i like reading joel on software, not because i always agree with him (because i don’t) but because he always makes me think and i always want to argue with him just to figure out what i think about the issue that he’s talking about.

he writes that good software takes ten years and i instantly don’t agree with him. it seems too much of a generalisation to me. so i get thinking and i think something like ‘software takes ten times longer than you think’ is a better generalisation. if it takes you a year to write a reasonable version of something then it’ll take you another nine to polish it and persuade everyone to use it. i think the same applies for the software that takes an hour, a day, a week or a month to write. they will be decent in ten hours, ten days, ten weeks or ten months.

i can write a script that’s useful for me in an hour and i probably save myself a large amount of time in the long run by spending that hour writing a script. however if i spend ten hours on the script i can get it out as something i don’t mind other people using and as a result a number of people will save time using it and the pay off is probably better.

in fact i think i could apply this rule to damn near anything. is this why my house is never clean and tidy? why i buy far more books than i read?

i don’t think there is anything special about ten. it’s a convenient point to focus on in a decimal ten fingered society. but i learnt something about myself and about estimation and why i’m always behind where i think i should be. that’s good, even if the thoughts are rambling ;-)