Topic: Why do AAA games take so much longer to make today than in the past?

It's not the technology, it's the people.

We see this is other industries that involve technology, e.g. large IT projects (even small ones) inside corporates. The commonality is aversion to risk.

Levels of management are high and spans of control are low, instead of more optimal flatter organisations. So there are either a lot of managers, or people who should be doing real work doing management instead.

There's lots of compliance to consider, to internal rules (e.g. Microsoft's rules on representation) an external ones (e.g. get legal's view on things), but also committee decisions to design things. In the past, a developer/ artist may have an idea and take it straight to the lead producer / director who'd say "great, let's do that" and it would take a few minutes — these days they would take it to their manager who'll then take it to theirs, who'll then have to discuss it in a meeting with others, and then it might got forward to another process or two or checks and compliance, got back to group meeting to discuss again, then eventually up to a committee who might say yes or no and kick it back down or kill it off.

This is institutional drag, which is caused by aversion to risk, and it's incredibly costly in time and money — has absolutely no positive effects on anything. Is everywhere.

A lot of it will be gold plating, over spec, bloat etc. Materials, models, levels will be over detailed, they will use too much manual placing when procedural would work better, they will allow decisions by committee which slows things down and loses a singular focus. Something can be good enough with a X amount of man hours, but get to "perfect" will likely mean 3-5 multiple of hours already worked to get to good enough to get to perfect. If someone higher up decides everything needs to be perfect under every circumstance then the workload will increase dramatically.

When I say good enough I do not mean stutter or technical issues, I mean does there really need to be 200 hand crafted slightly differently weathered variations of the same poster to be placed around an open world, for 10 different poster designs, or would a base design for each then use procedural generation to age them all differently. It is a bit like in the early days when studios had to manually build trees or they all looked the same, then SpeedTree was developed and many used it, whilst others till manually built trees because they wanted them to be perfect. That behaviour died out, but took time, the same will happen with other things.

I know it will be an unpopular view but this is something that AI will actually be very good at, a workflow multiplier for the art and materials team, they will be able to design the model, have the AI apply PBR data and then they can review, design assets and have the model add layers of wear and damage to them, use it to place world clutter without that needing to be done manually etc.

RT is also going to improve things when they can finally get totally rid of baked lighting, it means placing lights is instant, the level designer can light a scene and know how it will work, at the moment multiple light sources require baking out to see the result even if they still have RT on top.

@MattGPT Which raises a good point - the solution to one popular gamer complaint (games taking too long / too expensive) is in part actually another set of common complaints (AI, ray traced lighting) plus perhaps shorter gameplay length!