Threads for aconbere

  1. 1

    I’m also interested in this idea of resilient computing. A lot of vintage machines are still up and running, the vintage computing and arcade restoration communities have been keeping them going for decades (and have pretty good intuitions about who built more or less robust technology). I think a more interesting and unexplored domain is designing for continuous operation.

    The best example I’ve come up with is something like the Voyager Spacecraft which has been in continuous operation for the last 44 years.

    As a strawman proposal imagine a computer with the following specs:

    • 100mhz CPU
    • 1 Gb of RAM
    • 250 Gb of storage

    If we target 50 years of continuous operation we’ll exceed the operating lifetimes of the RAM and CPU silicon (but honestly we’ll probably have power supply failures long before that). Anyway I think this is a very Long Now Foundation like question, but in the case of computing it’s hard to even get our design specs out to the 100 year mark.

    1. 4

      If we target 50 years of continuous operation we’ll exceed the operating lifetimes of the RAM and CPU silicon (but honestly we’ll probably have power supply failures long before that). Anyway I think this is a very Long Now Foundation like question, but in the case of computing it’s hard to even get our design specs out to the 100 year mark.

      I think we’re (as technologists) just really insecure about how ephemeral our field is, but most things aren’t permament. What’s wrong with being fleeting?

      In Japan, they solved the ship of thesus - they just tear down an old shrine and build a new one in its place. I don’t see what’s wrong with things changing and evolving over time - that’s just nature, and something the lasts a long time is an abberation.

      1. 2

        I don’t think there’s anything wrong with technology being ephemeral! But i think that’s the status quo, and so thinking about the alternative is interesting.

        I think it’s interesting to imagine what a computer designed to run for 100 years would look like, to consider what parts would fail first and what tools we have to work around those failures.

        1. 1

          In Japan, they solved the ship of thesus - they just tear down an old shrine and build a new one in its place. I don’t see what’s wrong with things changing and evolving over time - that’s just nature, and something the lasts a long time is an abberation.

          Usually I see the refrain that “bridges last for decades so why doesn’t software”, but that belies the reality that bridges are one of the few things that humans build that need to last that long due to the sheer capital cost in building them. Even then, bridges (like anything else) need maintenance. Everything else we build, from bicycles to combustion engines to single-family homes, changes as humans and human society does.

          1. 1

            but that belies the reality that bridges are one of the few things that humans build that need to last that long due to the sheer capital cost in building them.

            This is an interesting point, that I would have agreed with before reading some public documents on a few relatively simple IT projects recently. And I’ve seen similar public documents on some road construction contracts, including small bridges (but that have high weight limits for lumber movement.) Software projects aren’t as cheap as we think they are, unfortunately. :(

            The up front costs on these relatively simple software projects makes the bridges look cheap. And the software projects don’t last a decade before they’re replaced or overhauled.