Notes from the IET / BCS
Blog Home Page
2014 Turing Lecture:
"Beyond silicon: cognition and much,much more."
London, 24 February 2014
We are all data
I have rarely enjoyed a train journey so much. I was buried in “A Dance With Dragons” with some soothing Bach on my headphones. My battles with Windows 8 on my new laptop and its seemingly unstoppable attempts to contact the internet and set up accounts with the mother ship were all behind me. I had wanted a machine that would never be connected to the cyber hive so that I could relax the eternal battle with spyware. Windows 8 had other ideas. It came preloaded and surgically grafted to the new machine and insisted that I needed to connect to the internet to validate the operating system. It came with a message that by using the laptop and the installed software I was agreeing to the collection of data about myself. It told me that if I had a problem with this I should return the laptop and negotiate a refund with the retailer. Details of this suspicious behaviour would be noted on the retailer records next to my credit card number, address, purchasing history and underpant preferences. I was okay on the train now though. My apoplexy had abated and I was reading my nice Kindle which, I found later, was uploading my reading progress to its mother ship. At least I have resisted the lure of the iPhone and my iPod has no wifi, until I connect it to my laptop to synchronise at which time it tells its mothership all about my movements, listening choices and probably infers my personality type.
I was on my way to the 2014 Turing lecture at the Royal Institute in London. Turing was a great mind but I wonder if he foresaw the changes silicon has wrought on the world. I doubt it.
The speaker was Dr Bernard Meyerson. He is the Vice President for innovation at IBM. I was framing a question about HAL in 2001 A Space Odyssey being only one letter removed as I entered the impressive building purchased by the Royal Institute in 1799. Here Humphry Davy and then Michael Faraday conducted some of the most important experiments in the history of scientific enquiry. I, however, was stopped at the front desk. Despite booking, my name was not on the list. This was soon rectified and I was directed to the balcony of the lecture theatre. “Up with the physicists,” according to Dr Meyerson, because they have figured out that falling plaster loosened after years of explosive experiments will not have gained enough momentum to do any damage up here on its way to brain the chemists below.
Meyerson has the aspect of a fast talking used car salesman and is an incredibly entertaining speaker. I even forgave him his constant brand placement as he made everyone belly laugh at least once as the talk progressed. Breadth is as important than depth, he told us, because if you have only deep experts they drill deep holes and you can find yourself sitting at the bottom of one.
The first part of the lecture was a hymn to innovation. There are two types: discontinuous and continuous. People who make huge breakthroughs every now and then are worth all the time they do nothing of value. These are the discontinuous innovators. They work in huge leaps of thinking. These are the pop stars of science. There are other kinds of innovators though. These are the people who innovate in small increments that over time make all the difference. They make sure that as chips double their capacity according to Moore’s law that they not double their power requirements thus ensuring that turning on your laptop is not a brief and very exciting experience. They invented the disk drive so your laptop does not have to weigh a quarter of a million tons.
We need to encourage all kinds of innovation, according to IBM, and we need to encourage it with infrastructure and diversity.
Silicon hits the ceiling
This is not just a politically correct plug for IBM, Meyerson has some real concerns about how much further silicon can take us. After fifty years of the stuff he thinks the technology has run its course. We have hit the mechanical limits. We are measuring in atoms now and atoms do not scale. He did not mention quantum tunnelling as did my colleague on the way to work the following day when I recounted the high points of the lecture to him. At the quantum level a particle can pass through barriers that would have stopped it in classic physics.
Meyerson talked about the subatomic world where insulators become conductors. We have burrowed as deep as he thinks we can go. He has some predictions.
Prediction for a post silicon age #1: Silicon will dominate IT but contribute nothing to progress.
Silicon is not the only thing impeding our progress. Light is far too slow for Meyerson’s liking. Compared to the speed of modern processors light can’t move data about fast enough so that the processors have to wait and twiddle their silicon thumbs.
Prediction for a post silicon age #2: Light is too slow.
Here is where the innovators, discontinuous and continuous come in. One solution they have given us is to integrate everything. Reduce the distance data has to travel.
3D chip bonding
Three dimensional integrated circuits require you to spend decades learning how to create chips 50 microns thin, stick them together, drill holes for buses and thereby extend Moore’s law for a while.
Prediction for a post silicon age#3: Integration of hardware, software and networks will compensate.
The wave driving this need for better, faster, higher capacity is big data. We think we have seen big data but according to Meyerson we are about to see a tsunami. Much of this will be brought on by our increasing use of video for one thing and the appetite for analytics for another.
“Information is the oil of the 21st century, and analytics is the combustion engine”
Peter Sondergaard, senior vice president at Gartner
In our classic data centred IT model we take data and pass them through a processor. We collect the results on the other side. The problem we are starting to face is that there are too much data to move about efficiently, especially with slow light and the limits of silicon mentioned above. The solution is to move the computer through the data.
We have arrived at analytics. Data mining was like finding a needle in a haystack but analytics is making the needle out of the haystack. Analytics is the discovery of meaningful patterns in data with the extensive use of mathematics and statistics.
Meyerson talked about using it to predict systems failure by analysing warning patterns in complex systems or using it in medicine to diagnose complex cases using genome, medical history, personal history, family history etc. Here the good doctor frightened the living crap out of me. He offered his bleak choice for the future:
Relinquish your privacy or die from terrorism or cancer.
I noticed Ben Goldacre at the same refrain earlier in the week. Apparently we have nothing to fear from big companies. If we have done nothing wrong we have no need of privacy and companies like IBM and Microsoft and Apple and Google will take the decision of life over privacy for us in the brave new world. Dr Meyerson cracked later in the Q&A when asked about big data, trust and the NSA, that if we knew what the marketing guys held on us we would not be concerned with the NSA. That made me feel so much better.
Watson and cognitive computing
Watson was the cognitive computing system that won the US game show Jeopardy where the contestants are given the answer and have to come up with the question. The doctor (Meyerson, not Who or Watson) was part of the team and it was indeed a remarkable achievement. His description of the journey the team went on to deliver that victory was a wonderful bit of theatre.
He pointed out that the 20watt human brains in the skulls of the contestants although consuming less power than Watson’s 80kw and with considerably less training than the 80 man years lavished on Watson, were impressive to still be in the game. People like Dr Meyerson intend to change all that.
I was thinking that this might be a good thing. In the right hands this sort of technology can be used to extend human cognition. I was thinking about the sadly missed Ian M Banks and Culture novels in which a future humanity is cared for and encouraged to self actuation by omnipotent Artificial Intelligences.
On stage Dr Meyerson was talking about how this technology was being used to infiltrate and influence consumer choices in the, up till now, democratic and self organising post Amazon advertisement immune blog communities. He was telling us how giant Chinese banks were using analytics and systems of insight to work out where to best place their branches to wipe out the competition. He finished his talk by assuring us that he and his colleagues are not building Skynet.
Not building Skynet
On the way home I left my electronic devices turned off and as I thought of the last person who assured me that this will not hurt a bit and wondered why the speaker at this years Turing lecture felt the need to assure me that he was not building a computer network that was destined to overthrow humanity.
The talk is webcast by the IET at their web tv site
Blog Home Page