geoff@utstat.uucp (Geoff Collyer) (01/05/88)
Rayan tells me that total news flow is 3Mb/day, about twice what I
estimated. On the other hand, I largely ignored the effect of
compression on news transfer times: compression tends to halve the
time. These two effects should cancel each other, making my previous
estimates approximately correct. :-)
Compression does slow transmission of news on machines where the
(elapsed) time to compress a batch exceeds half the time to send the
batch uncompressed (similarly, on reception, if time to uncompress
exceeds half the time to receive the batch uncompressed). This is
unlikely to hold for any modern machine (utzoo is another matter).
After thinking a little more about traffic growth, I remembered that
someone has observed that Usenet traffic doubles each year. Assuming
100kb/day in 1982, that assumption yields 3.2Mb/day in 1987, which is
pretty close to reality. If traffic continues to double yearly, in
mid-1989, 1200 baud will not supply enough bandwidth to receive a full
news feed. In 1990, 2400 baud won't be adequate. Around the end of
1992, the Trailblazers (over 9600 baud) will run out of steam, with
daily news flow of 102Mb. If the phone companies follow New Jersey
Bell's lead, we may have end-to-end-digital dial-up data communication
by then, and be able to run considerably faster, and without modems. By
1994, daily flow will reach 410Mb, requiring a VAX Eagle *per day*. By
1997, daily flow will reach 3.3Gb, approaching one byte per person on
Earth.
The answer to my own question seems to be that "long-term" is about four
years. By then, we will need modems faster than Trailblazers, digital
dialup data communication, widely-available WANs (Wide-Area Networks),
or moderation of most groups.
--
Geoff Collyer utzoo!utstat!geoff, utstat.toronto.{edu,cdn}!geoff