The following is a transcript of a live panel recorded at the Harvard Litigation Finance Symposium, organized by the Harvard Law & Technology Society. It was moderated by Kate Boyd (Chief Engagement Officer, Validity Finance), and the panel included Jay Greenberg (CEO, LexShares), Sean Thompson (Director of IP Strategies, Parabellum Capital), Peter Zimroth (Director of the Center on Civil Justice, NYU Law) and David Siffert (Director of Research & Projects at the Center of Civil Justice, NYU Law).


Kate Boyd:
Hi, everyone. My name is Kate Boyd and I'm so excited to be here speaking with this group about emerging technology. A quick note on my background--I'm a big nerd. I spent 20 years talking to lawyers about how they can use technology to do things differently. And I'm really excited to start to explore how that is coming in, both with what we've just seen, really this opening of access to information, but we're also going to peel back a little bit about what happens inside actual funding operations with what we cover today.

What I'd like to do is ask each of our panelists to introduce themselves. We'll just run down the row and then we'll tee you off with a series of questions.

Sean, do you want to go first?


Sean Thompson:
Sure, thank you. I'm Sean Thompson, I'm the director of IP strategies at Parabellum Capital, which is a litigation finance fund in New York. We fund commercial litigation and I am primarily responsible for our patent investments. I'm also the fund’s General Counsel.


Jay Greenberg:
My name is Jay Greenberg. I'm the Chief Executive Officer and Co-Founder of the commercial litigation funder LexShares. Probably unlike everybody on this panel, I am not an attorney. I do not have a law degree. I come to the world of litigation finance from a corporate finance background. Prior to founding LexShares in 2014, I was at Deutsche Bank doing technology investment banking and admittedly stumbled across the asset class and really became infatuated with it. At the time, in 2013, there were folks out there like Burford and Bentham and when you looked at the public return profile of those publicly-traded funds, yielding 60-ish percent on an annualized basis by investing in these magically uncorrelated claims, and my question at that point was: "why weren't more people investing in litigation?"

After a bunch of research, my thesis was that there was really an education gap. And that was an education gap on behalf of investors. I thought I was a sophisticated investor. I didn't understand that a lawsuit was a capital asset that I could invest in. And also on the flip side, an education gap on behalf of plaintiffs and attorneys, the recipients of funding in the sense of, they may have heard of litigation financing theory but really had no streamlined way to obtain that type of funding. And so the idea at that point was to create LexShares which, first act as an educational vehicle instead of hiding behind the old opacity of a fund structure, creating this online forward-facing retail market place where we can educate all of those constituents and then, with what I would call securities deregulation in 2013 with the Jobs Act capitalize what was seemingly an undercapitalized market online.

Fast forward five years and we are still funding commercial legal claims on a single case basis online. And in parallel to that, we also operate a private funds business where we invest a pool of discretionary capital into similar claims.


David Siffert:
My name is David Siffert. I'm the Director of Research and Projects at the Center on Civil Justice in NYU Law School. The center is, much more broadly speaking, academic center on the subject of civil litigation. Of course, we have been involved in third party litigation funding because we are at this conference, but we're not industry insiders the way a lot of folks are. There's a real benefit to that in terms of the library we're working on because we can stay neutral. But it means that we're coming at a lot of this from a somewhat different direction. We have been involved in technology and a whole bunch of different areas and Peter might be able to talk about.


Peter Zimroth:
Okay, you do it.


David Siffert:
We recently hosted a conference on artificial intelligence and legal technology, talking about figuring out what it can do, what it can't do, how it's actually being used, and how it might need to be either regulated or internally monitored, to get the most out of it and also avoid some of the potential pitfalls. We've been thinking about legal technology in this way for a while though. A lot of those thoughts haven't been specifically applied to third party litigation fund.


Peter Zimroth:
Well, okay. My name is Peter Zimroth, I've already introduced myself. And just to add, one other technology-related project is that we have been attempting to find a way to make more publicly available, and this may relate very much to litigation financing, make getting information much more publicly available about mass claims, not just class actions, but aggregate litigation. And it's been a long-term project. We've been working... a lot of people including judges, and we've had recently a major breakthrough in the                Northern District of California, where they put out a new advisory, which we can talk about, making information more available, but it's been a major challenge because it's not so different from what we're hearing about in litigation funding area, which is that there's all kinds of information out there, but it's closely held by a few people, and mostly, claims administrators, not even the lawyers, and certainly not the judges.

The judges, of course, have access in their individual cases, no question about that, but in terms of the broad scope of information that's out there, it's the claims administrators. And for some valid reasons, they are very reluctant to give up that data. And when I'm talking about data, I don't mean names and social security numbers, all that would be... we're not interested in that, we don't need any of that stuff, but just some other basic data and they won't do it unless they're ordered to do it by judges which is why we're working with judges to persuade judges, what are the benefits of doing this. And that's where the major breakthrough came in California. And I hope, I think, once... That's a very tech-savvy jurisdiction and once they start doing this, by example, other jurisdictions are going to do it.

And in the Q & A, we can talk more about that project if you'd like.


Kate Boyd:
That's great. What Peter brings up is something that we will trip into again and again. Where does the technology hit limitations versus open opportunities, which is...? I love this panel. They've got such different perspectives on the things we're going to talk about.

Quick note on my background. I'm at Validity Finance. My role there is Chief Engagement Officer. That means I spend all of my day thinking about what it would be like to be a client.


Peter Zimroth:
To be what?


Kate Boyd:
A client. What is my experience? Where do I hit barriers? We heard earlier today... I'm sorry this keeps ringing... about the delay. Your case goes into exclusivity and you have to wait.

What I'd like to tee off for this team, and what we talked about initially, is we're going to start very general with a little daydream, a little thought experiment, if you will. And then we're going to get into some of the specific technologies at play here.

If we take the thought experiment. What I would love to imagine is that as a client, I almost have an Uber app for litigation finance and for litigation in general. Where there were some sort of marketplace where I can put in the details of my case, and at the same time as I hit that request my Uber X, the law firms are able to understand that... What's the scope of that case, what are some of the complexities, put together a bid using some standardized magical codes, funders are able to say, "Okay, with these different offers from the law firms, this is how we might be able to risk assess this and give some funding." And then the client is able to make a very informed decision.

As Tim was talking, I have this imagination that people are going to sitting there, and it's going to say, "You're an exclusivity. You have to wait 15 minutes before your Uber is going to be able to show up or not."

What I would like to propose, this same thing, about all the different technologies that come together for this crazy scenario. How far away is it that people could have that kind of transparency? And where are there slivers of opportunity, both for the beginning, where I request a case, and then through the life cycle of the case because I know our discussion is going to really span that whole thread?

And I'd like to open for anyone who wants to jump in on that free-me scenario.


Sean Thompson:
The tech would be there to do something like that today, but you'd run into such other barriers, like people have real confidentiality concerns, about telling us anything about... In the patent space, we'll have people call us up and ask for terms but not reveal the patents, the targets, the law firms they have, and it's like, "Well, I don't know what this will look like."

And some of those concerns are really valid, and some are ridiculous, but I think you'd run into stuff like that being a major roadblock. And on the funders' side, I don't know that I'd be so enthused about an effectively over-the-counter market for funding, since that would highly commoditize us, and maybe one day it gets there. But the industry is still driven by such sticky relationships that the funder might have and the law firm might have, or the claimant might have, I don't see that happening in the near future. Though I don't think it's for technological reasons.


Kate Boyd:
I think most lawyers that I propose this too, also have the same reaction.


Jay Greenberg:
No, I have to agree with Sean there. I think the technology exists. Right? There are plenty of reverse RFP platforms that exist right now to organize and optimize pricing bids, I think from a technology standpoint, we're there. I don't think you, I or anybody else that's sitting in this room see it in our lifetime. And I think the barriers are non-technological barriers. The reason why the question excites me so much is that if we are to the point where there is an Uber for litigation, that really means we've overcome everything else, we've overcome all other obstacles to make litigation finance mainstream, like regulation, standardization of documents, transparency.

Those are really the three biggest things that are curtailing us from really being mainstream, proliferating this asset through the marketplace. Uber of litigation, that brings us through all those barriers. Like Sean, I think the constituency that benefits the most from a product like that, is really the consumer of financing. It's really the recipient of the funding, the plaintiffs and attorneys out there, because obviously with that type of optimization and transparency, alpha in return pricing is going to become much more economical for the recipient of the funding and therefore a lot of funders, their alpha, is definitely going to get even too.


David Siffert:
I would just say that privacy and transparency is a double-edged sword. It's a transparent solution but as you said there're real privacy problems. There are ways that that can be overcome. If you have a really trusted intermediary, it might be able to help especially with an intermediary that somehow can maintain attorney-client privilege. I could envision a world in which the funders had to come to the table on something like this because it became a standard, but the non-technological challenges are just non-trivial.


Kate Boyd:
Yeah. Those big barriers. Fantastic.

Peter, anything you'd jump in?


Peter Zimroth:
Well, nothing, I was amused by your question, the way you put the question to us in the email, you said was, "Imagine yourself two and a half years." Well, it's been almost... the library we put up is really a... compared to what we're talking about, is like simple, is really simple. And today we're launching it after more than three years. And the hurdles are nowhere near as significant... Unless someone orders you to do it, are you going to provide information on your cases? I doubt it.


Sean Thompson:
No. Well, we would be in breach of any number of our contracts, if we did.


Peter Zimroth:
Yeah.


Sean Thompson:
Yeah.


Peter Zimroth:
Then if you don't have the information... It's like having an Uber app with no cars out on the street.


Jay Greenberg:
No, not only the information. It doesn't make sense for the parties that would have to participate to establish something like that, funders, a ttorneys, right? Those people that would need to participate in that ecosystem, they're probably not going to be so willing to.


Peter Zimroth:
Well, it might, depending, as you pointed out, depending upon what the regulators ultimately say or do. Or the judges. Regulators or the judges because it may come a time where the industry as a whole, or certain segments of the industry, believe that it really needs public imprimatur, in some way, otherwise it's just going to be litigated to death.


Peter Zimroth:
I'm not knowledgeable enough to know whether that's a real risk or not. Is it?


Sean Thompson:
It might have been. Parabellum principal started investing in '06, things were a lot less certain then.


Peter Zimroth:
I can tell you that on our board, I don't mean the board of the litigation funding library, I mean board of the center, we have some of the best top lawyers, and defense lawyers, plaintiff lawyers and judges and policy makers. And I tell you, that this is... you may not think it's a danger, but there's so much skepticism about what... on the part of very well-educated people, and people who are deeply invested in the civil justice system. Every time we raise... even when we wanted to have a conference on... there was a... you remember that... there was a huge... not huge, I mean we did it, but the... Why we're doing this and this is... It's all a big secret and that kind of thing.


Jay Greenberg:
I'm curious. Not to steal the question from you Kate, but why do you think that skepticism exists?


Peter Zimroth:
Because of the secrecy.


Jay Greenberg:
Because of secrecy. You think transparency would just alleviate that skepticism?


Peter Zimroth:
It would-


Jay Greenberg:
If they could see what was going on, they would understand the true benefits of litigation finance.


Peter Zimroth:
Or the defects.


Jay Greenberg:
Sure.


Peter Zimroth:
But when the defects were exposed, you would have a bigger voice in the solution and a bigger willingness to hear on the part of the people who're making decisions. I really do think that that is it and that's what I was hearing from all these people, "We don't know anything about what's going on there."


Sean Thompson:
Do you not think though that some of the skepticism, from say the defense bar, is driven by something other than a lack of transparency.


Peter Zimroth:
I am not talking about the defense bar.


Sean Thompson:
I used to be a defense lawyer.


Peter Zimroth:
I'm not talking about the defense bar.


Sean Thompson:
Okay.


Peter Zimroth:
I'm talking... and actually we've had people on the defense bar and some of those were least skeptical. Okay? Surprising to you, but that's the case. The most skeptical were the judges and the policy makers. And... look, it was a small sample, right? We had like 25 thereabout people on our board but it's not just those. I talk to a lot of other judges about it as well. But even that is a very small sample, so I'm just telling you what I hear. But I would bet that you would hear that if you'd get into a candid conversation with those... judges.


Kate Boyd:
I'm going to jump in because I love that all so often times when you start talking about technology, the technology itself to do the thing that you want it to, you can make that button, but it's all of the other things that have to change, and all of the data that has to line up, all of the behavior that has to change, all the transparency and regulation that have to come into it. I'm going to let us dive in. I'm going to... I think this question of transparency is going to keep coming up and it sounds so important.


Kate Boyd:
But I want to actually let this group here, and let us discuss, some of how we're actually using the technology that is working, not the mythical technology that in theory can work, but we're probably not there yet. We're having... and you guys for sure has been involved in this for years. Where have you seen technology really making leaps and bounds? And before I jump into that, I'm going to offer a little framework that they already know but that might be helpful for you.

What we laid out when we started talking about this amongst ourselves, was the idea that we were going speak really to technology in four buckets. How is it used for sourcing? How it's used for opportunities that come in the door? There was discussion only about being inundated. Jay's got online presence where people are submitting cases, so speaking to how does that work. The diligence, the underwriting. How is technology coming in to help risk assess this asset? Then monitoring, how are we making sure as the case is going forward, and there's some crazy daydreams I know people are having about how we can use a... dare I say it, to start to look at cases as they're going forward, and then finally, the transparency, the investor relations, the questions regulatory about how can we share information in ways that still protect confidentiality, but help make sure the industry as a whole has the same transparency as other industries that are above reproach.

With those four buckets, and as this group discusses, we'll try to make sure we keep track of those. Where have you guys seen leaps and bounds in the adoption of technology? And I'll keep it very broad and then we can get more specific.


Sean Thompson:
Sure. I don't want to take everything first but I'll just...


Kate Boyd:
You're right there, just keep taking.


Sean Thompson:
In the patent space specifically, you're really... and this is true of commercial but more true for patent. You're faced with these two problems of, you're inundated with a lot of really terrible opportunities which within there might be buried something good, and then you don't have a lot of really great opportunities. So, how do you go out and get them? It's two-folds issue and stuff that's coming in, we might get 3 or 4 people a day call us with a patent opportunity or a lawyer with a patent, or... and it's just me. So it's and that's true of most funds, there's not this huge infrastructure, like a law firm. And the question is how do you quickly look at this stuff and decide, "No, immediately no," and getting to know quickly is important. Or, "Okay, we're going to put this into deeper diligence.


Sean Thompson:
And we apply a lot of technology, it's because we have to. In a perfect world, you have someone look at the patents and assess the case but you somehow have bandwidth for that. We apply patent analytics, especially if we have a large portfolio come in. We have an exclusive relationship with a patent analytics company that's... was developed by a University of Kansas professor and it will give us some sense of, "Yeah, there might be something here, we should look at these patents more. Or know that they're worthless.

Applying analytics but, also... and this is not technology, but I would say it's technology-enabled. In the patent world there's a lot of conventional wisdom when you look at a patent, like what makes it good, what makes it bad, like claim one is really long so that's good for somethings, bad for others. It cites a lot of prior data that indicates it's good.

And it turns out that now that you've had enough development in the patent space, in a lot of machine readable, patent data has been available, you've been able to run more realistic statistical analysis of that stuff, like testing the conventional wisdom against actual trial outcomes or outcomes before the patent board. And actually, yes, some of the conventional wisdom is actually indicative of something, some of it was just wrong. I wouldn't say it's tech per se, but we try to stay on top of, and to the extent we need to get that more detail, do some of the analysis to figure out what our first level screens are going to be, and how we can move stuff to the no-pile more quickly and spend time on the other stuff.

And on the flip side, in terms of targeting... and this is not a particularly insightful point. But we use digital advertising to very specifically target, for example, university professors that might have patents. We will... using LinkedIn, all the other crazily-detailed targeting that you can do now, specifically direct our ads to people we think might have patents that are interesting. And more laterally do stuff in that space. For us that's how we generally use it. Beyond that, look, there are certain really interesting areas of law, profitable areas of law that someone will look at every case in that area, at some level of detail and assess whether that's something we'd want to be involved in but you should also... and those tools can be very interesting, useful but at the same time, if it's an already-filed case that is not funded, you're somewhat restricted because it's... that person's decided to file the case without actually obtaining funding. To some degree it'd be like calling people who bought a house for cash and asking them if they wanted a mortgage.

That would have some success at some level but it's not this panacea of, now it's filed, now we'll get in to finance it. Of course, we and everyone else does do some of that but a lot of the game now is the very early stage opportunities before they're filed and getting to the good ones quicker and moving the bad ones aside as quickly as possible.


Jay Greenberg:
I'd say that out of the four buckets that you outlined, I think technology has had a radical impact on two of those buckets at least from LexShares’ purview. I'd say technology has had a radical impact on our ability to originate deals. I'd say it's also had a radical impact on investor transparency. Underwriting, for LexShares at least, is still... we use human capital, we use formal litigators to underwrite cases. I don't think technology has had a major impact there in same with regards to underwriting.

I will hit on the two facets where I do think technology has had a major impact. One, with regards to origination, when we founded LexShares in 2014, we were completely reactive. We were truly reliant on what I would call our inbound channel, which is plaintiffs and attorneys looking for funding approaching us. And that was great, we were able to grow our business to a certain extent that way, but in 2016 we realized we couldn't just be reactive anymore, we needed to be proactive. And what we decided to do, was develop a piece of technology, in-house, with our own engineers, we developed piece of technology that we call the Diamond Mine.

And essentially what the Diamond Mine does, is get mines through both federal dockets and at this point, 24 different state jurisdictions, and for any case types that we like. We only fund commercial cases, breach of contract, breach of fiduciary duty, anti-trust, false claims act cases.

It goes ahead and actually downloads the complaint. I'm sure as Peter and David know, one of the biggest issues with electronic filings related to legal claims in the US is that they're filed in all different types of disparate file format. And so essentially what the Diamond Mine first does, the first step in the process, is to normalize that data, whether that's a PDF or PNG or JPEG. We take that data, we do optimum character recognition, we pass it down into raw text. And then once we have that raw text, so once the normalization has been down, which is the second step, we then algorithm over that raw text.

Essentially we designed a 17-parameter algorithm. First that algorithm will start to look at natural language strings in that complaint. How many times in this complaint does it say unjust enrichment? How many times in this complaint does it say breach of fiduciary duty? Then the algorithm will look at quantitative metrics. Is there a damages number in this complaint? Is that damages number in excess of five million dollars? All right. That will give you a boost in algorithm score.

And at the end of the day what we do is, we take those 17 different parameters and we force-rank them to what we call a "five diamond score". If it had all 17 parameters, that's a five diamond case. And essentially each week our business development team, and I see a few members here, thank you for coming and listening to me, they go ahead, and they'll receive a list of cases. We'll basically have an automated process coming out of that Diamond Mine which essentially reaches out to both the plaintiff and the attorney in that matter, explains to them the benefits of litigation finance.

The big part of origination, at least still for LexShares, is really educating both attorneys and plaintiffs on how funding works and then see if they're particular claim could be a fit for us. On the origination front, I will admit in 2016 I was an absolute skeptic, that deploying our resources to develop the Diamond Mine would make sense. I have been completely converted. 80% of LexShares deal flow, and I think at this point, at least, given the knowledge that we know, LexShares on a number of deals basis, number of single case investments basis, is likely the most active commercial litigation funder in the United States, and we definitely would not have been able to source that amount of flow without this Diamond Mine technology. 80% of our flow comes through the Diamond Mine so I'm definitely a convert in that regard, but I'll leave it there.


Peter Zimroth:
Can I ask a question about that?


Jay Greenberg:
Please.


Peter Zimroth:
What testing have you done to validate the algorithm, in other words... not just... maybe you've got success out of that, but how do you know that with a different algorithm, you wouldn't have gotten more successful? Well, that's my question.


Jay Greenberg:
No, it's a great question Peter. And I think the clear distinction needs to be drawn, that for us, the Diamond Mine technology is purely origination technology. It has nothing to do with actually underwriting the case. For us, it's strictly just to source that opportunity and then once we've sourced that opportunity, we qualify that opportunity. At that point in our investment funnel, we would take that opportunity from business development, we would pass that opportunity along to our underwriting team.


Peter Zimroth:
I still have the same question. I still have the same question because after you do the funnel and then you stop putting the human mind to it...


Jay Greenberg:
Sure.


Peter Zimroth:
... you're spending a lot of your own resources analyzing those cases or the situations that come up. And my question to you is, how do you know that that's the best use of your time? Meaning, maybe with a different algorithm, you would have found better opportunities.


Jay Greenberg:
Definitely. No, no, that question makes complete sense.

That's really where machine learning comes in to play. Essentially each time we receive a case through the Diamond Mine, that case gets passed along to our underwriting team. The underwriting team says, "Wait, why are you wasting our time and resources here? This was a poor choice for the algorithm to pick and for the business development team to go ahead and qualify." That basically goes into a machine-learning algorithm is a negative parameter. Each time that the Diamond Mine produces an opportunity that's good, and we pass that to our underwriting team, it's verified by a human that, "Yes, this complaint, this is a good opportunity." That positive feedback is also passed back into the algorithm, which over time makes it smarter essentially.


David Siffert:
But typically machine learning requires millions of data points.


Jay Greenberg:
It does.


David Siffert:
If you're going through a few dozen that your Diamond Mine's going through, I'm not... how are you sure that that's working... How do you know that it's working? Is your underwriting team finding that it's getting better?


Jay Greenberg:
Yes. From 2016 to when we started doing this, we have now mined 670,000 complaints and that has definitely made a difference in the algorithm's performance. No question.


Kate Boyd:
One of my naïve questions, which now I know they probably received my email and giggled, was, "Is technology giving you a competitive advantage?" And we've heard definitely from both Sean and Jay, that it is. Both with how he's got the patent assessment through portfolios and obviously the Diamond Mine which is such a great phrase.

I picked up yesterday morning, there was an article in Artificial Lawyer, with a funder in the UK investing in a company called CourtQuant which I love... Robert's nodding which means I might have hit on something that's of interest.


Robert Mahari:
Spoke to the founder last night.


Kate Boyd:
I was reading this... this is so timely and so perfect. Do you want to speak a little bit about what CourtQuant is?


Robert Mahari:
It's analytics software, from my understanding. I can't speak to the internal technology but I think what they're trying to do is predict the outcome of cases and that kind of stuff. Which is in the same vein as a lot of what you've been talking about. And again, the lack of data is an issue.


Kate Boyd:
One of the things that... Jay mentioned this. The human capital is still so important to the diligence, and it's expensive capital. We're putting the smartest minds in my organization, trust me, they're not sitting in client engagement, they're the lawyers. And they're the ones who really have to have years and years of expertise, seeing lots of different trials to be able to know all of the challenges that they're going to come across with any given litigation.

What I'm interested in is, as you're doing that diligence and as you hear about things like CourtQuant, how far out is it before we're going to be able to predict the outcomes of litigation? And are there specific practices that are going to possibly be leaders in this?


Sean Thompson:
I would say... Look, I've used CourtQuant and it's ridiculous. It's one of the least impressive system I've seen.


Kate Boyd:
Now I've got to black mark it.


Sean Thompson:
Yeah, sorry. Sorry. Look, it's really... it's hard to predict... couple of points, but one, let's say you came up with the world's most perfect prediction algorithm, or let's say you did it in a stock market, you had an algorithm predicting stock movements, that'd be awesome. You could easily make those investments and your money, you could make could be infinite until it's priced into the market.

If you had a similarly impressive litigation prediction tool, it still wouldn't get you there because you cannot just open each trade and invest in litigation. You still have to originate the investment. Let's say you had... I know this patent case IBM has that's going to be a billion dollar case, and it's great. I'd like to get invested in it. While you call up IBM and say, "Would you like some money in exchange for portion of the proceeds?" And they'd say, "No. We're IBM. We don't need your money." And you call up the lawyers, and you'd say, "Great case. I'd like to give you some money against it," and they'd be like, "No, thanks, because I know it's a great case, so I'd like to keep my 40 points in it."

It is hugely important on the human side as to how you originate, even if you have the best tools possible. While I think, externally in particular, there might be this view that we can only get AI to the point of making case predictions, industry would be good to go, but you still face as much, if not more, pressure on the origination side to actually originate the investment. Which is going to be an effectively permanent barrier, to take in this area because they're just part investments to originate, even if you have those relationships.


Jay Greenberg:
No, that's spot on. From an AI perspective we're extremely, extremely far away from being able to predict the outcome of cases and Sean hits on an interesting point, even if we were able to predict the outcomes of cases, even if we're able to predict the outcome of cases, and we're able to make any investments in those cases, there're still a number of external factors that AI technology would just simply looking, is this case meritorious? Do we believe this case is going to win? At what quantum are we going to win? You still have to structure that deal. You still have counterparty risk. Do I actually trust that I'm going to be repaid by this counterparty? Defendant’s ability to pay. Right? That's great. Even if this case is meritorious, and I win, and I'm never able to collect... Well, those are not the type of investments that we're typically looking to make. That's spot on.


David Siffert:
I will only add there, definitely several companies looking into trying to predict outcomes of cases, and they all make pretty aggressive claims, and I haven't seen the data either to back up or to refute those claims, so I can't really speak to it one way or the other.

I will say just in terms of natural language processing, more broadly... we're pretty far away. Right? You can't talk to your computer and tell it what to do for you in natural language. You need to use your mouse and give it precise instructions, so I think we're... it's a ways away.


Kate Boyd:
Robert I'm going to call you up and see... I know you spent some time thinking about this yourself, is there anything you would weigh in on this?

I'm embarrassing him. Have you guys embarrassed him for two straight days?


Robert Mahari:
No, I haven't been embarrassed yet, but you're doing a good job.

There's that matrix, determinate indeterminate optimist pessimist? I think I'm an indeterminate optimist when it comes to these things. And especially if you mill about MIT where that positive energy when it comes to AI machine learning is palpable, there's enormous and tremendous opportunity. It's, I think, a data issue, unless... We can talk to our computers, right? Siri does that all the time. I've essentially stopped writing emails because I just talk into it and then proofread.


David Siffert:
But Siri doesn't understand the content of what you're saying.


Robert Mahari:
That's true.


David Siffert:
It can put letters to what you're saying, but you can't actually engage with Siri substantively.


Robert Mahari:
But then the question is... that's true, but then the question is, "Is law a flow chart or not?" And if it's just the flow chart... I don't think it's just the flow chart. But if it's just the flow chart, then this is easy. It's just about where are the yeses and nos. But maybe reducing law to a flow chart can get us 70% of the way there. And then the question is... You were saying if we had a perfect model, then we get infinitely rich from the stock market. I wonder if you have a 70% model, you'll also get pretty rich. Right? I can just get rid of the bottom 30% of cases, the ones I don't want to touch, which is what you're doing with the Diamond Mine, right? And then automating the interaction with the... essentially an automated sales force for origination would be interesting.

We could do that, because the emails are, at least the initial contact, is pretty simple and you're explaining what litigation finance and then on the back end, you can think of securitizing this and letting people buy. You can do an ICO for litigation finance. I'm surprised no one's done that. But you can have people crowdsourcing into litigation and then you have an end-to-end platform that's very much technology based.


David Siffert:
Seems feasible.


Robert Mahari:
Seems feasible, but I'd love to get the reaction from the panels.


Peter Zimroth:
Make that easy. Okay?


David Siffert:
I will say only that it's not that... if we're talking 50 years out, I'm so far beyond ability to predict, so all these dreams may come true. I just don't feel like it's around the corner.

Like Peter said, it took us three years to build a really simple document database. We may very well get there but I don't feel like it's close enough to... that we should be altering their behavior yet.


Sean Thompson:
The fact that you need to pin it down with the docs from PACER...


Unison:
Is a huge barrier.


Sean Thompson:
And for most... not even that most, but many states, you can't download the docket. What you're going to do?


David Siffert:
The states are really tough and that's actually something that we're starting to do some work on. There's now the IDB, that the Federal Judicial Center has, and is pretty thorough on. With Federal Court data you can actually get a lot now, but the states... I'm currently dealing with that challenge.


Sean Thompson:
One of the real leaders in legal analytics, Lex Machina, basically they took a bunch of private equity money and downloaded PACER. And started attacking it. And it's super useful, it really is.


David Siffert:
But that's pretty all just patent. Right?


Sean Thompson:
No. No. That's all commercial.


David Siffert:
Have they expanded?


Jay Greenberg:
It has expanded.


Sean Thompson:
It started as patent though, yeah.

But it's a tall order, and it's not something some MIT students are going to do in the dorm room because you also need to have a few million dollars to get all the development data.


David Siffert:
It's an expensive dataset.


Sean Thompson:
And even once you got it, the stuff gets sealed under protective order is not on the docket, like in patent cases. That's actually the most relevant documents.


David Siffert:
While we're on the subject of that, privacy issues, right? I was just at a conference where one of the folks involved in the integrated database-


Peter Zimroth:
In the what?


David Siffert:
I was just in a conference where one of the folks involved in integrated database the Federal judicial Center maintains, and there's a lot of talk about... for example, they don't have the judges names on it. University of Michigan also doesn't have the judges name on it. You can get the judges name from the dockets but those cost money. And that's not even a major privacy issues, when you're talking about patents, all sorts of personal and valuable information might be in the attachments to the various things.

When you're talking about aggregate litigation, the proofs of your claim can have all sorts of medical records in them.


Jay Greenberg:
It hearkens back to your question. It gets to, at least from a funder's perspective, buy versus build. Right? If we're all using the same technology to originate these claims and to underwrite these claims, the question of, "Is it really... is the technology actually valuable?" I think that's only to a certain extent. The technology, at least, for a funder would need to proprietary to really create that barrier-to-entry mode if technology is going to be your differentiator.


Kate Boyd:
Fascinating. I'm going to do a plug. I apologize, she just stood up to go grab herself a water, but Dana and Isabel are going to be talking about arbitration, and some of the data that they have access to, so, stick around for the next panel because they're going to come back to this, "How can you predict?"

One area where I'm curious, and I want to make sure we get to investor transparency because that's something that Jay mentioned quietly as one of his points. But before we get there I want to talk a little bit about monitoring.

I had a drink the other night with a fellow who is talking about who are funders who had invested in the case, then monitoring the dockets so that as soon as a filing was made, they would then take that filing and put it through something like AVA and say, "How good is this to standard? Was it excited to the level of quality that I would expect from this firm?" And if it wasn't, am I going to pick up the phone and now have a conversation with counsel to say, "Look, we're investing in this case, and I have questions about quality of what you're doing."

It sparked of all kinds of crazy little hamsters in my brain about what is the potential when the case is already financed, that you're monitoring it and using technology to give you both those thumbs-up, those gold stars that this case is going well, and we're happy. Or there's something that's happened, we haven't actually heard about it, but technology has alerted us and now we're on alert whether or not we can do anything about that.

I was curious because I'm still fairly new to this industry. To what extent are these discussions happening? Do you have any stories of things like this? Being opportunities, is it still really mixed like AI predicting that the case is going to win?


Jay Greenberg:
I would put monitoring into the bucket of what technology has not had a radical impact on our industry, at least from LexShares’ perspective. We certainly... like the person you were speaking with, monitor all the dockets for the cases in which we're invested. We receive monthly updates from the attorney of record on cases in which we're invested, anytime there's a material update. With regards to the quality of the work being done or the quality of those filings, that's really... we're mitigating that risk upfront when we're making that investment. Right? We only want to work with highly-reputable counsel, that’s extremely trustworthy and that we truly believe in. That's where we try to mitigate that risk.

Yes, there's certainly ongoing monitoring of all of our investments but from a technology perspective it has not changed much over time for us.


Sean Thompson:
We get daily automated updates for what's going on, just the dockets of the cases. And if there's some major filing I will manual look at it. It would be cool to have something that could be like, "Actually, this was a terrible brief."


Jay Greenberg:
And be missed.


Sean Thompson:
Yeah. But on the other hand, I don't know how productive... most lawyers don't want to hear that they're terrible.


Kate Boyd:
That's not a relationship development.

We go back to that game theory of repeat playing.


Sean Thompson:
I think to the extent there is a problem to solve, it's solved, that's not going to... You can get a daily integrated update of everything that's going on in your cases.


Kate Boyd:
We're going to move on. I want to hear about investor transparency. I keep worrying I'm going to get this wrong. And when Jay first mentioned this on the telephone to me, my brain started thinking about other industries and I know that's where you're coming out of, so I'm curious to know especially as we... a lot of discussion has been about how nice this industry is. Where is transparency for investors coming in? Where is technology helping you lead this? What are your thoughts on this?


Jay Greenberg:
Absolutely, and somebody mentioned this on a prior panel, we do operate in an opaque industry and that's because of a lot of privacy and confidentiality concerns that are real concerns. LexShares, and the genesis of LexShares, since we didn't grow up... as a discretionary fund. We grew up on this electronic platform that helped to facilitate transactions between investors and recipients of funds.

From a transparency perspective, technology has had a major, major impact. If you're an investor on our platform, you can go LexShares.com, register. You can actually see the specific case in which you are going to be investing. Right? So you're seeing plaintiff vs. defendant, who's the judge, what's the jurisdiction, what's the actual background about the case. What type of arrangement is counsel on? Are they on a contingency? Are they on a hybrid? Are they being paid by the hour? And then you're actually able to review the court docket for that particular matter directly through the site.

And once you've actually gone ahead and made that investment, LexShares essentially keeps you updated on your investment. As material litigation events occur, we're continually updating, a litigation timeline for your investment. If you're an investor in that case, you see, okay, in case x, y, z, the motion to dismiss was adjudicated today.

And so technology, at least from a transparency perspective for our investors has made a large difference.


Sean Thompson:
We are a discretionary private equity fund, and our investors are large institutions, so they're not really on, "I'm curious to what's going on in this case, this timeline." We're probably as far away from Jay as you can get and our investor system don't want that level of transparency.


Kate Boyd:
Okay. I'm going to ask one last, very broad question. We're going to go broad again. If we were to look at 2, 3 years, obviously we’re not going to be at the Uber stage, where do you see... I'd love everybody... if they can give two thoughts. One, where do you see the biggest opportunity coming next? And where do you see the trough of disillusionment, meaning people have had a lot of hype but it's just not delivering. Do you have a feel for that?


Jay Greenberg:
I'd have to say the case underwriting AI to underwrite cases and tell you if that's a good or bad investment opportunity, that's pure hype, three years out. We will not be much further along than we are today.

I think on the origination side as Peter pointed out, I still think we have a long way to go in sort of developing our own algorithm to be better at originating these claims.


Sean Thompson:
I agree. I just don't think we're getting there on case predictions because the data isn't that available. And even if you had profitability in the industry, you'd still have such, such issues with sealing that you're not going to be able see the key stuff and have a realistic pred... I think all of those companies are just like not going to make it.

There's probably more opportunity on just more banal stuff like DocuSign versus paper signatures. This is like on some sense a meaningless thing, but on the other hand that's a real time difference. And I think more automated... not automated, but contracts that become more cut and paste from a template that might have some AI-based adjustments on deal terms, is probably... you can see that in a couple of years.


Kate Boyd:
Contract generation, fantastic. I'm going to put you on the spot, and my... my industry outsider.


David Siffert:
This is a little bit of a level instruction above, but for me, the opportunity is in the data availability because I... right now, the federal courts have made major progress and I really do think that within three years, four years, we're going to see major progress in the state court availability of data, not at least to the level that we currently are in federal courts, like the integrated database reaching the states.

And that can open a whole lot of doors. I'm hopeful that a lot of the states will be fully transparent with full text of everything on the docket and the documents in the docket. I'm hoping that with OCR, we'll actually be able to get into the... We'll just have a big database of language, legal language available from which we can put machine up to see what they can learn. So that's why I'm hopeful.

Where I'm concerned, is that where data goes and people get excited about data, there's always risks of bias. And I'm worried that we can perpetuate a lot of problematic biases that our system already has. It's probably unsurprising that... in criminal cases, it might not be directly applicable here, but defendants don't always do so well, and if you look at what language they're using and that gets fed into the system, it might look like... You might start to see some troubling things that we're internalizing into our own programs and having machine learning pick up a lot of the biases that started off with the humans.

I think that-


Peter Zimroth:
Not just pick up, but actually reinforce.


David Siffert:
And reinforce, yeah. Exactly.

And this is already going to be somewhat of a concern with the Diamond Mine type of a situation where you might see some folks only had access to certain types of counsel. The language in those cases might be of a certain type that looks like low quality, and it might not actually represent the case, but it might just represent the person's financial means, or the communities they come from. And we talk about access to justice, we just had a panel on it, but if we start relying more heavily on these Diamond Mines we may in fact, be cutting off access to justice.


Kate Boyd:
I want to do a book plug, if anybody hasn't read the book, Hello World, it speaks to how some of these... the data that's on record gets baked into AI and then we count on that AI to make decisions for us and this can actually have dire consequences of... to give just a little bit, the book starts with somebody following his car navigation off a cliff because it tells him that's the fastest way to get from here to there, is to make a sharp right, he does it and straight up he survives. But it's a great book, and it actually has entire chapters on the legal system and how AI might be coming in, in ways that we're not ready.

I love this panel has gone from DocuSign, which is actually one of my favorite, like super tactical. Why are we wasting time with PDF scanners and signatures being sent places? From DocuSign to the Diamond Mine and the future of what might be possible, I think that this has been super exciting.

I'd like to open for questions.


Speaker:
You mention the PACER fees and the cost associated with these searches and I was reading an article the other day about how PACER fees are ridiculous now with the internet. And if you see those going away, how would it affect your technology and then what you think would have an impact on just databases and seeking cases in general?


David Siffert:
The first thing I would say is that we've talked to the administrative office of US courts about this, and they're not about to go away.

We can start there. In terms of how it'd have an impact, I think the Federal Judicial Center has a lot of really good folks who are trying to minimize how much of an impact it would have by doing as much as they can with what they can do. But obviously there's a lot of stuff in those dockets and the documents attached to those dockets that would be quite useful.


Speaker:
My question is actually for Kate. Validity is a newer funder, you seem to have a very product and customer centric view of your firm and your customers culturally, which I think is rare for litigation funders and I'd like you to talk through Validity's perspective on operations and technology and what tools and strategies you're using to grow your business.


Kate Boyd:
Sure. I teed off these guys that I was going to make people do storytelling and I think I might have put myself on the line here.

My company was founded actually by a litigation finance pioneer Ralph Sutton, who's been doing this since 2006. He was at Credit Suisse. He met me, I'm a techie. I've done three legal tech startups. I've fallen in love with how technology can change the way people work. And he was looking to start this fund and I was leaving a contract analysis machine-learning software company and I kept saying to him, "I think what you need is people who are thinking about clients, and thinking about technology, and thinking about a lot of where a lot of us have common emphasis. Let's not just think about it in terms of being the litigants, being the lawyers, but think about the end-to-end. What happens in the life cycle of litigation and how can technology and expertise come together?

We're still fairly new. We are using DocuSign and we're trying to figure out where the latest technology is. I love Lex Machina, because even for a someone like me, I can open it and get a real good sense of "What's the track record of that lawyer? What's the track record of that jurisdiction?"

I know there are holes. Not only are there holes in my reading of it but there are holes in the data that's available. And so part of my frustration and part of... as any company is going to be weighing cost and benefits. Where do we put our investment and technology? Do we try, and fund other startups that may be able to get this state-level data. Is that worth our time and investment? Should we be thinking about origination? Is this really just a network?

I watch my team go out and so many of their cases come through relationships. And I had 15 years in law firms where's there a lot of the similar. Activity relationships often drive the business in ways that you can anticipate.

As AI has taught me, through studying and thinking about how it's going to be adopted, often you need the best technology paired with the best people. And I think we're going to be experiencing that as our industry evolves and we ourselves start to figure out how we can borrow from technology and then double-check it, as it sounds like you guys are doing as you look at Diamond Mine output.

But that's also going to mean as we see opportunities for technology to change, we're going to bringing together our own networks of experts and saying, "Should one of us invest in this? Should all of us think about this? Should we all get behind this because this might help our industry as a whole, be better, do better, do more?"

So, I'm excited about the potential. I hope that answers your question without too much rambling.


Speaker:
You mentioned the paradox that if you have a very good case, the lawyers want to keep it for themselves. And, of course, if you have a bad case, it's not worth funding especially with the ROIs that your investors are looking for. Keeping that in mind, how do you see the... how do you deal with this? And does it mean that with more and more technology, especially if we do get to better prediction through tools like this Diamond Mine. The Diamond Mine comes into the hands of the lawyer firms, perhaps you'll be kicked out of the market. Do you see that happening?


Jay Greenberg:
Yes. I think a lot of the times why there's still room for funding is... A lot of times due to institutional constraints. Right? You have a lot of large law firms AmLaw 200 firms that... They don't take cases on contingency, that they want to receive current income, they want to get paid by the hour. Along those firms will not advance cost so there's room for litigation funder to advance cost. A lot of the times when the funder is able to engage in a deal, that’s a fantastic deal and... the firm would like to take that on contingency, but they're institutionally constrained against doing so. It stems from the fact that good attorney are a lot of the times, risk-adverse by nature. And they should be risk-adverse. And so, it's just a different skillset. Litigating is a very different skillset than investing. And so some of those institutional constraints definitely enable funders to engage in deals where if attorneys were not risk-adverse and did not have those institutional constraints at their firms, those opportunities may not be available.


Kate Boyd:
I would just... one other end of the spectrum so I know a lot of the Am Law 100, 200, there is this... They're not prepared to take that much contingency in their books. The other end of the spectrum is the small firms, the single practitioner who's got a great case, who has already mortgaged his house for the other case that's still a couple of years in. They need to be able to see those cases through and they're the best lawyers to handle them and they're great cases.

Those might hit the Diamond Mine and give that lawyer an opportunity not to be terribly stressed out that his wife's going to leave him when he loses the house again.

This is real... we in a way talk a lot about address selection. That's just a natural part of this industry. But I also think the opportunity to help real lawyers with great cases who really can't financially afford it because of their profits per partner on one end of the spectrum, or their house on the other end of the spectrum to think it through.


Sean Thompson:
I think two part, many of the very... like at Burford, for example, which we can look at what cases they're financing in an anonymized way in their public filings, it doesn't really do much in a way of single-case financing anymore. They've moved on to large portfolio structures where they might have a deal with a firm, where they see every contingency case the firm takes, and they get to opt to fund or not.

People are structurally dealing with the adverse selection problem using basically portfolio structures or just large law firm finance structures that are not, "This is a good case, I'm going to invest in this specific case."


Speaker:
To this point we've talked about matter origination from the point of the complaint being filed going forward. Do you guys want to talk... I'd be interested to hear, three of you, your thoughts on technology solutions to potentially reach the claimant prior to the complaint being filed. Most of that is relationship based, but any ideas on the Holy Grail.


Kate Boyd:
You're anticipating wrong doing.


Jay Greenberg:
Most of that is relationship based and not only is that relationship based on the side of attorneys and law firms, that's also relationship based on the side of large corporates. I think a trend that we're seeing in our industry is, not only getting access to these investment opportunities through attorneys in firms, but also through large corporations directly, through general counsel, in-house counsel at those large corporations. It's definitely relationship based in that respect.

For LexShares, I would say we don't love pre-litigation opportunities. We're typically entering into an investment where there is really strong procedural history, cases advanced through numerous litigation milestones, counsel has a strong relationship with the client previously being retained. For us ,all those relationships are great. I don't think they would lead to a lot of direct immediate opportunities for funding.


Sean Thompson:
Probably most of what I do is more pre-litigation client directed stuff. The patent space is inherently different. Right? Because if you're... you've suffered some anti-trust injury as an individual, you might not even realize you've been injured.

If you're a patent owner, those people generally paid money to obtain their patents, whether they're corporates or individuals and most of them would like to make money with them. If you can identify interesting patents, and we do that algorithmically and then approach the patent owner. That's a major strategy for us.

But it's probably more prevalent in patent than it is in contract bridges.


Speaker:
And that's where that university initiative came in?


Sean Thompson:
Correct.


Kate Boyd:
And I would just add, Sean mentioned LinkedIn advertising, and I've become much more sensitive recently to the advertising that's hyper targeted to me, and my behavior, that I'm getting in my email and I'm getting in my LinkedIn. I can't remember who it was who said... I think it's the fellow who left, "We haven't seen ads on CNN yet for litigation finance." We're not there yet, but there is an education of the market, of business owners, that this is an option for them. That this is something that can help them just have a stronger negotiating position to get a settlement before they need funding, that they even know that it's there and that there's people they can trust, that are not sleaze balls and trying to rip them off. But there's opportunities, and that they can understand how it works. The same way they do with how their mortgage works. Right? And nobody says, "Well, I don't trust those mortgage people. I'm not going to do that."

It has become standard knowledge that you would of course do that. And it helps people make the right decisions. And it is technology that we're all going to be trying to reach out to people and say, "Be aware of this opportunity if you're a lawyer or if you're a small business owner so that you can be prepared to defend your rights."

I'm going to get off my soapbox, man.

All right. Well, thank you everyone. Thank you Robert. Thank you Harvard for hosting us. Thank you to this wonderful panel for all the time and effort they put into this.