The problems go a lot further than just the 32 bit coord system. The engine was made specifically for FPS games around the LAN Party era. The main loop managed networking, physics, player movement, etc. all in one. Touch one thing and the other parts start to crack and crumble. It was made for a time where CPU were single core and high end were dual core, and max player count was 16 on a lag-less LAN connection. Every part of this engine is a consequence of that. Also coordinate trickery will imply physics trickery, as everything will be off in density, weight, etc. Better not twist the original intent...If you were say embedded or involved with CryTek/CryEngine at a corporate/professional development level during 2012/13, then you would either have to be a complete idiot or wilfully negligent in believing you could produce a large scale flight or space sim, without a ground up re-write of CryEngine.
(...)
The sad thing is the folk who have invested $300million+ in this fantasy, have probably given enough funding to develop such an engine, provided they could sit on their impatience for a decade or so and have no playable demo or product.
As for the decade or so: as long as we are talking about a bespoke engine, it's 2-3 years for a competent team, not 10. It's what it takes in in the industry, so far. Would have been a game changer if that decision had been taken back in 2012. I'll further this argument by pointing at quite a few existing engines that are around, and the collected experiences from these (starting with EvE which clearly showed the limitations of a centralized, single cluster, transaction-on-write database system). Even back then we had a lot of insight about how to code an MMO and what were the traps. CiG did fall in every single of these traps, like complete newbies, just like if their lead developer was coming straight from the early 90's... hmmm.. who would that be...