I was reading that 1961 BRL survey, and found the user
reports
of the IBM 701 fascinating. They apparently had an average
uptime of about 2-4 hours. That doesn't seem like much by
today's standards, but then how often does Windows crash on you?
One of the often told stories of the early days of computers is that of
the completely unreasonable downtime, due to tube failure. There is even a
myth (pretty much disproved) about AN/FSQ-7 SAGE techs wearing roller
skates to replace the constantly dying tubes.
Basically, tube computers were not that bad, all of the time. The use of
"average" is being distorted here. When a 701 or 650 or whatever was
placed into service, you could expect tubes to die out (rarely from
"burnout" - loss of emission was the main culprit) at an alarming
rate. One every two hours is probably not unreasonable, as the marginal
tubes in their infancy are weeded out. Once these tubes were gone - I have
heard this could be six months - tube computers were quite reliable. Often
whn an old IBM or Burroughs module is unearthed, a quick look at the
datecodes on the tubes will reveal that most are probably originals to the
machine.
William Donzelli
aw288(a)osfn.org