1. 9

  2. 4

    I don’t understand this post. I can see three or four soundbites, but no connecting thread. I don’t know what it’s advocating or not advocating, what point it’s making.

    To respond to just the headline, you’re naturally “long yourself”, so betting against yourself is a good hedge. E.g. for a prototype/side project/etc. it’s good to use a piece of technology you want to learn. If the project fails, you’ve still got value in learning the new tech. If the project succeeds, you can afford the cost of having made a suboptimal technology choice.

    1. 4

      I think, like everything, the whole point boils down to the definition of “new, unproven”. How many years and companies does it take? How many human-decades of work? I’ve noticed the default definition of “new, unproven” basically is anything you don’t know now. Don’t know Haskell, it’s “new, unproven”. Don’t know Clojure, it’s “new, unproven”, even though plenty of places use them and prove them everyday.

      Basically, worrying about “new, unproven” is worrying about what everyone else is doing. It’s a conservative mindset that aims for productivity in the middle of the pack, rather than putting the burden on the technology decision maker to know their tools and make sure they will work as needed before using them. It’s a shortcut, to just say, “well, I won’t get in trouble if I just pick what everyone else is using”, rather than actually asking yourself what you need. Maybe you only need Java, that’s fine, but maybe time to market is a big deal and you are faster in Idris, well then better make sure Idris will scale and does what you need.

      1. 2

        Another way to think about it: how come whenever people ‘use the best tool for the job’ it always ends up being the status quo?

        1. 3

          Well, back up a question. How did the status quo become the status quo?

          1. 2

            Because the “best tool for the job” is usually just a post-hoc rationalization for personal bias in tools.

            1. 1

              Alternately, because “best tool” in many cases implicitly includes “is easy enough to hire expertise in.”

          2. 1

            It’s the same kind of aggressive parochialism you see in US politics. Anything different than the status quo is new and unproven, even policies which have been working for huge swathes of the rest of the world for a hundred years. Meanwhile the status quo is considered “safe” and “proven” even though hardly anyone is happy with it.

            Likewise, for many developers, any technology that is new to them personally is unproven, no matter how many people have been happily using it for years, and no matter how much they chafe at the shortcomings of the “proven” technologies they’re used to.