John Kozubik - john@kozubik.com - http://www.kozubik.com


US Court Bases Linking Decision on Flawed Assumption

February 10, 2002

On February 6, 2002, the US Court of Appeals in San Francisco found that,
while thumbnailed representations of copyrighted pictures placed on the
web sites of non copyright holders are protected by the "fair use"
provisions of the Copyright Act, the use of "inline linking" to display
the entire picture is not.

Normally, the code used in a web page to display an image points to the
images location on the same computer as the web page itself (or at least
one related to the web page itself).  However, it is a trivial matter
to insert an image source code into a web page that points to a
completely unrelated web site.  In this case, the end user perceives
the image on the page to come from the same place as the rest of the
page, but in fact, the web page code has instructed the web browser to
make two connections - one to the computer that the web site is
actually on, and one to a different server to get the image.

Usually this behavior is completely benign.  In many cases it is even
encouraged as an easy way to insert banner ads from advertisers into
the page of a content provider.  There is, however, room for abuse in
this core component of web and browser functionality.  For instance,
if a very heavily trafficked web site were to place an inline link to
a picture on a web site that normally receives very light traffic, the
latter would then be inundated with traffic.  Nobody new would be viewing
their actual page, but the images being served to the viewers of the
high traffic site would still represent traffic - and this traffic is
not free.  This is not necessarily a copyright issue, however, as
any picture on any web site is susceptible to this - copyrighted or
not.

My contention is that the reference to any resource placed on an
internet connected computer without access controls (such as passwords)
should be protected by the "fair use" provisions of the Copyright Act.

Note that I do not claim the possession of the resource should be
protected, but simply the reference to it.  A reference is precisely
what a link on a web page is - including an inline link of an image.

The flaw that has occurred here is the (incorrect) assumption of the
Court that any kind of linking features in a web browser are anything
but completely arbitrary behaviors.  Remember that linking in general
(a textual link that takes a user from one web site to another) is
legal, even if the destination does not like visitors to be
directed to them.  Consider also that it would be trivial for a
programmer of a web browser to alter the behavior of the browser such
that a normal text link that points to a file determined to be a
picture would result in the actual picture being displayed - exactly
as it would have been if an inline link were being used.  Further,
it would also be trivial to rewrite the browser to display the
picture in question given an english sentence on the web page that
described the address of the picture, thus prompting this (as yet)
imaginary web browser to, once again, display the picture.

This process could be taken to ever-increasing levels of absurdity.

The important point is that the behavior of any given web browser
as relates to the use of links (inline or otherwise) is in fact
completely arbitrary.  Therefore the prohibition of one kind of
linking over another not only demonstrates a failure to grasp how
the underlying technology works (and could be made to work) but also
represents a legal precedent that would require constant maintenance
and revision to remain relevant.

It is my opinion that if the Court were made aware, and would accept
the fact of the arbitrary nature of web browser behavior as
described above, it would have only two reasonable courses to
pursue.  First, it could continue down the current line, interpreting
certain types of linking as unlawful, and prepare itself to
support that decision with endless revisions as more and more
(and arbitrary) behaviors are exhibited in web browsers that
violate the spirit of the original prohibition.  Second, the
Court could place the burden of referenced copyright protection
in the hands of the copyright holder - if the material is
placed on a server connected to the Internet and made available
for use with web browsers without access control, then any
references to that material interpreted by a web browser constitute
fair use.

This "referenced copyright protection" is in contrast to "possessed
copyright protection".  In the case of Les Kelly vs. Ariba Soft Corp.,
the Plaintiff sought "referenced copyright protection" - the
plaintiff sought help from the law in guarding the references made by
another party to the copyrighted material that the plaintiff had
themselves placed on a publicly accessible computer connected to the
Internet.  If Ariba Soft Corp. had blatently downloaded the
material in question and made it available themselves, then this
would be a case of "possessed copyright protection", and my opinion
would be in favor of the plaintiff.

It is my conclusion therefore, that of the two reasonable choices
available to the Court once the arbitrary nature of web browser
behavior as relates to referencing content has been demonstrated,
the second choice is clearly superior.  Certainly the rights of
copyright holders must be protected in the case of "possessed
copyright protection" but in cases like this one where material is
merely being referenced, the burden of protection should lie on
the copyright holder(1).  Following the alternative path would
lead the Court to an increasingly absurd, growing list of
prohibited referencing behavior whose logical (and absurd)
conclusion would be the government mandate of web browser
behavior and/or ridiculous prohibitions on web page content(2).


(1) In this case, this is not much of a burden.  Not only are
    several forms of access control built into the Hyper Text
    Transfer Protocol (HTTP) that governs web page delivery,
    but certain configurations on Internet connected computers
    serving copyrighted material can disallow a range of
    referencing actions on the part of other web sites.

(2) Again, consider the possible (albeit absurd) case of a web
    browser that interpreted human-readable english sentences
    describing the location of an image on the Internet as
    instructions to display that image.