I have no access to the internals and my knowledge is limited to what a handful of developers or support staff in the know can discuss publicly, or on project supports lists, so my word shouldn’t be taken as gold either.
That said, I’m sure the actual architectural/graph eval side of ICE could be decoupled from Soft with relatively little effort if it’s been written for that, and I’d be surprised to hear it wasn’t.
There is a chunk more though that I’m sure would need work before it’d become as fruible as it is in Soft, and a number of things are so inter-related with the way Soft encapsulates and offers data, that they would feel utterly awkward in Maya. That would need work.
ICE is still a complex and powerful Operator encapsulation system when you really, really get down to it. And Ops the way they work in XSI are very different from how you work in Maya. Max, I wouldn’t know, but it always seemed to me like it was written by aliens (with a large hive mind where no two cells agree), for other aliens
I also have no doubt that since the acquisition AD has been looking into it, and maybe even enabled that decoupling work to go on in B time or in the background.
I was mostly saying that stating it was designed from the ground up to integrate with Max and Maya in the future is obviously incorrect, if for nothing else, at least for the fact that ICE pre-dates the acquisition by several years if you talk inception, and over a year if you talk deployment