Category essays

Dimensionalizing the web

What is a web page today? If you look at the average web page, it’s a compilation of a diverse set of data sources drawn into a construct that we think of as a concrete whole. It probably started with CGI — and the first commercial application was likely the ad banner — but today that simple web page is made up of a whole mix of things ranging from dynamic content, ad’s,  widgets, sidebar tools, gadgets — the frame that we think of as a web page is now constructed from data streams in from all these sources and more.   This componentization of the page was the first step in what is becoming a different architecture for information delivery. What we have today are the equivalent of early life forms – necessary building blocks that evolution will use as more sophisticated lateral services develop. The organization of data streams and how they are constructed relates to our understanding of the dimensions of the web.

Question?   What would the web look like if you picked it up and looked at the bottom? I imagine, what you would see would be a set of databases – with streams of data flowing between them, into these things we call web pages and between these things we call web sites. These metaphors we have applied to the web — pages and sites — are analog’s that helped us grasp and structure the web, yet like any proxy they also impose limits on our perspective. RDF/RSS started me thinking about a lot of these ideas but in the eight or so years since those standards were developed our understanding and approach to web sites as vertical businesses has barely evolved. The spacial assumption we imposed on the web — that a site is a discrete experience that a publisher can control — maps with both a human need to impose hard edges on a dynamic, complex system but also with how we have understood media for the past 100 years or so. I think those edges are been broken down and are offering a different view of the web, and therefore of media companies, one that is less structured around the hard edges of a web page or site, less vertical, less about data silos and more about dynamic, fluid use of data and connections between data points. Some examples.

Take a look at this picture of this post I found on tumblr last week. This person — Erin — is using tumblr to announce a meetup. In this case email and reblogging are the tools she is using to confirm attendants. Shouldn’t this person use meetup for this — clearly its their preference not to, but why?

tumb log

I would propose two theories: context and easy of use. First context — context is important, Erin has followers (an audience) on tumblr, she has an environment that is customized with a user experience she could control (nice background) — and so she wants her meetup to appear in that context. Ease of use — for a myriad of reasons it seems it was easier for her to roll her own meetup than use or to quote Pip Coburn the perceived benefits outweigh the perceived pain of trying to learn something new. So here is an example of someone molding a use case (creating a meetup) into another web experience to fulfill a need.

Example #2. What about Twitter. What is the web site   The first answer — the one I would tell a stranger in conversation — is that its a destination to access and use the microblogging service provided by Twitter, “want to try one to many micropublishing? go to”.   Sounds simple enough. Yet that conclusion isn’t supported by the data. I don’t have the exact number but I think its safe to say that more than half of the interactions with Twitter occur off — and the number is in all likelihood a lot higher than that. So is Twitter a protocol?, maybe.   Maybe Ted Stevens actually understood the web better than we thought — thinking about Twitter as a pipe makes more sense than as a  destination.   But its not a pipe in way that old media understood pipes — its different, im not sure i understand exactly what that difference is going to yeild but what is clear today is that each interaction that takes place on the network add’s value or context to further interactions.    As data chunks move around Twitter the get organized and collated into conversations and meme’s.  Similar to the Meetup example — each node on the twitter network is contextualized in form that makes sense for that particular interaction. But unlike Meetup, Twitter is powering all these interactions. The data becomes more valuable as it moves from interface to interface — not less.     There is something very powerful that is happening with the simplicity and openess of this network.   A network is the best metaphor I can think of for Twitter.

Another example.  Iminlikewithyou — the flash casual gaming site, started off as a destination (disclosure note, a betaworks company).    All of a sudden users started grabbing the code and syndicating their game on to their web sites.    But this isnt just the game — its not a widget model — its the entire underlying game net that is getting syndicated.     IILWY is closer to our understanding of old media but its contains some of the bizarre distributed breadth and possibilities that Twitter holds.

So where does all of this lead us?  I believe we need new metaphors to understand and place dimensions around what a web experience is. I don’t have an answer but I do have a few thoughts on how we can begin to frame and understand the shape of what is to come.

i) Think Centers vs wholes, think about networks vs. destinations

Pic by CALast week I was re-reading Christopher Alexander the Nature of Order . In the first book he has a section about wholes vs. centers. He makes the argument that composing visual structures as whole’s — thinking of buildings, things, windows — anything as a whole — fails to recognize the context in which the object lives. He builds the argument up starting with a dot on a piece of paper — he then analyzes how the dot divides and structures our spacial understanding of the piece of paper.  From this point he starts to frame up a way of looking at the world that is based on thinking about centers, zones of spatial activity vs. wholes.   An example he cites:

“On one occasion, I was discussing the concept of centers, as it applied to some bedroom curtains, with my wife Pamela.     She made the comment that the use of the word “centers” as I had explained it to her, was already changing her view of everything around her, even as we were talking: “When I look at the curtain in the room, and think of the curtain, the curtain rod, the window, the sky, the light on the ceiling, as centers, then I become so much more cognizant of the relatedness of all things — it is as though my awareness increases”

I think Alexander’s point and work here is profoundly applicable to the web. If you start thinking about centers — clusters of information — vs. destinations and vertical sites, for me at least, it gives me a frame of reference a metaphor that is far more expansive and networked than the one in which we operate today.   At Fotolog I learned that centers can form and cluster with remarkable speed within a community — now this is starting to happen with information moving laterally between domains.

ii) Think what can move laterally and encourage it to move

People, those things we often call users, want to take data and move it laterally across the web.   They want it to exist in context’s that make sense for a particular interaction. Whether its data portability standards, micro-content standards, people want to cross post and move data from one service to another. There is much that needs to be done here.   A year ago when F8 was launched it seemed that Facebook was driving headlong into this domain.   Yet a year later it now seems like Facebook might become known as the last portal, the last walled garden experience — data comes in but not out.   Openness of interface, api’s — letting data come in an go out of a domain is central to this thesis.    The Facebook newsfeed could be a web wide service — instead the way its articulated today is about retaining eye balls and attention — a movie we have seen before.  Last week we started talking publicly about SwitchAbit — SwitchAbit is a service that is designed to help drive this lateral movement of data across the web, while retaining context, its a small contribution we are hoping to make to this larger puzzle.

iii) Think about how to atomize context so that it can travel with the data

Dirty DataAtomizing content is one piece of the puzzle, the other is doing the same for context so it can travel with the data as it moves around the web from center to center. — Steven Johnson’s creation — trawls through blog posts and attaches geo context to individual posts. I sometimes refer to as a washing machine — dirty data comes in one end — scrubs the data set and ships out geo-pressed results the other end.   The geo scrubbed post is now more useful for end users, publishers and advertisers.   A bit of structure goes a long way when the data can then move into other structures.   The breadth of what geo scrubbing can do is staggering — think about pivoting or organizing any piece of information around a map — the spatial dimension that is most familiar to our species.  A bit of context goes a long way.   (disclosure note, / an investment of betaworks)

iv) Think Layers

There is a layering dimension that is worth consideration — there are services starting to emerge that offer functionality that is framed around exposing some separation between different layers of the web.   Photoshop is the software that first introduced me to the layer metaphor,  i still cant use photoshop, but I think I get the layer idea.   Google earth has applied a layering concept to mapping.   Similarly services like PMOG are experimenting with layers.   Back at betaworks Billy Chasen started working with layers about eight months ago.   He developed a simple navigational tool called Fichey that lets you navigate web pages independent of their domain – using a common navigational tool.    Want to flip thru the top digg stories? — fichey makes it fairly easy and fast.   This was just a beginning.    Billy has developed a service called firefly — it’s in testing now and over the coming weeks we will begin to preview it — but its all about creating a layer of interactivity that is contextualized with the web site you are on but its exists independent of that web site.

v) Accept uncertainty, keep it rough on the edges

What did Rummy say about the known unknown’s?    As we experiment and design these new forms of interactions its vital that we remain open to roughness and incompleteness on the edges of the web.   The more we try to place these services into the convenient, existing models, that we have used to structure our thinking thus far the more we will limit our ability to look ahead and think about these things differently.

This is just a beginning.   I hope these five areas have helped define and frame how to think about alternative data dimensions on the web.  Time to wrap this post up — enough for now.

beta working

The past few months have been fast and hectic. The focus has been getting betaworks to scale. We call betaworks a platform for seed business creation — let me spell out a little more about what that means to us.

We are creating a network of companies — some of which we are building and some of which we are investors in — that are threaded together by a set of common themes and capabilities. A set of loosely coupled bits that over time we will seek to connect in ways that are meaningful. Some background follows on our perspective and intent.

I started betaworks in 2006 seeking to develop a new methodology, and platform, for seeding businesses. We gathered a small team and for a year we tested a series of different approaches to seed business creation. We learnt a lot, we had successes and failures, we intentionally kept it small — permitting us to learn fast, fail fast and make small mistakes. In hindsight the most important thing we learnt was a design principle. When we started out we thought we could design or architect elements of betaworks. What I learnt was that given that we dont know all of the elements we are going to thread together we needed to start by working, by building, and then in the practice of work let the components between the companies emerge rather than imposing a top down view of what the connections should be.

Mid 2007 we flipped our approach — and went bottoms up. We stayed small, under the radar, focussed on our theme, seeking not to be distracted by opportunism. We started identifying standards and methodologies to scale our work and stopped trying to over think the design. By the fall of 2007 we had assembled our learning and formally started to build out the platform. Six months later we have four things that we have built and we have fourteen seed investments.

Central to the design of betaworks is the assumption that companies building services with a common theme can and should profit from inclusion in a platform or network of loosely coupled bits and driving context and meaning across these sites is valuable. Portals are gone but a new metaphor has yet to emerge to replace the portal concept. The portal was grounded in the same set of assumptions as media has operated with for many years — as it recedes something new is emerging. Something that is more distributed — something that is not based on the assumption that erecting a walled garden around services is the way to build a sustainable long term relationship with users or a long term business. Something that actually encourages the movement of data across edge services, vs. building silos — as data moves and is exchanged it actually gains in value becoming more interesting to users not less. There is a hairball we are decoding, bit by bit, its right in front of us our job is to figure it out and scale it.

Someone said to me last week that we are a reverse incubator. Incubators share the peripheral services things that I believe entrepreneurs can and should get from the market (legal, hr, accounting, office space) — betaworks is designed to share core capabilities – software / IP, knowledge, data, standards, analytics, leadership, tools etc… Someone a year ago called it a funcubator — maybe a reversobator, or outcubator? — a little less George Clintoneque. enough.