Nathan, over at Flowingdata made a post recently about some of the shortcomings he sees in the Swivel concept - Swivel is site that allows users to upload, share and comment on data. I've mentioned Swivel before, along with some of the other sites trying to make a name for themselves in this area (ManyEyes, Data360, BO Insight, Freebase).
I think Nathan hit on a few good points - poor visualizations, statistical validity and data quality.
I agree that the visualizations could be improved, but you don't need an expert in this area - there is a lot of literature and commonly held 'norms' that good sites and products adhere to. Follow these and you should be fine.
The whole issue of statistical validity and data quality is tied up in the larger question of what does Swivel want to be? Adhering closely to the world of statistics and enforcing strict quality requirements will make the site less accessible to the very people it wants to attract - non-statisticians. But not doing so makes it more susceptible to comments like Nathan's, and potentially undermines confidence in the site.
Is there a happy middle ground between validity and accessibility? Probably. But is it a compelling proposition? Not really.
It's not compelling because, fundamentally, DATA is not compelling. Accurate or not, DATA is boring. I know, I've been presenting data to people for years. They could really care less about it. What they do care about is what it allows them to DO. Data is just an input into a decision, a thought, a process. It's meaningless without a context or as an end in of itself. The guys at Swivel know this. They write about their featured graphs in a way to promote context. But unfortunately, the discussions don't seem to take hold. Why? Because if I want to debate interest rates, hurricane Katrina or suicide in the US, I am bound to find specialist sites with more interested parties elsewhere.
As a result, I see Swivel stuck between being a specialist site about data (which is not compelling to anyone but statisticians), and being a generalist issues site driven by data (which is mildly compelling to everyone else, but not particularly compelling to statisticians (although we don't really care about what they think :)).
So what to do? My two cents follow:
- I'd move to shore up the validity and quality of the data on Swivel, and put more emphasis on people embedding and blogging about it in their own forums, rather than trying to have the discussion on the Swivel site. I know this is a pretty big departure from what Swivel is trying to do. But at this stage, I think it's a far more compelling 'tool' than 'destination'.
- In keeping with this theme, I would really put emphasis on improving the visualization tools. It would be great to have Swivel as the default source for all data visualization in blogs.
- Definitely concentrate on the Official Data source program. I love the idea that there is a site I can go to get data that I know comes from a reputable source.