Good Software Takes Ten Years. Law of nature?
The Joel has made a good case that good software development takes 10 years - http://www.joelonsoftware.com/articles/fog0000000017.html. I don't know if any has tied these two articles together, but Peter Norvig makes a similar case that developing human expertise also takes 10 years -
I have made that connection too... when I first read Joel's 10 year postulate, I thought about that psychology thing where it basically takes 10 years to really be good at anything.
I've often wondered the same thing. I seem to recall reading once that developing guru-level expertise in a field, any field, typically requires 10,000 hours of effort. That would translate into 20 hours a week for 10 years, which is probably about a realistic estimate of the actual time one invests in a full-time vocation or very serious avocation.
Those of us who are more highly evolved only require 8.3 years.
Depends on available resources. There is no magic number in my opinion. Maybe ten years means just that:
That would explain why vi/emacs are the best development editors around.
I guess joel's products aren't any good yet?
son of parnas
Probably not unrelated. One way to think about it (which may or may not be true): It takes 10 years for any one person to get good at something; with a program, even if you don't have the same programmers for 10 years, the domain knowledge of the organization is maintained (we hope!), so there is 10 years of continuously building programmer-knowledge being put into it.
Fog Creek Home