I have already linked once to Monica Bay’s article Panel Debunks Predictive Coding Myths reporting on a panel discussion between Howard Sklar of Recommind and David Kessler of Fulbright & Jaworski. Both of them were discussing similar points at the Carmel Valley eDiscovery Retreat in Monterey a few days ago, and the article has just resurfaced on Twitter. It is worth repeating.
The introduction of new technology quite often spawns myths. My long-time favourite practitioner is Dr Dionysius Lardner who constantly challenged the work of railway inventor Isambard Kingdom Brunel, asserting, amongst other things, that passengers on Brunel’s Great Western Railway would suffocate if the train’s brakes failed on a downhill slope and that a voyage from New York to Liverpool “was perfectly chimerical, and they might as well talk of making the voyage from New York to the moon”. The first of these assertions ignored wind resistance and friction and the second failed on the technical ground that “the carrying capacity of a ship increases as the cube of its dimensions, whilst the water-resistance only increases as the square of its dimensions”.
The point of the parallels is that a myth which was potentially damaging to progress was a) answered by science and b) proved wrong by experience when the railway passengers survived and the SS Great Western steamed into New York harbour with 200 tons of coal to spare.
All sorts of inventions and discoveries have been faced with such doubts and myths – heavier-than-air flight is one which would comes frequently to my mind for obvious reasons; doubters have derided things as varied as the telephone, the home computer and the Beatles. Pioneers perhaps need doubters to spur them to prove, by experience as well as assertion, that their inventions and ideas will work.
That is the task which Howard Sklar and David Kessler set themselves in this interview. Those of you with doubts about the future of predictive coding may care to see how these myths are challenged by these two eloquent and informed proponents of the technology.
The predictive coding naysayers may not accept the effectiveness of technology-assisted review, but you have not hitherto found many of them actually trying to prove that it does NOT work. One of the greatest dangers on the early Great Western was the risk of bumping into Dr Lardner’s private experimental train (pictured) roaming the rails, oblivious to timetables, whilst its owner sought to prove his pessimistic assertions. The equivalent, I suppose, would be lawyers who obstructed the course of litigation, getting in the way of its progress and interrupting the timetable whilst seeking to prove that modern technology does not work. Oh, wait – where have I heard a story like that recently?