I am about to show you a pop video on YouTube. This not entirely a bit of Friday afternoon relaxation, although I know that some of you wind down on Fridays and even take some week-ends off. It has a serious point, relevant to the use of technology for serious business matters.
To what extent do you need to know how something works in order to appreciate it (in the case of the video) or trust it (in relation to your clients’ disclosure and your professional duties)? In practice there are multiple levels of understanding, from the superficial to the deeply technical. In some cases, you need only to be aware that it is possible to achieve something with some technology, so all you want is the phone number of someone who can bring it along when you spot that it is needed. In others, you may have to justify in quite complex terms why you chose to use one method or application rather than another or none. In very few circumstances do you need to understand the computing science or the technical skill involved, just some basis for trusting that it works.
First, though, the video which set me musing on this subject. It is not necessarily to everyone’s taste visually, musically or in terms of the lyrics – especially the lyrics in fact.
It is the video to Interpol’s Evil
For those whose IT department rules bar you from YouTube, it shows the aftermath of a car crash. As the emergency services rush about tending to the injured, a life-size figure sings Interpol’s song. It looks like a marionette, but you can’t be sure. The mouth moves (more or less) in synch with the words being sung, but it is the facial expressions which are the source of wonder, to me at least. Every relevant emotion – fear, sadness, pain, bewilderment – crosses it in a manner which is recognisably human whilst simultaneously impressionistic. The eyes bulge, roll, stare, close and look around; the mouth droops and gapes, all the while “singing” the song; at one point, a vein appears on the forehead, consistent with the pressure implied by the rest of the face.
You see the figure being tended as it is put into an ambulance, then sent down a hospital corridor on a trolley, the camera above him capturing the face as well as the floor rolling past and the feet of the person pulling the trolley. In the operating theatre, the surgeons do their stuff and leave, and the figure leaps up and dances wildly and jerkily, all the while singing and making appropriate facial expressions, before collapsing dejectedly.
“How the hell did they do that?” my son asked. Was it a puppet or a human in a mask? Was it all cgi? Well, I have looked it up and know I know – at one level of understanding at least. It was a life-size puppet. The head was animatronic (but how does that work exactly?) and programmed to sing the song automatically whilst a puppeteer controlled the other facial features. Six other puppeteers manipulated his body and were digitally removed afterwards. The other players were all human and the operating theatre was a real one.
Well, that’s all clear then. But I am still no wiser how this came together, how this combination of technology and human skill produced the result. Nor, frankly, does it matter beyond that surface level of understanding. It is all far removed from Cliff Richard standing in front of a microphone to a static camera.
One of the reasons why lawyers like paper is that, as with Cliff Richard and his microphone, what you see is exactly what there is. You can pick it up (the paper, I mean, not Cliff), and examine every detail. You know how the ink got there, can see who signed it and who it was to and what it said, or perhaps deduce with common-sense that all is not what it appears to be. It is rather different with electronic documents.
For a start, they may not be “documents” in any conventional sense, but an amalgam of electronic forms and bits pulled from databases. Instead of browsing through a filing cabinet, you have perhaps instructed someone to do a “forensic collection”. He has explained what this means and how he intends to do it, but frankly, as he sets to work with his little black box, you have not a clue what is actually happening.
The mystery continues when your “documents” go into an application which, you are told, will cull and filter the vast quantities down to something you can use. You do, of course, have a surface level understanding that the keywords or search terms which you have chosen are to be used in this process, but how? Does it matter how, or is this like my understanding of “animatronic”?
This is a big issue just now in the US, where battles are joined over the efficacy of one tool’s algorithm over another and where a distinguished Magistrate Judge has recently said
“Whether search terms or ‘keywords’ will yield the information sought is a complicated question involving the interplay, at least, of the sciences of computer technology, statistics and linguistics… Given this complexity, for lawyers and judges to dare opine that a certain search term or terms would be more likely to produce information than the terms that were used is truly to go where angels fear to tread.
Magistrate Judge John Facciola – U.S. v. O’Keefe
Such questions turn up all over the place. How reliable is OCR? What use is metadata? Do I need to collect slack space? What exactly is Bayesian probability and how does it affect my trawl of the MD’s mail files? How do I know I have missed nothing?
In some cases, it matters very much to know about these things. For the most part, however, what matters is to know that there are is more than way of doing most things, and to know who to ask – indeed, knowing that there is something to question is part of deciding what is proportionate.
The answer to most of the questions above lie in that word “proportionality”. The question to ask is “What sort of case is this?”, and to know that some cases do and others do not, warrant a forensic collection, close attention to the metadata, or a close understanding of the strengths and weaknesses of keyword searches. Then at least you know what questions to ask, and the only outstanding point is knowing who to ask.
But you do not need to know, usually, what the underlying technicalities are. To me, some of these search applications are near-miraculous in their power. You can learn enough about what they are doing to take a view on their results, at least to an extent which is proportionate for most cases. If you have to handle a case where it really matters, you can find out.
In the unlikely event that I ever want to make a pop video starring a six foot puppet who can sing and make subtle facial gestures whilst leaping about on an operating table, I know who to ask. Until then, I am content with my outline idea of what is involved. At least I know it can be done. Those engaged in litigation need at least to know what is possible.
As an afterthought, I think the whole video is about an unsuccessful outcome to litigation. The opening car crash is the end of the trial. The ambulance and hospital sections are the post-judgment conference, and you can see the barristers (here disguised as surgeons) striding off with a flick of their gowns and closing the door. The twitching and leaping around at the end is when the client has just opened the bill.