Bill Moyers on this administration, the press, and secrecy.
Torture's Long Shadow by Vladimir Bukovsky. Whoo.
And, finally, as an antidote: waaaay too cute.
Trustiness
A spot of hacking tonight produced gratifying results. In Python,
from net_flow import TrustNetworkcapacities = [ 20, 7, 2, 1 ] network = TrustNetwork()
network.add_edge("-", "el_seed") network.add_edge("el_seed", "test1") network.add_edge("test1", "test2") network.add_edge("test3", "test4") network.add_edge("-", "test4")
network.calculate(capacities)
for user in network: print user, '\t', network.is_auth(user)
produces
- True el_seed True test1 True test2 True test3 False test4 True
which looks more-or-less correct to me; c.f. http://www.advogato.org/trust-metric.html.
The net_flow Python module contains a class wrapper around a hand-written wrapper for mod_virgule's net_flow.c. Using net_flow with a scraped download of the current advogato certification network (sorry, raph...) I can reproduce much of the current list of certified masters. I get 602; the actual list contains 766 members. So obviously I'm still doing something wrong, but probably it's just a matter of reading the mod_virgule source code a bit better.
raph, if you happen to read this, could you verify that the capacities:
capacities = [ 800, 200, 200, 50, 12, 4, 2 ]
and the seeds
network.add_edge("-", "raph")
network.add_edge("-", "miguel")
network.add_edge("-", "federico")
network.add_edge("-", "alan")
are what advogato is currently using, please?
thanks,
--titus
p.s. I'll make net_flow available once I clean it up and validate it a bit more.
