15may2002

Unreliability


(with Cory Doctorow)

okay
so i like the way he uses "modulo". as in, everything has property X, module
noun Y [which doesn't have property X]

he says:
- we don't need high quality content
- we don't need mission critical

[hmm
can't really think of anything to say about this.
i think it's one of those talks that's going to just throw some ideas at me that
i need to go into perculation for, for about a week]

expunge
- optimal

[so cory's exploding some myths about what makes good technology. fair enough.
it's one of the gripes about soap. he's saying like: there's the thing Don't
Replace A Human Process With An Automatic One; but he's saying it in the future.
don't optimise and genericise ahead of seeing what the adhoc uses are.]

historical examples of unoptimised examples:
- turing machine is nasty, unplanned, suboptimal for any given task. but it's
better at doing anything.

[aha! what he's saying is that if you optimise you get rid of the rich
environment things can grow in. and if you have no growth or feedback loops, the
technology won't be used and won't stay.]

question about optimal from chairman of EFF: market keeps rejecting the combo
pda/cellphone/mp3player. why?
cory: wrong scale. WAP, highly optimised, died. sms, not optimised, lived in
ways we didn't expect.
ah, disagreements about optimised: sms is optimised! no says cory. wap opt for
particular type of info. sms isn't.

this is it:
you never close an avenue you don't have to close.

[AND THAT'S THE COMMON SECRET PROPERTY:
the popular technologies are the ones that aren't necessary generic, but they do
a thing and don't close doors to using it in any other way. and that's a
property that "small pieces loosely joined" has, almost by accident: not that
the avenues aren't closed, but they're being created at all stages. and it's
*this* that makes technologies adapted.. that there are all these hooks of flow
to join into the ecosystem. aha. so that's why "small pieces loosely joined" is
a successful strategy.]


====

rumblings:
- all these collaborative things aren't going to work. why? because you'll get
in to building a CMS, and trying to emulate the workflow etc. so the 2-way web
won't get a foothold *until* we make the abtraction layer in the correct place
to have just the right amount of collaboration in the system, with adhoc
workflow that can be built on top, by various systems; in the same way web
design software sits on top of html. so: how do you tweak something like html to
make it, A, collaborative, and B with the correct permissions system so that
*any* workflow/cms can sit on top of it?
- this 2 way web thing is interesting. nobody's talking about it at all.

- IMPORTANT: i really need to explain this SECRET PROPERTIES thing is. the point
is, there are loads of successful technologies out there but we made them almost
by accident. they're popular for reasons we're trying to reverse engineer to
make making the second generation easier. these properties are things we didn't
know about till now: secret properties. these properties are the things making
the rich environment that the feedback loops grow in, or create the flow hooks
to join to the ecosystem -- and we're trying to emulate the *public* properties
that are obvious, like the feedback loops. but that's trying to *build in* [ie
make automatic] these adhoc things. it's not possible. we need to go a bit lower
and build a rich environment [or "loam", as jones is saying] that *makes* these
secret properties possible -- even though we don't know what they are.