I've been following the Data Portability initiative with interest. There is lots to read on their site, and the sites of many participants. (As you should expect, the debate is pretty portable across sites).

I think about data portability in terms of a simple use-case:

I'm a user on a photo site, and I want to edit my photos. I can configure the photo site to launch a web photo-editor of my choice, and pass the photo to the site. I can edit the photo, then pass it back to the photo-sharing site. Neither site was aware of the other in advance. They just supported a common way of passing data about.

(This is not my idea, but I can't find the website where I first read it…)

It is like a supercharged version of the way that plugins work in desktop software. A very exciting development from a technology and user perspective, but also exciting in terms of industry development.

Taking a macro view, this kind of interaction is bound to increase competition, which means better services for users. As an illustration, I'm struck by the strong support of Strands for Data Portability. Strands are well resourced by start-up standards, with some very experienced engineers and scientists working there. It seems that most of the effort goes into a serious recommendation generation technology, which has every chance of being world-leading. They have a consumer-facing product for recommending music (and music videos), and a newer financial recommendation product. In music, at least, they look like an also-ran compared to the market leader, last.fm. That is a shame because their underlying recommendation engine is probably better.

Data Portability offers the prospect of picking software components in the same way that you pick products at a supermarket. I can choose the last.fm music community, but log in with a Google account and use the Strands recommendation engine. Or I can just stick to Google brand for everything, if that's how I like to do things.

It might sound great for users, technologists and macro-economists, but we're not very far along, and there are many obstacles to overcome. I reckon this world will come eventually, and quicker than we think, because the Internet is fundamentally a competitive place, with barriers to entry much lower, and speed of innovation much higher, than in most physical industries.

So, the question: Where is the money?

The Data Portability group have some initial thoughts, but it is still early days. One of the main participants has a specific interest in broadcast, which looks interesting. Social Times also has an interesting perspective.

I don't have a firm answer to the question, but it is important to us at Meta Broadcast to understand what might happen, so I'm spending a fair amount of time trying to figure it out. I plan to blog my thoughts and join in the discussion.

This is where I got to so far:

The business of running a 'social network' (user accounts and some very important information like links to friends and demographics) is likely to be critical to on-line advertising networks such as Google, Microsoft. Possibly Facebook can also join this gang, if Mark Zuckerberg stops doing things in a careless way. There are likely to be strong economies of scale in this business because ad buyers like to go to one place to buy large audiences. It's no place for a start-up, unless they have a market-shaking idea AND lots of cash.

As for other apps? I wonder how far the supermarket analogy can take us? In a supermarket we choose on distribution, brand, price, perceived quality, and perceived convenience. You can build a nice business if you own a good brand, which you build through a mix of good communication and doing a good job. That suggests that someone with a big brand can market multiple components to consumers. The Internet equivalent of Nestle - a safe but basic choice for lots of products.

What about a niche brand? I'm a huge fan of Tunnock's, a Scottish biscuit brand. They're very successful, and all about their Victorian working-class roots. There will be space for niche brands too. But, just like Tunnock's, a niche brand will need to get good distribution. I wonder whether that will get cheaper or more expensive? Niche Internet brands will also need to ensure compatibility with the consumer's other choices. That is a problem that Tunnock's don't really have. Maybe this is a place where incumbents will be able to trip up newcomers?

Brand clearly won't be the only helpful asset. Some data will also remain a competitive advantage. If you have a nice enough brand you can attract people to give you privileged access to their data, which you can carefully use to build the competitiveness of your service without causing privacy concerns. How far you can go is another big question.

Data Portability could make a massive difference to the way that Internet businesses create value, and bring real benefits to users. Now is the time for businesses to think carefully about how it might affect them.

I'm looking forward to talking to some smart people about this stuff during the next months, and reporting back here. Do let me know what you think.

UPDATE: There's a fair bit of criticism at the moment about the ratio of discussion to action among members of the Data Portability group. I'm inclined to agree that more implementation should come before detailed discussion (that is certainly the approach we are taking with URIplay). However, the principles behind Data Portability are important, and the practical effects are likely to be felt sooner rather than later in some application areas.