Adding is favoured over subtracting in problem solving

Article note: Interesting how general it is. People point to incentive structures ("lines of code == productivity" "new features get rewarded because we can't measure maintenance" etc.) in software, but it all comes down to Chesterton's fence - you really have to understand how something works to shrink it without breaking it, and the potential cost of removing and maybe replacing is generally much more than the (more predictable) cost of adding.