Info Feed Weblog
A Distributed Product Review Data Standard
The following is a very rough write-up of some ideas I have brewing in my head, so please excuse its preliminary character.
I have been thinking a lot about how the opinions of consumers can fit more coherently into the picture of a distributed information society. Through my work as lead product manager for my former employer dooyoo I had been working on related issues, and it hasn't left me since. One could call it a desire towards the perfect commerce infomediary.
The summary is: I think we could all profit from a decentralized way of standardised product reviewing, something like a standard for product review metadata. The data would be created by individuals decentrally on their own sites or other sites, and via the data standard it could be made accessible to centralized services and back to those individuals again in an enhanced way.
The ecosystem value of services like dooyoo (or Epinions or Amazon or AllConsuming) comes from them (a) grouping the information around a standard product catalogue (thus allowing for comparability of products) and (b) standardizing the various consumers' opinions into a common scale (thus allowing for comparability of opinions). Well, there's more value in those services (like the web of trust, quality rating of the opinions, standardised question sets to the opinions depending on product category, etc.), but let's stick to the two above-mentioned points for now.
The problem with all the information scattered around the Web in weblog entries, on private homepages, and in services like dooyoo/Epinions/Amazon/... is that they are not at all comparable to each other. I have some opinions at dooyoo, I have a few at Epinions, some at Amazon, and quite a few on my own weblog. This is not "scalable", as we say.
What could now happen if there was a way of combining decentralized product reviews with centralized comparability and identification features? Potentially, this could be achieved via a global data standard. Something like FOAF for product reviews. Such a standard would have to allow for quite a few things, thus it will be a pipe dream as of today, but let's list them for fun:
- It would have to allow for individuals tagging their decentralized product reviews with standard data of (1) rating and (2) product identification.
- Preferably, it would allow for the user to be uniquely identifiable, so that recommendation abilities via collaborative filtering and such could be employed.
- It would have to hang onto some (actually many) centrally available product catalogues. Via using open-source catalogues like the evolving MusicBrainz, one might be able to provide for longevity of the data. Only unique product identification (like ISBN numbers for books) will make sure the data is not cluttered.
- There would have to be an agreed-upon rating scale for the product ratings. Either 1 to 10 stars, or 1 to 5, or whatever. Just something that unites everybody.
- One would have to be able to have unique IDs for each review, so that they are tracable. Thus, there needs to be some kind of NIC for that.
- Ideally, and that's the hardest part, it would not only be done via active tagging by the user (most would probably lack the accumen and desire to do so), but it would be nice if this could be embedded into all the common tools of web publishing, like weblog software authoring tools.
A common usage scenario might look like that: I go to my weblog interface, type my product review, and I give it a rating via some button that is pre-configured in my weblog software interface. When I submit the entry, a (or "the") central review ID server is pinged: It will hand an ID to my review and propose the product I am writing about. I confirm or change this, and submit again. My review metadata is being written into my template by my weblogging software, and the central ID server stores a reference. When Google/other search engines/other Internet services in general are indexing my site, they find the metadata. Via the standard, they know what product I have rated and what exact rating I have given. Doesn't sound like a lot of information, but that's almost all there needs to be centrally. Just IDs basically.
Now, how could this be used? I could find out what other people think about the current mobile phone via going to any of the services utilizing that metadata. And I could thus tap into the opinions of not just a few people that use this particular site, but the whole collective knowledge on the Web. Perhaps later, I could drill down to users that are like me, or that are from the same country as me, or whatever. The value of product information and recommendation increases dramatically with the number of participants. How many books don't have any review at Amazon, yet have reviews at dooyoo, or on weblogs? Combining all this collective knowledge would be mind-blowing. Google plus dooyoo plus weblogs plus Amazon on speed.
Services like dooyoo could add to their own data with these collective, decentralized reviews. Other services could build product catalogues adding more information to the products, which combined with all the reviews, could be a hugely valuable information to people wanting to make a shopping decision. Thus, this would also be a boon to eCommerce shops who could enhance their services with this data. Marketing could become more interesting (not going to dive into that one, but let your mind work). Or to go completely wild: Think about this much-talked-about concept of augmented reality, where people add data to the real-life environment: I go to a museum, and just when I'm there, I send an SMS to my mobile operators review server with a simple 3/5 rating of the place. Via location-identification, they can pick up where I am and add this data automatically to the global knowledge collective.
Whoa, OK, I am going on a huge limb here, dreaming up potentially unrealistic stuff like so many others have done before, but doesn't that sound like a nice vision?
It seems most of the technology pieces are coming into place. Web services and XML-RPC interfaces, proliferation of decentralized publishing by the individual user, intelligent parsing and data extraction technologies, and all those other nice buzzwords all of us hear about all day. Well, they've been coming into place for years, but at some point they will cross the border and make such a service possible. When that is, I am not sure. But it seems close.