Allen and Gates redux

The New York Times reviews Idea Man, Paul Allen's autobiography. (Such a nuisance that these links now disappear behind a paywall.) The reviewer states,

Allen had the original idea for Microsoft. The two worked the same breakneck schedule in those feverish weeks when the company was born. Yet Gates wrote the program that served as Microsoft’s first product, whereas Allen did the less glamorous work of creating the tools Gates needed to do his job.

I am not sure what that means. As I blogged earlier, the listing states which parts of the code for the 8080 Basic interpreter were written by Gates, Allen, and Monte Davidoff. I'd say roughly a third each, and if not an even division, Allen wrote less. But to be sure there was tool-building. The cross-assembler for the 8080 to run on the PDP-10 would not be much work -- basically just defining op codes as UUOs (Undefined User Operations). But the UUOs would trap at run time, and somebody wrote the 8080 emulator code so that the PDP-10 would at runtime do whatever the 8080 would have done. This emulator made it possible to use the PDP-10 debugger (DDT) to debug the Basic interpreter. Moreover the emulator would have been an important and delicate piece of code -- it really needed to handle arithmetic overflows and the like exactly as the 8080 would handle such exceptions. So who wrote that? This sentence suggests Allen did. But the listing states that "Paul Allen wrote the non-runtime stuff." That is in reference to the interpreter itself, not the emulator code, which I don't have. But the emulator would certainly be runtime stuff if the pattern were consistent.

Now that Allen has made a fuss over this, I think there is a nice little project for somebody to figure this out.