[Prev][Next][Index][Thread]
Re: Your contribution to AEC Memesis
---------------------------------------------------------
· · · · · · A E C F O R U M - "M E M E S I S" · · · · ·
· · · · · · · (http://www.aec.at/meme/symp/) · · · · · ·
---------------------------------------------------------
Joe Wheeler to Francis Heylighen, in reaction to my memesis posting:
>I have written this to you personally rather than to the memesis list, as I
>refer to specifics of your writing with which the other receipients may not
>be familiar. Do with this message as you like.
I have the impression these comments are general enough to interest other
people on the memesis list, and that is why I cc my reply (including your
original comments) to the list. Anyway, the list has been rather quiet
recently, so I suppose people won't mind the additional volume of mail.
>The review of Stock's "Metaman" book you give as the reference
>"http://tcp.ca/July94/MetamanNR.html" - it has changed (as Web addresses
>are frequently wont to do); it is now
>http://www.tcp.ca/July94/MetamanNR.html
I still find it on its old address. Probably these are two synonymous addresses.
>I deeply appreciate your making such a substantial contribution to the
>growth of knowledge on the Internet. Your scholarly comments on the AEC
>memesis listserv are a solid counterbalance to some of the more fanciful
>imaginings that lack the research background you bring to your writing.
>
>The generosity of your sharing your work on the web is also appreciated as
>it is a substantial and impressive compilation of work.
Thank you for the compliments :-)
>Working from a rather different perspective, media theory, I have come to
>many of the same conclusions as yourself. I think we may differ in some
>ways that I would like to discuss with you sometime, e.g.: (1) referring to
>the Web as a "superbrain" must be acknowledged as a metaphorical concept.
>What the Web really is, first and foremost is a medium of communication.
>One could engage in reductionism and say it was the interconnection of many
>computers facilitated by various software and physical connections. I
>suggest that analogies/metaphors like "brains" or "superorganisms" ought to
>be approached cautiously.
Of course, at present the Web is not a superbrain in any real sense, but
only in metaphor. What I argue is rather that it could *develop* into a
superbrain if a number of adjustments are made. The subtitle of my paper
"The World-wide web as a super-brain"
(http://pespmc1.vub.ac.be/papers/WWWSuperBRAIN.html), which I referred to
in my message to the memesis forum, is precisely "From metaphor to model",
meaning that it is a metaphor now but could grow into a model of how a
future web would look like. By "model" I here mean a plan or design, used
as a guide to build a more intelligent web.
>Although not a computer scientist by training, I have worked my way through
>Hinton, Rumelhart, etc. to connectionism and the present deliberations of
>the Santa Fe people (and back to Prigogine). What I think it comes down to,
>is that the "bottom up" approach of AI, while coming nowhere near
>replicating human consciousness, has been utterly pragmatic in testing
>cognitive theories (and in the process dispensing with many traditional
>philosophical conundrums). Here I would also be cautious of too much
>enthusiasm for "associationism" if it has the connotation of Hume, as that
>leads to a line of reasoning which then leads to Pavlov, Watson and
>Skinner, the dead-end of American behaviourism. I presume you are familiar
>with Karl Popper's scathing refutation of this approach, which he called
>the "Bucket Theory" of learning.
No, I don't know Popper's criticism, but am familiar with other criticisms
of behaviorism and associationism. What I believe to be lacking is the
connection between the associationist or connectionist, subsymbolic level
of cognition, and the cognitivist, symbolic level. What we need is a model
of how higher order concepts emerge out of networks of connections. My own
work tries to address this issue to some degree, but I don't have a
satisfactory solution yet. In my paper mentioned above, I explicitly try to
connect the associationist level ("learning") with the cognitivist
("thinking"), by introducing software agents as symbolic "thoughts"
travelling through the associationist network.
>Don't get me wrong, I'm not challenging your basic work, it is substantial
>and thoughtful. I have nothing yet in publishable form to match it. I
>simply have a few ideas I think are worth considering. For example I don't
>entirely agree with Popper (or Campbell), particularly in Popper's notion
>of demarcation. You provide an idealized view of the WWW in your work. The
>actual content of the WWW is rife with "folk psychology", raw sentiments,
>weak justifications, etc. I suggest that what the Internet shows us is how
>people really think and the way knowledge really grows; there is no
>demarcation as Popper asserts, there is a continuum from the most banal to
>the most thoughtful, but there are characteristics which are common across
>the spectrum: the natural tendency in thought is not inductive as Hume
>would have us think, but abductive, a contextual guessing that is then
>justified after the fact (it is only more obscrured in scholarly work).
I suppose that by "demarcation" you mean the strict boundary between
justified, scientific knowledge, and all the rest. I would agree with you
that that is at best an idealization. "Scientific" knowledge also to a
large degree depends on intuition and context, while intuition or folk
knowledge often is based on long experience of things that work, and is in
that sense often quite reliable. But it does make sense to try to test and
validate knowledge as much as possible. This makes it possible to
distinguish more from less reliable ideas or beliefs, though there is of
course no absolute boundary between reliable and non-reliable.
The fact that there is such a continuum, and that most ideas in society are
poorly justified does not mean that good ideas don't spread. There is a
kind of natural selection which makes that wrong, useless or low quality
ideas sooner or later are forgotten, while the really worthwhile ideas
(arithmetic, Shakespeare, the wheel, physics, Mozart, computing, etc.)
survive. Though fanciful ideas may be very successful at the moment, they
are much less stable than deep, well-tested ideas that provide insight in
important problems.
It is the same with the web. Though there is a lot of low quality content,
it is the best web pages and best web sites that will eventually get linked
to and used most often. The rest may be much larger in actual number, but
if nobody makes links to that stuff, it simply won't be read by many
people. Even the things that are low quality on content but popular for
other reasons (say, galleries of supermodels, tabloid-like news, etc.) will
not have the same lasting impact as sound, insightful information. The
"learning" system I propose for the web should accelerate this natural
selection of the more useful knowledge.
>The proliferation of memes and their quality have no necessary relationship
>- if anything, some of the more pernicious memes (e.g. racism, creationism,
>imperialism, chauvinism, survivalism, various religious fundamentalisms)
>have exceptionally strong replicative power and resist refutation. I feel
>that the associative mechanisms of hypertext links via software agents
>which you speculate about in your paper could as easily reinforce
>pernicious memes as lead to a growth of objective knowledge, a kind of
>"squirrel cage" effect.
I don't know about the "squirrel cage" effect.
Of course, the system I propose is not magic, and won't solve all
problems. It is likely to accelerate all evolutions, positive as well as
negative. However, I am intrinsically optimistic about the direction of
evolution, and I do believe that "pernicious memes" will lose out in the
long run, even though they may be very succesful in the short term. It is
our responsibility to criticize all dangerous ideas, so that other people
are fully aware of all the negative effects they may have and so that their
superficial attraction is counteracted.
Of course, many people on this list will think that the ideas I propose
here such as *natural selection of ideas* are already dangerous. However,
it is their task to convince other people by rational arguments, and not by
gratuitous comparisons with Hitler, Stalin and Social Darwinism, that there
is indeed a danger involved.
>Those of us in academia too easily forget that the vast majority of
>participants on global networks are not particularly thoughtful
>individuals, but often alienated, bigoted and looking for distraction and
>confirmation of their prejudices. I would like to think that we are going
>through an initial reaction stage, and that content will mature as people
>tire of the mindless amusements offered by multimedia, 'virtual reality',
>'3-D' and all the rest of the garishness. If TV is any precedent, they may
>not. We have to face that it is this sort of people who will be paying the
>money for Internet access, and will in large part determine the content.
I'd rather compare the Web to the medium of books than to TV. Since there
are many more books (or websites) than there are TV stations, there is much
less need to cater to the largest common denominator of the mass audience.
So there are plenty of difficult, high quality academic books published, as
well as mass market novels. After ten years, few of the initial bestsellers
remain, and it are usually only the really good books that are remembered
beyond that period. For the web, it is easer to start up a web server than
to publish a book, so there is less initial selection of material by
editors. But the natural selection processes I sketched earlier will make
sure that after the network has settled bad quality material will simply be
consulted rarely.
>There is an evolutionary selection process at work, but I'm not entirely
>sure it will proceed to the "Superbrain" that you envision, perhaps equally
>likely a "Mediocre-brain." Evolution itself operates in a sub-optimal way,
>if you agree with the "Panda's Thumb" argument of S.J. Gould (Bully for
>Brontosaurus, 1991). Sub-optimality is also a freqent theme of the Santa Fe
>Institute writers.
Optimality is only achievable if you have a complete model of all possible
alternatives and their effects on the value you try to maximize. Evolution
does not have such a complete information, and neither do we. So
suboptimality is a part of life. Evolution does have the capacity to
improve existing designs, though. "Bettering" is a more realistic strategy
than "optimizing".
>The one saving grace is, the history of all media shows that there is a
>tendency to prefer abstraction and over time and that trial and error and
>selection do eventually favour the more thoughtful and enduring memes which
>support human function. There is, however, no real precedent for the
>current state of affairs in the world or this new medium. It is an open
>question whether we are ascending to a superbrain or decending to a new
>dark ages of electronically engendered ignorance.
I tend to favor the first option, as you already know. There never is a
complete precedent for anything in evolution, yet evolution still manages
to overcome practically all obstacles put on its way.
>><><><><><><><><><><><><><><><><><><><><><><><><><><><><><><><><><><><
>>< J O E D. W H E E L E R ><
>>< snail mail: #414-310 Robert St. Tel and Fax: (604) 388-5712 ><
>>< Victoria, B.C. e-mail: jwheeler@pinc.com ><
>>< Canada V9A 3Z4 ><
>><><><><><><><><><><><><><><><><><><><><><><><><><><><><><><><><><><><
________________________________________________________________________
Dr. Francis Heylighen, Systems Researcher fheyligh@vnet3.vub.ac.be
PESP, Free University of Brussels, Pleinlaan 2, B-1050 Brussels, Belgium
Tel +32-2-6292525; Fax +32-2-6292489; http://pespmc1.vub.ac.be/HEYL.html
----------------------------------------------------------
· · · · to subscribe to the Forum just mail to · · · · ·
· · memesis-request@aec.at (message text 'subscribe') · ·
· · · · · · · send messages to memesis@aec.at · · · · · ·
----------------------------------------------------------
