The first is that, notoriously, if one programmer working 50% more time won't advance the project 50% faster, that is even more true if you try to increase by 50% the number of programmers. Software development is incredibly non-linear (and as a project manager I always have to struggle with my manager about this - it is incredible how much linear thinking is ingrained in so many people's minds).
The second problem is that, also notoriously, software developers vary incredibly in individual productivity, much more than hourly wages do.
So, where engineering (not just software) work prevails on other kinds of work, a few geeks working very long hours can really be more productive than more people working less hours. But this does NOT work in most other environments, and even in software development:
- there is a lot of repetitive work to do that real geeks aren't going to do
- there aren't enough geeks to do all the software development that our societies require
In the end, one size doesn't fit all, and solving this situation with regulation would be very difficult (even if some better regulation could help, especially if it could make unpaid overtime much more difficult/risky for employers). The best chance would be to have one or more very convincing examples, like Ford was almost a century ago. Any candidates?