Civilization 5 Announced

[Dantski]

If a mod was released which allowed multiple units to occupy a tile, would the AI also need be reworked again to make use of it?

[Kristian95]

If a mod was released which allowed multiple units to occupy a tile, would the AI also need be reworked again to make use of it?

Kael has already released such a mod (no more 1UPT)

[Kylearan]

Hi Sirian,

glad to hear from you again! And congratulations on your marriage - looks like you found someone who also likes games, which is quite rare. All the best to you!

(Originally I wanted to do something else this evening than posting, but your surprise appearance made me change plans... )

In defense of the Civ5 AI, I feel compelled to point out how dramatically the rules change to One Military Unit Per Tile (1UPT) increases the load on the AI. [...] The design and implementation challenges associated with this are orders of magnitude more difficult than anything that has come before. I would ask all serious analysis of the AI keep this fact foremost in mind

I'm a computer scientist, and even though my daily work is about malware research, I've done some hobbyist AI programming and have thought a lot about how to do an AI for a 4X game. To me, AI is one of the most important aspects of a (single player) strategy game. Even though it can be beaten quite easily, I'm still very impressed with the Civ 4 AI. From what I've seen and read so far, I'm also impressed by what Firaxis accomplished with the Civ 5 AI - the question is, if it's enough to make the game fun.

I'm not sure who it was, but I think it was Soren Johnson in his Google Tech talk on AI who said how important it is to design game mechanics in a way that the AI can handle it. If a feature is cool but the AI will never be able to handle it, then don't add it to the game.

I like the new 1upt rules a lot, I really do. But if it makes the AI considerably weaker even though it received more development resources, then something is wrong. In the end, only fun counts, not good intentions. The scientist in me applauds what Firaxis accomplished with the AI in such a complex environment. But the gamer in me is frustrated that he can plow through whole AI armies without losing a single unit or steal workers easily even during wartime. So even if I take Firaxis' accomplishments and good will into account, my enjoyment of the game is still detrimented by a weak AI.

The shift to 1UPT and its enormous new set of burdens on the AI render comparisons between Civ5 difficulty levels and those of past games incongruent.

I don't compare the difficulty levels of Civ 4 and 5. I only see that from all the numerous difficulty levels in Civ 5, only the two highest (and for some people, even the highest) pose a challenge, and that after 3 days of playing. And that sounds wrong to me. But maybe that can be changed by recalibrating those levels, as you said.

Nor is work on Civ5 "over with", as I am still on the project now and the new distribution via Steam will allow Firaxis to do things not possible via retail-box-only distribution.

This on the other hand gives me hope, if true. My experience with Civ 3 and 4 was that patches only fixed bugs and tweaked game balance, and that only once was the AI improved significantly by including Blake's work - in an expansion pack, not a patch. And as I had written before, I really wonder if that would have happened at all if there hadn't been a Blake in the community.

If Firaxis improves the AI via patches (for free, not in form of DLC), then I'm very happy and will again buy the next Civ game by them without hesitation. If the AI stays the same, or will only be improved if I pay for it, I will feel cheated, especially after the hype they did regarding the AI.

One positive thing (for me) about Civ5, having NOT been working 40-hour weeks on balancing its elements as I did for Civ4, is that I am still somewhat fresh to the game as it stands now. (I have been too busy with map tasks to be free to have played much of it). So I am not tired of it yet

This is great to hear! Have fun, hopefully in part here at Realms Beyond.

-Kylearan

[sooooo]

Hi Sirian again and congratulations!

I have a nitpick about one of your points:

The change to one unit per tile requires that AI units be aware of one another and attempt to operate as a group. The design and implementation challenges associated with this are orders of magnitude more difficult than anything that has come before. I would ask all serious analysis of the AI keep this fact foremost in mind; a look only at the final result of AI performance, comparing to past Civ titles as if apples to apples, is bound to be unfair and perhaps counterproductive.

I disagree with this because it was the Civ 5 team themselves that chose to change to 1upt. They chose to turn to this system despite knowing that the AI as a result would suffer.

[MJW (ya that one)]

I crushed the Prince AI. I had to play two games because the first one was odd. I started with the English all alone. They build the Great Lighthouse very early and it messed up their growth curve. I beat them and got a lot of land for myself and won easily.

My second game was plain and allowed me to see that the AI is very poor. No wonder why Civ4 on Noble cheats. It only had 2 cities while I had 6! I did it with the sub-subhara leader. Its very annoying to have to raze everything after you won because you can not get around the happy cap.

In Civ5 food is not the limiting factor (at least in my games where I built a lot of cities). The happy cap is. You get around it by using GOLD and HAMMERS. A city-state that gave hammers instead of food would be uber.

The tech-tree gives a lot less buildings during the end. This gives you time to build what you want (the end of the tech tree gives you lots of production bonus). If you win fast you don't have time to build everything.

The AI is poor at 1UPT and so it should be given a production bonus while building units.

[Bullstrode]

Sirian's post has me shedding my long-time lurker status to ask this question:

Why is the AI for 1upt so difficult to engineer? After all, the mechanics of Civ 5 combat are largely cribbed from Panzer General, and that game's AI did okay, and that was, what, ten years ago? Why is Civ 5 combat so much more complicated for an AI to handle than Panzer General, or Advance Wars? These are not new AI problems they're trying to solve. Are they?

Come to think of it, once they decided to go down the Panzer-General-style road, why wouldn't Firaxis go out and hire the actual guy who wrote the AI for Panzer General? I bet he's available. Turn-based war games are such an established genre (or at least they were way back when) that there must be quite a few 1upt AI veteran programmers out there they could have hired to help them out. Did they? If not, why not?

[sunrise089]

Instead of throwing around the labels of "King" and "Deity", is there concrete information about what the difficulty handicaps actually are? "Deity" doesn't mean anything by itself - it should be listed out in terms of AI cost factor and starting units and the other AI boosts.

This is not correct. This time around Civ5 specifically says players beating Deity are among the very best in the world. Does anyone here think the game is not lying about this?
__________________

@Sirian - First, glad to hear from you and congrats. I have nothing but the utmost respect for you as a game designer and a founder of this site. That said, I think some of your post is nonsense and colored by your close involvement with and employment by the developers. It's frustrating to see the degree of "circling the wagons" and groupthink from the insiders going on in the face of some real and thoughtful constructive criticism.

a look only at the final result of AI performance, comparing to past Civ titles as if apples to apples, is bound to be unfair and perhaps counterproductive.

Why? Did you charge me less for Civ5 than Civ4? As others have said, this was entirely a voluntary choice by the developers. If they made a choice that required more resources to undertake and then didn't provide the resources, it isn't my job to feel sorry. As a consumer, it is my job to call them out for what is certainly a mistake.

The shift to 1UPT and its enormous new set of burdens on the AI render comparisons between Civ5 difficulty levels and those of past games incongruent.

Fine, then care to comment on the above "Winning Deity makes you one of the very best players in the world?"

Civ5 cannot draw on these resources from Civ4 because the 1UPT gameboard obsoletes and nullifies any lessons learned under the old game mechanics.

Ok, then get more and better testers, not fewer and worse.

I am not sure how much I'm allowed to say (because I am not on top of what is or isn't out in the public domain at the moment) but I have only the highest regard for both the intentions and ability of my Civ5 teammates.

I have no reason to question intentions. Every piece of evidence I can see does cause me to question either the ability or the integrity of the testers though. I can say specifically that the MP testers are inferior to those from Civ4. Why?

In Civ4, balance engineering was my primary task, with the maps innovations being something I was able to slip in during the last few months of development. For Civ5, handling the map scripts was my primary duty, with balance input something I was only able to do on occasion.

Ok, and honestly that makes me confident that as I learn the game more I won't have room to complain about the map scripts. But strategically, I don't think a developer of your talent was best utilized here. We have a discussion above in this thread about how rare a guy like Blake is, and judging from the finished product balance guys are just a rare/crucial.

One positive thing (for me) about Civ5, having NOT been working 40-hour weeks on balancing its elements as I did for Civ4...I will try to get in to an SG or two where I can fit them in.

To end a fairly hostile post on a positive note, I will again say I'm pleased about this as well as your successes in your personal life.

[Speaker]

Sirian, congrats on your marriage! It sounds like you found someone who compliments you well, if she's willing to play Civ5.

The rest of your post, defending Firaxis, was pretty disappointing, however. The development of Civ4 was far more thorough than was done for Civ5. The exclusion of Friedrich Psitalon and the loss of Soren Johnson (and Jesse Smith? Was he still involved?) was a huge detriment, in my opinion. Reading through all the 1000s of words they wrote in the original Civ4 beta forum was very impressive. Those guys cared so much about game balance, and never would have allowed the crap that was Civ4 Corporations and Civ5 City States, nor the design decision to spend the absolute minimum of any effort on multiplayer.

[pocketbeetle]

The official 2K Games 'Multiplayer' forum is funny.

One thread's title: So is this where we report issues for the Civ V MP Alpha Test?

[Jowy]

The origins of Giant. Death. Robots.