Even for a programmer, it is really hard to know how much might have been rewritten. A total rewrite might not be a good thing anyway, since the more code you write, the more bugs you introduce (no matter how great a programmer you are). Better would be if they carefully profiled the existing code (meaning taking measurements as to what parts of the code use the most memory and take the most time) and then "spot fix" the biggest offenders. That way you minimze risk of introducing new bad bugs, but still improve the customer experience. For a new feature, you are also better off making small changes to existing code, so it seems at least reasonable to think that they might have achieved spot render by modifiying the existing code, rather than a total rewrite, but that is probably a good thing. Constantly saying "this sucks, let's throw it away and redo it from scratch" actually produced worse code in the long run than "this sucks, let's see if we can make small changes to make it better". That's also why version 1.0 software almost always gets patched early (because the patch represents that "small change to make it better").