Canadian Lawyer

w w w . C A N A D I A N L a w y e r m a g . c o m J U L Y 2 0 1 7 19 All these mental shortcuts help us when we don't have time for a careful assessment of all the evidence. And the explosion of data is making it harder and harder for experts to assess meaningful patterns across the huge data sets of evidence they're faced with. So we rely on our "gut feel" instead, backed up by anecdotes from our deep experience. These hunches on outcomes can work pretty well most of the time. But in the realms of professional expertise such as medicine, insur- ance, law and sports, some of the early success in "predictive analytics" has been in removing these inherent biases from the decision-making equation altogether and instead taking a more statistically driven approach. So where does that leave the expert? The surprising learn- ings from Philip Tetlock's studies (and others) are that while experts are able to systematically identify all the data points on which they make a decision, in the moment of actually making the decision, human biases intervene to interrupt this systematic and rational process. The power is in human plus computer (and with a great process in place if possible) rather than either one alone. We need experts to build the initial model; we need the speed and depth of an algorithm's processing power on our large data sets; and we then need experts to sense-check the processing outputs and interpret the outliers in the data set. A lawyer's deep expertise is often at its most valuable when they can predict outcomes for their clients. The grey area of the law means that the guidance of an expert to an element of certainty is invaluable. At the moment, this is largely done by hand. Lawyers scan databases looking for agreements and deals based on similar facts and arguments or they pull together reams of case law to see how previous judges have decided similar points. An algorithm, however, can process hundreds of cases looking for specific patterns in the blink of an eye. How use- ful is it to know how often a certain judge grants or denies motions to dismiss or motions for summary judgment? How might that knowledge impact the case strategy that you're developing for a client? Imagine using a set of algorithms to specify only those cases on similar facts across a much wider data set and that have been decided by a specific court or judge. And then imagine diving into the winning arguments and reasoning of that set of cases to develop a case strategy for your client. By focusing on only the disputable facts of a case that impact the outcomes, we can move to settlements faster and save both court time and client costs. Algorithms can do this churn task in a fraction of the time it would take us to research and analyze each motion and case hoping to find a pattern. They can apply regression analysis to masses of data and weigh a set of known variables in his- torical data. This then calculates a confidence level based on probabilities that a certain outcome will or won't happen. (I don't profess to understand much of how this actually works. Instead, I recommend that anyone interested should track down any of the white papers by Benjamin Alarie of the Uni- versity of Toronto. He has written several illuminating papers that explain what the technology can do, how it is being applied to law and the possible future of predictive analytics and machine learning.) This evidence-based approach to prediction, rather than relying on hunches alone, has the most appeal for law and, indeed, for clients. Using historical data at our firms along with our deep expertise, we can start to predict different outcomes and a full range of costs, along with an element of confidence in that prediction. Data really is the new oil and it will have a lawyer's hunch for lunch (to mix two genius quotes of Ed Walters) if we don't start properly tracking good data and building the models ready for the analytics engines that will help us predict out- comes better and more often. Kate Simpson is national director of knowledge management at Bennett Jones LLP. The opinions expressed in this article are her own. THIS EVIDENCE-BASED APPROACH TO PREDICTION, RATHER THAN RELYING ON HUNCHES ALONE, HAS THE MOST APPEAL FOR LAW AND, INDEED, FOR CLIENTS. CORPORATE COUNSEL Connect with Find almost 4,000 corporate counsel and over 1,500 organizations along with fresh editorial content, information on deals and links to important resources. Lexpert.ca/ccca ntitled-1 1 2017-06-13 7:57 AM

• Cover