One common complaint we get from the community is that so long as there are defects in Empire: Total War, we shouldn’t be working on any new products. If there was just one of us, or all of us could work on any issue that would make sense. As it is, we’ve had Empire: Total War patch work as the top priority for everyone. The campaign AI team has worked on nothing else at all since release. The other programmers have dealt with their patch issues before moving on, and get dragged back to them if they resurface. Most of the content team have not been able to help with patches – artists and designers can’t code and most issues are code issues – and have moved on to new stuff.
Patch 1.5 has just been released. This is the last planned major patch for Empire: Total War, and attempts to sweep up the remaining AI issues that for the hardcore gamer take the shine off of the Empire apple. The previous patches have dealt with the most common crashes and tidied up a lot of bugs, and 1.4 dealt with a lot of the AI issues. What is left at this point are a few minor issues spread around the game, and the last big campaign AI problem – the aggression level.
Battle and campaign AI are completely different systems and teams. I’ll talk about battle AI another time. The Empire campaign AI has been way too passive for me, and the community pretty unanimously shares that view, so it’s not something I need to explain. It is however interesting that a good proportion of the more casual gamers – and they are probably more than half our customers – actually like the AI to be fairly passive. The US casual gamer in particular likes a more sandbox-like experience, where he can make and execute long term plans and not have them constantly disrupted by an aggressive AI. This is a play style thing rather than a level of difficulty thing – they still want a challenge, but they want it to be their game, not the AI’s.
Making a passive AI may have sold us lots more games in the US, but it wasn’t intentional. Maybe we’ll have a play style setting in the future, but for now our intention is to challenge the player with an AI that is as aggressive and varied as human players would be.
So it is campaign AI that was the main focus for 1.4, and that I think we’ve finally got sorted out on 1.5. It’s worth talking a bit about how we ended up with an AI that didn’t have the play style we intended on release, and has taken 6 more months work to get there. The short answer is an excess of ambition.
This AI is not like any other we have written. It’s a beliefs – desires – intentions based planning system, and it’s also by far the most complex code edifice I’ve ever seen in a game. I wrote much of the campaign AI for Shogun and Medieval I (Ah… those were the days…) and I know that even quite simple “static” evaluate-act AI’s with no plans or memory can be complex enough to exhibit chaotic behaviour (we’re talking about mathematical “butterfly effect” style chaos here). It does what it does, and it’s not quite what you intended. This can be a good thing – you cull out the bad behaviours and are left with just what is good, and with a simple system that’s not too predictable.
Well, the Empire AI is way more complicated than any of our previous products, but the team is bigger and has more talent that we had in my day – PhD’s, and coders sharper than a box of razor blades. It’s a V12 supercar compared with Shogun’s 50cc moped. When it’s firing on all cylinders, it will be way, way ahead of anything we’ve seen in any PC strategy game before. It thinks about everything. It thinks of everything, it plots and it plans. As we approached release, bringing more subsystems on line, it was looking amazingly good, but at some point the level of chaos reached a tipping point and we lost control. Our AI did a “HAL” on us and gained the AI equivalent of multiple personality disorder. The net result is an AI that plans furiously and brilliantly and long term, but disagrees with itself chronically and often ends up paralysed by indecision.
We’ve had it on the coder’s couch for 6 months now, and it’s finally feeling better. It’s more aggressive, it uses naval invasions, and it doesn’t dither much more than most humans I know. It should now be well ahead of Rome/Med II’s AI, but it’s still only firing on two or three cylinders and had much untapped potential.
One thing I am sure about – I don’t regret having the ambition that led to this. This AI will I think astound in the long term, but I am gutted that we didn’t get the AI we wanted for the hardcore fans on day 1.
I had 6 copies of Empire: Total War sat on my shelf intended for close gamer friends that I didn’tsend out because I was too embarrassed about the flaws. Old friends are the harshest critics. Well they’ve gone out now. I think the game now meets my personal unreasonably high quality threshold – not just good but great. Hopefully my friends will agree.