On 2 August there was posted on YouTube a speech by Robert Elliott Smith, promoting his new book, Rage inside the Machine. It is subtitled How to Stop the Internet Making Bigots of Us All. It was under the imprimatur of Talks at Google.
Robert Elliott Smith is quite a lot of name; so if no one minds I shall refer to him as Rob, which is actually what I call him. He is a trainee of mine, so at least I won’t have to bother with any script-reading nonsense. I know he will shoot this from the hip.
I approach this talk with the words of Albert Einstein ringing in my ears –
If you can’t explain it to a six year old, you don’t understand it yourself.
Rob’s book is about algorithms, on which he is expert. I am very definitely not, so I should make a good judge of his powers of explanation.
Nevertheless the Q&A which begins at 42:15 shows that his audience compromises people who got algorithm, in fact they seem to be Google employees. His dilemma therefore is how to set out his stall so that video-watching dumdums like me can grasp it while not alienating the experts in the room. He solves it in the first seconds by metaphorically tipping his hat to the audience’s expertise. It’s a simple device, but effective.
The other side of this dilemma is that I have urged him never to spoon-feed his audience, that people engage more thoroughly with your message if made to think. Therefore he will have to tread, between dumdum and expert, a path made narrow by the need to engage both without spoon-feeding either.
Having explained to the dumdums that algorithms are the ubiquitous electronic calculations that, for instance, cause us to receive targeted advertisements through our computers, he moves into where and why they make mistakes. In particular he addresses the interesting concept that algorithms are prejudiced. This resonates with me. My having turned seventy two algorithms have wrongly concluded that I am in urgent need of a range of geriatric products, thus causing me much hilarity but not helping the supplier client.
Obviously this comes down to the alchemy whereby incoming data are transformed into outgoing conclusions, and Rob addresses the prejudice question by comparing it to human prejudice. He is well-placed, being a native of Birmingham, Alabama, and having been born as the Jim Crow race segregation laws were beginning to collapse.
Thus we have a section describing his growing up surrounded by racial issues; and that subsequently morphs into drawing parallels with data juggling by today’s computers.
I felt he got slightly bogged down in the autobiographical details, which can easily happen, and he needed to be using broader brush-strokes there. But though a card-carrying dumdum, and a geriatric one at that, I still felt that I grasped the essential message that algorithms still in their infancy, relatively blunt instruments, are constantly being made sharper, and here are some concepts whereby the sharpening process can be improved.
What I find particularly impressive is how many comments, not only on YouTube but also LinkedIn where I first saw the video, come from those who got algorithm and now want to read the book. Ultimately the reaction of the market is far more important than my opinion.
And, much more than before, I got algorithm. Who could ask for anything more?