Print

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

[Ultimuh]

Gilgamesh, the original Chuck Norris.

[Flying Dice]

Yeah, for all that I loved Civ III, the AI doomstacks can stay fucked off.

Still probably not going to buy VI though, because my money is on their AI still being dumb as a bag of bricks and the difficulty modes still just being extra multipliers to brute-force it into outproducing the player. Honestly baffled that nobody has managed to copy what Stardock managed all the way back in GalCivII.

They managed it there for three reasons:
1. The game was relatively simple. Computers are good at simple games.
2. They use many threads effectively, so the AI can make expensive calculations in the background.
3. The guy who does AI is the CEO, and at the time the company was pretty small, so he could do what he wanted, even if that was continue working on AI constantly for a game after it was released, and incorporate player strategies and responses.

Civilization is trying to be less simple, although it's not super super deep. They don't necessarily have a top-tier multi-threaded AI coder, because that's such a niche skill that the people who are exemplary at it are quite rare. Even if they do, they probably don't have one working on the game constantly and also playing and following community strategies.

3's a fair cop. The second is only fair because apparently they can't even manage to make the core game elements run efficiently/insist on catering to people with potato-tier CPUs--see the absurdly low map size limit in Civ V, both immediately available and the hardcoded one that you run into if you try to edit the map types to be larger. The first I would heavily debate. Civ is not complex, either relative to GalCiv or in the general melange of 4X-TBS-Grand Strategy games. I would in fact argue that the series has become increasingly simpler from a peak somewhere around Civ III or Alpha Centauri, dumbing down core gameplay in favor of pretty graphics and little bullet-pointable features for what once would have been back-of-the-box blurbs.

Look at what diplomacy and espionage are in V. It's fucking pathetic. Especially in the complete version there's no nuance, it's just everybody building all the goddamn buildings for 200+ turns, a hundred or two turns of war determined at the start by who has the most land and best production, then the long bloody war between the survivors. Unless you're playing a non-domination victory, most of which manage to be even simpler than "stomp everyone's shit until you own the world", usually in the form of "turtle and spam [resource] until game over". Even the influence victory in GalCiv II was more nuanced, because you had to stick yourself out into the galaxy to be effective at it and people wouldn't just blindly accept you rushing for it. They'd DoW and blow your goddamn starbases off of their borders.

It's a feedback loop at this point. Civ has shit AI so they need to make the gameplay really simple, otherwise they'll flail and be more brainless than Civ III workers on auto-construct. Civ gameplay is really simple so they don't need to make good AI, they can just give them resource cheats to bulldoze players.

[Cruxador]

Civ is not complex, either relative to GalCiv or in the general melange of 4X-TBS-Grand Strategy games. I would in fact argue that the series has become increasingly simpler from a peak somewhere around Civ III or Alpha Centauri, dumbing down core gameplay in favor of pretty graphics and little bullet-pointable features for what once would have been back-of-the-box blurbs.

Perhaps I should have said "mathematically straightforward" rather than "simple". But you're looking at the wrong thing. Look at how many numbers and variables each GalCiv planet outputs compared to a Civilization city. Notice that space stuff is node-based rather than (aside from movement) being functionally tile-based in a resource management sense. And city placement doesn't really exist, just prioritization of colonization targets, which is pretty much a two-factor decision in GalCiv: Accessibility and how good the planet is. In Civ, you've got accessibility, strategic value of the placement, nearby tile values, and resources. And all of those tile bonuses and resources play into quite complicated systems for efficiency within a city. Meanwhile, GalCiv is more complicated only in the basic combat paradigm, since you have three attack stats and three defense stats, but they're all interchangeable and work the same, so it's a tiny change.

Look at what diplomacy and espionage are in V. It's fucking pathetic. Especially in the complete version there's no nuance, it's just everybody building all the goddamn buildings for 200+ turns, a hundred or two turns of war determined at the start by who has the most land and best production, then the long bloody war between the survivors. Unless you're playing a non-domination victory, most of which manage to be even simpler than "stomp everyone's shit until you own the world", usually in the form of "turtle and spam [resource] until game over".

Depth of gameplay and mechanical complexity aren't the same or really related.

Even the influence victory in GalCiv II was more nuanced, because you had to stick yourself out into the galaxy to be effective at it and people wouldn't just blindly accept you rushing for it. They'd DoW and blow your goddamn starbases off of their borders.

Again, it's not more complex regardless of being deeper gameplay. It's just taking nodes, and the AI can use very simple value calculations to decide the likelihood of winning the game via ascension, balanced against the same availability/defensibility calculations it uses for planets and bases and everything else. With all that out of the way, it becomes a lot more reasonable to add a little "increased aggro" feature to the AI, meanwhile Civ has to build their whole AI for that stuff basically from the ground up. Look at tourism especially. What does that have in common with any other system? Almost nothing. Tourism needs a whole shitload of AI all to itself. Tourism in GalCiv is... Just a form of happiness modifier. And happiness is a raw multiplier on tax income. Super easy for a computer to optimize.

[Flying Dice]

That's because tourism is two different things used for two different purposes. You're painting it as some sort of complicated system in BNW. It's not. It's the offensive culture resource. You collect culture tokens and they generate tourism, which is pushed out in the exact same offensive culture paradigm as GalCiv II, except it's entirely self-driving once you start gathering it.

I'm also questioning whether you've actually played GalCiv II with what you said about city placement vs. colony placement. Literally all of those considerations exist in both. The only meaningful difference is that tile values are a binary yes/no rather than range considerations (as they're specifically tied to planets rather than securing them over X turns - not to mention that the tile modifiers are actually meaningful throughout the game rather than +1 gold/-1 shield/whatever; just having a 200% research modifier tile on a planet can drastically change what it is used for. Strategic placement is highly relevant, especially earlier in the game when ranges are shorter and the layout of the map requires strings of colonized worlds to reach different areas. Access to strategic resources (both the ones you need to starbase and mineable asteroid clusters) is determined by colony placement.

Granted, you probably notice those considerations less from the AI because the game world isn't the size of a peanut, so they've usually already begun spamming colony ships by the time you find them.

You're also understating the consequence of the combat system. Yeah, it's rock-paper-scissors, but it has a ship designer. You don't see literally the exact same units from everyone in every single game. You can't just go "Oh, they rushed cav, I'll build spearmen," because there's more nuance to it than that. Not to mention that there are considerations beyond the direct combat effectiveness, things like effective range, sensor range, speed, all sorts of things that can be radically altered. You can make units that can travel twice as fast as anything your enemies can build, but are so short-ranged that they are useless offensively. &c, &c, -- I dislike the simple nature of the combat itself, but by Civ-type 4X standards being able to control the stats of your units is massive. And the AI accounts for that shit, mostly.

[Cruxador]

That's because tourism is two different things used for two different purposes. You're painting it as some sort of complicated system in BNW. It's not. It's the offensive culture resource. You collect culture tokens and they generate tourism, which is pushed out in the exact same offensive culture paradigm as GalCiv II, except it's entirely self-driving once you start gathering it.

From a gameplay perspective yes, it's simple and intuitive. We're talking about AI though. You are aware that computers think differently than people do? It's much more mathematically complicated to account for.

I'm also questioning whether you've actually played GalCiv II with what you said about city placement vs. colony placement.

If I hadn't, why the fuck would I deign to speak on it?

The only meaningful difference is that tile values are a binary yes/no rather than range considerations (as they're specifically tied to planets rather than securing them over X turns - not to mention that the tile modifiers are actually meaningful throughout the game rather than +1 gold/-1 shield/whatever; just having a 200% research modifier tile on a planet can drastically change what it is used for. Strategic placement is highly relevant, especially earlier in the game when ranges are shorter and the layout of the map requires strings of colonized worlds to reach different areas. Access to strategic resources (both the ones you need to starbase and mineable asteroid clusters) is determined by colony placement.

Again, stuff that matters way more to humans than to AI. The AI can judge planets very effectively based on those things, because they're just mathematical modifiers on what the planet already does. Totally different from the civ resource system, where you have a ton of variables that have a more complex relationship, and claiming each one has, besides adding a city, geographic and defensibility connotations. Not to mention that the city has downsides you need to count against. That means that the AI needs to not just evaluate the base tile (where the city would go) based on what's there, but it needs to bring up the entire city development (building) and tile improvement AI structures and play forward a bunch of turns, just to get a comparable understanding of a location's value to what the GalCiv AI has to work with by default. And of course, it can't do that, because when you click "end turn" you don't want to have time to go make dinner before the next turn comes in.
And this is before even getting into defensibility. In Civ, you have water and mountains, both of which are impassable in some ways and not others. So you have these quite complicated geographic things that, to get a good placement, the computer has to think out ahead of time. GalCiv just measures distance to dangerous things, distance to safe things, and assigns each of those things a weight. Now, if a place is dangerous, GalCiv just passes it off to another part of the AI which will station some ships and possibly build a military station. Civ needs to keep looking for chokepoints and hills, and those should be considered before placing the city, none of which GalCiv has to worry about because one bit of space is pretty much like another for defensive purposes.

You're also understating the consequence of the combat system. Yeah, it's rock-paper-scissors, but it has a ship designer. You don't see literally the exact same units from everyone in every single game. You can't just go "Oh, they rushed cav, I'll build spearmen," because there's more nuance to it than that.

This is unimportant, or potentially a point in GalCiv's favor. The AI perceives an army and fleet, and sees, oh, here's something that looks like this, and builds a numerical profile because it's a computer and that's much more easy to work with than unit names. Then Civ crunches the math and has to build an army that beats those numbers, using premade units. GalCiv just designs units to counter what it's up against. There's more CPU cycles, but not actually hugely more, and you're getting a more effective AI for comparable dev time.

Not to mention that there are considerations beyond the direct combat effectiveness, things like effective range, sensor range, speed, all sorts of things that can be radically altered. You can make units that can travel twice as fast as anything your enemies can build, but are so short-ranged that they are useless offensively.

Yes, these are more complex things to judge for. They're also things that the AI doesn't actually need to do much with. It has scouts and science ships, which are optimized for different things than combat, and the basic ships can make do without a lot of it. Exceptions to this in GalCiv are artifacts of the greater dev time and CPU availability, rather than parallel to Civ.

[pedrito]

Civ 5, while not hugely complex, is still a deep enough game to challenge any current game AI. Even if they invested heavily in AI it would not even come close to presenting a serious challenge to any human with 1 to 1 production values. It's just too much data to process - in under a second, considering we players want all of the up to 11 AI players to take their turn and for the game to process the end turn in a reasonable amount of time.

That said, the Civ 5 AI is just amazingly stupid at things which are relatively simple from a computational point of view, like...

City placement: The AI doesn't need to recalculate often where it wants its next city, and if we completely ignore some more advanced calculations a human would do (how defendable is the city against my most likely enemies, what kind of specialization can I do in this city), we are left with some really simple stuff: how many early growth and production tiles does a city have? Does it have a river or mountain? How many resources are in the greater city area, excluding those resources that are already in another city range? This isn't rocket science, the AI can do these calculations once and assign values to hexes accordingly. There is literally no excuse for the blatant mistakes the computer opponents make so often in this aspect, except the developers/producers did not care. Cities found on ice without food, cities a single tile away from a river AND a mountain for absolutely no tradeoff, cities 3 tiles from their capital without a single resource that isn't shared with another city... the list goes on.

Tactical unit management: When the game came out, the operational unit management was a disaster, the AI would routinely declare war without even having the means to reach your cities. They fixed this somewhere around the first major DLC, and now the computer will bring on serious masses of troops in meaningful attempts to take a certain city. That's good. What they never fixed is the actual tactical unit behaviour. Units will rarely if ever retreat unless the whole attack is blown off. Archers that have a clean shot will move into often worse positions and miss their opportunities. Scouts will never attack, not even undefended settlers. All units when earning a promotion will always take healing even when only slightly damaged and in no danger at all, losing the chance to earn valuable upgrades. AI completely neglects pillaging. Etc etc etc. I'm leaving out the more complex and situational blunders like failing to trap units with control zones or placing forts smartly. Those things are actually hard to code properly, but all of the previous stuff is really simple checks that the game does not do.

Which brings me to my primary problem with Civ 5, that the developers/producers simply chose to abandon further development of an otherwise great game because these problems and the many multiplayer related problems are mostly noticed by hardcore players like myself with absurd amounts of hours into the game. The average customer is probably not bothered by them, and the company certainly won't lose many sales because of it. Civ 6 will be a commercial success. But unless they really change their stance towards continued development and listening to the playerbase, like other companies do improving their games year after year, they will again be stuck with a game that is good but could be great. And in my eyes there's nothing worse than falling just short while knowing that the potential was there, but wasted.

Let's hope they do better with Civ 6.

[Levi]

Apparently RockPaperShotgun likes it.

[umiman]

I believe Firaxis also likes to release demos for Civ games too, so we can try it out ourselves and see how the changes are in person.

[Neonivek]

Apparently RockPaperShotgun likes it.

Ehh they would like anything.

How about someone who was really critical of the previous game.

[umiman]

This is what nukes look like now: http://ca.ign.com/videos/2016/09/29/dropping-the-bomb-in-civilization-6

[Levi]

Apparently RockPaperShotgun likes it.

Ehh they would like anything.

How about someone who was really critical of the previous game.

Like these guys?     I guess they were only somewhat critical.

[umiman]

Totalbiscuit has been streaming it for quite awhile here: https://www.twitch.tv/totalbiscuit

He seems to really like it and I think he's losing track of time haha. The One More Turn effect in action.

He's also getting owned by barbarians.

[Cruxador]

Apparently RockPaperShotgun likes it.

Ehh they would like anything.

How about someone who was really critical of the previous game.

Like these guys?     I guess they were only somewhat critical.

Saying they're critical of Civ because they weren't 100% in favor of BE is like saying someone is critical of Germany because he's not 100% for Hitler.

[IWishIWereSarah]

It seems as if the prerelease has permanent raging barbarians or something.

I haven't watched any stream, but from what I gathered from earlier interviews, barbarian camps will produce units regularly and try to send them by packs to yours cities once one of their scout have found them. so that means that killing barbarians won't help you in the long term if you don't raze the camp. Is it what's happening here ?

[sambojin]

I think it might be a 150 turn demo thing. FilthyRobot, who takes his MP Civ5 very seriously and has made anti-barb tutorials etc, had a few hassles with them too.

So it's not just a casual gamer thing. They really are tougher. Whether this carries over to the full release or not, I don't know.