Chief Magistrate Judge Young Kim on Reasonable Discovery, TAR and Proportionality

You're very welcome, and thanks for having me.

I think, first of all, the federal rules of civil procedure would be our anchor. I mean, I think that judges really have to rely on the current version of the rules. Even if the technology is sort of surpassing what the rules contemplated, we do have to sort of stay within the confines of the rules. And in terms of ruling on discovery issues, that really depends heavily on what attorneys are able to offer us. So oftentimes, you know, it may be the first issue. that is going to be dealt with by the court that can sort of set the trend for other courts. there is a huge responsibility on the attorneys to ensure that you do, in fact, check off all the boxes and you produce the best evidence possible so that you get a good ruling moving forward.

I think it's one thing, you can see the impact that Judge Peck's decision had on TAR. And I. think that my decision in Livingston has been cited often in terms of TAR 2.0. So we know that Gen AI review is coming up and whether attorneys can simply rely on Gen AI in terms of meeting the 26G responsibility. That remains to be seen. So that's a long way of answering that, you know, whether the technology surpasses the rules, we still have the same role of sort of… reviewing the evidence before us and making the best possible decision. I mean, one thing we don't do is rewrite the rules. We can't, we don't have that power.

Yeah, that's a great question. So let's start with the old fashioned way of doing discovery review and production. That's when I used to litigate. Everything was paper driven. You know, you actually went to the boxes and looked through the documents, made copies of the documents that are responsive to the requests. That's how it was done. And no one ever challenged that was a reasonable method to use. Actually go to the boxes. review the documents. Now, it's a little more complicated because we are now using programs where the attorneys may not necessarily understand all the ins and outs and saying, hey, you know, I use this program and this is the way it came up with. And can you rely simply on that program and say, I've done my due diligence?

I think it depends on the reliability of the program. And I kind of go back to the. Sedona Principle 6 that defendants are in the best position to figure out how best to collect, review, and produce. And so there has to be some level of deference because especially the producing parties are usually the defendants and they're in court unwillingly. They didn't volunteer to be in court. So although there's an asymmetry in information and data, there's also asymmetry in terms of desire to be in this forum. So those things I do take into account. So in terms of the process, as long as defendants, when challenged, of course, as long as when challenged, defendants are able to explain the basis for the process and the reliability of the process.

I'm not going to push back too much. Frankly, our discovery system or discovery rules are really based on an honor system. I mean, we're. we don't really do discovery on discovery. So when we have document production, We accept it as complete. We accept the signature of the attorney that he or she did indeed conduct a reasonable inquiry and that the response is complete. Now, I don't know in what context that requesting parties are going to challenge the process unless, It came out in some other way that defendants used a certain particular process for reviewing and producing. Oftentimes, parties will have protocols, right? So you know the process that is going to be used. But if there is no protocol, I'm curious as to how the requesting parties are going to even learn what system the other side is using.

But at the end of the day, it really comes down to reliability. Can the producing parties explain to me in plain English what that process is and why it's reliable? Sometimes they may bring in experts to explain because the attorneys are not well versed in explaining the technology. And I have had situations where experts appear by phone to explain the process.

So there have been times, and this usually happens when somebody is being deposed, right? Documents have been produced and an employee is being deposed and it turns out the employee discloses that he used his phone for many of the work messages or on assignment or the employee used other messaging tools for communicating with others. but discovery responses never included any of those things. Now we have a problem, right? How did you go about finding these responsive documents and why did you, miss out on some of these really simple yet important discovery responses? Now we have to go, in and figure out, okay, what process did you use? What keywords did you use? And who's who, which custodians did you look at? And so that's going to be the discovery and discovery.

And it doesn't happen often, but it does come up from time to time. For me. Now, this is one of the unfortunate aspect of federal practice in discovery. Discovery issues are quite discretion. There's a lot of discretion that the court enjoys. In passing judgment and rulings on discovery issues. So unfortunately, it's going to be very different depending on your judge. For me, I do not, I am not in favor of discovery and discovery, unless there is solid basis that the producing party may not have comply with Rule 26G. I think that's a good way of understanding when discovery and discovery is necessary.

for what it is? Yes. So recently, I had an issue come up in a ESI protocol, the parties were able to agree on all aspects of that protocol, except for one item. And that item, was whether the producing party can go, ahead and thread all the emails so that when they review the email communications, it will be much easier. You have that one single chain to review and decide whether it's responsive and also whether. it's privileged. So the producing party wanted to thread all the emails, which would save them a lot of time and then, of course, a lot of money. Plaintiffs objected to this particular format because, number one, it's not native, right? And number two. And by the way, this is all public record, so I'm not giving you anything that hasn't been written on or talked about in court.

Plaintiffs took the position that when you thread the emails, you're losing a lot of the metadata. And plaintiffs argue that they are able to harness the power of metadata to get other critical information about various witnesses, who might be the most important person to talk to, etc. I'm not going to give you a list because they weren't so much specific in terms of how they were going to use that metadata. But when we talk about a form of production, I always go back to the rule, right? And when I said that the federal rules of civil procedure have to be our anchor, I really do mean that. And I think that the rule says, you know, 34 says it's a native format, whatever format you use.

And that is not threading the emails. That is not the native format. And so unless there is good reason, and this is where proportionality comes into play. And in the case that I wrote on, the… At least I wasn't persuaded from the producing party defendant that the burden of reviewing non-threaded emails is so great that plaintiffs should not be able to harness the metadata for their own use with their own programs. Now, it's still an open issue because I did deny the motion to use threading without prejudice, and if they are able to show me more information about the cost and the burden, I need to sort of weigh that with what benefit are we really getting from that metadata, and that remains to be seen.

I don't know whether it's going to be refiled, but we'll see. But that's how I would go ahead and deal with the form of production. I think that it's important to be efficient. I mean, rule one does say let's go ahead and interpret these rules for efficient adjudication of all claims. So I don't mind that defendants, or I should say producing party, wants to use the most efficient format and the form of production. I just need a good reason why we should deviate from the default rule.

This all comes down to balancing techniques. Right. We have to do a cost and benefit analysis. And that's what proportionality standard is in Rule 26B1. I mean, we are told that we need to do the weighing test. But when we weigh, we can't just accept your conclusory argument. Metadata is important. Yes, we know it's important, but how are you going to use it? I need concrete examples how that information is going to be used during the prosecution or defense of this case. So when we talk about burden, I'm still getting briefs where the producing party is saying it's overly burdensome without telling me exactly why it's overly burdensome. I did have one attorney who did a great job of showing burden. So he took, he submitted a declaration from an employee who is going to be responsible for locating the responsive documents.

And this declarant actually laid out, detailed the steps that she was taking to find these documents. And she laid out and also timed herself finding one document. And it turns out so many hours were spent to locate this one document. I'm not saying that detailed specificity is always required, but us judges, we do have to have some specificity so that we can use that as a basis to say. this particular request is not proportional or the form that you're asking for is just a waste of money. Without specificity, I think that most judges are going to be very reluctant to say, the requesting party is not entitled to the information the way they want it. And I think. that because it goes back to asymmetry of information, plaintiffs are, They don't have the information that defendants have.

And in order to level the playing field, we need to give plaintiffs, or I should say requesting party, access to that information and the way the plaintiffs can use that information. And that's where this form is coming into play, how the information is actually used. But again, for me, I think the requesting party has to do more than simply say, oh, metadata can be used this way or that way. Tell me how you are going to use that metadata in this particular case. That's the kind of specificity I think the courts are looking for when we are doing the balancing test.

So I think that 502(d) is really misunderstood. I think it is just that, produce. and don't worry about waving privilege. That cuts down on the review that you have to do for privilege screening. I think the way I read 502(d) is that, You know, you don't have to worry so concerned about producing privileged documents, you can always call those back. There's no minimal effort that you have to make, but I'm not sure that the parties are using 502 D, they're not taking advantage of the 502 D. And so in my, whenever I get a consent case, or if I am referred discovery supervision, my first, the first thing I do is when I issue a scheduling order for written discovery, I include the 502 D order in there, letting the parties know that you have this protection.

You don't have to worry about waving your privilege. If you disclose. How often do you. Hear arguments from the producing side, you know, it's going to be overly burdensome, because we have to do a manual review for privilege, and we have, you know, 50,000 documents to go through, I did. I had this come up in an employment discrimination case where the producing party didn't want to produce the documents because they said it was going to cost 400 hours of review to make sure that we have the privilege documents pulled out. And I said, that's perfectly fine. You have the right to go ahead and do a manual review for privilege screening, but I'm not going to then use that cost in doing the cost and benefit analysis because you're doing something that's.

for your own benefit, at least in my opinion. You have the right, you have the privilege of using 502d, and yet you're not willing to use that. Now, an argument can be made that, you know, We have every right to be reluctant to produce privileged documents, even if you're not waving privilege, right? So that smoking gun document, albeit privileged, is going to be seen by the other side, and it can have consequences. But I still resort to the rules that we have. Those are the rules that I follow when I am doing the balancing test.

It's been a while since I actually had a hearing in that one case that dealt with TAR 2.0. I do remember the expert coming in and testifying, going through the process, and how the review was going to work, and how that particular work was going to be verified. I think it depends on the program that the producing party is using, whatever program that might be. But because each program and platform, they have their own validation process. The key point, though, is that they are actually using those features, right? And during the RAND conference, one of the things that came up is, The cost that goes into validating, the cost associated with double-checking and whatnot, and is there a way for the industry or among the vendors to come up with some standard that can be followed in each case when you use technology to review documents?

But at the end of the day, it has to be plane English information that the expert can tell me that I feel good about the fact that this process is indeed reliable. That's really, for me, that's what it comes down to. I don't know that I would necessarily place too much faith in attorneys telling me, because they're just going to be repeating or parodying. What the experts said, but. Many things can get lost in that process. So I'd rather hear from the experts themselves who are familiar with the platform. And I'm not saying that the attorneys are not familiar. In fact, it's their obligation to be familiar with the platforms. But I much prefer to hear from the experts themselves to get, it's just been my experience, they explane things much better in terms of their own technology.

So I've never been in private practice. And so I don't know what it's like to have that pressure of the client saying, why do we need to give away all this information? This is all confidential. It's proprietary. I don't care that there's a lawsuit. Why do I have to give away our stoor here? When you're in your government practice, in terms of discovery, and I'll just kind of focus more on my days as an assistant lawyer. I was an attorney representing the U.S. government in civil matters. My philosophy was, you want it, take it. I just had an open book position, not only because it cuts down on the disputes, but also because I was representing the U.S. government. I felt that I had an obligation to be open with the other side.

But I need to sort of put that in context. Back when I was litigating, we were document-driven. So we're not talking, what is it now, zettabites of data? We're not talking that scale. I mean, the amount of data that is out there is just tremendous. So I had a very different practice. But that still stays with me, to the extent that… If you want it, take it. Do what you want with it. I've got nothing to hide. But I do understand that private practitioners have a very different concern and very different guidelines and pressure from their clients, because to produce everything that the other side wants, it's going to be very expensive, and you don't have the U.S. government paying for it. But aside from that mentality when I was practicing, there's nothing else that really influences me other than really, again, going back to the federal rules of civil procedure.

I think that one aspect that I do look at that perhaps other judges do not is this idea that although the requesting party and the producing party should be treated equally, they are not in equal position. They are different. One party wants to be in court, the other party does not. I do take that into consideration in terms of accessing burden, in terms of accessing how much burden is reasonable or unduly burdensome.

I don't know that it's going to make my job difficult. I think it's going to make the litigators job very difficult because not only do you now have discovery disputes to deal with. I think that during discovery, you are going to have authentication issues to deal with. I can't remember the last time where I tried a case or presided over a trial that we had an authentication issue. It's just not an issue. A document was produced. That's the document. No one questions its authenticity. But with Gen AI produced evidence, it's going to become more challenging. And I think the attorneys are going to have to sort of dig under and figure out how did you get this evidence? How was it created? And you could start challenging the process and start challenging the program itself.

And of course, what that means is it's going to be more expensive. I mean, I don't know whether GenAI in this regard is going to be helpful. But again, you know, going back to what RAND Corporation was trying to do was to come up with a plan, some standard that everyone can follow. And I hope that smarter people in the room can. formulate something or a process system that we can all sort of gravitate towards in terms of authenticating, in terms of showing reliability, so that we don't have to have these mini trials over authenticity and over reliability. I really do hope that's the case, but I don't know if anyone is really grappling with this issue of coming up with some standards.

Yeah. One thing I'm kind of excited about is the idea of both the requesting producing party using Gen AI in order to get the documents they need to prosecute and to defend. At the end of the day, this whole litigation process is a truth seeking process. Sometimes we forget that, right? Sometimes we look at it as a way to, it's a process for me to get my favored evidence. But it's really a truth seeking process. And at the end of the day, whether that process favors the requesting party or the producing party, it is what it is. With the cost of Gen AI going down and with the level of trust that we may build over time, Perhaps rule 34, you're not going to be detailing documents you want, you're going to be detailing prompts for the program to run.

And if the parties can get on board and say, this is the process we're going to take, then whatever gets, whatever is actually produced, you live with that, you know. Now, of course, there are going to be this and that, that you need to carve out, for example, privileged documents. Again, I think that there are, I mean, I've never used it, but I have read articles and have seen, Descriptions of programs that will in fact generate a privilege log. So if you can marry these features together, we can use 34 as a way to deal with all written discovery. Can you imagine a world where we don't have motions to compel? Well, for me, that will be an ideal where producing party and the requesting party agree on these prompts.

Now, maybe that's where the motions to compel will be, where the judge has to step in and say, that prompt is overly broad. That prompt is not proportional to the needs of this case. But I think that's my forecast, that's where we're going to go. Use of rule 34 is going to change dramatically.

All right. Thanks for having me. And thanks for listening to me. Take care.