# Build systems, not hacks.
A hack solves today's instance of a problem. A system solves the entire class of it, then keeps solving it while you sleep. Hacks feel faster because the cost is deferred — you pay it every time the problem recurs and you patch it again. The test I use: if this exact situation happens ten more times, will I have to think about it ten more times? If yes, I haven't built a system yet, I've just postponed the work. Systems are more expensive up front and cheaper forever.