But I am not irritated by the cheats Father Beast mentioned At least these are things you can control.

Invincible AI dips? Sink everything that comes towards your shores, double stack or go demo, create choke points with forests and the problem is solved 'cause the dips won't even see the outskirts of any of your cities. And from my experience, one of your 40 or so dips will eventually succeed even though you have become USA or Russia creating spies

I suspect Father beast knows about these things too.

What really buggs me is when I am obliged to follow specific rules and the AI isn't AND I CAN CREARLY WITNESS THE WHOLE THING HAPPENING.

The despotic democracy, declaring war every 2 turns without collapsing and refusing all GANDI like approaches from you, is one of the things that irriatate me in Civ 2.

It's ok with cheats really. But the AI should cheat us in a way that does not become too obvious and thus pisses us off. (which is always a bad thing for the AI, and primarly for civ itself).

Long ago, I actually took the liberty of sending an e mail to Firaxis pleading for just that in civ 3. (It's propably one of the 100000 emails they got but still)

quote: Originally posted by Zanzin on 04-08-2001 08:52 AM Seriously, how hard can it be to code a decent AI? It's all numbers run through various algorithms in the end. Each number has a certain response. It's a simple matter of IF....THEN...IF ELSE...IF ELSE.....ELSE......

A lot harder than you think. Leaving aside the tech difficulties though, lets just look at the cost. A napkin costing for a decent AI for Civ as described here (parametric coding for the macro stuff, combined with a neural net learning engine and an online strategy sharing room) would probably look something like this:

2 gurus: 250K / year
3 leads: 225k / year
10 code grunts: 400 - 500k / year
facilities: 120k / year
hardware: 250 K up front - 100 K per year
office manager and other support stuff: 100K / year
other (insurance, accounts, legal, trips to E3): 100k / year

Assuming an 18 month development cycle it would cost about $2.2 million.

It's just a guess, but I'm thinking they're planning on CivIII development costing in total less than $3 million with a development cycle of two full team years. Game coders cost a lot less than serious AI coders. Their code and art people are probably making about $30K on average and their leads are likely averaging in the $50 - $60k range. Now that Brian's gone I'd be surprised if they had more than two $100K plus salaries. Looking at the team photo I think $1.5M a year is in the ballpark. They might be able to swing a bigger budget from Infogrames due to the "Sid factor" but I think $3M is probably pretty close and maybe even high.

Who wants to bet that no one is willing to almost double the CivIII budget to build the AI you guys are talking about?

quote: Originally posted by Lord Maxwell on 04-10-2001 07:49 AM Hmm, the question whether or not to focus on AI when I get back to school after my wanderings as a programmer has just been answered.

Yeah, it's pretty crazy what a top flight guru level AI guy can make. Of course, before you start pulling down $125K you'll probably need at least one PhD and ten years under your belt (being a genius won't hurt either ). And if you have that, you could probably get a decent academic/research position somewhere with no real obligations to produce workable product. Thus the reason they have to be paid so much - the lure of just being able to tinker for a not much smaller salary in a cozy environment is pretty strong, at least it would be for me - but then no one is offering me the choice.