nikolaj

my first thought: woah, where do i get a .io?

I am only halfway through, but I love this gem: "I was flattered by his interest, so I thought it over. That is, I indulged in what young men frequently mistake for thought. In imagination, I saw my name over the door and myself in a fine glass office looking out and watching clerks taking in money."

His conception of brilliance is different than mine, but there is clarity in what he is trying to communicate.

of all the attempts i have made at estimating.. only one works pretty much spot on.

Essentially, take the gut feeling, double it, double it again, and add an increment.

Example1: gut estimate is "yeah, thats a five hour task".. so, 5 -> 10 -> 20 plus.. oh.. 2 hours, so 22. Seems absurdly large, yes, but factor in all those little thingsl, testing, deploying, etc, and lo and behold, two days plus some is gone.

Example2: Hmm, 2 weeks of solid (40hr week) dev time. So, 2->4->8 + 1 = 9 weeks. Woah, yeah that seems a bit large for project X.. but wait, client drags on assets, unforseen issues here and there, run into a snag for a couple days.. and boom!, almost two months have gone by.

I know it feels dishonest to so drastically overestimate what you Really think it will take, but the truth is the reason there is so much distrust of programmers by business people is that we so consistently miss deadlines. I know the estimate seems huge, but it needs to be, because ultimately it is very realistic, and if you beat it (doubtfully by much), everyone will be happy.

An oversized estimate finished early is better than an undersized estimate finished late. Always.

how academic to claim that programmers only implement ideas of others.. is the insight in the new theory, or the revolutionary application of the abstract theory?

I liked his criticism of big archaic architectures built up by armies of consultants.. until he said this:

"This new predictive system is the service-oriented architecture (SOA) I spoke of earlier. We call it that because it serves you and your customers better."

It's tempting to dismiss the experience gained at college as being not useful or applicable, but from my experience, there is generally a noticeable difference between those who learned all the fundamentals (e.g. college level math, physics, language theory), and those who are self taught.

There are alot of really talented self-taught coders out there, but for the rest of us, the breadth of understanding from slogging through those boring lectures on BNF, multi variable calculus, and linear algebra gave us an incredible analytical base thats easy to assume we always had.