I have found that the best way to update old software is to disassemble it into 68k and then sit down and study it, then re-write, modify, optimize and/or add new features that suits my own needs.
If that works for you, then that's great, but it's pretty much useless for people who'd want native PPC versions or want to use them on AROS, for example. And most people write asm much slower than C.
Personally I find updating C versions to be quite simple and straight forward - most of the changes that needs to be done are relatively simple. Even moving old Amiga C programs to AROS on x86 is usually quite easy despite having to deal with endianness issues, asm (e.g. I'm working on porting FrexxEd, and while the largest batch of work was simply getting it to compile cleanly on modern compilers, there was also snippets of inline asm etc.) and the occasional missing feature or API incompatibility.
This is especially true for software that was originally written in C (yuk) as the optimizations achieved in size of the code can be as much as 40% and in speed as much as around 30%.
With old programs compiled with SAS/Lattice or StormC etc., I don't doubt this - I did the whole disassemble and optimize bit with a few programs back in the day too, and for those compilers you often end up deleting most lines of a program due to horrendously bad register allocation which leads to massive amount of unnecessary stack usage for example.
With programs compiled with relatively recent versions of gcc, on the other hand, the margins are much, much smaller. A recompile (when the C source is available) is a far more efficient use of time to optimize most programs, though. If you then want to disassemble and optimize further, you'll still save a massive amount of time by having a better starting point.
The companies/authors that hold onto this software have nothing to lose by open sourcing old Amiga programs, they no longer make money on them and probably never will now, but for whatever reasons they have for not allowing the Amiga community access to improve or update this software I don't understand and never will.
A lot of the time I don't think it's down to not being willing to, but simply that nobody has asked or pushed them enough to put in the effort to dig out the old source. We'd get a ton of source if enough time was just spent tracking down authors and asking.
