Should an attorney forge an alliance with a blood enemy if it means a chance to win a larger victory on behalf of a client? "Game of Thrones" is HBO’s sweeping epic of medieval noble families battling for domination of the Seven Kingdoms of Westeros and the coveted Iron Throne. Any fan of the show knows that strategic alliances with a mortal enemy can be important to winning the ultimate prize.

At the end of season two, House Tyrell and House Lannister temporarily set aside their roles as sworn enemies, allowing each family to better position itself for a later victory. Could this work in the law? Should an attorney sidle up to opposing counsel with a thumb drive of documents and say, "I’d like to show you these battle plans my client, Stannis Baratheon, has drawn up"?

Forging alliances to go to war against a common enemy or achieve a common goal, à la the Tyrells and Lannisters, is akin to forging the working relationships necessary to make predictive coding a useful tool. It’s not fantasy: Three key 2012 rulings indicate that predictive coding may require a degree of transparency that litigators could find unsettling.

Why should litigators step out of their comfortable roles as clashing warriors? Just as the lure of the Iron Throne can make strange bedfellows, the promise of cost savings for clients may be the catalyst needed. Significant savings are possible, provided that mortal combatants (ahem, opposing parties and counsel) let down their guards and share sensitive information in the pre-discovery phase.

Parties in litigation spend untold sums for veritable Dothraki armies of flesh-and-blood lawyers to perform exhaustive document-by-document review of discovery material. That’s because the prospect of inadvertently exposing a client’s vulnerabilities and privileged communication terrifies litigators.

The typical commercial lawsuit generates mountains of data that invariably contain documents irrelevant to the dispute at hand but sensitive enough that defendants are willing to stomach the escalating costs that come with combining traditional and electronic discovery in the hopes that these documents will never turn up in the discovery set.

Businesses of all types now store enough electronic information to fill Xaro’s vault in Qarth, thanks to the low cost of computer memory and lax data management. Not surprisingly, the price of discovery continues to rise at a significant clip. In-house counsel increasingly seek technological solutions that can reign in these mounting litigation expenses.

Predictive coding technology represents the latest hope for chipping away at discovery costs. The technology employs machine-learning concepts, in that lawyers teach computers to identify relevant documents and, importantly, exclude sensitive, irrelevant material. If done right, computers quickly and economically sort through discovery data and segregate the relevant material.

But obtaining these savings takes more than tapping out commands or keyword searches, and predictive coding is not a silver bullet. For predictive coding to really work, it requires investment in good technology, readiness on the part of litigators to depart from their scorched-earth, Kingslayer personas, and willingness by both sides to show some vulnerability by coming together early to collaborate in the discovery process.

Trio of Thrones

Three 2012 cases appear to open the door a little wider for broader acceptance of predictive coding in litigation. U.S. Magistrate Judge Andrew Peck of the Southern District of New York issued a ruling in Da Silva Moore v. Publicis Groupe & MSL Group that marked the first time a federal court formally encouraged the use of predictive coding in discovery.

In Global Aerospace Inc., et al. v. Landow Aviation, et al., Virginia Circuit Court Judge James Chamblin approved the use of predictive coding over plaintiff’s objections.

And in In re: Actos in July 2012, acceptance of predictive coding by U.S. District Judge Rebecca Doherty of the Western District of Louisiana was based on an unprecedented level of cooperation.

These three cases also share something else in common. In each, lawyers only used predictive coding after litigants volunteered or agreed to a significant degree of pre-discovery transparency. Collaboration is necessary because parties have to work together to teach the computer system to find relevant documents by exposing it to sample sets that include relevant documents. Importantly, the computer also needs to know what kind of sensitive and irrelevant documents it should not retrieve.

In Da Silva Moore, Peck wrote, provocatively, that predictive coding technology is "acceptable in appropriate cases." He also noted "[Codefendant] MSL’s transparency in its proposed ESI search protocol made it easier for the Court to approve the use of predictive coding." Unfortunately, however, the prolonged squabbling over the protocol following Peck’s ruling erased much of the savings promised by predictive coding.

In In re: Actos, the case management order required the parties to cooperate, make joint decisions and work collaboratively to determine the relevance of discovery documents during the training period for the predictive coding system.

For example, in a significant departure from the way lawyers typically handle discovery, each side nominated three people to work together to review documents and train the predictive coding system. Giving up their traditional role of independently reviewing their own documents for relevance, this team had the authority to make one relevance decision for the documents.

In Global Aerospace, Landow Aviation agreed to log any sensitive, irrelevant documents and privileged documents that it withheld, enabling opposing counsel to evaluate and possibly object to the coding decision.

Is it too risky for lawyers to lay down their swords in such a way? Is a little more pre-discovery transparency and collaboration worth the risk if it means potentially enormous savings for clients?

These are difficult tradeoffs, indeed. If a lawyer faces opposing counsel such as the likes of the sinister Lord Tywin Lannister or his conniving daughter and Queen Regent Cersei Lannister, he certainly can use good predictive coding technologies to cut costs without full collaboration from the opposing party. But if predictive coding can deliver on significant time and cost-saving promises, in-house counsel may soon be demanding such unusual alliances.

Traditional discovery costs only will continue to increase. A technological solution, and the transparency requirements that the successful use of predictive coding require, may be something that all attorneys must accept, especially if they hope to win the Iron Throne of victory on both costs and merits. Just keep your eyes peeled for dragons.