User:Mstone/Higher protocols

Discovery

As described by our network principles and by other presentations, our peers need a way to know what names are interesting. We propose the following prioritized list of techniques:

typing in a name

This mechanism is the most robust and most amenable to "use by determined humans".

proactive local discovery

See http://cerebro.mit.edu

url-handling

See Network principles.

global techniques

e.g. Facebook integration.

We think this is the right order because we think it serves our most important network scenarios the best.

Presence

Scott's thought was that peers should run a stripped down jabber server like, say, Twisted Words.

Health

Networks break regularly and humans need to fix them. In order to do this, humans need measurement tools, a mental model, and some hints about what to look at. Consequently, a middle-level health system might start with the low-level diagnostics and self-test algorithm mentioned above and extend those diagnostics and that algorithm with additional tests based on measuring:

• is there any divergence between what the human has told me is true and what I actually perceive?