The way they have been working on this has been interesting. Rather than a from scratch approach, they started with the original game and over time replaced component by component the original game bits with open source replacements.
That sounds like that wouldn't hold up in court, it's not a "white room" implementation by any means. I hope it won't be tested though.
Contrast that to OpenTTD or OpenRA where they (AFAIK) re-implemented the engine while still having the same approach of re-using the main games assets.
Edit: taking a look at the source code, I don't think what you're saying is actually true. RollerCoaster Tycoon 2 was written entirely in Assembly but I cannot find any Assembly files in OpenRCT2. So unless they already finished porting everything, it seems incorrect.
I looked at the source a few years back and saw more than a few functions that looked like the result of decompiling machine code to C. I didn't spend much time trying to see if that's true or not, but in the end it doesn't seem to matter. Still, if you like the game, you should keep a local fork of the repository, just in case.
I think this is the go-to approach for most of these kind of projects. You decompile it and then start refactoring it into C-functions and such that make sense. That this game was originally written in assembly probably helped a lot making the disassemby readable.
