Patent applications are a rare cash cow for the federal government, and I suppose it's good news that they soaring and are being processed more rapidly. But how does this square with the thesis, advanced most recently by Tyler Cowen and Peter Thiel, that the middle class is suffering because of a drought of technological innovation, whether caused by the exhaustion of simpler, more profitable ideas, or by the decline of the nation's educational system?

So there's a paradox. Patent numbers are often treated as a proxy for invention or technological creativity. There are many more now than there were at the peak of technological optimism in the late 1990s. So in addition to Tyler Cowen's low-hanging fruit metaphor, perhaps there might an alternative seismic or ketchup-bottle model, in which an accumulation of relatively unnoticed ideas unleashes a torrent.

There's one intriguing trend in the New York Times report on patents:

Applications have also become far more complex, said Douglas K. Norman, president of the Intellectual Property Owners Association, a trade group mainly of large technology and manufacturing companies.

"When I was a young patent lawyer, a patent application would be 20 to 25 pages and have 10 to 15 claims," Mr. Norman said. A claim is the part of the patent that defines what is protected. "Now they run hundreds of pages, with hundreds, and sometimes thousands, of claims."

Nobody really knows why this is happening. If the fruit is higher on the tree, maybe it requires more scaffolding to observe and describe it, much less harvest it. And if there are thousands of claims, is none of them very important or -- as some speculate -- is there a tendency to obfuscate the significant ideas with chaff, defeating the patent's rationale of disclosure? Of course there are other factors, like more extensive collaboration, and the rise of newer industries like information technology and genetic engineering. There's a European technical paper on the subject here.

Complex as the subject is, I have a bad feeling about humongous patent documents, based on one of the most notorious and the longest at least until 1918, the Paige compositor. You can read its 64 pages of text and inspect the 271 figures here. Mark Twain, an experienced hand typesetter, lost his personal fortune by investing in the machine, which appeared to be the Watson of nineteenth-century mechanical engineering, duplicating the work of the most skilled humans in a fraction of the time. Unfortunately for Paige and Mark Twain, it was so complex that one part or another was always breaking down. So increasing complexity may not be such a good sign. (Ottmar Mergenthaler's comparatively simpler Linotype, which was nominally slower and sprayed a bit of molten lead now and then, defeated it in the marketplace and dominated until the rise of photocomposition in the 1960s.)