I’ve been involved in the world of legal tech—specifically eDiscovery—since it started in the early 2000s. I’ve seen this industry and its technology evolve from the early days: starting with Z-Print and Discovery Cracker; Summation and Concordance; Attenex and Ringtail. Nowadays, the names have changed, but my question for us old-timers is: has eDiscovery technology really changed that much in the last twenty years?
When I started, Friendster was the king of social media, Amazon was only selling books, and the iPhone was still five years away. It’s remarkable how much technology has revolutionized how we work, communicate, date, and so on…but one thing that doesn’t seem all that much different is how we do eDiscovery.
I mean, when you really stop to think about it, what’s new in eDiscovery? Relativity certainly works better than Concordance, but the vast majority of users use them in exactly the same way. All the professional-grade processing/indexing platforms were introduced before 2010 (Nuix, eCapture, Invariant, Mindseye). And has anyone really improved Brainspace since it came out in 2012? I could go on.
“Disruption” in eDiscovery
I recognize how much better the modern suite of technology works compared to its digital predecessors, but is it iPhone better? Nah.
Yet, every year about this time, I start hearing about the next disruptive technology that’s about to break through, which subsequently flatlines. It’s an obvious cycle in eDiscovery. I’ll give you a few examples:
- Early Case Assessment (2007 – 2015)
Sometime around the mid-2000s, Clearwell made a huge leap forward with their eDiscovery-in-a-box solution, giving law firms unprecedented DIY power over their $1,500/GB vendors.
This product did an amazing job of solving what, at the time, was a huge problem—getting through data quickly. Copycats followed: Allegro, Digital Reef, and some others I can’t recall. These tools promised ultra-fast processing in exchange for limited search ability and bare-bones review. The idea was that if you removed 98% of the data with so-called “ECA” tools, the cost of eDiscovery would plummet. It didn’t work because, as it turns out, limited search ability and bare-bones review are requirements, so everyone reverted to doing it the long way (with some modifications).
Although this period did have a lasting impact, the promise of ECA never really materialized (and gave ECA a negative connotation). I would argue that the “proportionality” rules introduced in 2015 were much more effective in limiting the scope of data collection, and therefore, the cost of eDiscovery, than ECA workflows.
- Predictive Coding (2008 – Current)
First introduced as “predictive coding” (PC), then “technology-assisted review” (TAR, or the subsequently rebranded ‘TAR 2.0’), and now “continuous active learning” (CAL). Regardless of the acronym, the promise was (and remains) the same: eliminate the expensive, tedious, time-consuming work doc review attorneys perform using artificial intelligence.
Since document review is the most expensive part of eDiscovery, it makes sense that we’d try to automate these decisions, but alas, almost fifteen years later, we still have armies of lawyers reviewing documents. The robot lawyer never materialized.
- End-To-End (2012 – 2018)
Everyone loves the EDRM, especially when your job is selling the holy grail of eDiscovery technology software: a cradle-to-grave, end-to-end, fully integrated, automated solution that will make all your eDiscovery dreams come true. No more exporting and importing. No more file transfer. No need to use multiple vendors, in fact, no need to use any vendors at all—the software can do it all. The easy button.
The list of failures here is very long, but the most spectacular remains Autonomy.
- SaaS (2016 – Current)
Cloud-based software has disrupted countless industries, so much so that venture capital is on high-alert looking for the next industry to liberate for a tidy profit. The legal industry fits the bill: Luddites galore. RelativityOne and Reveal steal the headlines these days, but before them, we thought that Disco, Everlaw, and Logikcull would displace the “on-prem” hosting model that, despite their best efforts, persists today.
I was in the room when Andrew told his top clients that their hosting businesses would be eliminated by RelativityOne within five years. ‘Five years’ from then was two years ago. Disco, for their part, went public with this story to the tune of a billion-dollar valuation and Wall Street has not been impressed with their progress since.
Why Isn’t eDiscovery Changing?
Software may be taking over the world, but not the world of eDiscovery. Why not? The short answer is: because every project is slightly different, close enough isn’t close enough, and there’s just not enough money in it.
- If every project was more or less the same, if every dataset contained the same data types, if every production had the same parameters, if the issues were the same in every case – we would have automated the process by now. “Slightly different” means someone needs to look at it.
- Lawyers are notoriously slow to adopt new technology. There’s a lot of research on this, but it amounts to two things: first, lawyers get hired to secure legal outcomes, not innovate new methods of discovery (i.e., nobody gets fired for following tradition). Second, they bill by the hour.
- If we can get a computer to beat a world-renowned chess master, you’d think we could train one to make a privilege call. Nearly anything is possible if you throw enough money at it, but the sad news is that the cost of eDiscovery isn’t a big enough problem to attract the big investors or the large teams of engineers required to solve it (they’re focused on more important things like crypto and the metaverse).
This Holiday Season
Don’t get me wrong, I love stickers, laptop bags, and hoodies as much as the next guy and I’m a regular at the big parties with free food and loud music. I’ve also built my career on mastering the technology I seem to be criticizing, so please let me be clear that I have tremendous respect for the men and women that build and support these technologies. Good people, but honestly, they get plenty of credit, they’ll be fine.
You know who doesn’t get a lot of credit? Data analysts, project managers, doc reviewers, and system administrators. All the people that work for the LSPs doing the work, day in and day out, often behind the scenes. Real people make this industry work—not robots or algorithms—regardless of what you heard at the last tradeshow.
These are the real heroes of the industry. Big parties and hip swag are awesome, but everyone knows that when the shit hits the fan, it’s not a 1-800 number that you’re calling, it’s your trusted service provider.
Chief Executive Officer