Big Data is one of those books written by experts who you heard talk somewhere ( Start the Week , BBC Radio 4 in my case) and thought, yeah, that's interesting. And I've got a hangover professional interest in data as well from HFEA days. The trouble is, that the book just restates, (certainly in the first three chapters), what he managed to say quite well on the radio. Basically, we have such huge data sets ( and I mean if you think the number of little triangles in a Risk set is big, then you're thinking way way too small) that accuracy, and meaning aren't important anymore. Google found a correlation between bits of their data in searches and the spread of flu in the United states that predicted outbreaks two weeks before the medical authorities were aware they had a problem. Big messy data, unstructured, incomplete, but containing within it correlations that allow you to make real time predictions. Women buy certain kinds of goods when they first know they're pregnant, so you can bombard them with baby stuff months before anyone else sees them ordering a babygro. Vital signs monitoring in premature baby units show, counter-intuitively, certain very stable patterns just before they go down with an infection, so you can predict ahead of time. Don't know why, just know what is going to happen. Don't know what the causal link is - just know the prediction.
Well it sounds like epidemiology to me. Give a million people statins everyday and less of them will die of heart attacks and strokes. Monitor a million people who smoke vs a million who don't, the smokers die earlier. We don't really know why, pertinent for me in the case of statins - if I take them I'll halve my chances of a heart attack / stroke in the next ten years. Why? Don't really know.
I am getting bored writing about stats because the final awful certainty of it is that Amazon, Google and Facebook are going to sell a lot more stuff a lot more successfully to you, and you are going to buy more material goods than you ever had before. Ain't that a good idea. Oh well. We have the data and the processing power.
But at the Hayward Gallery there were three rooms, one brightly lit in green, one rose pink, one blue. Walk from the green into the rose room and that rose is the deepest most intense rose pink you'll ever see. All you green cones have been saturated with the green light - all you can see is pink. It was really powerful almost overwhelming. The attendant outside the room was doing a whole sing and dance act, attracting people in, a sort of random rap, "All you gotta do, is take off your shoe, it's your birthday in this room, you won't have no gloom". He was kind of freindly and weird and joyful at the same time.
Later on, I overheard the head attendant saying to another, "Can you hear him? What shall I do, shall I just tell him to shut up?" I couldn't resist intervening - I said it's OK, it's a bit of harmless joy, nobody seems to mind.
dreamthinkspeak were as atmospheric and elegiac as ever. Instead of the department store that covered one floor of the Cherry Orchard in Brighton, we were invited into the headquarters of Fusion Inc, to see the PetBots (actually built in Brighton), or the flame something, the demonstrators speaking the same mixed argot of intalian/spanish / german /russian and nonsense language that they spoke in the department store. This will mean nothing to you if yo have never seen dreamthinkspeak. Just see them, whenever you can. The Somerset House event is sold out to July though. Nothing that new in it though if you've seen them before, but still a dreamy, immersive experience, with moments of power and beauty. They seem better when they are working with a myth or text like the Cherry orchard, which gives a more unifying and balanced theme. Lacking in meaning. We know working in call centres is dehumanising. We know that some high tec companies value the tech above all everything except whether it will sell or not.
I was a bit snotty about the New Statesman in a previous entry so bought one to read on the train up to London. It kind of met my snotty expectations. A bit local (though very European - article about how we always needed/ resisted the Germans!). Best article was Will Self writing about Costa in terms that you might expect, but so so stylishly.
So not a great day for reading, = a bit of a boring ramble of a blog. In my view.
No comments:
Post a Comment