46 THE ADVOCATE
VOL. 77 PART 1 JANUARY 2019
other electronic documentation, many large cases now involve far more
material (and far more irrelevant material) than ever before. In circumstances
where a counsel team may be inundated with well over a million
documents, it may be neither time-effective nor cost-effective, nor perhaps
even possible depending on the time available, for even a large counsel
team collectively to review every document.
One response to this new reality is to look to outsourced review
providers, where teams of reviewers (often legally qualified in other jurisdictions)
are briefed on the case and then do a “first-level” or “1L” review to
filter out irrelevant documents. The counsel team to whom the filtered set
of documents is provided after this first-level review then has a more manageable
pool of documents with which to work. The process is the same as
it would be within the law firm—documents receive an “eyes-on” review,
one by one—but more human effort is applied to the task, and outsourced
rates are more economical for the client.
In an “algorithmic” or “iterative” document review, a small team uses
bespoke software to conduct a series of searches across an entire pool of
documents, irrespective of volume. The team “teaches” the software what
characteristics make a document more or less likely to be relevant or
responsive to a particular issue by identifying key words, date ranges,
names and so on. As counsel review the initial rounds of search results,
they provide the software with feedback, identifying patterns, relationships
between documents, false hits and so on, following which the software
incorporates the fine-tuning and re-runs the searches. Proponents claim
that by repeating that “sifting” process again and again, with sampling, spotchecks
and audits throughout, the software completes a proportionate,
defensible and high-quality review of a large-scale pool of documents, more
quickly, more cost-effectively and—crucially—more accurately than a traditional
review. However, before adopting this approach the counsel team
must get comfortable—perhaps aided by a well-drafted claw-back agreement
enabling counsel to require, in certain circumstances, opposing counsel
to send back and not use documents accidentally disclosed—with the
fact that not every document disclosed has had an “eyes-on” review.
In recent years, “keyword” searches in online databases—or Boolean
searches for the truly savvy—have become a common feature of legal practice.
The traditional model of online database research has relied on identifying
key words or phrases (including previous cases or particular rules or
statutes) and relied on software to comb through information to identify
instances where those search terms occur.