Doug Austin on Hyperlinked Files, TAR, GenAI and the State of the eDiscovery Industry

Thanks, Daniel. Thanks, Brandon. Thanks for having me. It's great to be here and looking forward to having some fun talking about eDiscovery trends, challenges and everything else. Fantastic. I do miss the days when we used to actually just call it Discovery, Doug. So when you say you've been in this business for 30 years and you really have seen it grow, there's no question.

Well, so from a standpoint of what drew me to eDiscovery, I think like a lot of people, it was kind of more happenstance than anything. I had years ago, it was so long ago, they were considered the big eight consulting firms back then. I started working for Price Waterhouse on a large litigation and basically joined their litigation support team group and worked on several large litigation, some large bankruptcies over a few years. And just really, you know, that was my beginning is the lit support background that really, I think, spawned eventually eDiscovery. Daniel, as you mentioned, you know, we could use the term discovery. And at this point, most of the evidence is electronic. So if it weren't for the fact that it would really mess up Google searches to try to search for Discovery instead of eDiscovery, it would probably be appropriate to change it to Discovery at this point.

But, you know, from the standpoint of evolving from those paper days where you had all these steps to go through to first copy documents, then scan them and, you know, remove state remove fasteners and slip sheets and this and that and base labels, all that sort of stuff, to moving it towards, you know, electronic evidence and treating it electronically. It's been very interesting. I've been daily blogging for 15 years now. And I was thinking about this. You know, when I started in daily blogging, it's September 2010. The Zooblake cases weren't, you know, we're just a few years old. The EDRM model was like five years old. eDiscovery was still a relatively new discipline or, you know, only a few years. And we still pretty much equated it with a litigation.

Most of the solutions were on premise. The processes for, you know, most of the evidence we were dealing with was email and office files, and it was mostly manual processes moving from step to step through the EDIM model to get to them. We didn't have TAR yet. There was no, I mean, there was no really acceptance of TAR. Just over more case, Judge Peck's ruling was like a still a year and a half away. And then when you look at kind of where we've gone today in terms of eDiscovery is about all these other use cases, investigations, and incident response, and data privacy requests, and even HSR second requests. It's really almost equivalent to data discovery in terms of the things that we really apply eDiscovery workflows and technology to.

A lot of our manual processes, at least for the traditional data sources, have been very automated. And really, I think we have very established practices for those. Our data sources have exploded in terms of now we've got all these sources in the cloud. We've got collaboration apps. We've got mobile devices. We've got even things like data from IoT devices, or I just covered a case involving somebody being sued over. Their ex-girlfriend had an Apple Watch had given back to him when he broke up and gave it to the son. And apparently the ex-girlfriend conspired with a cop, or at least the guy alleges, conspired with a cop to put him in jail for a hearing. And he was getting all the text messages on his son's Apple Watch because it hadn't been cleared yet.

So stuff like that is just kind of amazing where we're at today. Now Tara's kind of like almost old hat really to us. And we're talking about all this Gen A I stuff. So it's really, they say that in the short term, progress seems slow. And in the long term, it seems very fast. And when you look over a number of years like that, we've really seen a lot of things change. So one of the things that's for me is interesting about eDiscovery is it's never boring. There's always some challenges to deal with. There's always some new things to learn. And that's why I love what I do.

Yeah, that's interesting because there's a couple of things you could point to. Obviously the evolution of TAR, the introduction of Gen AI, which is really, we're just starting to get a sense of what we can do with a Gen AI technologies and what have you. Certainly automation, we've come a long way with automating processes and streamlining workflows. But probably if I were to pick one, I would probably say it's been the cloud. Because we've evolved from on-premise eDiscovery solutions to cloud-based eDiscovery solutions by as the predominant solutions of choice these days. And our data sources are so much more in the cloud these days. I mean, we've moved from Office on-premise to M365, and Google has evolved and grown as an Office platform, it's email and Google Drive and Google Vault.

And then of course all the collaboration apps with Slack and Teams and what have you. So to me, I think the cloud, both from a standpoint of where we conduct eDiscovery, and also the sources we're pulling from eDiscovery, and any eDiscovery platform today is... pretty much expected to have links and APIs and integration with a number of cloud sources. To me, I think that's a must. And I think that's really probably transformed how we conduct eDiscovery more than anything else.

Yeah, I mean, you were in the big data era and I think when people think about big data, it's the three Vs, it's volume, velocity and variety. And we're certainly feeling all of those. I think 15 years ago, there were two Zettabytes in the world. Today it's something like 182 Zettabytes. And by 2028, it's supposed to be something like 394 Zettabytes. And when you mention all these file types and different sources of ESI, it makes sense and there continues to be more and more. And I think GenAI is gonna boost that way up because we're generating all this new content and they don't, and they still need more content, which is crazy. Definitely, I feel like we're in progress on a lot of them in terms of how we're addressing them.

I think we probably come the farthest with mobile devices because, I mean, 15 years ago, the iPhone was just a few months old and we were a long way from really getting our arms around mobile devices and a lot of litigations, the parties would say, okay, let's agree we're not going after the mobile devices because it's just too much effort, time, cost, what have you. And today, I think we've got a number of different options that can either collect forensically or collect much more streamlined, the data that's typically needed in a lot of cases. So I think we've definitely come a long way there. I think with certain collaboration apps, I think a lot of the platforms have developed robust collection capabilities to pull in data and what have you.

And I think things like Slack and Teams and what have you, there's getting to be established processes for those, even though those platforms are changing things, which of course means that then the eDiscovery platforms have to change to accommodate that. You mentioned hyperlinked files, that's definitely one of the challenges that's really causing us to stumble right now because it is so complex. It's the moving from traditionally, we've had the parent-child relationship with the email and the physically attached files in an email, which basically was a good static copy of everything we needed right there. And we had no worries about it. Now you've got a hyperlinked file. that may no longer even exist or if it exists you may not be able to get to the version that was sent or you may not have rights anymore or what have you and there's a whole lot of questions about you know what's required is it proportional when is it proportional what do you have to do and so forth I just covered another ruling in the inray uber case this morning and there's continuing to bring up you know questions about well how do we do this and we never talked about this in the SI protocol so that's definitely at the top of the list of I think things that challenges that have caused us to stumble and then I also think that you know one of the things we're going to continue to see is challenges I mean in a sense hyperlink files is another example of potentially ephemeral data but then the whole ephemeral messaging at consideration is another one that right now I don't think we have a lot of good answers for and I think one of the things that's going to you know that I don't know if we'll ever see it but it would be nice to start to see some of these source CSI solutions start to think about eDiscovery by design where they consider eDiscovery you know issues as part of the design of their product to begin with instead of having to she horn it in the you know the back end but until then it's going to be a continual minefield of challenges out there for eDiscovery professionals and certainly is going to provide job security for all of us right.

Doug you mentioned in there that you are looking at an update to specific case law and you know one of the things about your blog is that it's not just about the technology it's not just about you know current goings on in the industry itself but everything that's driving this is coming from the rules and coming from the bench and so you cover case law as well in different core cases that you've seen in 15 years there's been a lot that's happened in the eDiscovery industry as far as updates the rules you know how those rules have been interpreted how different you know judges have looked at those rules and applied them to different challenges that we've had within the industry at the very outset you mentioned a number of those that were you know relatively new cases when you first started your blog 15 years ago so in the time that you've been writing the blog you've seen a lot of that evolution what would you say has been the most interesting of the evolution from the bench that you've seen over the course of the last 15 years it's really guided what we're doing in eDiscovery today um well certainly a case that I think that tends to come up again and again and I would be remiss if I didn't mention the case of the honorable and of jpec who I do a case law webinar with every month on the edirm webinar network but certainly the disolvah more case where you know that was the first case that approved the use of technology assisted review and you know I think has really kind of established some expectations there that I think also will really establish expectations when it comes to the use of generative AI because I you know I think that in disolvah more and also in a couple of other cases judge peck really established you know the understanding of hey you know sedonal principal six says responsible responding parties are best suited to you know to respond to discovery in the manner they see fit and I think that's you know I think that's a good philosophy and I think that was truly a landmark case that we've referred to again and again interestingly enough we still see disputes about whether a party can't even use TAR still which I'm amazed about because we've had a couple of cases we covered last year one of which the judge actually granted the plaintiff's way of going about it to say use TAR only after trying keyword search first which is crazy to me but you're you're still seeing stuff like that out there you know some of the more recent cases I think one other I think you know some of the recent cases that I'm seeing that I think are indicative you know mobile device data discovery has really become much more predominant and a few years ago we had the Rossbock case so that what I call the case of the smoking emoji where the plaintiff sued the defendant for sexual harassment and had produced this text message exchange that looked really bad with a you know that was supposedly from her supervisor with a hard-eyes emoji and it turned out there were several things when they started getting into the metadata of the text exchange which she only produced an image of they really saw a number of issues with it but one of the biggest issues was that she had an iPhone 5 and the version of the hard-eyes emojis she used wasn't possible on that version of the iPhone so it was like a case where she was it was literally a smoking emoji that the version of the emoji for the iPhone, iOS version that supported the iPhone 5 was different.

And so those were among the many things that doomed her case. And we've seen a few fabrication of evidence cases. And I think that's one area where we're probably gonna see a lot more, especially with deep fakes and considerations like that. So I have the feeling Ross Bach is a case that's gonna probably come up again and again in the future more and more.

Well, I do also, one of the things that I do is a state of the industry report. And if we get a chance to talk about that, I think that's still to be seen just how fast we're adopting GenAI. But yeah, I mean, to your question, beats the heck out of me. Unfortunately in our industry, it seems like and a couple of other people have mentioned this, not just me. The approach to use keyword searches and manual review seems to be above review. It's just like accepted as a normal way of doing things and doesn't have to be justified or verified or what have you. But with TAR, we have to have transparency. And I don't understand why, because I mean, I've known a lot of attorneys who do not understand keyword search like they think they do and definitely don't know how to really get to the documents that they, to construct searches that would be maximizing recall and precision effectively.

And I think we still have a lot of disputes over keyword searches. But from a standpoint of the fact that it seems like anytime somebody wants to use TAR, there's still some disputes about either can they use it or how transparent they have to be when they use it and what have you. And honestly, I think the focus should be on the results. How are they validating? How are they confirming, from my way of thinking, whether you use keyword search and manual review or TAR or Gen AI or a divining rod, what were the results? If you can verify that the results were a high degree, high percentage of recall and a good percentage of precision, that's to me is what matters. And I think we're looking at the wrong things.

We're looking at the how they're doing it as opposed to what the end result is. In your last two answers, you've said so much, we could spend another hour just unpacking your last two answers. Sure. One of the things that you're talking about, the smoking emoji case, I want to go back to that for just a second.

Well, of course we're not ready for it. But I do think that we're doing some things. Certainly one of the things that I know that has been looked at and Dr. Mora Argrosman and Judge Cronin we're proposing some new rules, or rules changes to the federal rules of evidence related to deep fakes and what have you. Cause one of the things that, and they noted this in a paper, is not only do you have the potential problem with deep fakes, but now you're gonna have the potential deep faked defense where people are gonna say, oh, this evidence against me was deep faked. It's not legit. So then, you know, you're gonna have that challenge. To me, you know, I think some evidence is some potential evidence rules changes will help.

And then I also think that this may be one of the biggest drivers to how we change form of production in terms of producing evidence. Because I think now metadata and forensics is gonna be much more important than it ever is. Getting the data, having a way of validating that the data is legitimate. Metadata is one of the best ways to do that. And forensics professionals have the knowledge of understanding, you know, the metadata, understanding how to collect and preserve the metadata and what have you. And I think that to help ensure that the evidence we're working with is not deep faked, I think those are two disciplines that are gonna become much more important than they already are. And they're already, in my eyes, pretty important already.

It's interesting you mentioned that. We actually interviewed Dr.

So for everybody listening, it's what we're talking about over here is it's a lot. It really is a lot. And it equals a ton of dollars and cents as well. Just being able to forensically scrape this data, and especially with federal data, federal data you're not gonna get the messages, but maybe you can get the metadata, right? And so it was just the metadata to tell you something, something like the smoking emoji matter. So where do all of us lay people in the world, right? Listening to this podcast, where do they go to get up to speed? And if you say you're blocked, I understand, but other places. I, yeah, of course. Well, I will throw my blog in there, but I'm not, you know, it's a big universe out there.

And I certainly try to cover what I can, and I would certainly mention the discovery today as a resource to check. But there are so many good resources out there. EDRM has a great blog. EDRM also has a great webinar channel and has several webinars a month. ASETS has really good webinars and educational resources. There's conferences you can go to. NeLI is one of them by the way, Dr. Yeah, yeah, NeLI's one of them. Yes, I was gonna get to NeLI last. I was gonna save you guys the last. But I'll mention you first then. NeLI, the Masters Conference has local events all throughout the country throughout the year. And then of course, League of Weeks coming up soon and what have you. And then of course, really, the companies, the players in the industry, the smart ones really are smart about we need to have a good educational program because educated clients are the best clients.

And so there's plenty of great resources from a lot of the providers out there in terms of regular educational content, regular webinars and what have you. It's pretty hard not to find some resources that get into a lot of the things you mentioned, Daniel. And I think that's one of the things that you wanna do is continue. I tell everybody set aside, five to 15 minutes a day to read something related to the technology in the industry and set aside at least one hour a month to attend a free webinar. And there's plenty of free webinars out there. And that can push you further along being able to. do that. I'll also mention, Daniel, while you're writing in terms of that 40 of the 50 states have now a technological competence requirement in their rules.

Only two states, as far as I know, still have a COE requirement for technology. So not backing up. It's one thing to say, you need to have it, but then saying you need to train on it is something else. And I think that needs to change. And I think that would be the biggest thing that I think would start to help push the industry to learning more about the technology, is forcing them to get COE involving the technology.

I guess really some of the things that I would say is you really want to stay current with what's going on, which is what I try to do. I try to keep the moving around the various topics. So it's not like I would say there's one topic any more important than any other, but I definitely say it's important to stay informed. I also would say that, for example, anybody who's reading any of the blogs out there by now, who's a lawyer, legal professional, cast to know that there's been a bunch of lawyers that have filed cases with fake filing, fake case filings that were generated by chat GPT. If you read that, if anybody read one of those, then install that, they'd be like, okay, I got to make sure I don't do that.

I got to make sure I'm checking my stuff. In fact, they already should be under rule 11, checking their filings anyway. But that's an example of something where the more you educate yourself on some of the issues, the more you know, you know, in that regard. One of the case and one of the hyperlink spouse cases we covered was that I've covered is the in race sub hub case, where they agreed to an ESI protocol, where they agreed to produce hyperlink files as modern attachments before they realized what the chat, the technical challenges were in doing that. So, you know, ultimately, for fortunately for them, they had a provision in the ESI protocol where they could change it with, you know, with due cause or justified cause and the court let them do that.

But understanding before you go into a case, what the landscape is, what the, you know, what the collections of your of your client is, and you know, what sources, the ESI sources they have, you know, getting that getting that information and understanding stuff like that in advance is really key to getting a leg up. If you are trying to start learning this stuff, when you start the case, you're already behind. I think that's a, I think that's spot on. I like the way you said that.

Thanks Daniel. Yes. So the State of the Industry report is a report that eDiscovery today publishes every year. It is designed to be a, a, a comprised of two components. One, an industry wide survey of legal and eDiscovery professionals on, on important topics in eDiscovery. And then the second piece is thought leader observations on those same topics. So it's a combination of a couple of different things. We've been producing the report for five years. The 2025 report was released at the end of January. We had 551 respondents to the survey and 63 thought leaders providing observations. And, you know, we get into topics like Gen AI, TAR, mobile device discovery, collaboration apps. We asked for the first time this year should hyperlink files be treated as modern attachments?

Interestingly enough, there wasn't a 50-50 split like we thought it would be. We asked two yes and two no questions, but it was almost three out of four. You said yes they should, or at least in some case, in most, in many cases they should, but of course should and can are two different things. and of course we struggle with that. One of the questions we always ask every year is what eDiscovery challenge are we not talking enough about? In all five years, the answer has been lack of eDiscovery confidence in the industry. So we definitely have ways to go in terms of education still despite everything we're doing. Some of the findings that I thought were interesting this year, one was that for all five years, we've asked what percentage of cases are you using predictive coding on?

TAR, CAL, what have you? And this year there was almost 31% who said they use it in all or most of their cases and there was about 25% who said they use it in very few or none of their cases. And close to half were using it in half their cases. Not bad numbers. And then when you look at the five years of the history of the survey, the last two years are the first two years where the all or most cases groups were higher than the very few or none cases groups and these are the two Gen AI years. So people saying that, oh, Gen AI is gonna replace TAR well it hasn't yet. We'll see it's just getting started. We'll let down the road.

I don't know. But that's definitely one of the things I thought was interesting. I was very interested in the hyperlink files result because I was expecting more of a 50-50 split. And another thing that I would say that was kind of interesting in eye opening is while people are gung-ho on Gen AI in terms of their expectations for it this year, I basically like four and over a quarter expected it to have a transformative effect this year and four out of five expected it to at least create new workflows this year. Only 12% were using it on cases, all or most of their cases so far and 64% were using it on very few or none of their cases. So they're not using it on a lot of cases yet that may change rapidly, but that was very interesting.

So I think it's going to be interesting to see especially in the Gen AI area, just how quickly we make those leaps in an industry that is historically not quick to make technological leaps.

Yeah, so I mean, I guess it depends on, if there's really coming into the industry cold, obviously, we've talked about the educational resources. And then one of the things I think that I've seen some of the people that I've seen coming into the industry in recent years have really, I think, you can't underestimate the value of making connections in an industry. And of course, in LinkedIn, I've seen a lot of people who connect and a few people have said, hey, I'm new to the industry, what resources can you recommend? I've gone into a few places. I forgot to mention Complex Discovery, another excellent blog that I should have mentioned in the previous question. But I think that's one of the things, making connections, trying to get yourself out there.

Another thing that I would mention is if you're really serious about becoming eDiscovery proficient, consider taking the CEDS certification program from ASEDs because that is a very thorough immersion into all aspects of eDiscovery. It's a very good program. And it's certainly something that is, while there's a cost for the program, it's very valuable in terms of giving you the leg up and really understanding a whole lot of different aspects of eDiscovery. And I do think that the CEDS certification has started to become something that lawyers, that organizations look for when they're looking for people with eDiscovery experience. So you have the CEDS certification and I think that can be a difference maker. So those are some of the things that I would say in terms of getting into the industry.

A lot of people who get into the industry start out in IT or as paralegals. And I mean, those are the unsung heroes of our industry. They're the people who are doing the stuff that needs to be done and then it's kind of evolving into becoming eDiscovery. every expert. And so that, you know, these things, if you're in that situation, help it even more.

Thank you. Thank you, Daniel and then I really appreciate it. It was a fun discussion.