Message-Id: <199706081343.JAA07774@mathlab.math.ufl.edu>
To: memetics@mmu.ac.uk
Subject: An experiment in memetics
Date: Sun, 08 Jun 1997 09:43:27 -0400
From: Omar de la Cruz <odlc@math.ufl.edu>
In my previous post, I wrote:
>So, when I claim that memetic evolution has a lamarckian component, I don't
>have to prove just that there is change from the expression of the replicator
>back into the replicator (that would be easy). What needs to be proven is
>that the change has direction (is not random). As I argued in the previous
>post, the direction does not need to agree with the direction of "adaptation",
>we just need non-randomness. And as I claimed in my first post about
>Lamarckism, this can be studied experimentally. Please see my next post.
The goal of the experiment is to check if the variation between replication
events for a meme can be non-random, and if this particular non-randomnes
can be atributed causally to the expression of the meme.
Consider a meme easy to handle; especifically, a joke. We choose a random set
of individuals, to which this joke will be told. After some reasonable amount
of time, this people will be asked the following set of questions:
1) Did you tell the joke to someone else? How many people? Who?
2) Did you modify the joke? Tell us the joke in the modified form, or forms,
trying to be as exact as possible.
3) If you modified it, why did you do it?
4) How hard did people laugh?
Now we track down the people in the list of names that we get from answers to
question (1). These people will be asked to answer the same questionnaire.
There we get information about the next generation of hosts of the joke, and
this process is repeated as many times as possible.
by checking how well the new versions of the joke spread. We can then compare with
the (conscious) goals of the host that changed the joke before spreading it,
and also we can somehow assess the "quality" of the new versions, using some set
of standards. However, the first measure is probably the most significative.
Why the need of the experiment? Well, it could happen that when one host tells
the joke to two new hosts, one of the hosts has a talent for comedy, so he
changes the joke and improves it (in the sense that the new hosts will find the
joke so funny that they will tell it to more people). But the second person
could have a bad talent for comedy, so when he modifies the joke, he ruins it
(by making it too long, or giving away the punch line), so that the new hosts of
his version will consider it bad, and not tell it to anyone else. In this case,
the variation would be equivalent to a random variation; if this always happens,
the bad variations cancelling the good variations, then there would be no
noticeable lamarckian effect. The experiment would be trying to find a case
where the variations don't cancell out.
Of course, to get significative data, this must be carried on a big enough scale.
How big? Well, that's a matter of statistics, wich I don't like ;) However,
I think it is in no way out of the possibilities of a social science field
researcher with some resources (that is, graduate slave labor force! ;) )
One modification to this setup could be to use some engineered "urban legend",
some false story appealing enough to make it be told once and again. But
there are some ethical issues: we would be literally spreading lies. And,
for what I have seen about urban legends in the Web, they seem to cause harm
to someone one way or another. In any case, the alternative is to use
already collected data about naturally occurring legends, since it seems that
there is quite some volume of data. It remains to be seen if the
relevant information (similar to the questionnaire above) was collected.
If a non-randomness in the variation is observed, caused by the expression of the
meme, in this case, how well it performs (if it makes people laugh, or other
observable effects), then, for me, the existence of a Lamarckian component in
memetics would be proved (that doesn't mean that it is a universal feature of
memetics, of course, but it proves that Lamarckism is a force to be reckoned).
However, if the experiment
fails to show such a non-randomness, that proves nothing, since lamarckian effects
could be found somewhere else (in fact, non-Lamarckism cannot be *proved*
experimentally; even in biologic evolution the proof is empirical). But enough
experimental data could lead to the empirical conclusion that memetic evolution
has no lamarckian component at all.
A final note: the people that don't agree with my arguments about Lamarckism in
the previous posts (because of imprecision of the terms used, or other reason)
will find that this experiment would prove nothing in any case. Well, that's
O.K., but I think that the empirical data collected would be interesting, anyway.
So.... is there anyone out there willing to do the dirty work?
Omar De la Cruz C.
===============================================================
This was distributed via the memetics list associated with the
Journal of Memetics - Evolutionary Models of Information Transmission
For information about the journal and the list (e.g. unsubscribing)
see: http://www.cpm.mmu.ac.uk/jom-emit