Re:Random Acts of Optimization
by hinkley on Thursday October 29, 2015 @05:43PM
(#10474563)
I came to a deeper understanding of this problem earlier this year.
A team that 'made a mess' may think they know all the mistakes they have made in the code, but they have never proven it, and they haven't been practicing doing things 'the right way'. They haven't even learned how to push back on bad strategic decisions that made things worse.
How do you know that the rewrite will be profoundly better than the current version? I think you have rather a lot of evidence that it won't.
Take a team that cleans up their messes as they go. They know whether their new ideas are better or worse. They have proven they 'deserve' a better code base by building one. By the time they know for sure what's really wrong with the code, they no longer need a complete rewrite. They need a partial rewrite that they can string out over the course of a couple years.
If any of that is true, then the people who want a do over won't take advantage of it, and those that would benefit most wouldn't feel comfortable doing it.
A team that 'made a mess' may think they know all the mistakes they have made in the code, but they have never proven it, and they haven't been practicing doing things 'the right way'. They haven't even learned how to push back on bad strategic decisions that made things worse.
How do you know that the rewrite will be profoundly better than the current version? I think you have rather a lot of evidence that it won't.
Take a team that cleans up their messes as they go. They know whether their new ideas are better or worse. They have proven they 'deserve' a better code base by building one. By the time they know for sure what's really wrong with the code, they no longer need a complete rewrite. They need a partial rewrite that they can string out over the course of a couple years.
If any of that is true, then the people who want a do over won't take advantage of it, and those that would benefit most wouldn't feel comfortable doing it.