Long Tails and Geodata The whole "long tail" idea comes up again and again these days. I'm
not sure everyone has the concept in hand, so I want to start there
before going any further. The "long tail" coined by Chris Anderson
(editor of Wired, who followed up with an article
in that publication recently and published a book by that name this week, which was
in TheNew York Times) refers to the phenomenon whereby
there's lots of demand for a few popular products (movies, books,
albums/songs) and hence they produce a lot of revenue for companies
(NetFlix, Amazon, iTunes), but there's little demand for lots of other
similar products. For example, many people are going to want to rent
Stephen Spielberg's Munich (like my brother and my parents and
others), but only a few will want to rent Running on the Sun,
about a long race through the desert (like me and other crazies who
think that's interesting/fun).
So, Blockbuster will make a lot from Munich and other
"blockbusters." It'll make less from Running on the Sun, but
and this is the key thing it doesn't cost Blockbuster (NetFlix,
Amazon, fill in your favorite Business 2.0 company here) that much to
stock those films that are in "less demand." In fact, by making
available all of those "long tail" films for a long time and serving
them to the few who want to see them, it can make a pretty penny.
(Want a better discussion? Listen
this short piece from National Public Radio's On the Media,
where Brooke Gladstone interviews Clive Thompson from Wired.)
Now, why does this work in movies, books and music? It works for a few
practical reasons: (1) these are not things you need to touch and you
can buy them sight unseen with confidence theyre not stale; (2) those
companies can store the stuff cheaply in an ugly warehouse in some
place you never have to go; (here's the big one for us in geospatial)
(3) the seeker can easily find the product using an online catalog.
But, back to geospatial. The point I want to make is that the long tail
should/could/would exist in geospatial data products, but it does not
yet. Think about it. A handful of data providers have lots of customers
and make a good deal of money. That part of the curve is solid. I'd
offer up TeleAtlas, NAVTEQ, DigitalGlobe, GeoEye as top sellers at the
head of the curve.
At the other end, there are lots of folks offering up very specialized
databases of specific locales (some are local governments, for sure),
specialized coverage (oil and gas, wind, come to mind), and real time
data (traffic and weather). They are out there, but are often smaller
niche players, so instead of large ads you might find them in our
Directions on Data publication and noted in blogs. That group, so
far as I can tell, is not yet a "long tail."
Why? There is no giant index where those offerings can be found and
purchased. There's no "Amazon" for geospatial data yet. Many portals
exist where you can search (Geography Network, MapDex, Geospatial
One-Stop and others index, and some make available for purchase, some
data), but theyre not as complete as those of the Web 2.0 businesses I
noted above. Why? There are lots of reasons, among them the fact that
creating such a database is far more complex, due to the nature of
Whats happening instead or perhaps in parallel to the creation
(someday) of such a search tool is the proliferation of new tools that
lets any of us offer up our datasets to share with friends. ArcGIS
Server will let you publish out your data (and services). A network KML
file will allow many to see your favorite sailing spots. Virtual
Earth's collections allow dataset sharing on that platform. Save
ArcGIS, these examples are, for now, more tied to the "social
networking" and advertising businesses. They have an informal quality,
rather than a business/science focus, though of course, they are
certainly used for those things. But, they do address a "long tail"
sort of problem. For the 200 people who care about the 10 best places
to fly kites in Massachusetts, a dataset is available and shareable.
The challenge of making that information findable and accessible is not
That leads to another question: Should we expect to see a long tail in
geospatial data? Would that be a good thing for our industry? I think
it would. I think geospatial needs niche data providers. And, I think
the industry needs an "Amazon" to organize and make all these datasets
findable and purchasable. Several companies have tried that, most
notably GlobeXplorer, which acts as a broker for many data providers.
It offers its data up in an open standards way via support for OGC's
WMS, which makes it more Amazon-like than other providers. It has a
vested interest in one data provider, as it now owns AirPhotoUSA,
though it continues to partner with many players. Amazon (and other Web
2.0's noted above) do not create their own content; they broker it.
That may or may not be a key part of their success.
Clearly, there are a few preconditions for such a long tail geospatial
data business. First, there needs to be a viable business model. There
need to be sufficient metadata and/or samples to confirm that the
product meets consumer needs. (This is a bit more complex than selling
a book.) There needs to be assurance that data will be compatible with
the software the client uses. (We are getting better at that!)
There need to be businesses comfortable with the idea of working
through such a broker. And a few other things folks with MBAs would
I will say this: once we get such a business up and running and
successful, it'll be another sign that geospatial technology industry
has "made it."