Can big data us help us reach the right decisions? Will big data give me a competitive advantage? How do we adopt and implement big data?  


questions about big data? we can help

Just as businesses have taken advantage of massive datasets and fast computing, our law firm has embraced big data disruption. As engineers, programmers, and scientists, we have downloaded and wrangled patent data to extract numbers such as pendency and allowance rates. We regularly incorporate such statistics to help our clients make decisions. But, we also know the limits of statistics, and understand that the true potential of big data is not in counting notices of allowance, but in accurately predicting trends and identifying hidden classifications and associations in the patent data.

Blindly accepting statistics without understanding what they mean can be misleading. Confirmation bias, false correlations, and over-generalization are just a few of the common statistical traps. For example, does an attorney's high allowance rate reflect high quality or low quality work? What if the high rate is the result of taking prosecution short-cuts at the expense of breadth? Or, what conclusions can we draw from an Examiner's low average allowance rate? Does it reflect a stubborn Examiner or might there be another explanation? What if the Examiner was recently assigned a large number of new cases in the past year that raised his caseload by 60%? Finally, how are some of the Examiner statistics being sold on the internet better or worse than the USPTO's own Enhanced Patent Quality Initatives (EPQI) and metrics reflected in the EPQI Master Review Form.  

These are just some of the real-world examples we have seen that seem self-explanatory at first glance, but in fact, require careful consideration. As famously warned by Darell Huff in his 1952 book "How to Lie with Statistics," statistics are often misused to "sensationalize, inflate, confuse, and oversimplify." This is not to say that we do not see the utility of stats in getting a clearer picture.  We are strong proponents of using numbers to help guide decisions, so long as each assumption used to support the decision-statistic is understood and carefully tested. But rather than just talk hypothetically, let us show you how we applied predictive modeling to Information Disclosure Statement (IDS) analysis, and how we can do the same for you.