This likely won't be everyone's cup of tea, but there is an interesting academic paper by Professor Colleen Chien of Santa Clara University School of Law entitled "Predicting Patent Litigation." You can download a free copy of the paper here [PDF] on the Social Science Research Network.
The paper is interesting because Prof. Chien identifies a number of objective markers that identify a patent as being more likely of being asserted in litigation than the general population of patents. The really useful aspect of her analysis is that many of these markers can be discerned prior to the patent actually being asserted.
Some of these markers are what Prof. Chien refers to as "intrinsic characteristics," i.e., characteristics that are in inherent in the patent itself. Such intrinsic characteristics include:
More important than intrinsic characteristics for predicting the likelihood of assertion are the patent's acquired characteristics - those things that happen to the patent after it issues. These include:
- they have more claims than unlitigated patents;
- they have more prior art citations (indicating that they have been more thoroughly vetted at examination);
- they are typically found in families of related patents;
- they are more likely to be originally assigned to the inventor or to a small company; and
- they are more likely to be issued to a domestic U.S. company than a foreign company.
Many of these characteristics are indicia of the perceived value of the patent and are not surprising.
- the number of times the patent is reassigned to another party other than by merger or acquisition of the company holding the patent;
- successful completion of a reexamination, particularly a reexamination sought by the patent holder;
- continued payment of maintenance fees on the patent;
- whether the patent holder has borrowed against the patent (securitization); and
- forward citations to the patent (the more later issuing patents that cite to the patent, the more important the patent is likely to be).
Having identified these numerous characteristics, Prof. Chien then tests them against a random sample of litigated and non-litigated patents. What she found is that these characteristics dramatically distinguished between the two groups. The one problem is that some of these characteristics only show up after the litigation has ensued. The information is not of much use at that point.
So Prof. Chien worked her way back through the data to determine what could have been known before the litigation commenced. What she found was the markers were still just as dramatic and statistically significant in identifying patents likely to be litigated.
The model is not perfect. It still only correctly predicted a patent's likelihood of being asserted in litigation about 54% of the time, but that compares to only 25% with no model and 35% when only looking at intrinsic characteristics.
While more research into this subject is needed, I found the paper useful and thought-provoking.