Wednesday, 27 April 2016

Algorithms and the generation of nonsense: introducing All Prior Art

Our attention was recently caught by a curious new project, All Prior Art ("Algorithmically generated prior art"), which got a mention on beSpacific earlier this month. According to the All Prior Art website:
All Prior Art is a project attempting to algorithmically create and publicly publish all possible new prior art, thereby making the published concepts not patentable. The concept is to democratize ideas, provide an impetus for change in the patent system, and to pre-empt patent trolls. The system works by pulling text from the entire database of US issued and published (unapproved) patents and creating prior art from the patent language. While most inventions generated will be nonsensical, the cost to computationally create and publish millions of ideas is nearly zero – which allows for a higher probability of possible valid prior art.

Further, a large institution could dedicate many servers to this task, along with developing more advanced techniques such as deep learning, to flood the prior art space. It is not unforeseeable with current technology (along with sufficient cash for fees) to flood the actual patent application process itself with sufficiently advanced patent applications based on this concept.

A sister website All The Claims is attempting the same thing, but with the use of claims and a more verbose alternative.
This project seems totally misguided and, if it had any substance to it, it could be quite harmful to the innovation ecosystem. 

For one thing, the patent system is actually designed to democratize ideas, in that it ensures that all patent applications are published, irrespective of whether they lead to grant; patents that do not sufficiently disclose an invention to enable those skilled in the art to understand and make it can be invalidated; the facility of filing a patent is open to all and many individuals have benefited from its accessibility.  The fact that so many patents today are complex, difficult to understand and incapable of use except in conjunction with other patents is actually a sign that the system has worked extraordinarily well and has enabled teamwork and cooperation to produce advanced technologies that lie beyond the skills set of any individual -- it's a sign that innovators share their ideas and their analyses at an early stage, before a patent application can be formulated.

Another point to raise here is that an attempt to publish algorithmically-constructed prior art would not just work against non-practising entities: it would work against all prospective patent applicants, including small-scale and poorly resourced private inventors, whose intellectual creations would be just as unpatentable as those of trolls.

One might also wonder how any funding would be secured in order to support further research, development, health and safety testing and the other costs incurred by bringing any innovation to market. The existence of a patent is no guarantee that this funding will be protected, but it is at least a comfort and an incentive -- and a way of attracting money that might otherwise head into real estate, the purchase of art works or other less risky forms of investment.

The initiator of All Prior Art concedes that most inventions generated by algorithm would be nonsensical.  The same word can equally well be said to describe this initiative.

No comments:

Post a Comment