CWD--PORN-O-RAMA [part 1]

[David Farber <farber () central cis upenn edu>]

CyberWire Dispatch // Copyright (c) 1995 //


Jacking in from the "Point-Five Percent Solution" Port:


Washington, DC -- Time magazine's credibility is hemorrhaging.


The magazine's recent "Cyberporn" cover story has ignited a fire storm of
criticism owing to its overblown coverage of a statistically
inconsequential study,  written by a university undergraduate.


Time's story is being assailed as "reckless," "shoddy work" and an outright
"fraud" by academics and civil liberties groups.


Martin Rimm, who as an electrical engineering major at Carnegie Mellon
University took 18 months to complete the study,  says 90% of the criticism
"is junk."


The writer of the Time story,  Philip Elmer-DeWitt, characterized the
attacks as "a lot of rhetoric from a professional lobbyist and a professor
who called it reckless and criminal before she had read" the study.


Besides the pejoritives used to question how academically rigorous the Rimm
study is, Time's critics also are chaffing at the veil of secrecy that has
surrounded the study.


Time, the Georgetown Law Review (where the study was formally published,
despite the fact that it only deals with points of law inside footnotes)
and ABC's Nightline, in a kind of media collusion, refused to let anyone
outside those organizaions do an independent review of the study before
publication.  Each cited secrecy and a prior arrangement with Rimm as the
reason.


At least a week before publication, Time magazine was alerted to several
potential problems in the study's methodology.  "I rasied what I thought
were several red flags," said Donna Hoffman, an associate professor of
management at the Owen School at Vanderbilt, and one of the most respected
researchers on Net access issues.  "Those concerns were apparently
ignored," she said.


Further, at least two legal experts, Mike Goodwin of the Electronic
Frontier Foundation and Danny Weitzner of the Center for Democracy and
Technology, were refused access to the study, despite being asked by Rimm
to review the report's legal footnotes.  Both declined to provide any legal
analysis, issuing warnings that such analysis was impossible without seeing
the footnotes in context.


Time magazine, aware of all this, ran its story without noting any of the
criticism.




The .5 Percent Solution
====================


One of the most egregious spin elements that Time used on the story was
hyping Rimm's claim that 83.5% of all images on Use net are "pornographic."


That 83.5% figure has already been sized on in by some members of Congress
looking to bludgeon the First Amendment by placing unconstitutional
constraints on Internet content.  This figure is likely to become a
rallying cry of the First Amendment impaired;  it has been trumpeted in at
least one Senate floor speec.


Small problem:  That figure -- and the study which ejaculated its results
to a select media group under the cloak of secrecy --  is severely flawed,
according to several academics and civil liberties groups that have since
obtained and analyzed a copy.


By Rimm' own admission, the 83.5% figure is derived from a seven day time
slice of the postings to only 17 of some 32 Usenet groups which typically
carry image files.  Usenet is comprised of thousands of newsgroups, the
vast majority of which are text based.


Further, Rimm's own figures show that his so-called "pornographic" images
comprise merely ONE-HALF OF ONE PERCENT (.5) of all Internet traffic.


Time reporter, Philip Elmer-DeWitt did report this fact.  Sort of.  But
readers of the Time story have to wade nearly 1,000 words into the story
before stumbling across this passage: "As the Carnegie Mellon study is
careful to point out, pornographic image files... represent only about 3
percent of all the messages on the Usenet newsgroups, while the Usenet
itself represents only 11.5 percent of the traffic on the Internet."


DeWitt would later claim during an online discussion on the WELL that he
didn't finish the math, citing the .5% figure, because readers tend to get
lost when more than two figures are cranked into a paragraph.  (See, Time
takes care of you!)


Cooking the Books
================


To juice the coverage, Time also cited that the study had "surveyed 917,410
sexually explicit pictures, descriptions, short stories and film clips."
These files, however, were dredged up from adult BBS systems, not Internet
newsgroups, a point that is not entirely clear when reading the article.


The 917k figure is further misleading because even Rimm admits in his paper
that he winnowed out so many files that his analysis is based on
merely 294,114 files.  And that STILL doesn't tell the whole story.


To analyze such a huge number of files, by visually verifying that
something called "Naked Bitch with Mardi Gras Beads" is actually a woman
and not a hoaxed picture of a female dog (which actually happened), would
have taken years.  Instead, Rimm's analysis is based overwhelmingly on file
*descriptions* only, not actual viewing, using an artificial intelligence
program.


Yet a reader of Time's cover story gets none of this analysis.


Walking Back the Cat
===================


How did  major magazie like Time get roped into reporting as "exhaustive"
such an apparently flawed document?  It was likely a combination of several
factors, including errors in judgment, fatigue and the need to scoop the
competition on a hot button issue of the day.


The intelligence community often debriefs its operations through an
exercise called "walking back the cat."  During this exercise, the major
players are gathered and the mission is examined in detail.


While not all the information surrounding the events that led up to the
Time cover story are known, let's walk back the cat on what we do know:


Early 1994:


Rimm assembles his "research team" to begin trolling some 68 adult BBSs.
His team is instructed to try and obtain as much as possible data on the
BBS customers through a kind of "social engineering."


Dispatch interviewed 15 major adult BBS operators to ask about their
participation with Rimm.  None of them remember ever having spoken to Rimm
or a member of his research team about the study.


Dispatch asked Rimm:  "Did your team go uncover, as it were, when geting
permission from these [BBS operators] to use their information?"  He
replied only:  "Discrete, ain't we?"


When asked how he was able to obtain detailed customer profiles from
usually skeptical operators of adult BBSs he says:  "If you were a
pornographer, and you don't have fancy computers or Ph.D. statisticians to
assist you, wouldn't you be just a wee bit curious to see how you could
adjust your inventories to better serve your clientele? Wouldn't you want
to know that maybe you should decrease the number of oral sex images and
increase the number of bondage images? Wouldn't you want someone to analyze
your logfiles to better serve the tastes of each of your customers?


October 1994:


Eight months before the "exclusive first look" that Time touts about its
story on Rimm's findings, "people involved in the study were pitching it to
the media," reports  Michael C. Berch, editor of INFOBAHN magazine, in a
posting to the alt.internet.media-coverage newsgroup.


Berch said he took a flyer on the story because he had "other coverage of
Internet erotica" in the works.


Rimm says he has no knowledge of the exclusive offered to Infobahn or any
other publication before shopping it to Time.


During this time,  Rimm also shops a draft of his study to the CMU
administration, according to a Time magazine report last year.  Shocked at
the findings, the school scurries to implement a full scale censorship of
alt.sex groups from the school's Usenet feed.


November 1994:


All hell breaks loose. Word gets out that Carnegie Mellon University has
decided to make public its policy to censor all Alt.Sex newsgroups from
flowing into its computers.


The ensuing turmoil surrounding the CMU decision draws media attention and
Time is there.


Time reporter DeWitt hooks up with Rimm and using sparse stats drawn from
the Rimm paper, he writes in the November 21, 1994 a story headlined
"Censoring Cyberspace."


In the story he refers to Rimm as only a "research associate."  DeWitt's
story says the CMU administration acted on a draft of Rimm's study "about
to be released."  In actually, the study doesn't see the light of day until
some seven months later and only then under a secrecy agreement between
Time and Georgetown Law Review.


DeWitt writes in that November article that Rimm has "put together a
picture collection  that rivaled Bob Guccione's (917,410 in all)."


In reality, Rimm had few, if any, actual images.  The 917k figure then, as
now, refers only to descriptions of images.  And when the data was finally
washed, only some 214k of those image *descriptions* were valid.


[continued...]