> This would only be a risk if the difference between the local optimum and the global optimum is large, but there is no evidence to suggest that is the case
Agreed, it's been elusive for the software engineering community to objectively assess and quantify productivity effects of software tool chains, in particular in comparable stacks.
Comparison of programming languages or stacks is not what I was (trying to) talk about though. I have no idea whether hack compares well to $RANDOM other toolchain. I am curious as to whether a continued incremental investment in PHP made sense, compared to choosing a whole stack and propping that up to meet your requirements. Even assuming your final PHP-based stack is identical to whatever else you'd have chosen, the paths do have differences in effort.
The local optimum in that context is that it's locally optimal to just add that one feature to your existing solution vs investing in a different solution that in the long term will cost you less to build and maintain.
I don't have an answer to that question though, as I'm not familiar with the effort required or comparisons made. I think it's a genuinely hard call to make. Even post hoc and with all the knowledge it'd be hard to judge, as you need to discount effort by the relative risk factors.
Agreed, it's been elusive for the software engineering community to objectively assess and quantify productivity effects of software tool chains, in particular in comparable stacks.
Comparison of programming languages or stacks is not what I was (trying to) talk about though. I have no idea whether hack compares well to $RANDOM other toolchain. I am curious as to whether a continued incremental investment in PHP made sense, compared to choosing a whole stack and propping that up to meet your requirements. Even assuming your final PHP-based stack is identical to whatever else you'd have chosen, the paths do have differences in effort.
The local optimum in that context is that it's locally optimal to just add that one feature to your existing solution vs investing in a different solution that in the long term will cost you less to build and maintain.
I don't have an answer to that question though, as I'm not familiar with the effort required or comparisons made. I think it's a genuinely hard call to make. Even post hoc and with all the knowledge it'd be hard to judge, as you need to discount effort by the relative risk factors.