Team size. (Although, at scale, you're actually talking about dozens of teams.) The problems that occur at scale have to do with dividing work, cross-team collaboration and coordination, and the sheer volume of work being too large for any one person to fully track and understand.

I'm not saying Star Citizen development is perfect. They're doing a lot less automated testing than I prefer, the code they show in Bugsmashers isn't particularly well designed, and they would have benefited from adopting feature teams and timeboxed releases much sooner than they did.

But software development always looks easier from a distance than it is in reality. The problems are fractal, and don't make themselves known until you're deep in the middle of implementation. (This is why so many projects are under-estimated.) Your criticisms lack that nuance. You say they're writing crummy software, but you offer no evidence other than 50,000-feet hand-waving assertions. It comes across as either inexperience, or an axe to grind.

The issues that concern me the most aren't ones arising from production scale.

I think the project was just fundamentally flawed at the beginning because of the choice to use cryengine, and they never corrected it. They doubled down and poured huge amounts of great work after bad. At the very inception of any software project you need a very small handful of experts to decide some fundamental things eg. What language / framework / engine to use. What platforms it will be on, what networking model it will use.

In the case of star citizen, Chris Roberts picked cryengine himself (probably just cus cryengine games tend to look nice), even though he hadn't produced a game in years. It did make sense for the demo-video. As I said, it worked, he got funded, but after getting funded he should have scrapped that visual demo and hired a few experts to draw up a plan for how to realise the actual game. I would have been looking to headhunt people who were architects for battlefield or planetside. Instead it seems like CIG just started hiring everyone like crazy and dogpiling work ontop of the visual-demo.

It seems from their dev updates that the team is putting in a heroic effort trying to work through the issues, though it also looks like a perfect realization of the sunken cost fallacy.

Here's what you've said about Star Citizen elsewhere in this thread:

"An unmitigated clusterfuck." "A mess." "Crummy software."

And now you're saying they chose the wrong engine? That's a Far Cry from being the Crysis you described, which makes it sound like choosing CryEngine makes a game Prey to any Monster Hunter Online. Sure, Crytek's in a State of Decay, but the CryEngine Kingdom's Come: Deliverance in the form of Amazon. The engine is still maintained, and still used for major games.

Okay, I'm resorting to bad wordplay now. I guess that means I'm done. I'll let you have the last word.

I wish I could give you points or karma or something for that excellent punmanship!