memristors
Can I tell you what they are? No. Can I tell you what they do? Not really. But they appear to be a pretty incredibly big deal for the world of electronics. When something exists theoretically for decades before it becomes possible to fabricate; when it has implications for the fundamental ways a discipline explains its workings; and when, so shortly after its announced discovery, it's already been integrated into functioning systems: these are all reasons to be excited. It'll be a while before we come up with metaphors that can describe what the hell these things do in a way that makes intuitive sense to you and me, but it's not too early to recognize that, if the claims associated with memristors are true, their discovery is of immense importance.
Incidentally, it may seem as though I'm being a bit credulous about this seemingly huge upheaval in the world of fundamental electronics. Maybe I am, but I have a reason: the way we teach and understand the physics underlying electronic systems is totally screwed up.
It's all Ben Franklin's fault: not only was he a total jerk to John Adams (apparently), he also guessed the direction of electricity wrong. He thought it went from positive to negative, but in fact the electrons involved travel in the other direction. But by the time we had figured that out, enough progress had been made under the original explanatory framework that everyone just said "what the hell" and plunged onward. We still talk and teach and diagram as if current flows from positive to negative. And it works! But in order to make it work we've had to fool ourselves into thinking virtually every component in a circuit works the opposite way that it actually does. Fortunately once you reach a certain level of abstraction it becomes easy to ignore the situation, and eventually it recedes from your attention and ceases to be confusing.
But it's still a bit appalling, which makes it easy for me to believe that the discipline has overlooked fundamental truths in the way that Prof. Chua alleges in the article. Memristors: sign me up.
Comments
"In the analog domain, we want to build memristor-based devices that operate in a manner similar to how the synapse works in the brain -- neuron-like analog computational elements that could perform control functions where decisions must be made involving comparisons as to whether something is larger or smaller than something else. We are not building a neural network yet, but we think that using the memristor in its analog mode with our crossbar is a pretty good representation of a neural net."
I for one welcome our new memristor-based robot overlords.
I wish I had a better sense of what new kinds of things they can build now. I guess I'll find out when one of them is implanted in my brain.
It's just differential equations, Ogged. Work it out.