I've done it myself several times, but of course on projects small enough for a single developer. I also think the approach of rewriting bits and pieces over time works. You can identify pieces of the old code base that has well-defined behavior and can be split off with a well-defined interface between the new and old code. This has the added benefit of both needing to understand the old codebase and what you'd ideally want from a new codebase.

many years ago I worked on the windows firewall team at microsoft. it was common practice to get to a certain milestone of a project, then purposely ditch the code and start from scratch. I haven't seen this practice since then, but that part of windows was one of the least hated-on at the time.

As a second thought, I think the move from .NET Framework to .NET 5+ could be considered a rewrite, and that has been rather successful on many fronts.

Facebook has done it with their iOS Messenger app

https://engineering.fb.com/2020/03/02/data-infrastructure/me...

Actually its not going to be that.

That's when you've got something small that works and the rewrite is a bloated mess of every feature you could dream of.

This is the case where you've got an accumulation of over a decade of technical debt and Chesterton's Fences and someone who has no idea why the code is the way that it is inherits it and decides to rewrite it, and not knowing which constraints are necessary and which aren't they throw away most of them and find out the hard way that the problem isn't actually that simple.

It is almost the inverse of the second system syndrome.

But it is abundantly clear that Elon doesn't even know how to have this conversation.