so what happened instead


andrew odlyzko's excellent "myth of Internet growth" study (nov 2000)

'traffic doubling every 90 days'
maybe for a few months in 1995-1996
in reality, no real data since 1995 (nsfnet sunset)
more like every 12-18 months for rest of 1990s
financial markets (at least in US) believed (bubbly!) estimates
over 6 years, that means a factor of 16 million
assume (generously) 500M users, 1.5Mbps per user around the clock
and yet we're mostly still using 28k modems, & only for an hr/day, and ave 5k bits/sec then
the math just does not work out
it took 5 years for true traffic growth data to finally manifest itself 
                (since providers would not release data, if they even had it) 
         via other metrics (hardware and bandwidth sales)

that's actually an embarrassingly pathetic willingness 
to just ignore real data (or just invent it)