Dear p2p-hackers:
Our open source project, Allmydata-Tahoe, has garnered a few
criticisms which I thought might interest the larger p2p-hackers
community.
David Barrett posted to this list his observation that there's
something to be said for real simplicity versus "simplicity through
encapsulation" [1].
Also on this list Alen Peacock suggested that designing a system
exclusively for a specific use case has value [2].
Over on the tahoe-dev mailing list, first Wes Felter [3] and then Jim
McCoy [4] questioned whether decentralization was a good strategy at
all. Jim McCoy wrote: "I can't see any reason why central servers do
not beat complex distributed architectures in every possible axis of
comparison.".
Nick Szabo posted on his blog, Unenumerated, about a possible
architecture for decentralized, automated economics [5]. (This is
not so much a criticism of the Allmydata-Tahoe design as a suggestion
for future improvement.)
And a few years ago Bram Cohen posted to his blog, saying that
erasure coding was rarely warranted [6].
I'm glad to be getting lots of feedback about the Allmydata-Tahoe
architecture. If I'm wrong I want to know it, of course, and also if
I'm right then answering criticism helps me to think more precisely
about why I'm right. I owe some of these people answers...
Regards,
Zooko
[1] http://lists.zooko.com/pipermail/p2p-hackers/2007-June/001082.html
[2] http://lists.zooko.com/pipermail/p2p-hackers/2007-May/001055.html
[3] http://allmydata.org/pipermail/tahoe-dev/2007-June/000022.html
[4] http://allmydata.org/pipermail/tahoe-dev/2007-June/000023.html
[5] http://unenumerated.blogspot.com/2007/06/nanobarter.html
[6] http://bramcohen.livejournal.com/1416.html
_______________________________________________
p2p-hackers mailing list
[email protected]
http://lists.zooko.com/mailman/listinfo/p2p-hackers
