I was happily surprised on the gender balance front. Many women were there, speaking as well as attending. The panel I liked best
was two-thirds female. I had been struggling to pin down what it is
exactly that information architects cultivate as a chief virtue. Based
on what I saw there, the answer would appear to be empathy.
In a few short years this emerging area of practice has burgeoned into
a full-fledged community that actually feels like a community, more so
than the scholarly fields that I have frequented so far. I got a sense
that the people there cared about what they do and about each other.
I've learned quite a bit, made new meaningful connections, and strengthened others, so the trip was worth it. Plus the freezing rain had gelled around the branches on the way back to New Brunswick - it looked quite beautiful.
I'm a loyal customer of Amazon, but I have some criticisms.
Amazon doesn't *really* know you. It knows what you bought from it.
Pieces, but not enough of the whole. It is a siloed technical platform
which distributes, but doesn't consume. They maximize what they have,
but are approaching the limits of tradtional CRM.
Some components of Web 2.0. Social networks, desktop information,
browser history, tastes and preferences - getting them documented and
un-siloed, analog information
LinkedIn launched job postings. This is the beginning, but it's a lot of work. It is not enough.
Blinkx offers access to the desktop. People don't remember where they
saw something - web or desktop? Google now offers access to the desktop. [SP: and Copernic, too.]
Music on the desktop. Fire up iTunes, it's all about you. It's great.
Until you get into the Apple Store, and oh, it wants to sell me music.
Audioscrobbler have a tracker that records what you've played [SP: he's blider].
It recommends other people with similar tastes. There's no commerce
component. There was an undergraduate project that offers to pick up
your iTunes XML record and offer recommendations. An undergraduate
project. There's nothing more being done with this. !?
A9 integrates browser history data. A9 tells me what sites I visited recently.
Web-based bookmarks: del.icio.us. Web-based archiving: Furl saves you
from linkrot. Physical history data. Who, what topics? Evite has this
list of all the parties you've ever been invited to, and a list of
people who have been invited along with you. This is your data, why is
it Evite that does this?
Delicious Library lets
you curate your material goods, e.g. DVD collections. We need to
increase the ROI for things like this if we want anyone other than
hardcore collectors to start using them.
What is Amazon (or any eCommerce site) doing to bring these concepts to the eCommerce experience?
What if I want Yahoo, Amazon, etc., to consume *my* stuff? [SP: this sounds a lot like DataLibre]
C2B is the empowerment of customers with their past purchase and
preference data, allowing them to share it when and with who they
I'll present a centralized C2B system, but it could be made decentralized. C2B is one customer to many businesses.
It inverts the pyramid: the user is at the top. Vendor-vendor
"coopetition" for customer data is key to getting access to the
superset resulting from the combination of what each vendor knows.
C2B is important. Consumers will demand a more and more personal
relationship from businesses over time. Micro-markets, narrow-casting.
Harvard Business Review just ran an article on C2B, hailing it as the
heir to CRM.
Say Jane, knowledge worker, gets stuff from different online retailers,
brick-and-mortar stores, friends, ... Recommendation-based
merchandising is crappy right now. The C2B way? If the customer can
send over more info about herself she can get much better
recommendations, and a better relationship.
"Future" C2B: Trusted third parties, anonymizing filters.
Better Bookshelves idea. Snap a pic of your bookshelf, upload it, book spines are OCR'd, your book collection is digitized, voilà! Recommendations get better.
If customers have RSS feeds, Walmart has an incentive to rip Netflix's RSS feed and compete.
C2B is coming. It has a community of business academics around it.
People are actively thinking about it as a business. It gets businesses
out of data warehousing.
Amazon could be so much more. The same goes for eCommerce on the Web in general.
By being on top of this, IAs can be agents of some serious innovation. Thank you.
Q. It's difficult to find a trusted third party. Any one that is commercial or government is kinda suspect.
A. It's about the business model, you need to have one where earning trust translates into success. Financial services perhaps.
Q. This sounds like Google.
A. Just to be clear, I've just been hired by Google, but this stuff predates my hiring. [SP: I really hope they won't ask him to work behind closed doors now that he's there.]
Q. You're really saying that the customer should own their own data, and leverage that.
A. If the customer can't manage their own data, [missed the last part]
Q. And if you don't give me my own data I won't want to be your customer.
Q. [me] Who's going to jump first in this arena? Not Amazon, I presume.
Financial services, people like Quicken, might be willing to move in that direction. Web portals, and startups.
Q. In higher education, you could recommend courses based on what you and o
Q. I work with a company that's in the business of selling *huge*
databases of customer data. This information is already circulating. [SP: but the customers are not involved and little aware]
A. The alternative is indeed much scarier.
Q. Diminishing returns on recommendations?
A. It's hard to guess what the limits are to recommendation
Q. [Lane Becker] I think this
was one of the better presentations, but I think it falls down at the
example level. The bookshelf picture OCR examples, there's no way I'm
doing this. There's a lot of other stuff beyond recommendations.
A. Recommendations is the existing metaphor. Think about RSS feed aggregation, this is all happening.
Q. Will vendors be able to look at what other vendors do with a customer's data?
A. I don't have the answer there, it will depend on business rules and such.
Q. [Thomas] Some providers I don't care to deal with because I hate their service.
Q. Are there standards for purchase history?
A. AFAIK, hmm... There are XML, RSS, RDF, Atom, in that general area; more specialized formats may exist, I don't know.
Most of MAYA Design funding (full disclosure here) comes from DARPA. I
come from the tree-hugging, lefty side of the road. My friends were
very puzzled. I'm sorta the business guy, hired to turn swords into
Most of the command-and-control software in Iraq right now is actually being powered by the technology I'll be showing you.
What have been the most effective ways we've had in the past to
effectively spread information? We've been looking at public libraries
in the US. That system was an exceptionally efficient way of providing
universal public access to information. It provided perpetual storage
and distribution of intellectual property, while respecting the
intellectual property of publishers. They've helped with distribution
and dispersion tremendously. Imagine trying to erase "Moby Dick" when
there are copies in libraries everywhere. The concept if actually
pretty crazy. You walk in and get out with a book without paying
anything, and the publishers aren't screaming out.
In the digital age we need universal digital access to all types of
information from anywhere, perpetual storage and distribution,
flexibility in maintaining intellectual property (IP) rights, and
logical but flexible organization structures.
The Web comes short of this. Access is limited by the original format
of the data. It's difficult to combine data from multiple sources. Web
pages/databases are only available as long as their owner keeps them
up. No comprehensive system for dealing with IP rights. Useful info is
frequently hard to find among billions of disorganized web pages.
(Note in passing that most nonprofits currently copyright the info they put on the Web, which violates laws on nonprofits.)
Look at a Google results page. Where's the IP information on the items
listed in there? Look at this recent headline: "Chief Exec Pulls Plug
on County Assessment Website" Truly public information was taken
offline, to citizens' detriment, because of a working conflict.
Solution: Build the next generation info commons.
Note that I haven't said open source. I've talked about open standards
and open architecture. We would like you to buy some of our services if
you want to. It's a very interesting twist on business models. This is
an interesting experiment in how a private company can help build a
With this commons you'll want to be able to take all the information in
the world about, say, the American Goldfinch, and tag it with creator
and IP information. One rule should be to completely separate your data
from the user interface that will display it. Peter Lucas wrote "The
Trillion-Node Network" in the late eighties, which was perceived as
crazy. But now Rich Internet Applications are on the horizon, and very
large numbers of devices no longer so crazy. The idea is that items
might be displayed on different devices, with varied user interfaces.
The core architecture. We spent years and millions designing one large
distributed peer-to-peer architecture which we refer to facetiously as
"the last database". Not centrally controlled so you don't have the
Napster effect. [SP: they call this replication information-liquidity.]
You can't delete something once it's gone into the Commons. One layer
is made of "shepherds" who move the information artifacts around nodes.
We can have pipelines out to various data sources. E.g. we regularly download stuff from the EPA.
The EPA doesn't actually know about it, it's FTP activity. We have
nonprofits updating stuff locally which gets transformed and cleaned
and stored in the Commons.
Projects we've implemented. We've downloaded the US Toxin Database all
the way back to 2000. In the future someone can go to that copy to use
it for their local community. People can start connecting the dots,
e.g. correlating production facilities and diseases in their locality.
Just like the Library of Congress doesn't care if you connect Nietzche
and some other author, this should be possible.
Our persona, Julia, is concerned about a smoke-spitting facility in her
neighborhood. She fires up a map-based application, finds the plant,
and sees all the toxins released over time by the plant. Figures out
the health risks.
Susan James is recovering from breast cancer and pulls out cancer
incidence info for her county, looks at various related info, etc. But
this official info sheet has a flip side. On that other side, you could
potentially rate what you saw on the front, comment, find others who
also care about that item you were looking at. [SP: group-forming around common interests.]
I went to five cancer sites in the Pittsburgh area. They did not know
about each other. By linking them all to the Commons, they can still
exist but become connected.
Rudy Brown is a concerned consumer. He pulls out his Info Commons cell
phone that scans barcodes, and gets a report on the products, based on
datasets such as Material Safety Data Sheets, user-contributed
ingredient lists, product bar codes, and chemical substance and health
Attribution is sacrosanct in our system. So you get accountability for
the info. What about spam? Yes, this can turn into the Wild West. We
are betting on community moderation.