Well, if we are looking for games without a curve that equals “I played it, I had fun, then I stopped having fun and moved onto other games” we’ll never come up with any titles. My point with Warhamer is I have yet to hit that spot where I feel like I mastered it and it no longer challenges me. YMMV.
I think this is key. @rhamorim mentioned AI War, you’re throwing out the asymmetry of Xcom. I add in Paradox Dev Studio games.
In those the starts are asymmetrical, there is no one fixed ‘win’ condition, and the challenge is to use the spectrum of mechanics to achieve some self directed goal. The AI feels like it is playing the same game as you, utilizing the same tools. But it isn’t directly trying to achieve the same goals. So it’s not like it needs to achieve some fixed goal, and that it is obvious that going to war with England is not helping France in beating the Spanish in colonizing. Instead it is plausible that there is some strategic objective there.
So the asymmetry and open objectives make it so tha imperfect AI play is not only forgivable, but reasonable. Sure it may not be perfectly logical for Austria to attack France over Savoy, but it’s also not historically implausible. History is replete with insensible decisions. And it works there.
But the symmetrical defined goal oriented playnof a Civ? Yeah, it ain’t gonna work currently.
Sorry, I’m ramming my head into a strange vSphere issue at work and I can’t parse your post/question.
Are you saying competative mulltiplayer is a counter point to “I played it, had fun, and moved on?” If so, that isn’t a counter point - there are tons of competative multiplayer games people eventually get tired of and move on from. I mean, all of them, eventually. Isn’t PUBG pretty much dying these days? Fortnite is no longer the hotness, now it’s all Apex Legends. It’s not really my scene so I’m going by whatever I pick up from comments online and from my son, mostly.
Also, are you asking why AI is bad as a concept, or why it’s bad at playing games? I don’t agree with the first, and for the second I have to say AI is mostly bad due to the complexity of the systems and the time/resources it would take to make a really good one. I’m generally okay with AI that is decent at playing its own game (as opposed to amazing), because I tend to never dig deep enough to exploit it (except by accident).
Sorry. Just saying games that seem to be able to bypass the “learned mechanics/got bored” seem to rely on competitive bits, or randomness (aka roguelike, shuffled decks/tech trees/win conditions). I guess to generalize everything is boring because it gets all-samey. So the question is how to make things not the same. Randomness via mechanics or people.
Don’t kill me, but the Kingdom Rush tower defense games are pretty good at this. They do have tutorial levels, but you have all your tools available several levels in and then you need to use your complete knowledge of the towers/abilities/enemies/systems to win a bunch of the levels. Eventually you beat all of the levels, but I don’t think that means it parabolas at the end; it’s just the type of game you can finish.
No mention of Warhammer 40K: Gladius in this thread? The AI there seems very good. Not sure if it evades the Chick Parabola, but it’s certainly one of the better (turn-based) strategy games to come out in a while.
It’s also worth mentioning that Slay the Spire (and Into the Breach, though I haven’t played that) are both very much asymmetric. In Slay, you’re playing a card game but the computer isn’t. They even made the intentions of the AI visible as part of the mechanics (something Breach did as well). The AI is predictable and plays a different game than you, and if that’s how the game is laid out, it works.
Asymmetry is the only way you can write truly effective AI, since you can design a much simpler game for the AI to play. For symmetric games, my opinion is that you want human players and short games that real, average humans can actually play regularly. My go-to for this is Chaos Reborn, which surprisingly has very good AI as well.
In reading the original post, for some reason the Seven Kingdoms games come to mind. Really great AI, wonderful economy, the best espionage in any game of its kind and so on make it pretty varied for each playthrough, I thought.
Indeed. It shouldn’t be a great AI, it just should feel right and provide a challenge. And game mechanics should help with that. Looks like most people think that making an interesting AI opponent is the same as making a true AI and that’s far from it.
In Civ5-6 they made it mechanics that make AI lose wars when he has three times as many units as you do. In Heroes when you win in such conditions it’s not just because AI does something outright stupid - really it’s rather hard to do anything really stupid in those games - but because you do something clever, like combining spells and unit abilities and attacking with the right units. In Civ5-6 your genius play is sending damaged units home to heal and standing on defendable terrain, and AI can’t even do that.
Thus in Heroes it doesn’t really matter that much if AI is dumb. It’s enough for AI to not do obvious mistakes and it’ll work fine as an interesting opponent.
Yeah, Heroes just doesn’t have that much tactical space for the AI to make mistakes in. Something like 80-90% of a combat’s results are from the math of smashing two stacks of units together. Letting the player feel smart for eking out that last 10-20% is the fun bit.
There’s something to that, I think. AoW3 is similar in that a lot of a combat’s basic outline is determined by the maths of how your stack measures up to theirs, and the AI is good enough at moving their dudes around that it leaves 10-20% of the outcome up to the player’s decisions. So you get to feel smart for outplaying the AI without rolling your eyes at beating a blind kid at dodgeball.
Civ V (at launch particularly, and the community patch also helps a ton) can feel a lot more like beating a deaf kid at Marco Polo than outsmarting a big dumb bully. And I am the resident Civ V defender around here.
e: and Beyond Earth is like smashing your dangly bits in a door while the albino kid looks on, horrified.
It’s been a long time, but I seem to remember that Seven Kingdoms was one of the rare games, especially among RTSs at the time, where the AI actually knew how to play well. I wasn’t a big RTS fan at the time but I do recall Seven Kingdoms standing out and keeping my interest in the single player game for a while.
I’ll submit another gem from 1997 for consideration. Sid Meier’s Gettysburg. I found the AI to be excellent and the game to be endlessly re-playable because of it. I’d come back to it for years after release to play a quick random battle.
My other favorite game from that year, Myth: The Fallen Lords, probably doesn’t qualify in this thread since while the multiplayer was endlessly replayable, the single player was a rather linear affair with the scenarios providing the challenge more than the enemy AI.
Edit: I realize the original post said “modern” but as Seven Kingdoms was mentioned it got me nostalgic for the amazing year in gaming that 1997 was.
I loved 7K but the AI was really bad at spying and reputation and had a habit of attacking you without declaring war, so if you destroyed the building their troops were heading towards, they would just mill around for a while and then go home. They were very good at building huge numbers of troops, I’ll give them that.
I kind of want to say Crusader Kings 2 fits the mold of continuing to be interesting to play once you feel some mastery of the mechanics. Though that game feels like one is more up against the map than the AI.