Patent information - a dodgy statistic?

The IPKat's scholarly friend and colleague Nicola Searle, from the northern climes of the University of Saint Andrews (WITHOUT an apostrophe) writes to him with a very real question:

"I have read, many times, the following: “It has been said that more than 80% of the information in patents cannot be found anywhere else.” This is the kind of statement that you would expect to be accompanied by a reference or at least some support. However, I have not been able to find any.

The statistic, without support, can be seen at the following locations (among others): The UK Intellectual Property Office, the European Patent Office, Trinity College, Dublin and IP Frontline (Patent Café).

Any ideas where this magical 80% comes from? Perhaps your readers might know?"

Above right: Like Pooh's search for honey, patent information is easy to locate outside the patent system if (i) it's there in the first place and (ii) you know where to look for it.

The IPKat doesn't know, though he seems to recall that in his younger days, when he was just a little kitty, the commonly-cited but never substantiated proportion was a good deal lower. He suspects that the 80% figure might be the result of morphic resonance with the intellectually respectable-sounding 80% of the Pareto Principle.

Left: Pareto - is he the innocent culprit?

Merpel adds, I'd expect the proportion to be high because

* almost by definition, any patent that satisfies the criterion of novelty is going to contain material that isn't in the prior art;

* the result of any imperfect search for patent information outside the patent system, where the person looking for it just plain misses it or can't pinpoint it in the language(s) it's in, will inflate the percentage further;

* patents are a good deal easier to search than anywhere else, because of the accessibility of convenient computer search and the use of the International Patent Classification system;

* information contained in patents that is downright wrong or off-the-wall - and it seems that there is a fair bit of that - is less likely to be found elsewhere than that which reflects or is based upon the conventional wisdom;

* it also depends on when you make your measurement: on the date on which a patent application is published, the chances that it consists largely of information not found elsewhere are obviously going to be a lot higher than if you do your search five years later.

Thoughts, please? The IPKat and Merpel suspect that their new friend Tufty may have some insightful comments here.