Machine Learning: A Newer Version Of A Lean Thinking Tool?
If you already are a lean thinker, then you know ML
Machine Learning is a hot topic in the computer science universe and it has filtered into our everyday existence. With tools like Alexa sitting on our desks and Siri in our pockets, and with online marketers predicting our every move, ML (let’s be cool and not use the full term) is helping computers and companies predict our every move and thought.
If you have data and want to make predictions, then ML is your jam. But, as much as ML sounds new, it is old. ML is the new way of doing what statisticians have done for a long time. The difference is that with computers, we have the power to work with larger data sets and to do the math phenomenally fast, and of course we were using the tools to teach us rather than a computer. While the scale is different, the fundamentals (data →statistics →prediction) remain the same.
If you accept the premise that ML is a new way to do an old thing, then the connection between ML and lean thinking is very straightforward. In lean thinking, we gather data, apply basic statistical analyses, and use the results to predict future behavior and events. We go further, obviously, since lean thinking involves much more than the prediction. We change things, measure the results, and repeat many times.
The Deceptive Allure Of The Word “Data”
“If you have data” packs more than you think into those few words. In the Big Data era, it seems like data surrounds us and all we need to do is reach out and scoop what what we need. The truth is different.
One of the most challenging parts of doing many ML projects, and the same is true of lean thinking events, is data acquisition. We still have many places where collecting data — useable data — is the exception. Since I am a lean thinker and I work in the legal industry, I will explore my little corner of the world.
Most countries proclaim themselves “Rule of Law” countries and certainly the U.S. is first among most. Let’s take a look at how easy it is to get law as data.
Federal law — cases, statutes, and regulations — is easy to get. At the state level, cases are iffy, statutes less so, and regulations impossible. All of these materials require substantial work before they are useable. This last part isn’t unusual in the ML world, but that doesn’t make it any less annoying. So, you can kinda-sorta get to the law in the U.S., if you are willing to work for it and probably spend a bit of money. But that isn’t where the real action lies.
Assume you want to know what happens in that iceberg below the water line. What happens in all the courthouses where filed cases are resolved without the court publishing an opinion. What happens in all the other encounters among parties that involve lawyers and legal issues. Here, you are out of luck. Data collection is not state of the art — for 1980! Data has never been a priority in the legal industry and our lack of interest in it is now coming back, in spades.¹
The Shared ML And Lean Thinking Data Challenge
This is where the connection of ML to lean thinking gets interesting. Lean thinkers often find themselves without data. If you don’t look at what you do as a process (ahem, lawyers) you won’t think about tracking what you do with associated metrics. Now, someone like me comes along and starts explaining that everything you do is part of a process. We can map what you do and show where to improve, but without data we are chasing ghosts. So again, lean thinkers need to find the data.
ML folks like big data sets. Variability due to things you can’t control gets smoothed out in big data, whereas in small data it sticks out. The trick is to know what you want to predict, how much data you can feasibly get, and whether that is enough to separate the signal from then noise.
Lean thinkers seldom get big data sets, they must live with small data compilations. This is especially true in law and a challenge for ML in law. How do you get people and computers to learn from small data sets?
This is a particularly relevant question as computers move into law. Let’s do a hypothetical. Assume you handle a particular type of contract and you complete one per week. We will give you two weeks off for vacation, so you do 50 contracts a year. If you work for 30 years, you will do 1,500 contracts. That is a drop in the bucket compared to the data sets ML eats for lunch. Sure, ML often must deal with data sets that size or even smaller, but it took us 30 years to get your data set.
Now that you know data is valuable, you might start saving the data. But you probably do not have a data collection system in place. And, that assumes you know what data to collect.
Then, of course, there are the confidentiality and privilege questions. You can cleanse the data so that you can use it, but you still must be careful. Anomalies can stand out, especially in smaller data sets, and give away relationships you thought you had carefully masked.
The Data Punch LIne
Data is the tie that binds ML and lean thinking, and the challenge for both computer scientists and lean thinkers in law. There are many data sets that both can mine and help lawyers better serve clients. But, equally, there are many data sets that don’t exist or remain deeply buried. The common point we all share is this: if we don’t start working towards developing tools to capture data and ways to integrate data into the professional services we provide, clients will find another reason to replace lawyers.
¹ “The American journalist and writer Damon Runyon used the expression that way in a piece for Hearst’s International magazine, in October 1929:
‘I always hear the same thing about every bum on Broadway, male and female, including some I know are bums, in spades, right from taw.’ ”
If you enjoyed reading this article, please clap (click the the clapping hands). The more you click (or the longer you hold down the icon), the more you enjoyed the story.
About: Ken is a speaker and author on innovation, leadership, and on the future of people, process, and technology. On Medium, he is a “Top 50” author on innovation, leadership, and artificial intelligence. You can follow him on Twitter, connect with him on LinkedIn, and follow him on Facebook.
