12.0335 WWW manipulations

Humanist Discussion Group (humanist@kcl.ac.uk)
Fri, 15 Jan 1999 18:55:30 +0000 (BST)

Humanist Discussion Group, Vol. 12, No. 335.
Centre for Computing in the Humanities, King's College London

[1] From: "Gary W. Shawver" <gshawver@chass.utoronto.ca> (39)
Subject: Re: WWW manipulations and Meta-jacking

[2] From: Toby Burrows <tburrows@library.uwa.edu.au> (31)
Subject: Re: 12.0328 WWW manipulations

[3] From: Peter Evans <evans@i.hosei.ac.jp> (24)
Subject: WWW manipulations? (12.0328)

Date: Sun, 10 Jan 1999 14:45:15 -0500
From: "Gary W. Shawver" <gshawver@chass.utoronto.ca>
Subject: Re: WWW manipulations and Meta-jacking

> An attentive Humanist, attending a dinner-party recently, heard about a
> technique by which clever people conspire to raise the probability that
> Internet search-engines will place their pages at the top of the list for
> any search on the relevant key-words. The example he gave was for "Darwin or
> evolution", which in the results of a Web crawler produced a list the first
> dozen or so of which were references to creationists' pages. I just tried a
> search on these terms in AltaVista and did not get the same results, but
> this hardly matters for the point to be made. I would assume that the
> technique in question is to use the META element to set multiple instances
> of the relevant keywords in order to run up a high score on the number of
> hits/page for these keywords. I've seen this done before.
> A clever person certainly could get around any simple checks based on the
> contents of the META element or proximity.
> What are the implications for one's reliance on the Internet as a source of
> information? It certainly doesn't take much of an imagination to see how an
> unscrupulous person could in effect drown out a competing view of some
> subject, such as evolution, by making sure that the first 100 hits would
> always be to his or her own.

Perhaps the most infamous, and most thorough-going, manipulation of this
search-engine vulnerability was the Web-page for the Heaven's Gate cult. They
not only included hundreds of repetitions of the most commonly-requested
words (sex, etc.) in the META element, but also repeated these words in the
body of the page using 0-sized font. In both cases, the words were invisible
yet exerted a powerful influence on search engines. Apparently, this practice
(shall we call it "meta-spoofing") is so common that it has generated another
practice, meta-jacking, the copy-paste of such code from one page to another.
Has anyone else heard anything more on the vocabulary of this practice?

> Is this a real worry? If so, what do we do about it?

Actually, I just glad to see SOMEONE using the tag. Many sites, created by
people whom one presumes would know better, contain no hint of content in the
META tag and seem to rely on the vagaries of search-engine algorithms and the
gritty determination of its potential audience to be found by same. The
practice of meta-spoofing at least implies an awareness of the need to be
and an understanding of how search-engines work. I only wish I saw it applied
more often, and more honestly. We should not abandon speech for liars.


Date: Mon, 11 Jan 1999 10:59:58 +0800
From: Toby Burrows <tburrows@library.uwa.edu.au>
Subject: Re: 12.0328 WWW manipulations

Dear Colleagues,

Stephen Pinfield ("Cure for hype on the cards" Times Higher Education
Supplement 9 October 1998, 12-13) also drew attention to this problem of
"Web page stuffing".

Doctoring the <META> tag is only one way of doing it; tinkering with the
text of the page is another. Pinfield claims that some Web pages have these
artificial keywords written in the same colour as the background of the page
- the electronic equivalent of invisible ink, perhaps?

As Pinfield says: "The problem with self-created metadata is that it can be
cynically fixed."

We need people other than the authors compiling metadata for Web pages. We
don't get authors to create their own catalogue entries for their books, or
their own database records for their journal articles...

We should avoid using Web search engines which are compiled by robots and
have no powers of intellectual discrimination. We should use searchable
databases which contain entries selected on consistent principles. The
results of a search won't be nearly as extensive, but at least the quality
of the entries will be reasonably consistent.

Toby Burrows

+ + + + + + + + + + + + + + + + + + + + + + + + + + +

Dr Toby Burrows
Principal Librarian, The Scholars' Centre
The University of Western Australia Library
Nedlands 6907, Western Australia

E-mail: tburrows@library.uwa.edu.au
Facsimile: + 61 8 9380 1128
Telephone: + 61 8 9380 2358
Web: http://docker.library.uwa.edu.au/~tburrows/

Date: Mon, 11 Jan 1999 17:17:22 +0900
From: Peter Evans <evans@i.hosei.ac.jp>
Subject: WWW manipulations? (12.0328)

Willard McCarty wonders about how it is that, for example, creationists can
boost their twaddle to the top of a search engine's hit list for "Darwin or
evolution", and says:

> I would assume that the technique in question is to use
> the META element to set multiple instances of the relevant
> keywords in order to run up a high score on the number of
> hits/page for these keywords.

Er, no. Well, possibly, but if this worked, it would indicate that the
designers of the search engine hadn't taken even the simplest of precautions
against what might be called search engine spamming.

For a brief summary of how search engines rank the pages they find, see
<http://searchenginewatch.com/webmasters/rank.html>; and for a lot more on
closely related subjects, spend time looking on other pages of
<http://searchenginewatch.com/>. (I could attempt to summarize what's said
there, but there seems little point as it's said so well.)

There is certainly a danger that a fanatical and energetic faction can go some
way to elbowing out alternative, intelligent points of view from the top of
lists. I can't offer any solution, but suggest that the danger isn't as great
as it might have been back at the genesis of some horror stories that still
make the rounds of dinner parties.
Peter Evans evans@i.hosei.ac.jp

Humanist Discussion Group
Information at <http://www.kcl.ac.uk/humanities/cch/humanist/>