Hacker News new | past | comments | ask | show | jobs | submit login

There's no single definition of "best"; your is implicit, but what specifically do you imply?

Straightforward, idiomatic imperative code can be "best" if you have to maintain a legacy app and can hire developers with little experience and desire to learn.

Code which is less like Basic can be "best" if it allows your team to move twice as fast, and have 10x as short a backlog of defects to fix. It may take a more expensive team, and some onboarding time for new members, of course.




> Code which is less like Basic can be "best" if it allows your team to move twice as fast, and have 10x as short a backlog of defects to fix. It may take a more expensive team, and some onboarding time for new members, of course.

This is true in theory but it’s famously hard to measure, having both very strong religious beliefs and notorious confounds. For example, is any particular anecdote telling you that a niche language like Lisp or Haskell actually makes people more productive or just that you’re filtering for developers who would be above average in any language?

Actually measuring these sorts of claims scientifically is expensive enough that it’s rarely attempted, but that rarely causes people to taper the degree of certainty people express. That doesn’t mean that new languages are bad but I think Fred Brooks was right to recommend skepticism of big claims rather than more incremental improvements.


I wasn't really clear sorry. I am really advocating for idiomatic code at the implementation level. Of course your architecture will be novel, and it's there that you build efficient and easy to use systems. You want people to be focusing on that, not being distracted by arbitrarily different unidiomatic details.




Guidelines | FAQ | Support | API | Security | Lists | Bookmarklet | Legal | Apply to YC | Contact

Search: