I’m just wrapping up the first day of Defrag, and it’s been exhilarating. It really does feel like several days at most conferences packed into one. You can see some great in-depth coverage of everything on AltSearchEngines, Charles was live-blogging in front of me in the morning.
First up was David Weinberger, billed as "Everything is miscellaneous", but actually renamed on the fly to "What’s unspoken between us". He’s posted his own notes from the night before, but what really struck me was how useful it can be to step back from the technological bubble I live in, and use the wider world to get a fresh perspective on what we’re doing. He drew on the poetry of Rilke and Rabbinic teachings to explain both how much meaning is implicit in our world, and the hope that computers will gain a soul, through their close association with us.
Clay Shirky couldn’t make it to give the next keynote, so instead there was a round-table discussion loosely based around "Social intelligence". Jerry Michalski moderated, with Joshua Schacter of de.licio.us fame, JB Holston from Newsgator and JP Rangaswami, the CIO of BT. It was unstructured, but full of interesting nuggets, such as Joshua never using RSS, or the idea of deliberately avoiding automatic spam blocking so that the community steps in. What I found most relevant to my work is the descriptions of the existing ‘attachment culture’ in most companies, where collaboration is done by emailing around Word, Excel and Powerpoint files. Mailing lists are a related ubiquitous big collaboration tool.
The general tone, being tech visionaries, was pretty derisive about these approaches, but I’m a contrarian on this one. I think if you can’t understand why people are resisting moving to the latest techniques, you’re probably overlooking some important advantages to the traditional tools. I developed some of these thoughts in a later open space discussion about user acceptance. Compared to a wiki, emails have a much simpler and more explicit security protocol. You make a decision about exactly who sees what you send out, with a clear chain of accountability if one of those recipients decides to make it more widely available. On a wiki, the visibility is determined by somebody else in an opaque way, and it’s a lot harder to understand who’s to blame if something does get wider exposure than it should.
Michael Barrett from PayPal then gave a mini-talk, "A message of warning". He’s worried that we’re merrily building tools with little thought to security, and that we’ll end up like telnet, impossible to use in any situation that requires secrecy. It’s hard to argue with that idea, but it’s also hard to see what the solution is. I’ve yet to see a startup wow its investors with a security demo, and there’s never a easy time to devote resources to securing your software. It took years of bad publicity before even MS moved away from fudging security in favor of user-facing features.
I’d never been involved with an UnConference-style open space, and it was hard to choose amongst all the topics. I picked the user acceptance theme, and to my surprise found myself the only one without a collar. It seems like the number one cause of failure of new tools is lack of user acceptance, so I was expecting a lot more techies and small companies. Instead it seemed dominated by people looking from the business side of large companies and trying to work out how to sell their employees on these ideas.
Andrew McAfee of HBS was one of the leading voices in the discussion, and one of his most interesting ideas was that the security argument used to block use of wikis internally is hogwash. I’m no stranger to bogus security concerns being used to veto change, but as I said above, I think there’s at least a kernel of truth in this case. One of the unspoken ideas behind Defrag is that ‘information wants to be free‘, and a corporation will be a better and more productive place if only we can enable wider sharing.
This is probably true on the macro scale, but on the front lines there’s both winners and losers. Someone whose power is based around being the holder of arcane knowledge will fight tooth and nail against this. As a more benign example, a line-manager may not want other departments taking up his star engineer’s time to answer questions, if it’s to the detriment of the project he’s responsible for. Fundamentally, people’s performance is usually judged by the progress they make on their own tasks, not the overall benefit they provide to the company as a whole, since that’s a lot harder to measure. It would be nice to see this change, but until it does, there’s going to be resistance to collaboration, and the security argument is a useful tool in that resistance.
At lunch-time, I ended up sharing a table with Doc Searls, JP Rangaswami and Andrew, which was pretty heady company.
I had another tough choice to make after eating, since the conference split into two tracks. I wanted to see the panel with Adam Gross and John Crupi, especially after yesterday’s discussion with John about how to do something better with email, but Charles Armstrong from Trampoline Systems was speaking in the other room. They’re a really interesting company living in the same space I’m interested in; analyzing email to automatically figure out relationships within an organization, and then doing something useful with that. Dawn Foster and Aaron Fulkerson from MindTouch joined Charles on a panel themed "Social networking the enterprise".
Charles started with a description of his background as an ethnographer, and the inspiration for his work coming from his study of the communication techniques used by small communities. He’s using the insights he gained to write tools that analyze email and IM data, and use it to find experts within a company, or visualize the way people actually communicate within the organization. These are both areas that I’m really interested in, and it was great validation of the opportunities in this area to see how many customers Trampoline had picked up.
Alex Iskold from AdaptiveBlue/ReadWriteWeb then gave a talk on "A look at structured attention". He focused on the benefit to users of being able to control and share their own activity streams. As a practical example, if NetFlix could access your Amazon buying information, it could provide a lot better recommendations. He was proposing a model where there was some central, company-agnostic
data-store that all the services contributed to and pulled from. I’ve long been convinced that this would be a big leap forward, and allow startups without their own user-activity logs to do really interesting things, but I have a hard time understanding how to persuade Amazon to give up their competitive advantage.
Alex asked whether the big players would open up, and in the discussion at the end, I pressed him about what his answer was, and what we should do if they do keep saying no. He seemed cautiously optimistic that it’s possible to produce some client-side approaches instead, which is something I’m betting on too.
Dick Hardt gave hands-down the most entertaining talk of the day, packing in 450 slides in 12 minutes, on the topic of "Defragging identity". I really need to try something similar for one of my corporate presentations, just to keep everyone awake! He’d obviously practiced like crazy to get the talk spot-on, it was a virtuoso performance. The content was good too, a primer on the history and evolution of trust, and how it was all based on past behavior predicting what someone would do in the future. He took us from the village where you knew everyone’s past first-hand, to cities where you had to trust strangers. After urbanization, people turned to third-party institutions to provide certificates indicating past behavior, for example a doctor’s qualifications. In the same way, online we want identities that we can attach tokens demonstrating past behavior to. These may not be a single, monolithic identity for everyone, we may use different identities in different situations, such as online games.
After this quick talk, Esther Dyson stepped up to the plate with "Discussing attention". It was really useful to hear her perspective on targeting advertising using individual consumer’s behavior, as someone who’d been involved on the marketing side. She had just returned from some FTC hearings on the same topic, and proposed a solution that had definite resonance with Alex’s ideas. The proposal was that users get access to the composite profile information that services generate from on the raw click-streams and buying habits, as hey can do with credit reports. This would allow consumers to escape from being labeled in incorrect or insulting ways, the "My Tivo thinks I’m gay" problem. Esther didn’t have a fully-formed proposal, but it was an interesting approach, and she was looking for feedback and improvements from the audience at Defrag. It raised some questions about who actually owns that data, the user or the company that captured it. With my client-based bias, I’m still pretty convinced that we’ll never persuade those firms to open up, and we’ll need to run on the user’s own machine to give them more control of that information.
In "Customer reach versus vendor grasp", Doc Searls was on very similar territory. He’s rebelling against the constant obsessive measuring and pigeon-holing that’s behind personalized marketing. He asked "Who here wants to be better targeted?", and only one brave soul stepped up and said they did. Doc used a Walt Whitman poem to drive home the uniqueness and irreducibility of every human out there. That led to the idea that we should be able to control how companies see us, with him using the term "Vendor Relationship Management" to describe his approach, in opposition to the traditional customer relationship management that’s run by the vendors. He’s taken up the challenge of actually creating something like this with Project VRM, aimed at producing some practical software and standards to implement this vision. One of the compelling ideas he threw out is reversing the usual passive data model, where vendors pull information about user’s desires, and instead allow people to broadcast something they need, and see who can come up with a product that matches those requirements.
As somebody said in the discussion at the end, I’d love to lock Doc, Esther and Alex up in a room for a few hours, and see where their visions of the future match, and where they clash.
Ross Mayfield of SocialText gave the final talk of the day, and I was intrigued by the title, "Things to do in Denver when your corporation’s dead". Unfortunately he switched to "Made of people" instead, so I’ll never know how he would have lived up to the original heading. Ross’s talk covered a lot of ground, talking about the Radiohead album sale model, and how that approach could be used with other businesses, SocialText’s search for a new CEO and pulling in your customer’s expertise. The common thread with all of these is the active engagement of people all over the world in achieving your goal.
This just covers the formal talks, but of course some of the most interesting conversations happened in the hallway. I had a great chat with JC Herz about the work I’m doing on graph visualization, gave a demo to Robert Reich from me.dium, and received a demo of HiveLive from Greg Schneider.
I’m looking forward to another great day tomorrow, especially JC’s and Matthew Hurst’s talks on social visualization.
Funhouse Photo User Count: 2,099 total, 75 active.
Event Connector User Count: 105 total, 5 active.
Pete Warden's blog 