Hacker News new | ask | show | jobs
by prodigal_erik 5036 days ago
Knuth's code has bugs. NASA's code has bugs. I don't think our species has produced even one nontrivial correct program, much less a programmer who's competent (which I take to mean a large portion of their work is 100% correct). Our profession is passing through the mercury-and-leeches phase medicine once went through. Someday our descendants will look back from their error monad formal proofs or something and cringe at everything we did.
2 comments
smartkids 5036 days ago
Yes. This is why I defined "incompetence" quite carefully.

Finding bugs in Knuth's or NASA's code might be beyond "a reasonably skilled hacker". To find bugs in that code you would likely have to be "highly skilled", above average.

Everyone makes mistakes. Even professionals who are licensed. The idea is to minimise them to achieve a reasonable, expected level of "correctness". Competent does not mean "perfect". It means no stupid mistakes.

In my biased opinion, there's a high tolerance for stupid mistakes in software.

link
pilgrim689 5036 days ago
>I don't think our species has produced even one nontrivial correct program

Off the top of my head: seL4, a correct microkernel...pretty nontrivial!

link
prodigal_erik 5035 days ago
Wow, I'm pleased to see that's no longer totally beyond us, though the price is still very high (LtU says fifty dev-years!)
link
pilgrim689 5034 days ago
It was actually about 28 py, but that number would come down to 10 if they were to do it again.

source: http://ertos.nicta.com.au/research/l4.verified/numbers.pml

link
smartkids 5034 days ago
This is nothing compared to the number of py spent writing crapware. Think of how much effort has been spent writing lousy software. It is enormous. (But then most consumers of software don't know any better, so from a sales perspective, maybe writing crapware makes perfect sense.)

IMHO, these guys are heros merely for undertaking the task, let alone completing it.

link