Gabor on how engineering decisions get made

I feel that #1 usually happens not necessarily because a developer is trying to show off but because #2 and #3 aren't working well. If the process for coming to design decisions is too cumbersome and slowing down development too much, developers will simply circumvent that process entirely.

I happen to prefer 1, combined with an attitude of ruthlessly throwing away things (e.g. the prototype) that have been proven not to work.

the basic tenet here is that #1 produces something that allows improvement and iteration, where #2 and #3 result in decisions based on abstractions.

It's certainly a difference if you decide to use Ruby on Rails, or Django. I would take the second route then.

But when you need to get something done in one of the framworks, the first route probably is better. The second one is very dangerous as you can spend so much time discussing things that you end up spending hours not having done anything productive.