Er…no? There’s certainly a difference since cross-compilation front loads that complexity onto the author, rather than making providing games for different platforms painless. Does the average IF author want to deal with compiling separate binaries for each platform? No, probably not.
It’s a trivial process if you get past the nontrivial hurdle of setting up VMs or actually physical machines for each environment, notwithstanding the complications coming from the specifics of that (e.g. how old of a Windows/OSX version are you going to support, what libc are you using on Linux, etc). Even if this was as smooth as possible, adding support for a new platform would be a nightmare. But you don’t need to do that to provide a story file, all that complexity turns to interpreter writers.
Aren’t TADS 3 lists limited to 13094 entries before they start causing runtime errors? I guess you can say that’s a limit that isn’t likely to be run into in practice, but I suppose that still seems a strange choice for a limit.
EDIT:
Except Lynx, or people using NoScript, etc.
But (barring those cases) does all JS work across all browsers? 