[brlewis]

There have been a lot of posts about failed startups. This is about a successful startup and the problems it survived. Doing a diff between this and the other posts might prove useful.

One thing I notice is the failure stories always include problems with their toolset. This could-have-failed-but-didn't story does not. I think the "programming languages don't matter" crowd should pay attention.

[nostrademons]

This is probably selection bias. Everybody has problems with their tools, because all non-custom-made tools are compromises to fill the needs of multiple users. When you fail, there's always something where you can say "I wish we hadn't used this tool and had tried something better." While if you succeed, the tool headaches kinda fade into the background and just don't seem that important.

When I wrote up my postmortem (http://diffle-history.blogspot.com/), I wished I'd prototyped things out more. This was intended as a general statement about process and not an indictment of any particular toolset (though I still wouldn't use JSF for any Web2.0 site ;-)). I think tools matter, but doing your homework matters more, and the best tool usually depends on the job.

I noticed that a lot of this guy's points seemed to concern doing your homework up front and not feeling time-pressured to act immediately. That's something I noticed a lot in my startup: you always know far less than you think at the start, and it's worthwhile to validate your starting assumptions before you go off the deep end on them.