Re: [Add] My principles for discovery

[John Levine <[e-mail]>]

In article <CAOdDvNpVNbEgy3djk5aOD2FK+Tz7Bh9=[e-mail]> you write:
>I think this is a pretty strong summary and a useful contribution -
>thanks!. I might add that their view may include the scope of their
>recommendation. (e.g. I'm a general recursive, I'm for this split horizon,
>I'm for this authoritative, etc...) and, if we were making mission
>statements, emphasize that authentication is something they are implicitly
>conveying during presentation (even if its the absence of authentication).

This is reminding me of the way people used to build spam filters, and
not in a good way.  When the filters didn't do the right thing with
messages, the nerdy response was to add ever more knobs and buttons
that the users could control to manually adjust the filtering rules.

It didn't work -- most users didn't understand the controls, and for
the ones that did, or thought they did, it was a support nightmare as
they demanded ever more finegrained knobs they could try to tweak.

What everyone does now is to have a very small set of standard
filtering policies that work pretty well, typically one per system or
one per enterprise in shared systems, and to the extent users can
control their filtering, they do it implicitly by putting addresses
they know in their address book and moving messages in and out of the
spam folder.

This isn't a perfect analogy, but I'm quite confident that the fewer
decisions we expect users or their proxies to make, the more likely
this is to work.

R's,
John