Reading all the above, I comment:-

1) having a degree in Computer Technology of any kind, might help with
designing a compiler, but is not essential. I don't see a lack of such
in the gfortran team to be worthy of complaint.
Being qualified to think is foremost.
To be a long-time user of Fortran and C compilers certainly is
important. Having time to write a good, clear installalion manual for
the compiler, is much, much, more important. I mysel DO have an MSc in
Comp. Tech and other degrees in Maths and Physics, etc, (and
commercially write compilers using Fortran) but I occasionally get
scorn heaped on my Fortran advice for being too conservative. I
wouldn't think of designing and writing a Fortran compiler; I bless
those who do for the general good.

2) I worry about the mentioned stages of "compiling compilers". Why
not specify some recognised c-compiler as a base-point and offer a
start from there; those who don't have such a commercial or otherwise
starting point can download a required first-stage compiler.. But
again, clear instructions are vital.

3) It seems obvious there is a beginning divergence in option coding.
Perhaps a nested structure is needed for the user/installer to first
define a target environment, and within that, the particular form of
the option codes to be used are now pre-defined, and taken from a
replaceable table.

4) There is always my sore point. Fortran needs to interact with the
operating system environment, especially screen, keyboard and mouse,
and possibly USB and other user ports to external devices. This should
be tackled by defining an interace; then the public can help in
provided run-time operating-system dependant device access libraries.
I don't see that the Fortran committees have ever touched this area.