Re: h2 priority

In message <>
, Patrick McManus writes:

>To clarify my statements, I indicated that the world is moving beyond
>HTTP/1 - the data shows that.

I think you suffer badly from selection bias when you make this

*Your* world may be moving in that direction, but there are huge
swatches of HTTP traffic that doesn't seem to follow you and your
little red wagon.

IMO this effect has been present throughout, where representatives
from a small number of large sites seem to confuse their site and
other sites of the kind with "the world".

The whole "Mandatory TLS" is probably where we saw this most clearly:

The advocates of "Mandatory TLS" all seem to be from sites which
require user login and which have privacy data and concerns.

But those sites are not even close to carrying 50% of the HTTP
traffic in the net today:  The majority of traffic is public
without privacy concerns that is not already revealed by the
existence of the TCP connection in the first place.

I share the anti-NSA sentiment as much as the next guy, but there
is never going to be a sane case to be made why CNN or BBC's
frontpage has to suffer the huge overhead of TLS.

Nor is there any sane architectural argument to be made why
the emergency-services web-pages of national governments
should risk DoS'ing themselves with TLS during a catastrophy.

And there is no way the porn industry is ever going to fork out
money for TLS hardware for the 30+% of the total HTTP traffic they

So yes, you may *think* you have felt the world move, but I think
it is just the faulty suspension of your office-chair.


Poul-Henning Kamp       | UNIX since Zilog Zeus 3.20
phk@FreeBSD.ORG         | TCP/IP since RFC 956
FreeBSD committer       | BSD since 4.3-tahoe    
Never attribute to malice what can adequately be explained by incompetence.

Received on Tuesday, 2 September 2014 17:38:21 UTC