[tahoe-dev] knowledge management Re: Tahoe-LAFS is widely misunderstood

[Zooko O'Whielacronx]

Hey folks:

I just want to say that I appreciate everyone taking the time to
contribute their thoughts here, and I'm reading everything you write
(whenever I have time -- sometimes I get a few days behind due to work
and family and all), but I wonder if we shouldn't start using some way
to encode ideas into a more stable form.

I can easily imagine that the outcome of a lot of discussions like
this is that afterward everyone still has slightly different opinions
and ideas from each other, but within a few weeks most people have
forgotten most of the details, and then when someone is ready to
really set to work improving the docs or writing new code or whatever,
they won't have that knowledge to draw on because it is buried in
mailing list history.

Let's try to gain more benefit out of this somehow -- issue tickets?
"LAFS Enhancement Proposals"? Maybe for the broad philosophical
manifestos that don't fit into issue tickets, the author could write
them up as documents and we could link to them from the wiki?

I guess I'm thinking of a few specific ideas that could be encoded: 1.
Chris Palmer's philosophical manifesto in favor of medium (nvi'ish)
complexity, 2. James Donald's argument that users need a tree
structure to manipulate, 3. The question of whether it is okay to have
secrets in "names" (or in "pointers" or "inodes" or "identifiers" or
"handles") and whether there are secrets in "names" in Tahoe-LAFS, and
which parts of Tahoe-LAFS are "names", 4. The comparison of
replication vs. erasure-coding and the recent and related discussions
about what the default settings of K, H, and N should be. I probably
left something out.

Regards,

Zooko