For testing the release candidates of GalCiv II 1.3 I've been playing all kinds of different sizes and scenarios.  In the process, I've been making tweaks and enhancements to the computer AI, uncovering mild logic bugs in my code and adding new abilities to try to keep the AI evolving as players evolve.  If you play at the lower intelligence levels, you won't see much difference but at the higher levels, players will likely see a significant change in the computer players with the final 1.3.

1.3 has a lot more re-balancing in it than previous versions.  Over the last several months, based on player feedback, we've made adjustments to how the game plays to improve the player experience.  But this has made some fundamental alterations of how the mechanics work that are sometimes subtle.  For instance, with 1.1, we made it so that you could only have 4 starbases in a sector. This change hasn't been really reflected in how we balance the starbase modules.  In 1.3, rather than starbase modules becoming increasingly more powerful (to encourage fewer but more powerful starbases) we now have a system where we know players won't build more than X starbases per sector so we could re-evaluate the numbers for the starbase modules in a way that we think will make the game more fun (it's certainly made it more enjoyable for me).  For example, building a starbase factory assist module no longer pumps out a piddly 2% improvement. Instead it does 5%.  But at the same time, the higher end factories aren't nearly as powerful.  But the important thing is that early game you can start to get an early boost.

What you're going to see is not going to be a complete game because it's being played on a Gigantic galaxy.  Instead, the goal is to watch how well the computer players play and see if they're doing things more intelligently now.

So what does "Allow AI to use CPU-Intensive algorithms" do?  Basically it allows the AI to design new ships more often and allows it to do some predictive branch analysis on what will happen in future turns. Does it make a huge difference in difficulty? Hard to say. It will mean more in Dark Avatar when I start to put in a lot more AI work. But I think it probably does make a mild difference. It doesn't affect Metaverse scores either way (so if you're really into maxing your score, you don't need to use it).

Here's an area that I did make a change today. It wasn't a bug fix per se but it might as well be one given how unintuitive the behavior was. The difficulty level was based on the intelligence level of smartest player.  So you could play a "suicidal" game where 1 player was really smart and the rest were set to completely brain dead. Then, once in, just set the galaxy so that the really smart one was in an isolated pocket while you fed on the easy players, then mop up.  One wonders how many "suicidal" games were done that way.  No more.  Now it's based on the average of the intelligences.

In this test game, I'm going to see how the Drengin, the Terran Alliance, the Altarians and the Torains do.  These are the 4 main AI personalities. The other players have derivative AIs (i.e. subclasses) of these 4. It's also why it drives me nuts when someone says "The AI did X" since it usually tells me nothing since there's literally 4 independent  (and even the others have their own code)  AIs in the game.  They ain't going to do the same thing. When I wrote them, I intentionally wrote them with very different algorithms. They may do similar things but not because the code's the same. 

It's a large galaxy after all...

Tight clusters, very large galaxy. I'm going to be cheating the whole game using the debugger so I can see everything they're doing at all times.

Opening Moves

The Drengin cranked out a factory and started a second one along with a scout and then started on a constructor since there's nothing good nearby (the constructor could be used to extend the range for the Drengin).

The Torians built a lab. Not what I would do but..

The Altarians have power built. On turn 3, they've got 2 factories and a research lab.  I'm looking into why they have 0 social spending right now.

The humans fast-buit a factory and then started on a second factory. They fast-built a scout and are now building a constructor.

As for me...

I fast-build a factory and a lab and start on a factory while building a constructor and a colony ship.

Watching the AI blindly send out scouts to see if it can find planets to colonize.

Turn 25

By turn 25, the AI has done a pretty good job on its home planets. Playing at "tough".

The Terran Alliance

The Torians

The Altarians

The Drengin

and for comparison -- me:

Their choices, other than the Drengin who built an Embassy, are similar to mine. I'm making a tad bit more money than they are but the Altarians are doing the best overall.

Me vs. Top players.  The Torians building that manufacturing planet really helped things as you can see.

End of first year

At the end of the first year each side is on its own unique strategy. Let's start with the Torians.

A pretty boring home planet but pretty effective. They could be making more money if they moved their taxes further up.

The Altarians have the right idea. This is probably the best home world right now overall. The Altarians don't get a second starting planet but they're doing well with the class 12 they get.

The Drengin are doing pretty well too. The embassy was an odd thing but still pretty impressive. They could stand to raise taxes further.

I have mixed feelings about the Terran Alliance.  They have their taxes up and are making some good money but they're clearly out of money. They could also stand to focus more on colonizing or else they'll lose some real opportunities.

And for comparison. I think it's safe to say that in terms of home planets, I'm doing the best. I have the highest income of any of them right now. Of course, I also made some different set up choices to have better morale.  The challenge is always translating "duh" strategies into a fast algorithm that doesn't do "stupid" things in other circumstances.

I'm doing pretty well overall.

End of Year 2

By the end of the second year, I'm doing pretty well. (gray line).

The Terran Alliance AI is clearly the weakest at this piont. It had the best starting position and is still blowing it.  So I'm looking into that.

Now I'm going to sit back awhile and see how the players handle themselves. Warfare should be particularly interesting because a good portion of my weekend went into making the AI more effective at waging war at higher levels. So in theory, they should be better at effectively wiping each other out.

Much later

By 2231, the Drengin are long gone.

The Terran Alliance have a small niche and the Altarains have their own corner of the galaxy. They all have friendly relations with each other at this point though the Torians could easily crush the humans.

Doing some analysis on what the AI has done..

While the Torians are dominating, I'm disappointed in a couple of the things they did.  First, while they are making a lot of money, they could be making a lot more. Their tax rate is lower than it could be. And they have a planet that they're not doing anything on. Why is that? (turns out they were just between turns).

Moreover, no starbases. I shouldn't say no starbases but very few. Resources are maxed out but there are very few economic starbases around. That's silly. With such a powerful military, the Torians should be shoring up their production.

I am generally satisfied with the AI's handling of their planets, however. While they may make a few different decisions, the overall results are pretty good.