I listen to a lot of audio during the week, much of it from The Conversations Network. Here are a few shows I found especially interesting this week.
- Put all your data in one place. New correlations and ways to analyse the data will emerge.
- With separate data spaces, you will miss important, valuable relations.
- Save queries (again, in the same space) and notify of changes and updates.
- Stored queries can connect to other queries.
- Stream data into the system and learn and fix errors (repair history) along the way.
- Sequence neutrality. End state is the same regardless of the order the data arrives.
- “… when you dream, you’re doing deep recontextualization … to remedy some things you actually have to go offline for.”
Mandating coherence and structure doesn’t work.
- “Valid” means different things to different people and groups.
- Imposing structure in separate realms can and does work, but designing for interoperability between realms creates all kinds of new value.
- Just get all your data together and let structure emerge/evolve/develop.
- Don’t throw structure away, defer it until you really know what you need.
- RDF and graphs. I’ll admit right here that I don’t really understand these (yet ), but I may have to look into them more.
- Use peer pressure and self-interest to build interesting, valuable, open data sets. It’s the only way to make this stuff happen.
- “There’s no such thing as quality of metadata.”
- “… there’s that perception that coherence is quality.”
- “Data first, instead of structure first.”
- “Just start off, and write down whatever you want, and then you can incrementally add structure to it, and make value out of the structure as you build.”
- “There’s a lot of structure in email, if you want, it just depends on what kind of structure you want to look at.”
- “A lot of the semantic web research was based on the big hypothesis that was ‘If whatever, then we could do this.’”
- “… when they show up, we can party.”
- “It’s clear now that it’s all about data and loose pieces connected together than it is about uber ontologies …”
- Keep things really simple to start. Just get basics working so we can build on them.
- The potential of OpenID endpoints for service discovery. I hadn’t thought about this, but it’s really obvious once you hear it.
- These technologies are actually great for big companies. They can make use of them without having to be the initiator and deal with the suspicion and other associated problems
- OpenID takes a big load off of developers by handling their whole authentication system for them.
- Interesting balance between design-by-committee and just-build-some-stuff with DataPortability and DiSo.
- Keep things loosely joined. Don’t depend on one tech for a task, make it all swappable.
- Being the repository for “master profiles” is valuable.
- Focus on people who already get it (to whatever degree), and build momentum. You can’t convince other people without that base.
- Building a personal reputation based on trust of opinions and knowledge is extremely valuable.
- Monetizing too early can be dangerously limiting. Tough balance.
- “Simple usually wins in the long term.”
- “Help people discover new things.”
There’s definitely a theme there. Some other stuff:
- Jon Udell interviews Fernanda Viegas & Martin Wattenberg about Many Eyes, an interesting social data visualization project.
- Whedonesque reported that Joss Whedon would be on the GeeksOn podcast. And he was. I checked out a couple of other episodes as well, and it seems like a really interesting, well done podcast. I’ll keep listening.
- Chris Messina on DiSo is a nice introduction to the DiSo project.