Re: Techdirt: Net Neutrality Debate Again Descends Into Shouting, Farce

[David Farber <dave () farber net>]

Begin forwarded message:

From: k claffy <kc () caida org>
Date: June 26, 2007 2:53:44 AM EDT
To: Suresh Ramasubramanian <suresh () hserus net>
Cc: Dave Farber <dave () farber net>, Robert Berger <rberger () ibd com>,  
Gordon Cook <cook () cookreport com>
Subject: Re: [IP] Re: Techdirt: Net Neutrality Debate Again Descends  
Into Shouting, Farce




i'm rolling my eyes at how misquoted i am in this article.
i guess that's what i get for speaking at mach 10;
reporters just reconstruct words they didn't hear.
someday i will learn.

i never said 99% of traffic was spam, how silly.

i said, as an example, 99% of my email was spam.  (i might
have said 'probably', i was trying to make a point about how
little we understand the nature of the Internet because we
have so little data.  david moore actually analyzes (opt-in)
spam for folks at caida, so we could get rough numbers on my
email, but the spam filters don't get everything, and as suresh
points out, there is no spam bit.)

it's also not clear that we've figured out what units spam is
usefully measured in.  universities are spending 6+ figures per
year dealing w spam, which makes me think the cost of spam,
among other even costlier pollution on the network, will come up
sooner or later in the broadband debates.  as will even more
important issues, like copyright.

wrt costlier pollution, i'm surprised the thurs morning panel
of supernova wasn't covered, where they gave figures on the
underground economy, e.g., one panelist said you could buy an
entire identity (inc credit card numbers, checking account #,
drivers license, social security, etc) online for $14.  i have
no idea whether that's true, but have to admit it was more
quantitative data than anyone offered in the 700Mhz tussle
everyone considered so signficant.  fwiw, i went on right after
john kneuer, and it seemed less like hostility and more like
both sides were contributing passion to the debate rather than
quantitative empirical data.  i did warn the audience not to
yell at me like that or i'd leave the stage. :)

in all seriousness, some important decisions about our
communications fabric will be made in the next few years that
will change the rest of the century, 700MHz usage among them.
i'm still hopeful we can make them according to the best
available empirical data, including international comparisons.

k

On Sun, Jun 24, 2007 at 07:36:42PM +0530, Suresh Ramasubramanian wrote:
 |
 | "David Farber" <dave () farber net> wrote:
 |
 | >investigator KC Claffy said.
 |
 | >"We need numbers on spam, but where do you get numbers on spam  
from -
 | >anti-spam vendors. These aren't the people you want to be getting
 | >numbers from when setting policy," she said. "Let's look at what a
 | >public network is really used for. We cannot answer that. And the
 | >carriers are about to ask us to pay for traffic, 99 per cent of  
which
 | >is spam!
 |
 | Easier said than done, unfortunately.
 |
 | There are two problems with what KC asks for.  Numbers on spam  
are going
 | to differ widely depending on who is doing the measurement, and  
what the
 | target audience being measured is.  Oh, it also depends on what is
 | actually defined as spam and measured.
 |
 | Corporate mailboxes that get filtered by antispam vendors?
 | Large ISP / webmail providers like us, yahoo, aol etc?
 | IETF members who have the same email for years and get 4K++ spams  
a day?
 |
 | One set of metrics you could use is the metrics cited by the OECD
 | antispam task force -
 | http://www.oecd-antispam.org/article.php3?id_article=244
 |
 | MAAWG (http://www.maawg.org) which is an organization that has  
some of
 | the largest email providers / ISPs in the world as members (us,  
yahoo,
 | aol, roadrunner, bell canada etc) has been collecting spam metrics
 | contributed by their member ISPs, and this metrics effort was  
setup in
 | response to the OECD antispam task force's call for usable spam  
metrics.
 |
 | The MAAWG metrics cover a basic set of criteria, evolved after a  
long
 | period of discussion to reach consensus on what should be measured -
 | number of mailboxes, number of dropped connections & blocked/tagged
 | emails, number of unaltered - as in, untagged and unfiltered,  
delivered
 | emails.
 |
 | http://www.maawg.org/about/EMR has all the reports, and the latest
 | report has the summarized results of the last 4 surveys (Q4 2005  
to Q4
 | 2006)
 | http://www.maawg.org/about/MAAWGMetric_2006_3_4_report.pdf
 |
 | Of course, this measurement is by very large email providers, and  
the
 | combined number of smtp connections those providers handles runs  
into
 | the billions .. so these numbers may well vary substantially from  
the
 | numbers other reports show.
 |
 | Trying to average or otherwise normalize these reports will be  
rather
 | unproductive due to different mail flows, different filtering
 | methodologies, different measurement criteria etc.
 |
 | Measurement with statistical rigor applied is something CAIDA is  
very
 | good at, and I think this would be an excellent project for them  
to take
 | up.
 |
 | regards
 | srs
 |
 | ps: 99% of all traffic is spam - not sure, given that smtp doesnt  
even
 | register on the radar compared to other traffic (p2p, http,  
etc).  If,
 | by spam, kc meant something like "junk traffic", there are lots of
 | measurements on that try to estimate backscatter levels, traffic to
 | darknets etc, that'd get factored in ..


-------------------------------------------
Archives: http://v2.listbox.com/member/archive/247/=now
RSS Feed: http://v2.listbox.com/member/archive/rss/247/
Powered by Listbox: http://www.listbox.com