager, and you can slap it to teach it not to do that. Or you could pet it and it'll start eating villagers all the time. You could also teach it to not eat anything and it'll get all starved and emaciated, or you could teach it to eat constantly and it'll get fat. This system also applies to things like helping your villagers with the harvest, smashing up enemies, fighting enemy creatures, and so forth.

[dorf]

How much fun is NERO? It sounds cool, but it also sounds like a royal pain in the ass.

Attention! I've last played NERO ~4 years ago. There might have been updates.

NERO is/was a research project which used a cool neuroevolutionary algorithm to demonstrate its practical value.

Is it a game? Well, it can be!
You and your buddy can each teach your units how to behave separately and pit them against each other in a custom scenario/map. You teach your units by setting up obstacles, enemy units etc. As units die (by the hand of the enemy unit OR the player), new, evolved units are spawned.

The really, really basic principle of the algorithm is that the units will try to remember the "good" patterns and forget the "bad" patterns. Death = bad, life = good.

If you want to teach your units to follow the edges of walls, you place some walls and kill every unit that is not doing what it supposed to. If you want units to follow the enemy units, you set up scenarios and kill off the units that are not behaving like they should.

Was it a breakthrough in self-learning AI? Sure.
Is it interesting to watch your units learn stuff? To a degree.
Is it fun to play? Not really.

[a1s]

NERO is/was a research project which used a cool neuroevolutionary algorithm to demonstrate its practical value.

Is it a game? Well, it can be!

It's not a game, it's a toy. You have to make a game out of it (kind of like a bunch of plushies and plastic soldiers aren't a game, but you can (and probably did) make them into a game by expending creative effort). It's in fact impossible to turn relatively pure AI learning into a non-meta game, because there isn't really anything for the player to do- the best way to teach an AI to solve an external (pre designed by author) puzzle is to make it run through that puzzle a million times. And making players design impossible puzzles to make them feel smarter than an AI sounds kinda weird and issue-y.

[Virtz]

NERO is/was a research project which used a cool neuroevolutionary algorithm to demonstrate its practical value.

Is it a game? Well, it can be!

It's not a game, it's a toy. You have to make a game out of it (kind of like a bunch of plushies and plastic soldiers aren't a game, but you can (and probably did) make them into a game by expending creative effort). It's in fact impossible to turn relatively pure AI learning into a non-meta game, because there isn't really anything for the player to do- the best way to teach an AI to solve an external (pre designed by author) puzzle is to make it run through that puzzle a million times. And making players design impossible puzzles to make them feel smarter than an AI sounds kinda weird and issue-y.

NERO is about training an AI so that it beats other players' AIs in different combat scenarios, not about designing impossible puzzles.

And it's way more of a game than a lot of the garbage walking simulators that get called games these days.

[a1s]

first of all,

garbage walking simulators

Wat. (I understand all of those words, but no pair of them (let alone all 3) form any meaningful thing in my mind. )

NERO is about training an AI so that it beats other players' AIs in different combat scenarios, not about designing impossible puzzles.

Please elaborate on this. Especially on how you make it beat other people's AIs (what are some of the "meaningful decisions" that you make?).

edit:(in case you're wondering why I asked that, I'm arguing that NERO is more of a "toy" (like the misnamed "falling sand game") or "gaming tool" (like a plastic solider) than a game. I'm not saying it forces you to make puzzles, impossible or not)

[Virtz]

first of all,

garbage walking simulators

Wat. (I understand all of those words, but no pair of them (let alone all 3) form any meaningful thing in my mind. )

The genre of walking simulators (as in games where all you do is walk around an environment with very limited interaction, if at all), which I consider to be of a value close to garbage due to their simplicity and lacking gameplay.

NERO is about training an AI so that it beats other players' AIs in different combat scenarios, not about designing impossible puzzles.

Please elaborate on this. Especially on how you make it beat other people's AIs (what are some of the "meaningful decisions" that you make?).

edit:(in case you're wondering why I asked that, I'm arguing that NERO is more of a "toy" (like the misnamed "falling sand game") or "gaming tool" (like a plastic solider) than a game. I'm not saying it forces you to make puzzles, impossible or not)

You stick bots with your trained AI in some scenario up against someone else's bots with their trained AI and see which side wins based on the chosen ruleset for the match (such as team deathmatch). Your meaningful decisions boil down to how you train your AI prior - creating and modifying different scenarios for the AI to learn from, deciding what kind of behaviour to promote, etc..

[Spacefaye]

Well, as I said, I wasn't looking for a fully-fledged game, just something that displays this feature.

[Cormack]

PTW

[Shadowlord]

first of all,

garbage walking simulators

Wat. (I understand all of those words, but no pair of them (let alone all 3) form any meaningful thing in my mind. )

The genre of walking simulators (as in games where all you do is walk around an environment with very limited interaction, if at all), which I consider to be of a value close to garbage due to their simplicity and lacking gameplay.

Who needs gameplay if it gives you feels? Of course, if it doesn't, then sure, you can say it's a bad game. The Steins;Gate VN doesn't even have walking - the only forms of interaction are the 'go on' button, and choosing what reply to send to messages on your phone (which determines where the plot goes, but you need a chart/walkthrough to see what does what, because you're just clicking a word and then the game brings up the message you'll send back, and there's no way to back out and pick another as far as I could tell). And despite that, it's the best game I've ever played (and I've played a lot of games).

[Damiac]

Nero sounds like what you're looking for... but honestly it's only just barely a game.  It comes with some pre-packaged AIs, so you can train yours up and try to fight them on different maps or whatever.

As I recall, it's mostly frustration and waiting.  However, if you're more patient maybe you'd enjoy it.

Black and white was fun, but the AI, like anything else Peter Molyneux was involved in, was wildly overblown. 

[Urist McScoopbeard]

I believe Twilight Struggle does this.

Also, Twilight Struggle is a Steam game now... Neat.