Well I'm keeping an open mind on this one. I always read Steve Gibson with buckets of salt but...
It is easy to say this function was introduced in the days when internet was in its infancy and security and backdoors were of little import so it didn't matter if it was in there, or that MS have any number of other ways of introducing code onto a target machine.
But the fact still remains that this function seems misplaced as a printer Escape function in a metafile function unrelated to printing, the exploit depends on a special key (the one byte of record data) to run the arbitary code which is not at all the way the process should run, plus the possibility that it was only introduced with W2000 (since I checked my 9x machine with no patches and found the vulnerability was absent - at least according to Ilfak Guilfanov's tester).
There is more to this than at present meets the eye.