Premonition’s Toby Unwin: I Think the Fundamental Problem With Law Is That Financial Incentives Are Misaligned
The legal industry is evolving. The mechanic routine tasks are now more often automated, leaving legal professionals with more time to work on complex problems and make strategic decisions. Or at least that’s what law firms want us to think.
In order to figure out what’s going on with the modern legal industry and legal tech, we’ve reached out to Toby Unwin, co-founder and Chief Innovation Officer of Premonition. Mr. Unwin’s company is working with litigation data to evaluate individual lawyers’ success rates in dealing with certain case types before certain judges. We’ve discussed the main problems of the industry, as well as the potential solutions for them, and talked about the future of legal technology and the respective professions.
lawless.tech: Please, tell us about your way to becoming a litigation innovator and how the concept behind Premonition came to be?
Toby Unwin: The short story is that I got sued a lot and it really pissed me off. The US has a ridiculous amount of litigation. 95% of the lawsuits comes from the 5% of the world’s population. I’ve been sued for things like not having an elevator in a one-story shopping center that I owned. I used to do things like collect the lists of the “best lawyers”. There was one particular lawyer I’d hired, whom I meticulously researched and who should’ve been the best. But they were actually horrible and the only one I ever filed a bar complaint against. I realized that I really didn’t know what I was doing, and neither did anybody else.
There was no objective measure of how good a lawyer is. Case management software usually doesn’t even have a field where you can put in the outcome of a case. I was talking to a person from one of these companies that’s been around for about 20 years, and asked why didn’t they have this outcome field. They just said that no one ever asked for it. It’s crazy that as a profession law doesn’t keep track of case outcomes, wins or losses. I figured that this is a data problem and if I can get the data and crunch it up, I can find out who’s winning which cases in front of which judges. Getting that data was very difficult. There’s no centralized source for this, it’s spread out across thousands of different courts. So we had to build a system that would gather this automatically, keep it all normalized and metrical quality stays up and so on. That was quite a complex undertaking which the industry at the time regarded as impossible. Therefore no one did it.
We wound up in the odd position where we’re the largest online litigation database and have more coverage than LexisNexis, Thomson Reuters and Bloomberg combined.
lawless.tech: Was the introduction of AI a part of the initial concept or was it a subsequent addition?
Toby Unwin: Unfortunately, we’re in this kind of industry where people call things AI that really aren’t AI. In its very basic definition AI is just a simple algorithm. We do have some things in our system that involve machine learning and such, but 98% of what we do can be tackled by a simple algorithm. You can use machine learning for the same task, but it would be a lot slower and less precise. AI is something we added later, but most of our work doesn’t rely on machine learning and it doesn’t need to.
lawless.tech: In one of your interviews you’ve mentioned that legal tech is broken. Could you please elaborate on this notion? What are the persistent problems with legal tech and what are the potential solutions for those problems?
Toby Unwin: Legal tech is appalling. It’s a good 15 years behind fintech. I was giving a speech on legal innovation in Australia. On the second slide I had to say that there is not much of legal innovation at all. On the next slide I had the word “innovation” in speech marks, because people say that things are innovative, but they aren’t. We see all kinds of nonsense being touted around as being disruptive and innovative. There was one company that came to see us and they had the ability where they could videotape a deposition, and I was waiting for the punchline but there just wasn’t one, that was it. It’s just really really bad. There’s a number of reasons for that and if you want to get into detail, I wrote an article called “Why Is Legal Technology So Bad.”
The main reason I see for this is the law firms themselves. Law is still for the most part delivered on an hourly billing basis. The problem is that it actively discourages innovation and efficiency. There are people in law firms called “Innovation Officers” and they have a horrible job. Most of it is organizing window dressing type of activities, so the law firms can pretend to be the top of the game in terms of technology. Some of the firms even have incubators and make small investments in legal tech companies that just come out to reveal inferior products that no-one uses. After a few drinks on a Friday night they’ll tell you the reason. If law firms used a product that actually makes the process quicker and more efficient, the partners will refuse to use that product because their billing goes down.
Ironically, you wind up in a situation where a potentially successful product has no market. When we see legal tech projects and they are asking us for an advice,we tell them simply not to do business with law firms and focus on the real end users, such as corporate litigation departments.
A lot of people are stuck on the idea that the work can be done better, so therefore it will be, but that’s not how it is. Simply because the financial incentives are completely misaligned.
lawless.tech: In this regard, can you think of a different financial motivation system instead of the hourly billing, the one that would potentially incentivize innovation and efficiency?
Toby Unwin: On a plaintiff’s side it’s “they don’t get paid if they don’t win.” Also, I see flat rate as being a good incentive. People say that litigation is different from case to case and you never know for how long it’s going to be, but that’s not the real reason. Yes, cases are different, but if you litigate a lot, you know roughly how long it will take and how much it costs. Unfortunately, if you go to a lawyer and say “I have this lawsuit, how much will it cost?” they’ll tell you it will be, for example, $25,000. They know well that it’s going to be a $100,000, but they have to lie to you because they know that the truth isn’t going to get the case and they know the other lawyers will lie as well.
Every time I paid a lawyer by the hour, I was hugely disappointed. Every time I paid a lawyer a flat fee, I was very pleased. The other thing which we are working on with some of our clients is the TOBY number concept. If we know the case outcome and how much it wound up costing in terms of damages and such, and we know what the legal billing was, we can put all those numbers together and come up with what I narcissistically call the Total Operating Budget Yield, or the TOBY number. It represents how much value has that lawyer created for every dollar you’ve paid them. If you’re on a plaintiff side and someone gets a judgement and they’re taking a third of it, the TOBY number is going to be 3. They created $3 of value for every $1 they charged you.
This is great, because finally you have a defence lawyer, whose incentives are aligned with yours. Therefore a lawyer could say something like “I guarantee you a TOBY number of 1.8, and everything I get above that I split, how about that?” In this case they are incentivized to solve the case quickly so the billing is low.
Also, it means that if you have a case where the result was not good and you are looking at a TOBY number of 0.9, you gave them $1 and they created $0.90 of value, then you both know that that guy won’t be hired again. Let say, in a $10,000 case, it’s in their interest to call you and say “I don’t think I’ve had a good enough outcome for you as I’d like, so I’d like to give you $1,500 back.” Now they have a positive TOBY number and they can be hired again. The client’s and the attorney’s incentives are finally aligned for once. I think the fundamental problem with law is that financial incentives are misaligned and it causes practices which are against the client’s interests.
lawless.tech: What applications can Premonition’s technology have outside the legal industry?
Toby Unwin: There’s actually a bunch. Once, we were going to hire a lobbyist. My CEO and I both had very bad feelings about him, but from completely different areas, though the candidate was well-recommended, and from a good firm and so forth. I said that if we replaced the word “lobbyist” with the word “lawyer” we would be laughing at ourselves. After all, there’s data out there, so we simply needed a proper dataset and some changes to the algorithms to see if we can crunch it. It wasn’t that hard and the results were quite illuminating, even though we didn’t do much in this area, now I can tell you which lobbyist get which bills passed with which committees, agencies and politicians.
We’re talking to some clients about doing more analysis of their workers compensations cases. The interesting thing here is that it involves a lot of medical data, allowing us to do the same we do for law: to see which physicians, procedures, drugs and hospitals are the best. Is it worth for client to pay extra for a particular type of procedure, and so on. There’s a lot of things that can be done in this area.
We also work with insurance companies and they are starting to use us in risk and underwriting, figuring out how likely someone is to wind up in a lawsuit. Insurers underwrite risk based on historical data and most of the time they make a lot of money. But when the numbers change, things gets very expensive very quickly and that’s a big problem for them. With our data they can see what the risk is in real time.
I have some ideas that because things like individual policies are linked with risk pools, to insurers and reinsurers, and to insurance-linked securities. By tossing a bit of machine learning at it you can figure out how someone getting an accident down in South Florida will affect a particular risk pool, then affect a particular company, then a re-insurer and then insurance-linked securities. You can actually see into the future, because it takes time for these effects to get felt throughout the market, so we would know where price is going to move in advance. It might sound a bit far-fetched, but we’ve already been able to do this in the stock market, proving that the banks foreclosure performance in litigation was predictive of future relative stock performance.
lawless.tech: Premonition shows that certain lawyers have higher chances of winning a case before certain judges. Does it point to the fact that judges are always biased towards particular lawyers or certain lines of argument and are therefore somewhat unprofessional?
Toby Unwin: You could say that. But I’d take a slightly different approach and say that judges are human, like you and me. If we were judges we would have certain biases. I’m sure if I was sitting on the bench plaintiff lawyers would hate me, because I believe that most litigation in the US is frivolous and I’d be tossing cases like yesterday’s sushi. We also have people that are friends of ours and, as impartial as we try to be, we’ll always going to be nice to our friends. We also have a memory of what’s going on in the courtroom. If I remember a chap who is bullying and lying every time he is in the courtroom, I’d be less inclined to believe his so-called facts. And then there would be people that present in a certain way. So one of the things we look at when we are examining judges is their Pro Se win rate.
If people go before the judge to make an argument themselves, rather than hire a lawyer, how likely are they to win? There is a mantra in law that if you represent yourself you have a fool for a client and you should expect to lose. That’s not necessarily true. You usually have a Pro Se win rate of about 42%. It isn’t as good as hiring a lawyer, but it isn’t terrible. Yet, many judges have Pro Se win rates in the sixties. There’s a court in the UK with Pro Se win rate of 75%, so the non-lawyers are beating lawyers heavily over time.
If you are not a lawyer and you have taken a case through a court of appeals, that kind of says that you are quite an intelligent and organized person, but typically Pro Se people are going to make an argument from equity. They will say “this isn’t fair”, because they don’t necessarily know the laws, statutes, and precedents, whereas many lawyers will make an argument based on the law. There are judges that like to see arguments based on the law and there are also judges that are amenable to an argument based on equity, because they think it’s their job is to make sure things are as fair as possible. Measuring this plaintiff win rate can tell you whether that’s the way they think. Both of those arguments are actually true.
So, yes, the biases are in the legal system. We found that 30.7% of the average cases involved a relationship between the judge and the lawyer. Is that fair? No, but that’s where we are at the moment. I personally believe that every change of law will come from transparency in it. Eventually system like Premonition that bring transparency will make justice fairer. But it won’t happen quickly, it will take time.
lawless.tech: What do you think about automation and introduction of AI-based solutions into the legal industry? Will there ever be absolutely unbiased robotic lawyers and robotic judges?
Toby Unwin: Automation in law is inevitable. At the moment, law firms are kind of dragging their heels on this one pretending to be enthusiastically adopting it. Law is way too expensive for what it is. There are only a couple of legal technology companies I have some respect for. One of these systems, Do Not Pay, can do a thousand different things for free, that normally you would pay a lawyer for. You’ll see in-house counsel at corporations are rapidly accepting that. They can do the form-filling work internally, without having to go outside for a law firm. There are alternative providers that were able to raise funds to provide this kind of services to in house counsel.
I definitely see the administrative part of the work going away. The value of filling out a few forms or checking a contract is going to fall to practically zero in the future. People will pay lawyers for the interpretation of what these systems do, rather than for filling forms and ticking in boxes.
Will there be unbiased robotic lawyers and judges? I think there will always be some bias to them, but certainly far less than humans. We actually have a system called JustText, which can read a pleading or a motion and tell you who’s likely to win based on the precedents and statutes that are in it. When lawyers research precedents right now, the system will tell that a particular precedent has being quoted a certain number of times to this judge, but for some reason they haven’t kept track of what happens after that. So, it might have been quoted 70 times, but maybe the judge ruled against it 70 times, maybe the judge hates that. You need to know before you open your mouth. We came up with this persuasive precedent system that generates win rates for precedents, and applied for a patent for it. That system could easily be used to take a lot of the mechanic work out of the legal system.
I don’t see human judges going away ever, but a lot of their tasks could be automated so they can focus on the stuff with the most value. By bringing this kind of transparency we are setting more reasonable standards, so the same case won’t go one way before one judge and another way before another.
lawless.tech: How do you see further evolution of legal industry and legal tech in general? What will law firms look like in 5 or 10 years? How lawyers’ day-to-day job will change?
Toby Unwin: I think, in 5 and 10 years time the single-practitioner law firms will still be around. We are starting the see the death of brand in law. Most of our clients are insurance companies, because they get sued the most. We’ve come out with the new product called Panel Report. It looks at the panel, the lawyers that the firm is providing you. This is quite an interesting question to ask. Let’s say you call up your law firm and you say “I have a case before Judge John Kest in Orlando, Florida, it’s a contract case and I’m defending it. I want the best person for that case type and Judge” can your law firm give you that person? The answer is that it’s highly unlikely, because there’s over 5,000 lawyers in Orlando and over a thousand firms, so the odds that the best person is working for your firm are slim.
We actually have our own law firm now that can provide our clients with any lawyer on the local counsel basis, irrespective of what firm they’re in. That’s a big deal for large corporations because they don’t want to have to go onboarding thousands of new firms just to get the best lawyer. Getting back to the question, you can understand that the firm can’t give you the best person, but can they give you their best person? Well, they can’t do that either, because law firms don’t keep track of wins and losses, they keep track of hours and billing, so they have no idea who is their best person. They are going to give you the person who is sitting on a bench not billing enough hours right now. Is that person good? Maybe they are and maybe they are not. You don’t really know. What Panel Report does is it shows you the win rates of the lawyers in that firm and it shows who these firms have been sending over. Therefore, you are able to hire not by brand, but by people.
That is where I see law going. As for the individual lawyers, you will have people that use automation and tools to put together paperwork and for that they will charge you practically nothing, but what they are changing you for is their advice. That where, I think, things are going to go.
lawless.tech: In broad strokes, what plans do you have for Premonition?
Toby Unwin: Right from the start we had a three-part plan. Part one was to get the most data, which we have now done. Part two was to become the litigation benchmark, which we have now also essentially done. I gave a speech at PCI, the association of insurers, before 80 insurance companies CEO’s. These guys account for just over half of all litigation spending. Insurers basically own the litigation market. Afterwards they took a poll of the audience and asked what is the best way of finding a lawyer and 96% said it’s their win rate for that case type and judge. Before the presentation that would’ve probably be zero. There’s about 80 companies that do “legal analytics,” but they just don’t have a lot of coverage. If you have an expensive system, but whenever you type a case in there’s a 97% chance it’s not going to be there. That’s not that useful, no matter how pretty it is. We have overwhelmingly more data and coverage from everybody else. So Premonition is the litigation benchmark.
The third thing, which we are now starting to do in beta, is to become the litigation marketplace. Essentially, we’re the only people who know how good particular lawyers are for certain cases and certain judges. It’s foolish to hire a lawyer without knowing how good they are. You don’t have to pay more to get a good lawyer. In law there’s a standard referral fee for referring a case over. They pay an average of 25% for referring a case. We’re talking about a business where we could make $25,000 just for picking up the phone and referring a case which could also be automated. In the US that happens 41,000 times a day, so you can do the math. It is a substantial market and we’re the only player.
For audio version follow this link.