World IPv6 Day? [included bonus report on brokeness studies]

[Steinar H. Gunderson]

Den 13. januar 2011 06:15 skrev Martin Millnert <martin at millnert.se> følgende:
> I recently saw that Geoff Houston has measured 0.2% brokeness on v4 (!)
> on his site in a test he ran recently:

FWIW, any reasonable IPv6 figure is relative to (on top of) the IPv4
brokenness in some way. Of course any measurement is going to have an
error rate, e.g. if you do a JavaScript measurement, some browsers
won't have JavaScript etc.. (The user navigating away before the
background load fires is another common failure mode.) This would show
up as “IPv4 brokenness”, but in reality it's much more likely to be an
artifact of your measurement methodology which you have to compensate
for. The easiest solution is to just claim IPv4 is 100% working, and
then use IPv4 measurements as controls in your experiment.

> If for whatever reasons, v6 connectivity meant having access to a
> better-functioning Internet than v4, that would put this debate in a
> whole new light.. :)

This would then typically show up as negative brokenness (and would be
an unlikely result in today's Internet).

/* Steinar */
-- 
Software Engineer, Google Switzerland