[Mod-discuss] Not a one-way door

[Tim Bray <[e-mail]>]

I’ve read the current draft and have been considering what I’d like to
suggest adjusting and fine-tuning. Upon further consideration, my input
condenses into two words: “Ship it.” Here’s why.

Whatever you think of Amazon’s culture or ethics, its management process
has proven to work well; speaking personally, it was the best-managed place
I ever worked, and that includes the places where I was CEO.

One of the central theses of that process is about one-way and two-way
doors. If you’re facing a design decision, think about whether it’s
reversible; i.e. if you turns out to have been wrong, can you shift to Plan
B? One-way door decisions need to be carefully considered at whatever
length is necessary. For two-way doors, the cost and delay of arguing about
it pretty quickly becomes greater than the cost of getting it wrong and
then shifting to Plan B later.

Most IETF products are one-way doors, not easily reversible, if only
because they get blasted into hardware and firmware and there you are.
This justifies the pretty-heavyweight IETF process with draft adoption,
versioning, last-call, etc etc etc.

Modpod isn’t like that.  The chance that the current draft gets everything
right rounds to zero. It is however, in my opinion, pretty good and likely
to have got quite a few things right.  The chance that we can, by further
arguing, anticipate and fix all the problems also rounds to zero, because
nobody can predict the future, in particular the future of community
conversations.

I mean, who knows, maybe Bernstein is right and a cabal of evil BigCo
minions will use it to hurl thunderbolts of censorship to suppress anything
against their corporate interests. (I don’t believe this for a second, but
who can say it’s impossible?) There’s only one way to find out.

This is not a one-way door.  Finish up the current cleanup activities,
publish it, and when we learn what it really got wrong, iterate. It’s not
blasted into any firmware.

-Tim