[Editor’s Note: This article has been republished with permission. It was originally published November 9, 2023 on the eDiscovery Assistant Blog]
In Episode 123, CEO and Founder of eDiscovery Assistant, Kelly Twigger, discusses the scope of and whether search terms in ediscovery can be used to identify relevant social media posts in In re Tasigna (Nilotinib) Prods. Liab. Litig., presided over by United States Magistrate Judge David Baker.
Let’s dive into this week’s case. It’s a short one this week, but it raises two really important principles for you to think about. This week’s decision comes from us from the In re Tasigna (Nilotinib) Prods. Liab. Litig. This decision is from United States Magistrate Judge David Baker in the Middle District of Florida, dated September 18, 2023.
We are before the Court here on multiple issues on discovery disputes. Two of them are specifically worth noting that we’re going to discuss today. The underlying case here is a multidistrict litigation in which the plaintiffs allege injuries associated with the use of Tasigna, which is a drug used to treat chronic myeloid leukemia.
The first issue that we’re going to discuss is the production of social media from the plaintiffs. Novartis, the defendant here, sent requests for production to plaintiffs requesting social media profiles — the plaintiff’s social media profiles. The plaintiffs objected and we’re before the Court on the scope of what must be produced by the plaintiffs.
The plaintiffs argue that the ESI in their social media should be searched using search terms and only responsive materials should be produced. Novartis argues that the plaintiffs should use a tool such as the “download your information” tool to be able to produce everything that is pulled from that social media platform.
The “download your information” tool is not a commercially available tool that works across social media platforms. It’s simply the social media platform’s individual ability to allow users to download all their information. What’s important to know about “download your information” is that, particularly with Facebook, you don’t get anything that resembles what a Facebook profile normally looks like. It’s all broken up into HTML. You’ve got a section that has all the photos, you’ve got a section that has the context of the posts, but nothing appears the way it does on the social media platform. Its value is very, very limited from the “download your information” tool. We’ll talk about that a little bit more.
According to the Court here, although it doesn’t say how, it says that Novartis made the showing that, “search terms cannot be tailored sufficiently to capture responsive social media postings, particularly given the often casual nature of such discourse.”
That’s really valid. It’s virtually impossible to come up with search terms that you might use on a social media platform because you’ve got such a disparate sense of users adding comments, the plaintiff’s language, and whose writing those posts. This is not a commercial case in which you’ve got two parties that are continuously dealing with each other, so you can look at emails or other documents to see what language is from which to draw search terms. Social media is just something it does not lend itself to search terms very effectively. That’s why, as we continuously talk about on the Case of the Week, you have to approach how to identify, collect, preserve information for each individual source of ESI based on how that ESI exists in the current platform.
The Court ordered that “subject to the time frames and any other limits that the parties have agreed to, the plaintiff’s social media ESI will be produced in full.” But the Court did allow for each individual plaintiff to seek any further protection from the Court as needed — “if broad areas of the downloads have nothing to do with the matters in this case and there is a way to segregate such material, after the appropriate meet and confer, the plaintiffs can file for a motion for further limitation.” Essentially, the Court has said you can’t use search terms, but you’re still only supposed to provide what is relevant. If you conduct a collection of the tool, you find that there are broad swaths of information that shouldn’t be produced, come back to me and we’ll deal with it separately on an individual basis. That’s really the discussion of social media.
The second issue here is related to search terms to be used on ESI in electronic devices. As I mentioned, this is a sophisticated, complex multidistrict litigation. This decision today is a very short one that really, very briefly addresses these issues. I’m sure that the other decisions in the case, there’s one other discovery decision in our database that is similarly brief, but likely because the parties are before this Court on a very regular basis in this MDL, and the Court is not repeating the history of all that information. In this particular issue, the search terms issue, the Court doesn’t even specifically talk about what electronic devices are at issue.
The plaintiffs here told the Court that they intend to conduct a “reasonable manual search” of their devices for any responsive documents without the use of search terms. The Court looked at that and said, “I don’t think so.” Instead, it said:
“[T]o the extent that they intend to perform a manual search on their devices, unless they are going to undertake to certify that they have reviewed every piece of ESI in the device during the manual search and nothing responsive has been found, the devices should be put through a technical search process to apply search terms in finding the responsive documents.”
A couple of issues here. One, we don’t know what devices we’re talking about. It sounds like we’re talking about mobile devices. Second, you’ve got the search term issue again. We just dealt with that with regard to social media. Search terms in mobile devices are still going to be hard, and the Court doesn’t really talk about what’s going to happen here, whether they’re going to require them to conduct a forensic copying of the devices and then be able to run search terms off of the information that’s gathered. Some applications have to be downloaded separately. Others are going to be stored in a cloud somewhere. It really is very dependent on looking at each individual source, and the Court didn’t go so far as to do that here. But again, the parties didn’t seem to raise it.
The Court also overruled plaintiff’s objections to the defendant’s search terms that were proposed and ordered that those terms proposed by defendants used to be conducted that search. Those are the two main issues.
What are the takeaways here? Let’s start with social media. First, social media collection and production is still very fraught in eDiscovery. What’s interesting to me here, as I mentioned earlier, is that Novartis asked the plaintiffs to provide the export from the “download your information” tool, which is that native ability to allow you to download your data from each individual platform. I’m not sure if Novartis is looking for something specific that they think they could parse through those HTML downloads to be able to find, but there won’t be any context for it.
Social media collection really requires using a tool that is designed and dedicated to that purpose and that has collected from that specific platform — social media platform — whether you’re talking about Facebook, from Twitter, now known as X, whatever you’re talking about, you need to have specific metadata pieces that you’re gathering. You need to have specific views. You need to know whether you want to grab comments. All of those are really important, and you need to use social media collection tools to be able to get to that.
The standard, really, in social media collection is to copy a profile or the scope, date range of a particular profile, and then be able to redact non-relevant information if that’s permitted, or apply any other protections that maybe have been ordered by the Court or that the parties had agreed on.
The changes in social media lately — people leaving Twitter turned X, the formation of threads, the search for where Twitter users will land in multiple platforms — is all in flux. Companies that are collecting these platforms defensively, using a real social media collection tool, have to constantly revise and develop new tools for those different and changing platforms. We have talked previously on the case of the Week about how the social media platforms changed their algorithms after the 2016 election to prevent fraudulent use and displays of information on one side or the other to voters. That meant that social media collection companies had to go back and completely revamp their tools so they continue to work.
That means that there has to be an ROI for those social media collection tool companies to develop those tools. We know a lot of people signed up to use the Threads platform, but we haven’t seen any eDiscovery of threads in case law to this point, and there’s not a lot of discussion about it. How is it collected? Can tools that were used to collect Twitter be retrofitted, or modified slightly to be able to collect from threads and still be able to view those things? That’s possible. But what that means for you is that you shouldn’t wait to identify what tools you’re going to use for these types of collections, and you need to understand not only what tools you can use, but what those associated costs are so you can have that discussion with your client and know out of the gate what things are going to cost. We see a lot of social media collection in smaller cases, in employment matters, and understanding those costs and having a tool at your disposal to be able to identify and collect that information is really crucial.
The second issue on our takeaways is the search terms that are used to search devices. As I mentioned, there’s nothing from the Court on what devices we’re talking about. It seems like mobile devices. Search terms in that context are a little bit fraught. But frankly, the plaintiff’s approach to search terms here was baffling to me. Again, this is a really thin decision, so it could be that there’s a lot more information in the briefs. On Case of the Week, we just go from the context of what is included in the decision from the Court. The defendants here proposed search terms, and instead of proposing different terms, the plaintiffs said they would just look through the phone and find what was relevant. As the Court required here, unless they could say they would review every application and every single entry in every application, that was not going to cut it. What the plaintiffs should have done was to respond to the defendant’s search term after reviewing information on the devices so that they could provide real feedback as to search terms that would be viable. Instead of doing that, or by not doing that, they essentially lost their right to control the information to be provided from those devices.