“We won’t really understand the brain until we can make models of it which are analog rather than digital, which nobody seems to be trying very much”*…

… As Max Levy and Michael Moyer argue, perhaps that should change, if only because our climate future might depend on it…
Computing today is almost entirely digital. The vast informational catacombs of the internet, the algorithms that power AI, the screen you’re reading this on — all are powered by electronic circuits manipulating binary digits — 0 and 1, off and on. We live, it has been said, in the digital age.
But it’s not obvious why a system that operates using discrete chunks of information would be good at modeling our continuous, analog world. And indeed, for millennia humans have used analog computing devices to understand and predict the ebbs and flows of nature…
[Levy and Moyer survey the history from analog computing, starting with the ancient Greek Antikythera mechanism…]
… Analog computing reached its apotheosis in the differential analyzer, first built by Vannevar Bush at the Massachusetts Institute of Technology in 1931. The analyzer used a complicated series of gears and shafts driven by electric motors. It could calculate a huge variety of differential equations — the kind of equation used to model physical phenomena. But to modify an equation, the machine had to be laboriously reconfigured by hand.
When modern digital computing began in the late 1930s, it was clunky, expensive and inferior. But digital computation had benefits. Digital computers were easier to program and often more accurate than analog machines. And with the rise of the transistor and the subsequent advances fueled by Moore’s law, digital processing soon took over.
But as our digital world has exploded, its costs have as well. Every switch of a digital bit takes a smidgen of energy. And new artificial intelligence systems require huge amounts of computing power. To take just one example, news reports have revealed that Microsoft and OpenAI are planning a $100 billion data center that would suck about 5 gigawatts of power. That’s roughly the output of five nuclear reactors.
Analog computing offers an alternative. The neural networks that power AI systems make predictions by repeatedly blasting through a sequence of multiplication and addition operations.
In an analog computer that uses electrical signals — not gears and pulleys — a current could pass through a circuit that uses carefully chosen resistors to model those operations, at a significant power savings.
The advantages of digital computing are real, but so are the drawbacks. Perhaps, by reaching back to computing’s past, researchers will be able to steer a sustainable path toward our computational future…
You don’t need 0s and 1s to perform computations, and in some cases it’s better for the climate to avoid them: “What Is Analog Computing?” from @laxmevy and @mmoyr in @QuantaMagazine.
* Freeman Dyson
###
As we celebrate the continuous, we might send elastic birthday greetings to Peter Hodgson; he was born on this date in 1912. An advertising and marketing consultant, Hodgson introduced Silly Putty to the world. As The New York Times recounted in his obituary,
The stuff had been developed by General Electric scientists in the company’s New Haven laboratories several years earlier in a search for a viable synthetic rubber. It was obviously not satisfactory, and it found its way instead onto the local cocktail party circuit.
That’s where Mr. Hodgson, who was at the time writing a catalogue of toys for a local store, saw it, and an idea was born.
“Everybody kept saying there was no earthly use for the stuff” he later recalled. “But I watched them as they fooled with it. I couldn’t help noticing how people with busy schedules wasted as much as 15 minutes at a shot just fondling and stretching it”.
“I decided to take a chance and sell some. We put an ad in the catalogue on the adult page, along with such goodies as a spaghetti-making machine. We packaged the goop in a clear compact case and tagged it at $1.00”.
Having borrowed $147 for the venture, Mr. Hodgson ordered a batch from General Electric, hired a Yale student to separate the gob into one ounce dabs and began filling orders. At the same time he hurried to get some trademarks.
Silly Putty was an instant success, and Mr. Hodgson quickly geared up to take advantage of it…


You must be logged in to post a comment.