Combat AI

[Magil]

I'd say there's no benefit to having "tactical" combat if there's no competent opponent to be tactical against. How is methodically picking off the enemy with ranged units (while the AI units helplessly mill about) at all fun? It's absurdly repetitive, takes forever, and poses no challenge whatsoever.

Combat needs to be made more difficult somehow. Good Civ V players can routinely conquer AIs with armies five times as large as their own without losing a single unit. Maybe combat can still be made difficult even with 1UPT by some drastic measure, e.g. giving the AI a flat 50% combat bonus or somesuch. But I'd rather just put 1UPT on the shelf until someone designs an AI that understands it.

I 100% agree with this.

Again, there's no excuse. Modders have improved V's AI, so it's possible. We shouldn't assume there's nothing that can be done.

 

[stealth_nsk]

When people speak about AI being smart, they tend to forgot - one of the smartest thing vanilla Civ5 AI did were backstab attack.

The AI should be fun to play with/against, not "smart". Even without actual AI changes, Civ6 makes combat more interesting. You can't just sit in your cities, you need to protect districts from pillaging. This means more active play from human and much less archer turtling.

I don't have enough information about Civ6 AI yet, but the things I see make me feel quite well. AI tend to concentrate forces if it could take out enemy units; move a bunch of units effectively and retreat against superior human attack.

But that's not the situation where we'll be testing it actually. Most more or less active players are likely to be around Emperor level where AI will have significant bonuses and thus much more units (and more advanced ones). We yet to see how fun it will be to play AI with superior units.

 

[Magil]

When people speak about AI being smart, they tend to forgot - one of the smartest thing vanilla Civ5 AI did were backstab attack.

The AI should be fun to play with/against, not "smart". Even without actual AI changes, Civ6 makes combat more interesting. You can't just sit in your cities, you need to protect districts from pillaging. This means more active play from human and much less archer turtling.

Most people who want the AI to be "smart" are talking about it being more intelligent at the tactical level, rather than the strategic level. The AI surprising you with a backstab is a strategic move, the AI intelligently moving its units is tactical.

The only real benefit of 1UPT is increased tactical depth (debatable, but that's the most common argument). If it's not going to be utilized, then what's the point?

 

[stealth_nsk]

Most people who want the AI to be "smart" are talking about it being more intelligent at the tactical level, rather than the strategic level. The AI surprising you with a backstab is a strategic move, the AI intelligently moving its units is tactical.

The point was - AI being "smart" doesn't necessary makes the game experience better. I understand what the sample was from strategic AI, but for obvious reasons tactical AI don't have good samples of being smart.

The only real benefit of 1UPT is increased tactical depth (debatable, but that's the most common argument). If it's not going to be utilized, then what's the point?

Even not very smart "hide and shoot" tactics against AI is a way better than no tactics. And as far as we can see, fighting Civ6 AI is going to be more interesting than that.

 

[Acken]

I personally don't mind if the AI needs superiority to be competitive. People not wanting to ever lose can play whatever difficulty level for all I care. I just want it to be able to take cities and not shuffle its unit around doing nothing on the 2-3 top difficulties.

 

[Magil]

The point was - AI being "smart" doesn't necessary makes the game experience better. I understand what the sample was from strategic AI, but for obvious reasons tactical AI don't have good samples of being smart.

Not asking for miracles here. Just thinking the human shouldn't be able to rack up 12-1 kill ratios against the AI because it keeps throwing suicidal units into a meat grinder.

And as far as we can see, fighting Civ6 AI is going to be more interesting than that.

The entire point of the thread is that, in the videos, no, it doesn't. It's only somewhat interesting to us because the systems are new. But the AI's tactics are still pretty terrible, from what I've watched.

I personally don't mind if the AI needs superiority to be competitive. People not wanting to ever lose can play whatever difficulty level for all I care. I just want it to be able to take cities and not shuffle its unit around doing nothing on the 2-3 top difficulties.

Pretty much this. I think it's reasonable that the AI won't be able to outperform a reasonably skilled human player given an equal playing field. All I'm asking is for it to not constantly embarrass itself.

Though I'd argue that difficulty is largely irrelevant here, since it should be able to move its units and perform tactically at any level. It just stands up better against the human at higher levels because it has so many more units.

 

[stealth_nsk]

The entire point of the thread is that, in the videos, no, it doesn't. It's only somewhat interesting to us because the systems are new. But the AI's tactics are still pretty terrible, from what I've watched.

Well, timely retreat is already huge improvement. This means rough "hide and shoot" tactics will not work. You'll need either much bigger army (which is unlikely on higher difficulty levels) to finish enemy units right away, or you need to chase them to kill, which is a way more tactics to handle.

Also, the need to protect districts adds a lot to defensive tactics, even though it has nothing to do with AI behavior.

Both things are not miracle, but they seem to enrich gamin experience significantly.

 

[darko82]

From the videos I have seen, the AI is no challenge, no threat at all. It cannot take cities, organize a proper attack on a city etc. I will stick to Vox Populi unitll they can fix it. I know, it may take years, or it can take mods, or it may never happen in the worst scenario.

Otherwise, I like what I see.

 

[Aristos]

To me, it's embarrassingly simple. If Gazebo and Ilteroi can make a very good tactical AI for FREE and in their SPARE time, then there is no excuse whatsoever for Firaxis nor anyone that wants to try any defense in their name. Period. No excuse.

 

[darko82]

To me, it's embarrassingly simple. If Gazebo and Ilteroi can make a very good tactical AI for FREE and in their SPARE time, then there is no excuse whatsoever for Firaxis nor anyone that wants to try any defense in their name. Period. No excuse.

Why don't they hire them or something? They do deserve it. Everyone would benefit. I don't get this.

They also do have the code changed by the modders available. They should learn from it,

 

[stealth_nsk]

Why don't they hire them or something? They do deserve it. Everyone would benefit. I don't get this.

Just to make clear:

1. How many spare time they actually used? It's quite important question. I'm pretty sure Civ5 had less time invested in AI development.

2. There's a huge difference in developing AI for a game during development with constantly changing rules and developing AI for a game after release and patches, with already fixed set of rules. You just can't implement too complex algorithms during development as they are much harder to change.

Firaxis actually had great experience hiring top modders in the past. Jon Shafer, for example.

 

[Magil]

2. There's a huge difference in developing AI for a game during development with constantly changing rules and developing AI for a game after release and patches, with already fixed set of rules. You just can't implement too complex algorithms during development as they are much harder to change.

If Civ V's tactical AI was fixed in a patch after release, you might have a point here.

We all know that wasn't the case, though.

 

[stealth_nsk]

If Civ V's tactical AI was fixed in a patch after release, you might have a point here.

We all know that wasn't the case, though.

Full rewriting of AI from scratch after the last expansion would be nice, but gaming industry just doesn't work that way.

 

[Aristos]

Just to make clear:

1. How many spare time they actually used? It's quite important question. I'm pretty sure Civ5 had less time invested in AI development.

2. There's a huge difference in developing AI for a game during development with constantly changing rules and developing AI for a game after release and patches, with already fixed set of rules. You just can't implement too complex algorithms during development as they are much harder to change.

Firaxis actually had great experience hiring top modders in the past. Jon Shafer, for example.

1. It does not matter, because it was in their spare time and without pay. That means that anyone with both factors in favor (focus on a job to design AI for Civ 5, and payment for it) MUST be able to at least do something of the same quality, if not better.

2. Sounds like a lame excuse, and has no real foundation as an argument.

Jon Shafer was anything but one of the top modders. Ever. That is fallacious. You wanna talk top modders? Blake, Rhye, the BUG team, Afforess, kaltorac, the beautiful Lemon, and more recently the obvious: G and Ilteroi...

I am having trouble understanding why you are posting everywhere to defend an AI that does not exist yet (Civ 6's AI), while at the same time berating the amazing work of two REAL TOP modders in the series... what is your hidden agenda?

 

[darko82]

Just to make clear:

1. How many spare time they actually used? It's quite important question. I'm pretty sure Civ5 had less time invested in AI development.

2. There's a huge difference in developing AI for a game during development with constantly changing rules and developing AI for a game after release and patches, with already fixed set of rules. You just can't implement too complex algorithms during development as they are much harder to change.

Firaxis actually had great experience hiring top modders in the past. Jon Shafer, for example.

Sure thing. That is why I once talked about the support for the game and Paradox in comparison, which still supports CK etc. after all those years. Plus, they are still releasing DLCs etc.

What Firaxis does is, here you go, here is our DLL, improve and change whatever you want. Fix it for us. We are done with this project because it will not bring us more bucks anymore.

Great thanks there are people who want and can do it (mod).

I hope in such case DLL will be released on day one.

Why should we wait years to possibly fix that by ourselves? (sounds a bit shameful). They do not care to make it up to the standards (standards set by the modders at least).

 

[Magil]

Full rewriting of AI from scratch after the last expansion would be nice, but gaming industry just doesn't work that way.

To use Vox Populi as an example, the AI wasn't rewritten from scratch as far as I know. It was fixed and adjusted.

And for the record, that mod also changes a lot of V gameplay features, so the argument of it being "hard" to develop an AI concurrent to a constantly changing environment doesn't hold up. If that team can do it, Firaxis sure as hell can.

I understand that maybe the AI won't be perfect on release, as it's likely one of the last things to be fully developed. But actually fix it this time. And maybe, just maybe, make it a little less terrible for release? Would that really be so much to ask?

 

[darko82]

It is kind of silly because modders are supposed to mod, not fix things. But they need to fix it because Firaxis does NOT give a full support to it. They seem not to care to release outstanding AI, which lives up to the current standards.

I am not that demanding. All I want is to feel threat and tension on high levels of difficulty. I want to be afraid of their troops etc. I want to see them taking cities effectively.

 

[stealth_nsk]

1. It does not matter, because it was in their spare time and without pay. That means that anyone with both factors in favor (focus on a job to design AI for Civ 5, and payment for it) MUST be able to at least do something of the same quality, if not better.

So, every company ever made any encyclopedias are lazy and don't know what they are doing, because Wikipedia is better? I fail to see logic in this.

2. Sounds like a lame excuse, and has no real foundation as an argument

If you think about this in software development terms, it makes perfect sense. There are iterations. On each iteration gameplay rules are changing. AI needs to be in line with them. The more complex AI is, the harder changing things is. If you could react to changed rules by changing some numbers in AI priority settings, you'll have the AI following the change. If you'll need to rewrite hundreds of code lines, you just can't follow the main game development with AI.

I am having trouble understanding why you are posting everywhere to defend an AI that does not exist yet (Civ 6's AI), while at the same time berating the amazing work of two REAL TOP modders in the series... what is your hidden agenda?

I do neither. Modders work is great and helps many games a lot. Regarding Civ6 AI, I'm speaking about things I saw in the gameplay videos and not AI in general.

My hidden agenda is understanding of software development and AI development principles

 

[Aristos]

My hidden agenda is understanding of software development and AI development principles

Oh come on, you think you are the only one in the world that can claim the same? Honestly, that sounded a little too petulant. In any case, that is not a hidden agenda, but a hidden skillset (?)...

I don't think FXS needs more blind defense, but more pressure. Less conformism, and more demanding attitudes. Especially from those who pre-ordered.

 

[stealth_nsk]

Oh come on, you think you are the only one in the world that can claim the same? Honestly, that sounded a little too petulant. In any case, that is not a hidden agenda, but a hidden skillset (?)...

I'm surely not the one. Never claimed to be.

I don't think FXS needs more blind defense, but more pressure. Less conformism, and more demanding attitudes. Especially from those who pre-ordered.

The pressure can't make game better, it could only make it worse - if the pressure actually reach developers. Defense will not game better either.

Overall I find idea of affecting developers through forums quite weak. It may be possible to ruin some indie game where actual developers are on forums, but Firaxis have enough PR wall to protect their developers from both praises and reprehensions.