chain letter petitions

Phil Agre (pagre who-is-at
Fri, 5 Apr 1996 06:58:11 -0800 (PST)

I've enclosed a bit of text I wrote for my newsletter in response to the
chain-letter petition that was posted here earlier, which I regard as
very misguided. [Sorry for the early version of this that went out to
this list by mistake.]

Against Chain-Letter Petitions on the Internet

Phil Agre
pagre who-is-at
April 1996

Feel free to circulate this article for any noncommercial purpose.

I am worried about chain letters on the Internet, and particularly about
a type of political action alert that is structured as a chain-letter
petition. These petitions typically encourage you to sign your name
at the bottom, pass it along to many other people, and mail it to some
address if your signature numbers some multiple of, say, 10. Recently
this genus of chain letter has multiplied greatly. Most of the ones
I have seen have been for liberal causes such as public broadcasting or
gay rights. Although the Internet can be a powerful means of political
expression, I will argue that these petitions are far worse than useless.

Most of them, for one thing, have been very badly designed. They usually
have no cut-off date, source of background information, signature from
the organization or individual who is sponsoring the alert, or instruction
to post the alert only where appropriate. As a result, these alerts
have caused a lot of disruption and annoyance all around the net, and it
would not surprise me if the negative sentiment they cause outweighs the
positive benefit of the actions they encourage.

This type of chain-letter petition can also counterproductively annoy
the legislative staffers and other lowly individuals who are supposed
to open the petitions when they arrive in the mail. The problem lies
in the mathematics of Internet chain letters. The chain-letter format
creates a branching tree, and the petitions that arrive in the staffer's
mail represent paths from the root of the tree to some vertex whose
depth is a multiple of 10. This means that some signatures will appear
on large numbers of petitions. It also means that some signatures will
never arrive in the staffer's office at all, namely those which appear
on branches that die out before ever reaching the next multiple of 10 in
depth. It is quite possible that such a petition might generate thousands
of signatures that never reach their destination. Early on, the petition
will get forwarded to most of the relevant mailing lists, so that the
branching factor of the tree will be quite high. The result might be
hundreds or thousands of separate petitions, one for every person who
adds their signature. Every single one of those petitions will die out,
causing that signature to go to waste, unless it finds its way through
some chain of several more signatures and reaches the next multiple of 10.

This, however, is unlikely. The underlying problem is not the limited
number of individuals willing to sign the petition but rather the limited
number of venues in which those people can be found. Once an instance of
a given petition has been posted to a given mailing list, it is unlikely
that further instances of that petition will be posted to the same list.
Let's say, for example, that ten thousand individuals would be motivated
to sign the petition, but that the vast majority of those people subscribe
to one of five mailing lists. It seems likely that an instance of the
petition would appear on each of those five lists very quickly, resulting
in ten thousand signatures, each giving rise to its own branch of the
tree. Since the vast majority of the resulting petitions would not find
its way to any further, additional venue in which likely signatories could
be found, the result would be that very few if any of the signed petitions
would ever be received in a legislative staffer's office, and those that
did arrive would include a vanishing proportion of the actual signatures,
consisting mostly of the same few signatures repeated over and over.

Now in practice one *does* see instances of such petitions with dozens of
names on them. These are the lucky few survivors that have found their
way around the net; looking at the patterns of signatures, they seem to
have passed primarily through networks of individuals with small branching
factors. If these petitions were much better designed, and if they
carried instructions encouraging them to be passed to individuals and
not to mailing lists, then it is possible that they would produce more
efficient results. But even then, legislative staffers would receive
numerous duplicate signatures.

It would be much better, in my opinion, to circulate well-designed action
alerts that encourage people to take a more substantive action in support
of the cause, for example by supplying them with the facts and arguments
they need to write letters to the editor, call legislative offices on the
phone, organize meetings on the topic in their community, and so on. Many
fewer people may actually take these actions, but they will have a much
greater impact than simply forwarding a lot of e-mail down the rabbitholes
of the Internet.

For some suggestions about designing political action alerts that work,
send mail to rre-request who-is-at with "Subject: archive send
action-alerts". The very best action alerts are the ones issued by the
Voters' Telecom Watch, vtw who-is-at,