Jun 2

Tim O'Reilly

Tim O'Reilly

Ubicomp and Web 2.0: Connecting the Dots

I've been saying for some time that the next stage of Web 2.0 is the application of collective intelligence techniques to sensor data, not just to data input directly by humans.

Two stories this weekend illustrate this point nicely. The New York Times published a story on Saturday entitled Billboards that Look Back, about a new generation of electronic billboards that use cameras to track who looks at the billboards, and a story yesterday on Techcrunch about's contextual ads triggered by Facebook photos.

Most people will immediately recognize the first story as a ubiquitous computing (ubicomp) story: a next generation display equipped with sensors bringing computing to an arena that was previously analog and uninstrumented. But connecting the dots between that story and the second one is really important.

Many of the most important breakthroughs in Web 2.0 have come through finding new meaning in data that already exists, often through statistical methods and related algorithms, not by gathering new data, or adding metadata and structure to existing data. (Pagerank is the canonical example.) If really is able to do a good job of matching ads to photos via clever algorithms, they've effectively turned a wealth of existing user-generated photos into sensors for their application, without having to deploy a single camera of their own.

In my talks, I've long argued, following Dan Bricklin's Cornucopia of the Commons, that there is a hierarchy in architectures of participation, with the most powerful literally building a system in which participation is automatic, and driven by the design of the system itself rather than any explicit request for user contribution. Methods for extracting additional layers of meaning from activities that users perform for their own self-interest fall into this category.

Thus, it's important to include in the category of sensor data richer interpretation of photos and audio/video streams. So for example, photosynth is a great example of an application that, after the fact, extracts additional data from user-contributed photos. Similarly,'s audioscrobbler turns your playlist into a sensor, and Wesabe is effectively turning the credit card into a collective intelligence sensor. (Disclosure: Wesabe is an OATV investment.)

Take away two messages:

  1. Think about ubiquitous computing not just as the move from the computer to the cellphone and other mobile devices but the fact that those devices are becoming sensors for cloud applications harnessing collective intelligence

  2. Remember that "data is the Intel Inside" of Web 2.0, and that databases driven by network effects and applications deriving meaning from that data via statistical methods will continue to be the key to competitive advantage in the ongoing network era.

P.S. I've been calling this trend ambient computing, because I like the sense of computing encountered while walking around, and because I found Peter Morville's Ambient Findability so thought-provoking, but ubiquitous computing or ubicomp seems to be the winning buzzword.

Previous  |  Next

0 TrackBacks

TrackBack URL for this entry:

Comments: 10

  Tony Stubblebine [06.02.08 11:16 AM]

It's a good bet that whatever the next paradigm is called (i.e. Web 3.0), it won't have "Web" in the name. What's the lever that would drive success in an ambient computing world? "Data is the Intel inside" is a Web 2.0 meme. Will there be a new meme that takes massive amounts of data as a given?

I was thinking it might have to do with reach. Wesabe will have better financial data (and thus advice) than Yahoo finance because it reaches all the way to the credit card. Twitter is a rapidly growing social network because it reaches the most communication channels (web, phone, IM, RSS).

Privacy aside, Comcast would beat billboards at the looking-back game because they have more reach (person viewing hours).

  Trent [06.02.08 04:51 PM]

Tim, Adam Greenfield's been all over this for awhile now, particularly the ethics of ubiquitous surveillance. As co pared to 'Ambient Findability', his 'Everyware' is IMO the better book on ubicomp. I would strongly recommend having a look at it.

  Tim O'Reilly [06.02.08 05:22 PM]

Trent, I wasn't saying that Ambient Findability was a book about ubicomp, just that I loved the concept of "ambience" better than I liked the concept of "ubiquity" as describing the qualitative change that is in the air.

I'm aware of Everyware, but haven't read it.

  Falafulu Fisi [06.02.08 06:47 PM]

Tim O'Reilly said...
I've been saying for some time that the next stage of Web 2.0 is the application of collective intelligence techniques to sensor data, not just to data input directly by humans.

Tim, I am sorry to say, but you're late to the party. This has already been happening prior to your claim. BTW, I browsed the book table of content for collective intelligence (although I haven't read it) and it looks good for web developers who don't have a deep understanding of mathematical computing. I am specialist in data-mining/machine learning algorithm development and I am well verse with the algorithms described in the book and I think the author had done a good job of brining those complex & difficult topics to the masses (web developers).

Tony Stubblebine said...
Privacy aside, Comcast would beat billboards at the looking-back game because they have more reach (person viewing hours).

I had been involved with the expert group which is currently drafting the JDM 2.0 (Java Data Mining API) spec, and I raised the possibility of including a package in the spec to do Privacy Preserving Datamining, which is based on a signal processing technique called wavelet, such as being described in the following paper:

On the Use of Wavelet Transform for Privacy Preserving Data Mining

but I think that the group might be interested in such functionality in the next version , JDM 3.0, because JDM 2.0 is nearing its release date to the general public.

Implementing Privacy Preserving Data Mining means that different vendors can exchange data without breaking the privacy law, therefore important intelligence information, could still be mined, with some of the attributes required by law not to be revealed into a 3rd party vendor still being withheld.

  rektide [06.03.08 04:41 PM]

I am fascinating to see how Apple adopts to the challenge: for the most part I picture Apple as its own interconnected ecosystem, but the ambient challenge is all about allowing ad-hoc connections across standard protocols. It will be interesting to see if Apple adapts to the ambient connectivity model, or tries to preserve its closed ecosystem.

Similarly, a lot of consumer electronics companies are building devices on linux. But how many gain the main advantage of linux, the ability to inter-operate with a massive assortment of hardware and the largest most malleable repository of code on the planet?

Tim, you characterize ubicomp as "not just as the move from the computer to the cellphone," but one of the biggest drivers of ubicomp will almost certainly be near desktop grade ARM/MIPS chips with powerful GPUs on communication devices. What I see unique about this is that this makes more than a portable computer, its a computer you want to adapt to any network situation. There is unprecedented new demand for ad hoc network connectivity, the need to share and consume any data source, be it one's own personal webserver or a humidity sensor.

The ability to create these adhoc connections is two things. First is the protocol level. Second is the coordination systems. The first is reasonably well developed, and keeps a steady pace forward. For the second, its time to dig up tuplespaces, Linda, object databases & a host of abandoned research projects on coordination. Certainly there will be a lot of large centralized statistical AI systems churning through the data that gets generated, but the main coordination systems, the cloud, will be far more spontaneous responsive and enabling than anything we call "cloud" now.

  Jon [06.03.08 10:43 PM]

There are a couple of start-ups focused on this. Two in my town are Collective Intellect and Leximancer. Interesting challenge with a lot of promise.

  James Dellow [06.04.08 12:50 AM]

Tim, CSC coined the term "extreme data" around this emerging trend back in 2005. See for a bit of a summary - talk about the impact of data everywhere, time and place, social connections and meaning.

  Falafulu Fisi [06.05.08 05:21 AM]

Tim said...
...but ubiquitous computing or ubicomp seems to be the winning buzzword.

There are some computing peer reviewed journals that uses Ubiquitous, and it is not a buzzwords, since it has been called that way for almost 10 years to the best of my recollection.

- Ubiquitous Computing and Communication Journal (published by UBICC)

- Personal and Ubiquitous Computing (by Springer)

- Journal of Ubiquitous Computing and Intelligence (by American Scientific Publishers)

- International Journal of Ad Hoc and Ubiquitous Computing (by Inderscience)

... and a few others in journal special issues.

  J. B. Martin [10.12.08 07:18 PM]

UBICC, the Ubiquitous Computing and Communication Journal [ISSN 1992-8424], is an international scientific and educational organization dedicated to advancing the arts, sciences, and applications of information technology. With a world-wide membership, UBICC is a leading resource for computing professionals and students working in the various fields of Information Technology, and for interpreting the impact of information technology on society.

  J. B. Martin [10.15.08 11:09 PM]

Established in 2006, UBICC is a non-profit research organization that has been at the forefront of technological excellence. As a recognized leader in the computing and communication research institute, we will strive to be the best in the fields of computing and communication sciences.

In the 21st century, by combining knowledge and information with computer and telecommunication, we will open the door to the future ubiquitous age.

In order to become the world's leading R&D institute in the IT research field, we will pursue for the best intellectual property, the best in human resources, and the best per capita technology royalties.

At UBICC, our management philosophy is based on "Renovation Management", "Quality Management", and "Knowledge Management." We believe that new innovative technologies developed in our institute will make community industrially competitive thus contributing to the prosperity of human kind.

Key to our success is the ongoing commitment, innovation, and forecasting of ever-changing industry. To position UBICC as globally competitive technological institute, we are making continuous improvement and development.

The information revolution is constantly changing the way we live and making it possible to achieve higher standard of living. Our efforts are to create an atmosphere that stimulates creativity, research and development.

Looking forward to the high-tech world, we will continue to make strong efforts to maintain its leading position in the IT industry.

Thank you for your support.

UbiCC Journal

Post A Comment:

 (please be patient, comments may take awhile to post)

Type the characters you see in the picture above.