Wow. What can I say, Henry Markram is a man after my own heart. I just read an article by Jonathon Keats on Wired.com and I have to say the sentiments in it are quite a revolution in thinking.
I am not plugging my new book, The Animus Portal (well, I am a little, it’s on Kindle worldwide) but in it I make a few points that ring true with Henry Markram’s vision of how we will learn to understand the human brain in its entirety. I know we have learnt and are learning a lot about the functionality of the brain from semantic groupings of memories to cell structure and health. Problem is, even though what we have learnt so far could fill a warehouse, the majority of a map of the brain would have the words ‘Here be Dragons’ marked across most of it. We are still babes in the woods comparatively speaking.
Actually, I’ve decided, I am about to shamelessly plug my book, The Animus Portal (available on Amazon Kindle worldwide). In it I take a different approach to the same problem. I begin with a cure for dementia through the insertion of a chip that connects into the spinal cord just under the skull, that can rebuild damaged cells (you’ll have to read it to find out how) and restore memory function. Through the funding by a Russian oligarch the technology takes a different turn and in effect begins to use the brain like the massively capable organ it is.
So, what are the benefits of using a human brain versus building one? The first, and most obvious point, is that the brain already exists. It is compact and highly functional and yields little waste energy (although you might not believe it if you have ever sat in an interminable business meeting where lots of self-important people are expounding on stuff you already know, at which point you are suddenly startled into the realisation that the hot air to information ratio is huge). The brain also has massive decision making capability, each synapse representing a decision gate supposedly equivalent to those found in computer chips (or are they?) and already has the processing capability of more than one supercomputer. I’m not going to bore you with stats here.
Network a few brains together and you have a computer capable of analysing absolutely anything quickly and efficiently. They have built-in failover capability through inter-networking multiple brains, a much higher MTBF (Mean Time Between Failure) than any current hardware components and would create and self-manage the perfect neural network.
So, that’s the version 1.0 of the solution. For version 2.0, we decide instead to build one. First issue is the price tag, €1.3billion, although that in itself isn’t the actual issue, the issue comes when they put out the begging bowl asking for more money saying there is a 100% overrun on costs. So, if we can cough up the final €3bn price tag (they WILL come back to the table more than once as it will be a publically funded enterprise), we then have to work out a suitable language with which to programme it. Most current options are too buggy and inefficient to be satisfactory. Who among you has actually had a personal blue screen of death? Not me although there have been moments. Would the programming be ‘touchy-feely’, say something like The Matrix, or something a bit more prosaic? The whole project would be a mass of firsts; for example, just inventing memory that correctly mimics human memory would take a complete rethink of technology. Should we be happy with all of these firsts, the next problem is: which football stadium are we taking over to house it? We would struggle to contain it with current technology, short of buying and covering a small island. Have you ever seen pictures of IBM’s Watson or ICE and they fall a long way short of an actual brain?
All kidding aside, I like the way Henry Markram thinks, it is certainly outside the realms of what I’ve seen up to now. I also think he’s got a long way to go and his timetable of ten years is ambitious. Never the less, go Henry, go!