> I also try to do the same, just write crufty code that works, hopefully clean it up some as I go. I account for this in planning. POC -> POC validation -> F

> I also try to do the same, just write crufty code that works, hopefully clean it up some as I go. I account for this in planning. POC -> POC validation -> Finalv2ForReal -> Finalv2ForReal validation It gives two shakes at a decent solution with some built-in learning. If this looks like too much for business then well I tried I guess I'll do less work for the same amount of money (inevitably the stupid shit that got prioritized goes away). And I documented what I would have done and can point at it when POC as a service is ermmmm ... less than performant.