A few years ago, when theoretical physicist Michio Kaku took on the future of computing in his thankfully short lived Big Think series, I pointed out the many things he got wrong. Most of them weren’t pedantic little issues either, they were a fundamental misunderstanding of not only the existing computing arsenal deployed outside academia, but the business of technology itself. So when the Future Tense blog put up a post from highly decorated computer expert Sethuraman Panchanathan purporting to answer the question of what comes after computer chips, a serious and detailed answer should’ve been expected. And there was one. Only it wasn’t a reply to the question that was asked. It was a breezy overview of brain-machine interfaces. Just like Kaku’s venture into the future of computing in response to a question clearly asked by someone whose grasp of computing is sketchy at best, Panchanathan’s answer was a detour that avoided what should’ve been done instead: an explanation of why the question was not even wrong.
Every computing technology not based on living things, a somewhat esoteric topic in the theory of computation we once covered, will rely on some form of a computer chip. It’s currently one of the most efficient ways we found of working with binary data and it’s very unlikely that we will be abandoning integrated circuitry and compact chips anytime soon. We might fiddle around with how they work on the inside making them probabilistic, or building them out of exotic materials, or even modifying them to read quantum fluctuations as well as electron pulses, but there isn’t a completely new approach to computing that’s poised to completely replace the good old chip in the foreseeable future. Everything Panchanathan mentions is based on integrating the signals from neurons with running currents through computer chips. Even cognitive computing for future AI models relies on computer chips. And why shouldn’t it? The chips give us lots of bang for our buck so asking "what comes after them" doesn’t make a whole lot of sense.
If computer chips weren’t keeping up with our computing demands and could not be modified to do so due to some law of physics or chemistry standing in the way, this question would be pretty logical, just like asking how we’ll store data when our typical spinning disk hard drives can’t read or write fast enough to keep up with data center demands and create unacceptable lag. But in the case of aging hard drive technology, we have good answers like RAID configurations and a new generation of solid state drives because these are real problems for which we had to find real solutions. But computer chips aren’t a future bottleneck. In fact they’re the very engine of a modern computer and we’d have to heavily add on to the theory of computing to even consider devices that don’t function like computer chips or whose job couldn’t be done by them. Honestly, I’m at a complete loss what these devices could be and how they could work. Probably the most novel idea I found was using chemical reactions to create logic gates, but it’s trying to improve a computer chip’s function and design, not outright replace it as the question implies.
Maybe we’re going a little too far with this. Maybe the person asking the questions really wanted to know about designs that will replace today’s CMOS chips, not challenge computation as most of us in the field know it. Then he could’ve talked about boron-enriched diamond, graphene, or graphene-molybdenum disulfide chips rather than future applications of computer chips in what are quite exciting areas of computer science all by themselves. But that’s the problem with a bad question by someone who doesn’t know the topic. We don’t know what’s really being asked and can’t give a proper answer. Considering that it originally came from a popular science and tech discussion though, makes answering it a political proposition. If instead of an answer you explain that the entire premise is wrong, you’ll risk coming across as patronizing, and making the topic way too complex for those whose expertise is not in your field. That may be why Panchanathan took a shot, though I really wish he tried to educate the person asking the question instead…