A.I., Section 101 and Battleship

I was just explaining to an author working on a piece about litigation research, which he refers to as A.I. (“artificial intelligence”), and came up with a metaphor that I thought was worth setting out in a little more detail.

The question presented was how useful docket analytics are when evaluating the likelihood of success of a motion before a particular judge, and the assumption was that knowing that the judge has denied a certain type of motion 70% of the time was a reliable indicator that a future motion stood only a 30% chance of success.

I disagreed and took the position that the percentage was actually not particularly relevant – the utility of the statistical analysis was that you could then look at the holdings and see if there was a fact was crucial in the motions. If you could identify the necessary fact, then you could predict with a far greater degree of accuracy that if you didn’t have that fact the motion was 90% likely to be denied, and if you did, you had a 70-80% chance it would be granted. In other words, you could do far better than the 70% that the bare statistical analysis showed for past motions. (This is related to the pro tip of actually making sure you have the necessary facts before filing a motion, but assumes that had not occurred to you yet, and you thought that courts were like slot machines – drop in a motion and 60% of the time you’re a winner).

This is even more applicable in situations where the relevant caselaw has evolved recently, for example motions asserting lack of patentable subject matter after Alice or motions to dismiss for improper venue after TC Heartland. In such cases the predictability of an outcome for a future filing increases as cases provide examples of what the Federal Circuit thinks is and is not patentable subject matter or a regular and established place of business.

Still from Battleship – the best movie ever made. Don’t @ me.

A good metaphor – especially for these motions – is a Battleship board. Simple grant/deny statistics might tell you than 65% or 70% of shots failed, but after the board has been played for a couple dozen turns, the grant rate for future motions played using that board can be estimated with a much better degree of accuracy because you can often make a fair guess where on the board your claims are. So you should be able to come up with a better guess as to whether you’ll score a hit than simply doing a percentage analysis of the prior shots.

It’s always good to remember that most tools don’t actually do the work for you. They just help you do the work better if you use them correctly.

© 2020 Michael C. Smith. Use limited to individual subscribers. Further distribution prohibited.

This blog does not constitute legal advice. If you’d like to discuss a related legal matter, please contact Michael C. Smith via email or call 903-938-8900.