in spite of it all, amazing stuff has emerged 

topology mapping/inference: caida/ucsd, uw, and icir
simulation: ns (icir,isi...), renesys bgp/ssfnet
bgp/security measurement and analysis: renesys
tcp analysis: icir, cambridge (frank kelly, glenn vinnicombe), ucla (paganini), caltech
dns: gatech (ewz), mit (hari), caida
passive measurement/analysis: sprint, icir, caida, att
dragonflies/tortoises, self-similarity
macroscopic analysis: dns roots, dos attacks
invariance of rtt spectrum 
evolution of bgp system at AS, prefix, and IP granularities
introduce semi-global prefix and related taxonomy
dispel myths of bgp growth and churn (rates and contributors)
route dampening considered harmful
using bgp beacon (unused prefix announced and withdrawn at well-known times)
lots about topology/policy/event inferences from bgp data
patrignani, gao, schulzrinne, nicol, feamster
bgp convergence and scalability (anja&olaf)
udp/tcp performance during bgp update activity (avi)
comparison of routeviews and ripe data for bgp analysis (agilent, caida)
data sets seem congruent, both need improvement/coordination
traffic modeling