Inkwell: Authors and Artists

    
ckridge, we're at odds on this one, but in an interesting way. I agree
with you all the way up until your last paragraph. There is no neutral
classification scheme. The world consists of endless attributes by
which things can be said to be alike. Which attributes matter to us
depends upon our interests, our project and our background. There is no
one way the world is. Therefore, there is no one way to classify it.

Cue the entrance of the miscellaneous! Instead of having to settle on
a single sort, or even nominate one sort as the "major" or "real" one,
in the digital world we can have as many as we want, simultaneously.
Yours can put Che with the revolutionaries and mine can put him with
the dirty Commies. Yours can put acupuncture under healing arts and
mine can put it under careers to be investigated. You can put "Sounds
of Silence" on a playlist of "Favorite Folk Rock" and I can put it on
one called "Embarrassing Poetry." While some classifications may be
wrong, there isn't a single right one, and now we don't have to act as
if there is. 

(All classifications above for illustrative purposes only.)
  

    
> we need to develop new principles of organization.

Do you have any in mind?
  

    
Thanks for the set-up, jonl!

My book talks about four, although, as you'll see, they're not
principles so much as broad statements of the nature of the change. So,
with that caveat, here they are:

1. A leaf now can go on many branches. A physical book has to go one a
single shelf, forcing us to pick one main way of categorizing it. The
digital book can go in as many categories as we want. Count tags as
categories, and it may go in _lots_ of categories.

2. Messiness is a virtue. The digital mess gets enriched as more
content is added and more connections are made. Keeping things neat
works in the real world because it makes stuff findable by manifesting
their order, but digitally neatening can reduce the pile's potential
for significance.

3. Everything is metadata. Instead of having to reduce information to
what fits on a catalog card or a folder label, we can use the digital
contents of the book or folder as search terms, as well as using every
item linked to it directly or by inference, as well as using every
classification layered on top of it... 

4. The owners of the information no longer own the organization of it.
The users do. I'm not talking about legal ownership here, which is a
different question. Rather, the users get to sort and order the way
they want, sometimes despite what the information providers want.

Put 'em together and you get a strategy that says it's better to
include and postpone...include everything you can and postpone the
moment when the classification happens until the user decides how she
wants to make her way through the information.
  

    
By the way, Tom Matrullo has some beautiful examples of how the line
between data and metadata is being eroded: http://tinyurl.com/2gze2u
  

    
Who controls syndetic structure?

Synedetic structure is the set of relationships between tags. The
"See" and "See also" references in a thesaurus, index, or catalog are
syndetic structure. Subordination of one tag under another, with "See
also" references running down but not up the hierarchy is another kind
of syndetic structure. Scope and disambiguation notes explaining how
similar tags are distinguished from one another are still another kind.

So long as people are just slapping tags on things, you have happy
anarchy, and I can tag the Armenian genocide "myth" and you "history"
and we will both be happy. But if I can get control of the syndetic
structure, I can make sure that searches for information on the
Armenian genocide also pull up information on Atlantis and visitors
from space who built the pyramids. This will color the topic by
association.
  

    
ckridge, anyone can control the synedetic structure, can't she? (Never
heard the term before. Thanks.) So, flickr controls its clustering
algorithm that pulls together photos of the island of Capri and of the
Ford Capri. It's easy to imagine - through bad programming, or through
"flickr bombing" - the "terrorist" cluster including photos of Bush or
(if surrealists invaded the site) Bert and Ernie. (Yes, in the back of
my mind, I'm thinking about the Osama-Bert photos from a a few years
ago.) But someone else could derive and publish a different synedetic
structure, which is what happens in mashups. Likewise, the American
Medical Association could compile its own encyclopedia by picking the
versions of articles in Wikipedia that pass its muster. The American
Alternative Medicines Council could so likewise.

The miscellaneousness of information affords multiple, simultaneous
synedetic structures. That means that we don't have to have just one,
putting one view and the institition backing that view into control.

(And now I'm going to get on a train for a few hours, a highly
non-miscellaneous means of transportation.)
  

    
This reminds me of Peter Morville's comment in his review of the book... "we 
must all be more aware, as consumers and creators, of the incentives, biases, 
and weaknesses inherent in all sources and structures of authority and 
knowledge." http://semanticstudios.com/publications/semantics/000167.php
  

    
We must, but there is no easy escape. 

Look, take the simplest, weakest form of syndetic structure,
establishing synonyms. No authority terms, no broad or narrow terms,
just establishing two terms as closely related, so that if you are
interested in one, you are likely to be interested in the other.
Suppose a system like the one proposed above, in which tidy-minded
people can mark tags as synonymous, and other users can vote for or
against the association. So long as the association is in place, when
you use either term, you get everything under the other term too. This
is as much as we are likely to be able to get out of a user-run tagging
system any time in the near future.

Now, what is different about marking terms as synonymous is that you
can't have it both ways. Two terms are either synonyms or they aren't.
You can give a piece of information contradictory tags, but two tags
cannot be both synonymous and non-synonymous. This turns out to be
important.

Now suppose that someone, also as suggested above, notices that items
pertaining to the USA are scattered among "USA", "United States of
America", "United States", and "America". She marks all those terms as
synonymous. Fine, right?

Not if you are Canadian or Mexican, it isn't. So the Canadians and
Mexicans vote that association of terms right down. Then the US
citizens, likely not even understanding why one-fourth of their
information has disappeared, put it back. And so on, and so on,
forever, with one-fourth of the available information sometimes
available and sometimes not. 

Now consider a biographical database; there are a bunch of them on the
Web. If someone wants to tag the article on George W. Bush "War
criminals," that is tolerable, because people looking for information
on Bush who don't think he is a war criminal need not ever see the tag.
But what happens when you make the tag "Bush, George W." synonymous
with "War criminals", and you get all your friends to vote for it? That
is not permitting everyone to have their own view of the matter; it is
enforcing propagation of that particular view. 

Syndetic structure is metadata about metadata. Who controls it
controls how metadata are associated, and thus has disproportionate
control over how information is associated. Further, syndetic structure
is largely invisible, operating behind the scenes, and can thus go
unnoticed or unchanged for a long time.
  

    
I'm not clear why the association must exist (or wouldn't be optional), and
why we suppose that there's a vote to "elect" synonyms. That would seem more
a condition of top-down, second order, non-miscellaneous order, no? 
  

    
the idea that villains might gain control of the metadata about metadata
seems about as likely to me as nations adopting facsism for their own good.
that makes it pretty scary.

there will always be guerillas though, and over time the mood will change,
perhaps the media will shift and something new will come to be.

I think about tags as generational markers.   weed is no longer weeds in our
generation, but its not weed anymore in our kids' generation either.  I was
at a parents meeting before a high school trip to Cuba and there was a kid
there who'd been to Cuba the year before.  He had a different word for the
stuff you weren't supposed to bring to Cuba.  The kids understood but the
parents had to ask what the word meant.  They were talking about boo.

words adapt to the meanings we put around them.

so I wonder if Cheney's marking up a list of metadata about metadata for his
daily spin right now.  He'd like to own our syndetic structure.
  

    

All of this is beside the point.

In the future metadata will come from (be gathered from, in secondary
queries) from multiple _unlimited_ sources (some individual, some aggregated
and sources both controlled and uncontrolled), and will be aggregated
together with data brought in from primary queries.

This metadata will come in many classes, far beyond the simple keyword
metatagging that we see emerging today.

Furthermore these classes of metadata will be associated (or more accurately
associateable - this association itself being editable at a lower level)
with a wide array of visual, behavioral, and spatial attributes, each of
which will be freely interactive (Full On, Full Off, or anywhere in
between).

Such a system will be usable in a passive manner, or in an interactive
manner.  Sets of combined metadata types/associated attributes will be
saveable and passed around.  One will be able to issue queries (or have them
set up to run automatically (what's on YouTube tonight) and have these run
through one or more of these filters.

At any point a person could then, simply by interacting with a one-axis
control, cycle between full on and full off for certain attributes, yielding
"visual behavior" in the dense visualized field or swarm.

Other tools will easily sweep through these visualizations allowing easy
picking for opening up and examining source data or media.
  

    

And the idea of "bad" or "purposely misleading" metadata identifies an
actual problem, but the answer is in gathering metadata from numerous
independent sources to be brought together with data returned by primary
queries.  Then methods such as reputation filtering and source
identification can be used to examine how visualizations of large datasets
change when the metadata sources themselves are turned on and off or grouped
in different ways.

There's no such thing as "truth" in static datasets or visualizations.
Interaction equals understanding.  More interaction and more exploration
equals greater insight into the patterns and interrelationships in large
datasets and search results.

Another level of interaction comes from determining equivalencies of terms
or values encoded by metadata.  Disambiguation of keyword-type metadata can
be handled by presenting known meanings, and the metadata itself may
eventually come with disambiguation means (meta-metadata), which itself may
come from yet other sources (via pointers or aliases or registries.)

Some ambiguity can be avoided by allowing people to choose from large sets
of metadata types.

Let's say Fred is surfing the web.  Fred has installed the MetaHelper plugin
for Firefox.  Fred comes across a film he's seen.  Fred goes up to the
pulldown and selects MetaHelper, which pulls down a secondary long
scrollable list of types of metadata.  He selects "Ten Star Rating" and a
dialog appears asking him to select between 0 and 10.  Fred likes this film
and gives it an 8.  This metadata is then registered at a third location,
with links to the film.

Later on, Susan, whose swapped out her crappy flat desktop for VizSpace,
issues a query for Top Movies of Summer 2010.  Instantly the visualization
space that used to be her flat crappy desktop fills with an array or swarm
of hits.

Her query also automatically included one or more secondary metaqueries,
which ask the net something on the order of "Who's got something to say
about these search results?" and a great deal of (different types and
classes of) metadata are brought in (from multiple, independent sources).

If this were a type of query that might be repeated very often (like "What's
On tonight?"), Susan might have some standard metadata/associated attribute
filter sets already set up and going to work passively.  If so, her
visualization of returned search results may have already configured itself
into a unique form.

Susan spots several large spheres.  Susan knows because she's familiar with
this filter set that spheres represent action movies, and size/scale
represents the current box office returns.  The three biggest spheres in the
swarm of 10,000 in the visualization represent the big films everyone's
talking about.  She guesses this even before she cursors across the
visualization revealing their labels.

But Susan isn't interested in these blockbusters.  Susan wants to know what
the "blockbusters" are among her group of online friends.  She pulls the
levers down for the metadata/associated attributes tracking box office, and
jiggles the metadata/associated attribute control for films people in her
group of friends have rated.  Hmmmmm, looks like Susan's friends aren't such
big fans of these action blockbusters.  An entirely new group of elements
emerge from within the visualization to show prominence.

The combinations and permutations of this methodology are endless.  But this
is how much larger sets of data can be easily explored and sought-after
targets retrieved.

The keys to this powerful information future lies in this type of metadata
usage and interactive visualization capabilities that are built right into
the fabric of our computing experience.

<http://www.well.com/user/jleft/orbit/vizrev/slides/5.html>

<http://www.well.com/user/jleft/orbit/vizrev/slides/8.html>
  

    
Or we could just go with TiVo-style thumbs up / thumbs down feeding an
invisible collaborative filtering / reputation management / recommendation
system.
  

    
Jon Carroll's column today talks about master narrative.  This describes the
effect I suggested up there in 85.

http://www.sfgate.com/cgi-
bin/article.cgi?f=/c/a/2007/06/05/DDGITP18DH1.DTL&hw=jon+carroll&sn=001&sc=100
0

or for the tinyurlers http://tinyurl.com/2set45

his context is the OJ trial, and the medium was not of course web 2.0, but
the idea is the same.  who has the real picture of anything?  it depends.
  

    
Yikes. You get on a train and are off the Net for a day, and all sorts
of constructive hell breaks loose. It makes me think that my staying
out of this conversation would be the best way for me to contribute.
Wow. I love this thread.

Unsurprisingly, ckridge, my views are more in sympathy with jleft. I
disagree with you on two (semi-)factual matters: The best tagging
systems have done so far is generate synonyms, and things either are
synonyms or they're not.

Tagging systems allow us to discover clusters of related terms, not
just synonyms. Lots of sites do this well already, although it's an
area where there's always going to be room for improvement. E.g.,
search on a tag at Technorati and it will suggest "related tags," with
varying degrees of accuracy. The tag McCain" has as related tags: bush,
politics, iraq, republicans, democrats, 2008, republican, law,
torture, and immigration. This is an impressive list, but none is a
synonym. It'd be more impressive if technorati knew what the
relationships were, but I think over time, it will get better at this. 

As for synonyms: the ambiguity of language is (imo) its strength, not
its weakness. Oh, at times it gets in our way, but without it, language
is mere codebook. So, yes, we're going to disagree about who the
Americans are. But with tagging, folksonomies, etc., the system can
maintain multiple meanings, and allow each of us (or, better, multiple
groups) to view it in ways that make sense to them. Folksonomies are
not mere bottom-up taxonomies that give us a single way of
categorizing, albeit one from The People. Rather, the value of a
folksonomy is (imo, as always) that it knows not only that 75% of users
think America=USA, but also that 25% don't.

ckridge, you and I seem to differ on some fundamentals. E.g., I think
the world is fundamentally ambiguous, so clarifying and settling are
often reductions of truth, not enhancements of it. Is this where we
disagree? Or do we in fact not disagree all that much after all?
  

    
Well, here's my two cents' worth...

I looked for del.icio.us' methodology for determining which tags are related 
- evidently for any bookmark, they make suggestions based on tags others have 
used for the same item (or a similar item? not sure how they would assess 
similarity).  I think they also determine which tags are most often used with 
other tags to come up with even more related tags. Bottom line: they're 
looking at association, not meaning, in determining those relationships.

In a social tagging environment, it wouldn't strike me as a "best practice" 
to enforce primacy of a synonmym based on meaning, even if a preffered 
synonym was identified by popular boat (aka, sometimes, "tyranny of the 
majority").  Instead we show relationships based on adjacency. The good news 
is that, in tagging, you could use both USA and America, and they would both 
be meaningful because they're related, not because they're synonymous.

I've definitely been in discussions of folksonomy where the question of a 
need for an authoritative label has come up, and I think the consensus is 
usually, ultimately, that it's unnecessary and not a good fit for the kind of 
environment we've created.  Sometimes groups will determine a label that all 
will use to relate a bunch of tags... we do that for conferences, for tagging 
material relevant to a specific conference, but there we want a label that 
others are not likely to use.

Clay Shirky says "ontology is overrated" 
http://www.shirky.com/writings/ontology_overrated.html 

At the conclusion of his paper, Clay says "It's all dependent on human 
context. This is what we're starting to see with del.icio.us, with Flickr, 
with systems that are allowing for and aggregating tags. The signal benefit 
of these systems is that they don't recreate the structured, hierarchical 
categorization so often forced onto us by our physical systems. Instead, 
we're dealing with a significant break -- by letting users tag URLs and then 
aggregating those tags, we're going to be able to build alternate 
organizational systems, systems that, like the Web itself, do a better job of 
letting individuals create value for one another, often without realizing 
it."
  

    

Jon, I was saying something very similar beginning back in the early 1990s,
when I presented the InfoSpace paper at 3CyberConf (which is when I met
you), and througout the 1990s.  This was before metatagging emerged on the
web, and so my arguments weren't against controlled taxonomies of tags, but
against the "indexed view," or "centralized oganization" of data.  I
(wrongly) assumed that as soon as metadata appeared, that it would be (what
later would come to be called) folksonomies (in addition to being
decentralized and gatherable from multiple sources).



Topic 327 [wired]:  (jleft)'s Prophecy: The Visualization Revolution
#6 of 77: doing 'n somethingness (jleft)      Sun Jan  3 '99 (19:22)    80
lines


- - - cut - - -

Another big fallacy is the notion that in order to tame the information age
we must index it and establish an order for the interrelationships between
things.  I call this the Yahoo view of the infoverse.  Cone-tree
visualizers, hyperbolic tree browsers, Apple's Project X and the like are
all attempts to draw lines between symbols for things.  This is another form
of information fascism.  It's not that these might not be valid
categorizations of organizations of information, but that they infer a sense
 of "this is the way it is"-ness that is, in reality, only *one* way of
seeing things.

The phrase "information wants to be free" means something different to me
than I think it does to others.  To me it means that information wants to be
free of the shackles of authoritative index.  Through the use of
metainformation, both that attached to pieces of information and secondary,
decentralized sources that make reference to other things, it's possible to
construct informational views that do not eminate from centralized sources.

- - -


Topic 327 [wired]:  (jleft)'s Prophecy: The Visualization Revolution
#29 of 77: doing 'n somethingness (jleft)      Mon Jan  4 '99 (22:09)    83
lines


- - - cut - - -

Here's my basic take on the major difference between my InfoSpace concept
and most other information visualizers such as Xerox PARC's cone tre
visualizer, Apple's fly-through hiearchical database Project X, and I
*suspect* (but I'm not sure) the database app. you're referring to.

Most visualizers presuppose that it's a necessary/given thing that we've
indexed or structured/filtered the database in a hierarchical and
categorical manner.  Now while I understand the utility and validity of this
approach to organizing information, it's not necessarily the only way or
best way depending on what how you wish to utilize the dataset.  In fact, in
order to explore and examine all the infinite myriad of interrelationships
that exist in complex data systems, it's almost imperative that you unhook
(or free) the individual content-oriented data objects from a rigid index or
order, as these are generally organized from a singular point of view or
source.

In InfoSpace, which is actually *your* InfoSpace (as you control the space
in which you retrieve, collect, and visualize information) a fixed indexing
or ordering of information, while possible, would be only one way to
organize one's assembled data.  I discuss how such an personal InfoSpace can
work in conjunction with the shared/agreed-upon data
representations/structures that represent most of the information in
cyberspace today in my Metaverses vs. Myverses slide.

This slide is among a number of discussion slides that I've uploaded to my
WELL website.  These were created for a presentation and panel discussion
that I participated in at ISEA97, held at the Chicago Art Institute in
September 1997.  They help me describe visually and symbolically some of the
concepts and insights in my work.

- - -

Metaverses vs. Myverses
http://www.well.com/user/jleft/orbit/vizrev/slides/2.html
  

    
Whenever somebody brings up "information wants to be free," I feel compelled 
to offer the entire quote:

"On the one hand information wants to be expensive, because it's so valuable. 
The right information in the right place just changes your life. On the other 
hand, information wants to be free, because the cost of getting it out is 
getting lower and lower all the time. So you have these two fighting against 
each other."

Of course, information doesn't really want anything, including liberation. 
It's all about what we want. There's another old saying, "people don't want 
drills, they want holes."  Similarly, perhaps "people don't want information, 
they want meaning."

Part of meaning is order, and I think what David's been talking about really 
is liberating - and the liberation is of information from rigid ontologies 
and taxonomies. You don't have to have a bread crumb trail if you can store 
your locations and teleport.
  

    
>E.g., I think the world is fundamentally ambiguous, so clarifying and
settling are often reductions of truth, not enhancements of it.<

Also clarifying and settling can be a way of backing into essences, a
fairly obvious effort to project "structure" onto "reality".  OTOH,
humans can be confused to the point of adding more ambiguity to the
fundamental ambiguity.

>Similarly, perhaps "people don't want information, 
they want meaning."<

Which may be be a code word for power including power over the feeling
of disorder that must be lingering in the subconsciousness in an
ambiguous world. 

We may also think that if information can be useful to accomplish a
practical, limited objectives, we can organize information in some
grand, divine manner, arrive at ultimate meaning and banish ambiguity
forever. Such a program will be able to capture a following as long as
there are humans, as we know them, on the earth. 



 
  

    

> Of course, information doesn't really want anything, including liberation.
> It's all about what we want. There's another old saying, "people don't
want
> drills, they want holes."  Similarly, perhaps "people don't want
information,
> they want meaning."

It's true that people want meaning, and want ways to uncover it.  But it
doesn't follow that people *are only searching for meaning* when exploring
and examining information.  That's related to what I call the "Needle In the
Haystack" myth.  And that's that searches are limited to trying to find a
"needle" in a large amount of completely discardable "hay" (other results).
I suggest that exploring (visually, by means of metadata associated visual
and behavioral attributes) that one can learn a great deal about the larger
context in which the information exists by interactively asking different
metadata questions to the "hay."  And thereby not only see where information
is coming from, but what context it's coming from, and also distributed
sought-after targets that might not otherwise be popped up to the top of
some boiled-down search results list.

Bear in mind that there are many goals in interacting with information sets.
 Discovering meaning is just one of them.


> Part of meaning is order, and I think what David's been talking about
really
> is liberating - and the liberation is of information from rigid ontologies
 and
> taxonomies. You don't have to have a bread crumb trail if you can store
> your locations and teleport.

And I fully agree, and have for a long time.  That, and a number of related
concepts are essentially what I've been saying all along.  I took great
exception with the notion of the Metaverse, as described in Neal
Stephenson's book, "Snow Crash."  In my slide about the Myth of "Downtown
Cyberspace," I described the ludicrous notion that there's a fixed structure
to any cyberspace place or location.  A particular structure or layout to
any informational set can definitely exist, but if this itself is but a set
of parameters configured by setting the supporting metadata to a particular
way, then interacting with those controls can instantly *rearrange* one's
data.

It's not the map that's fixed.  It's the individual viewer's centrality,
with the data being able to be rearranged and re-explored along different
metadata parameters at will, and in an interactive activity.

In this type of model, a central, shared view can still be had by all, but
it comes from people running the same filters and filter settings.

It's order, without the need to claim one configuration of meaning is *the*
main one.
  

    


> Whenever somebody brings up "information wants to be free," I
> feel compelled to offer the entire quote.

But Jon, I prefaced my use of it specifically to avoid confusion with either
its real origin (which you've posted), as well as the common misconceptions
about the statement.  It was essentially a play on the words, if you will.

I should've made that more explicit, I see.
  

    

I'd like to address a bit deeper this concept that "people are searching for
meaning," which seems to be presented by Jon as exclusionary of other things
(I'm not entirely certain what "information" is in the context that he puts
forth that people don't want).  I think it's a bit of a strawman in this
discussion.

Among the many things I like in David's writings is the concept of
"messiness is good."

I'll postulate that two complimentary statements to that:

Both are related to the old wisdom saying, "It's the journey, not the
destination."

1)  The goal is not (necessarily) to arrive at a single
point/piece/truth/meaning, but rather gain ever more and deeper
understandings of the contextual interrelationships in which any particular
piece of information, media file, subject, object, or set exists.

(not particularly concise, but I want to fully capture the different facets
of what I'm putting forth).

2)  It's only through continued interaction (interactive questioning, via
manipulating the way metadata is used to support swarm or cluster
visualizations) that will allow us to progressively gain that type of
higher-level understanding of where things are coming from, and what their
interrelationships are.

It's only after we open up points from within these mass visualizations that
we're back at our historically familar level of dealing with information,
which is a file, a piece of media, an idea, etc..

We live today in a terribly myopic infoverse.  The reason everyone is so
obsessed with nailing down the definitive meaning of things, or
categorization/organization, or definitive keywords is that we don't have an
overview level of interacting with information.

Thus, we want authorities, or AI, or Google to "think for us," and give us
that needle, or "meaning."  In a certain sense, these are definitely valid
desires and goals for our informational activities.  But they are not the
only goals.

And meanwhile, our information age is not sufficiently giving us tools to
leverage the parts of our brain that we use *pre-cognitively* to navigate
and understand a fabuously rich and dynamic real world around us.  Our
visual cortex instantly, and without conscious cognition, processes the vast
complexity of trees, grass, buildings, environments, clouds, etc..  Billions
of times greater in complexity than our written language and current
computer technology has presented to us.

This is because historical language and symbology for encoding meaning
operates at a bandwidth far less than we've evolve to function visually in
the world.

And this is why I've stated that interactive visualizations will eventually
support a new level of "awareness" that's missing from the current
information experience.

And it will be build atop metadata, and will embody many of the things David
is discussing, but at a level much more complex than the simple keyword tags
that have emerged as the first embodiments of internet-based metadata.
  

    
> But it
> doesn't follow that people *are only searching for meaning* when exploring
> and examining information.

I may just be dense, and missing something in your posts, Jim, but I don't 
see where you've suggested what people are searching for in exploring 
information, other than meaning. Is the new level of awareness you suggest 
beyond meaning?
  

    
Yes, tags are just one embodiment. They happen to make for a
particularly useful example, just as Wikipedia does. But they are just
examples. Just for one other f'instance, I just blogged about a video
demo of Photosynth, which poulls together photos from Flickr to build
composite images of oft-photoed spots, and also pulls together the
metadata of those photos: http://www.youtube.com/watch?v=s-DqZ8jAmv0.
This is not much like tagging, except that it notices likenesses and
builds something greater out of the pieces. 

And, if nothing else, it's a great demo.
  

    
Wow, that PhotoSynth demo is astonishing.  Really excellent work.

The part that I think represents a good opportunity to probe some of the
visualization ideas I described was the cathedral surrounded by the small
points, each representing a point from which a photo was taken.

Okay, so those points are already spatially arranged in what's essentially a
realistic mapping in the area around that cathedral.  It appeared (hard to
tell), that as he cursored around through them, there were little
perpspective lines that reached up to the composite 3D cathedral image (that
was cool!).

But now let's assume that at the bottom of the screen there are also
controls that allow the user to vary other visual and behavioral attributes
based on other known/gathered (again, perhaps from other sources) about
those individual points.

For example, the color of the dots may be mapped to the type of camera.  Or
a metaquery to highlight Nikons may lead to various scattered dots
sparkling.

Another attribute may track metadata for how many known photographs are
available from the various photographers represented by those dots.  So
perhaps that metadata is associated with the visual attribute of height.  So
then we see the dots as thin poles of different heights.  That high pole
over there?  We sweep and open that source and see that it's a very prolific
photographer, and can then (if desired) go off in that direction to look at
his or her photos.  Or add that to a collection list to sort through later.

The reason I find it hard to get into a discussion of "meaning," is that I'm
not sure how meaning fits into this kind of exploration.  It's a useful kind
of exploratory activity, and reveals much that may be hidden *in any one
single view*.  And this is why I see this all leading to the conclusion that
it's all about ongoing interaction.

While the other demos in that video were very cool, and very useful, it was
the one that showed both abstract representations of the sources themselves
that were closest to the *additional layer* of metadata-supported
exploration that I advocate.

Great link!  So many great things shown at TED.