I think Martin's point is that if something doesn't work, it is much easier to fix if one has direct access to the system. And I think this is true.
The alternative for most RISC OS developers would be to have a beta tester with that machine try out iterations and correspond via email. Which is unwieldy at best.
Secondly bear in mind that a significant proportion, perhaps most RISC OS apps are written by what you term novice programmers.
Finally, I think it is fair to say that most programs have bugs, and it is likely that a large proportion of the "Well it works on my machine" variety.