Metadata have very interesting feature to organize information: .They enables a many-hierarchies classification : For instance the tags on this blogs enable as many a one level hierarchy per tag, aka a grouping. If I add one hierarchy underwhich to file entries, I would have to file each message under ONE node. Without it you end uup with the nightmare of having t choose weather a car's bill should be filled under the car node or the bill node, and you have all the chances that part of those items will be there, another part there : your information will be scattered, disorganised. .They provide with an extensible type system : You dont have to own the object to add some comments to it, or other datas. Metadata natively enable distribution wich is the underlying character of the internet... and of information itself ! So having a correct system to handle metadata is an essential piece of work in IT. Actually, regarding the first aspect, the french secret service, back in the time where everything was paper based, were famous at some point for having a cross-reference system which would emulate physically the tag system. Isn't it interesting to see how non-linear history can be?