史蒂文 :: Release 2.0

I recently read Scott White's (et al) discussions on Open Data and the Implications of Data Disownership and although I may not have caught the entire jist, it came to my attention yesterday during some discussions with a company i was talk with.

You see the positive thing (in this case) is that it is a government body. They don't make money opening the system via web services, but it helps third parties integrate and use their services. I suddenly started to see not only the open services that could be made available, but the data resuse that makes no sense to be duplicated everywhere.

We have NI numbers, company addresses, people addresses, qualifications and so on that are really just sematic data on top of the core functions of the system (they are each managed by a separate larger body), which is really about payments. So i think - why do we store and manage all of this stuff. I can say that most of the effort is around managing stuff they only need to know about and shouldn't really have to maintain. A company registration number can be used to tell us everything we need to know about a company, so why manage and store the entire company details. This extends to the other attributes. Opening the services would be great, but there is a snag.

You see they all want to open their systems, but charge to get to the data. Not only that, but the drive of services will see others use your service within their services - this is exactly a case we have just now. We provide an address lookup facility which is managed by a third party. However, the data itself is owned by the post office and the third party license that from them! So if we serve this data out of our own service then although the software people may be happy to open the API, the Post Office aren't going to be happy their data is floating around the web with a single license paying for it!!

The net result is duplication. And this extends to the other attributes and no doubt around the web there are many cases of this. So you get to a point where services can only DO things and the things you get become hazy. Or rather, what you can DO with what you get back is hazy - "Yep, I found 6 matches which may be the address they are looking for based on their query, but the license tells me i can't send it back to them."

In the enteprise I don't see an easy way out of this short term. However, if profiles were available direclty from the source (e.g. the person, the company and so on - ignore verification of that for just now), then maybe certain links in the chain could be ignored and we would be in a position where Open Data could be a reality.