Why it's impossible to become a programming expert

This really resonates with me, and is one of the reasons I dread giving estimates. 80% of the time, I'm just doing something I haven't really done before and some parts of the problem will either:

a) be solved by someone else already and take half a day to integrate

b) be solved in another language / framework whatever and needs to be ported.

I find that in project mode, it's just easier to bargain and try to drop any feature which falls into the b category, at least for version 1.

This way, the customer will have a working product minus a few features in a reasonable timeframe (2 weeks to a month).