i like the idea of this. however, in practice i don't know how to implement it or draw the line between "enough iterations" and too much time spent on iterations.

Right now, with a small, strapped team, my primary problem is finding enough hours to do all the tasks necessary.

As a result, while the idea of slowly evolving the design to let customers ease into it sounds good, I'm willing to shock my users into a radically different design in order to save development hours.

My userbase, in particular is the sort that (I believe) won't dramatically change engagement based on most designs (untested, and now that I write that, ludicrous sounding, should test). Additionally, for early stage startups (at least Rocket Lease stage startups), the total user count is very low and the alienation risk carries a small absolute penalty relative to the time sacrifice of incremental evolution.

tldr: article acknowledges that there is a time penalty for an incremental design evolution. this is in contrast to the typical time BENEFIT to iterative programming approaches. i suspect that many early stage startups can't afford the time cost, so interface iteration is an important tool, but not to be used by everyone.