Psybernet
Posted to the Psybernet mailing List on 26 December 1996.
"I Like This!"
Dan Randow and I go rock climbing together, and we do a lot of what
we call Geek Talk, fantasising about the possibility of the media, its
future and impact. Early this year (1996) Dan wanted an "I Like
This!" button on his screen, leading to an exchange of recommendations
with others of like minds. My contribution to this invention was
to say that it was not science fiction, but that we could implement this
with a database and a web server.
I mocked up some I
Like This! pages and Dan made a database in Access.
To actually make it all work we would need more expertise and more *time*,
already the talking and the planning and the emailing enthusiasm was eating
away the nights, driving us crazy. At that point about April, 1996
we discovered Firefly.
MIT and Agents Inc. had made "I Like This!" a reality,
with another name. They used the main idea we had of making recommendations
based on the choices people made who liked the things you like. They
have a great site, and grand visions, they predict accurately what music
I like, and have come up with artists and albums I have not heard of, just
to my taste! The site encourages connection between people, and offers
everyone a home page. On those pages you can see what items you like
in common with that person. This is a qualitative leap in communication
and the e media, still not fully realised, and certainly under exposed
or recognised. Not understood . . . notice Jaron Lanier's critique
of agents in Wired (Ideas Fortes Wired 4.11), where he seems
to miss the importance of *synthetic intelligence* a word used by Alexander
Chislenko for this sort of collaborative filtering. It is quite different
from agents that work on features of an item, as opposed to who likes it!
To proceed with our business plan seemed out of the question.
We felt disappointed and also elated. The end of plans to implement
but the joy of knowing we had anticipated what we predict is one of the
most important developments on the web.
The Internet, enables us to network *people* not computers. It
is about relationships not information. We see the value in automating
the fine tuned linking of people, not artificial intelligence, but enhanced
sociometry. Sociometry is the word used by Moreno
for the measuring of patterns of relationship in groups. (See sociometry
thread in the Psybernet mailing list.) Sociometry in Moreno's methods
is the foundation of personal and social revolution, Dan and I have experience
in these methods, we come at this work familiar with this radical tradition.
Reflection
We continued our Geek Talk and used words from sociometry that were
useful in our discussions:
Tele: The feeling --- positive, neutral or negative
--- that a person projects towards something or someone.
Criteria. Patterns of sociometric links are always with
respect to a criteria, we do not simply Like! someone, we like to go to
movies with them, or play golf, but not necessarily both.
We also came up with new words:
Taste-buddy. Coy, but describes the concept
well, someone who shares your taste in some way. Linking with them
increases your ability to sense what you like. They are a taste bud
in both senses of the word.
Taste pool: People who share taste with respect to
a criteria. I have a sense of being able to represent this graphically
in real time as people join and leave these pools, I envisage waves of
fractal patterns. I like the link with the notion of a "gene
pool" it gives the sense that these pools have a life of their own
and one can be in small ones, almost extinct, as well as overblown ones
that will never die -- e.g. the people who like the "Star Wars Trilogy"
-- but on the basis of which we learn very little else.
The Evolving Features of our FantasyWare
During the year we identified more aspects of our fantasy dream, "I
Like This!" it would have some great features:
- Focus on getting *any* recommendation, and being able to recommend
*anything*. Work around the need to store items in a database.
( Fire fly now works with movies, Music and Web Sites)
- Items could be of many *types* e.g. email, URLs, newsgroup
"news" (include a use by date)
- Interest categories could be named and a way of customising them developed.
E.g.: politics Internet psychology
- Having the option of recommendations come to the desktop, rather than
having to navigate to the site.
These concepts are important, Firefly seems to require that one rates
an item in their database. Why not focus on the relationship between
the people? Develop a way of sending, storing and processing relationship
information. Develop an ontology to define interconnection.
Items do not need to be in the database in our model. All that is
needed is the relationship between people. Person A likes what B
likes etc. Send me type <email> from <Mary> in category
<computers.psychology>.
A revolution in News
If the basis for recommendation was the sociometric patterns, then
one could recommend anything, via an url or an email. Using these
methods there could be user configurable types of information. Ones
know as *News* means the service would function as a news caster, like
PointCast, but flexibly, simply delivering items of interest to ones taste
buddies, no enforced ads, no need to be tied to news media barons, journalists.
A new form of discussion group
With items known as *Posts* the service could function as a type of
discussion group, in addition to Mailing lists and newsgroups. If
one indicates a willing ness to receive posts from peopl one tates highly
in a taste pool, that could lead to a group that was conscious of its own
existance, and boundaries, yet totally open to others who shared the interest:
Imagine, finding one belonged to a group of three who shared an interest
in < building , dwellings, straw, alternative >
Revolution in shopping
Send me type <forsale> <urls> about <widgets> that
I will <love>. Will have big impact on advertising. Just
post the information on the web and it will get sent to whoever wants it.
We thought of one more possibility:
- Having the option of storing the data on users desktops, so
the system functioned like a web over all the participating computers. This
is the Intranet friendly approach.
It was this latter idea that got us motivated again. Let's
build software puts the button on our screen, so we can rate our reading,
viewing & listening, and that will email it out like a chain letter.
In November We got the I Like This fever again . . . but not for long .
. .
Obstacles
- Still no time.
- Need to employ programmer.
- Need to get grant and finance.
- Need business plan
- Need time!
Inventors know that an idea is not worth much, making it happen is
the hard part. We are not a software company, and we like our jobs
to much to set one up.
Intel's CoolURL
Then we saw that Intel has something in the pipeline called CoolUrl, with
a Cool! button, cool indeed. Our distributed model was also being
developed much as we had envisaged it, once again we were behind where
someone else was leading.
Back to Basics
We are no longer looking to build software. Learning that we are perhaps
best to stick with this as our Geek Talk hobby, and to write about our
ideas, to explore their implications especially in the psychological aspects
of the Internet . . . in short back to Psybernet!
My enthusiasm for these things is high, and the thought I don't need
to *do* anything, other than write about it, is great. Already it
has led to some interesting discussion with people who are writing software
that focuses on *networking people*. Dan has been busy writing.
In the Psybernet mailing list I think the role is to experience the implications
on our psyche, their place in the battle for soul.
Walter