After I wrote Graph Databases: Information Silo Busters it occurred to me that it names what Google (as an interface) is missing:
- Declaring relationships
- Persisting declared relationships
Think about it.
Google crawls a large percentage of the WWW, peering down into first one information silo and then another. And by using Google, I can look down into the same information silos. Better than nothing but could be game changing better.
What if instead of looking down into one data silo after another, I can gather together all the information I find about a subject?
Much like you create a collection of bookmarks but better. I get to say what “string” that the various URLs have information about. Nothing fancy, no complicated, whistle out your left ear syntax, just a string.
See how people like that. If successful, bump that up to two collections with a limited set of relationships (think of operators) between the two strings.
Oh, that’s the other thing, relationships need to be persisted. Think of all the traction that Facebook and others have gotten from persisting relationships.
Unless you know a good reason to throw away searches and whatever declarations I want to make about them?
I can hear Googlers saying they already do the foregoing with all the annoying tracking information attached to all search results. True that Google is tracking search results and choice of sites for particular search requests, but Googlers are guessing as to the relevance of any result for a particular user.
So, rather than guessing, and remembering that making Google more (and not less) useful to users is the key to ad revenue, why not give users the ability to declare and persist relationships in Google search results? (Any other search wannabe is free to try the same strategy.)