Two day conferences with Geeks have their own challenges. Showering habits are not always that, well, thorough, so you get an idea of how the after party yesterday evening ended by the smell.
At the another hand, they are very nice folks, and pass ethernet connection cables around as if they were joints.
On with the show…
Robert Scoble started with a historical overview of networking online and then went on with what it means today. Interesting observation: networks start becoming interesting only after they get being used and shared. Twitter really sucks when you first log on and have no network (yet).
In fact we know this already since way back, in the ’90s, when every small-medium business was required to have their own forum on-line (user interaction, you know!) with two odd posts lingering around. How sad!
Scoble about social media: “The first experience is a crappy experience
Werner Vogels of Amazon
This keynote was in fact one big sales pitch for the Amazon Services like S3, E2C and friends. But it was a clever and enjoyable pitch.
He focused on Push & Pull models in contrast. Traditionally, we had only the push model, where companies produced goods and forces these top down onto consumers.
Relations have been reversed, customers find what they need and go after it (also in the B2B marketplace).
An example; about a Chinese company, that selects subcontractors on products they make already (no development, just build more of what is already there.
Push vs pull
- demand is anticipated – vs – demand uncertain
- top down design & control – vs – emergent design
- centralized – vs – decentralized
- procedural – vs – modular
- tightly coupled – vs – loosely coupled
- resource centric – vs – user centric
And then the Pull model life cycle, which goes like this in a circle:
find – connect – innovate – reflect – (find, new selection based on reflection)
So, swap out resources which don’t perform well enough instead of redesign new products yourself; syndicate the innovation at the product- or part level.
Statement: “you are the missing puzzle piece in a map of connected pieces; that’s your added value: you make the connections”
Resources is becoming a dirty word in this setting, because you don’t know when you will need them. The pull model requires that you can acquire and release resources on demand. Pay as you go and only for what you use.
Running a server infrastructure (as a startup) is waste of money. Focus on inovation instead.
“The eXtreme Amazon.com” organization is service oriented (in small teams). So they developed a service oriented, on demand architecture, which now forms the basis for the Amazon services s3, e2c and friends.
This – indeed – is a nice infrastructure of building blocks, and others are doing very nice things with it. Like this: Scalr – a fully redundant, self-curing and self-scaling hosting environment utilizing Amazon’s EC2.
One note from myself: there are alternatives providing on-demand resources. We are using slicehost, where you can order “slices” of managed servers – and upgrade them on-demand in a matter of minutes.
Werner Vogels: “Everything fails all the time” (after the fact that the internal Amazon infrastructure copes gracefully with constant loss of components, up to complete data centers)
The Future of Search and Discovery
Garrett Camp – StumbleUpon.com
His keynote was all about Search vs Explore, which in itself is an interesting theme. StumbleUpon is supposedly a real expert in discovery, but disappointingly Garrett talked about almost every theme around search, advertising and finally just a bit about discovery.
He foresees a great future for recommendation services like Pandora and Digg, without going into great detail of their characteristics.
Garrett Camp: “one-size-fits-all in search is history”
The last session: Khris Loux interviews Chris Saad about Dataportability
Chris Saad, Co-Founder and Chairperson at DataPortability.org and CEO of Faraday Media. Also Co-Founder at Media 2.0 Workgroup and APML Workgroup
A well done interview and lively discussion.
Chris raises the question: “If all services collect data about me, why not make it explicit, open it up (to me, the user – make me owner again) and merge it all together?”
This will lead to better search at google (e.g. books) and better recommendations at Amazon (context aware). And I’m in control to remove (part of) my data, if this is going to improve my experience. Guess what: if the enhanced services live up to their promise, I will leave my data in, in order to benefit from the better service! But at the same time I stay in control.
We (end users) should request from our vendors that we get open data. And OpenID for login, instead of yet another proprietary login on yet another service’s system. Which means that the provider will have to do another effort, in parallel to their traditional login system, if only for a transition period when not every user will have an OpenID account. If…
Also, open data is about open standards, so use OPML, Microformats, RDFa and so on and start using/producing open data. These technologies enable users to esport and re-import their data in other services and applications if they want to.
User lock-in should be based on great service, not on a technical / artificial barrier.
Oh, and we get another overloaded buzzword for free: data 2.0!