Skip to Content

The Wiki Man

The Wiki Man: Speaking to Siri

7 January 2012

12:00 PM

7 January 2012

12:00 PM

Why am I typing this article rather than dictating it via some wonderful voice recognition software? It’s a question worth asking. Twenty years ago, all Spectator writers would have written every article by hand (only two or three still do). In my office in the 1980s, it was frowned on to type your own letters, since typing was seen as secretarial work. Are we due another revolution in how our thoughts are transmitted from brain to machine?

Science fiction has always assumed that computers will converse — HAL-9000, C-3PO, Marvin the Paranoid Android and Thermostellar Bomb #20 (from the 1974 comedy Dark Star) have provided some of the most memorable dialogue in film. Now, 44 years after HAL refused to ‘open the pod-bay doors’, the new Apple iPhone contains a virtual assistant called Siri to which you can address short commands or questions. Given Apple’s record of inspiring trends rather than merely exploiting them, we can’t ignore this development. Yet for now I remain sceptical.

Yes, there are some circumstances in which voice activation is useful. If you are driving, say. Currently the only voice recognition service I use is a number I call from the car called Traintracker (0871 200 49 50) which gives me live departure times of trains whilst I am headed to the station. It can be slightly unreliable (I find it helps to fake a Mancunian accent for place names, pronouncing the ‘u’ in Surrey like the ‘u’ in ‘put’ not ‘putt’) but it is better than crashing the car while trying to type.


Intriguingly the technology behind Siri has its origins in defence research. Fair enough; the dogfighting pilot of an F-15 may already have enough to do with his hands to welcome the option of issuing additional spoken instructions. But the pilot has a first-class noise-cancelling microphone. On earth, once you have children or televisions or sirens or any other background noise, it’s a different story.

With any noise, the slightest glitch in voice recognition leads to responses so mind-numbingly insane you want to tear your hair out. If I mistype Westerham and am told the weather in West Hampton, I can cope with that. If I ask for the weather in Westerham and am told what’s showing at a cinema in Muncie, Indiana, I am reduced to a furious, gibbering lunatic.

So for now I’ll stick with typing. In ten years’ time, I’m sure, someone will have developed a device making the connection between brain and machine faster. But in a way I hope they don’t. You see, that 1970s system, where business letters were first handwritten (or dictated) and then typewritten by someone else, was spectacularly costly and inefficient. Allowing for revisions, each of which involved a return trip to the typing pool, it would not have been uncommon for the gestation of a business letter to span three days. With postage and paper it cost about £1 in modern money — or perhaps £50, allowing for the time involved. And in a way that was good. Because it meant that, when you received a letter, there was a reasonable chance that its author had something important to say and had invested some time and effort in saying it concisely and intelligently.

What technologists need to invent next is not a faster, easier form of communication but a more costly, effortful one. Something that imposes on the sender some extra cost and effort, thereby signifying the importance of the message and suggesting some thought has gone into its preparation. The email equivalent of the handwritten envelope. ‘So sorry to have sent you such a long letter,’ Pascal once wrote, ‘But I didn’t have time to write you a short one.’ Easy writing rarely makes for easy reading.

Rory Sutherland is vice-chairman of Ogilvy Group UK.


Show comments
Close