On Sun, Mar 7, 2021 at 5:43 PM Nico Williams <nico@xxxxxxxxxxxxxxxx> wrote:
On Sat, Mar 06, 2021 at 08:57:45PM -0500, Keith Moore wrote:
> On 3/6/21 6:52 PM, Phillip Hallam-Baker wrote:
> > One of the most frequent comments on new work proposals is 'no need to
> > re-invent the wheel'.
In the ASN.1 thread my point was that we should know well what came
before before we replace it.
Sometimes the old wheel has nothing worth saving other than the idea of
it, and when the pain of using that old wheel gets bad enough, we must
re-invent it. Doing it ahead of time is risky and has opportunity costs
such that the new wheel needs to be that much better than the old to be
worth pursuing early -- it's an economic analysis problem.
> > Well maybe we should do that more often. We have 40 years of path
> > dependence in IETF working groups. Rethinking a protocol from scratch
> > can have unexpected returns that are completely orthogonal to the
> > original idea.
>
> As an exercise, yes. But probably not with the expectation that the new
> wheel will replace the old one.
>
> Second-system effect is alive and well.
Insert standard "and now you have two problems" joke.
Also: https://xkcd.com/927/
