Low Resource Computing to the Fore

Now, this is the kind of thing we should hope for from DARPA: last week it issued an RFI focused on low resource computing (LRC), seeing as the “Department of War (DoW) typically operates in environments with limited power, communication, or physical space, where mission completion depends on the ability to sustain critical computation and adapt at the edge.” DARPA is firm about looking for breakthrough technologies only: “DARPA is interested in disruptive concepts at the material, component, runtime, and authoring levels that represent significant advances beyond current practice for fielding robust, ultra-efficient systems,” further emphasizing: “Approaches resulting in incremental improvement and description of existing capabilities or prior work without a clear vision for advancing the field are not of interest.”

You can see why DARPA is interested in low resource computing. Credit: Microsoft Designer

I love it. I mean, I don’t know what kind of responses DARPA is likely to get, but I love breaking the paradigms that govern most of our computing lives. We worship at the altar of Moore’s law and brag about the computing power of our laptops and smartphones, but AI is driving us back into the era of “bigger is better,” with lots of computing power needed to train and operate them. Your devices may be powerful, but without the cloud to help them, they’re more like expensive paperweights.

Most of us, of course, don’t operate under the kinds of constraints DARPA is worried about, but all of us stand to benefit from any advances its LRC initiative brings. We’ve been operating under basically the same conceptual computer model since the 1940’s, just with better chips, so it is time for a fundamental rethinking.

As DARPA says:

Decades later, the apex of the commercial computing industry follows this same resource-heavy trajectory. Modern data centers, built around advanced graphics processing unit (GPU) clusters to drive artificial intelligence, demand staggering, gigawatt-scale power requirements that necessitate massive, dedicated infrastructure.

While commercial enterprise focuses heavily on scaling upward to exascale capabilities, this trend leaves significant, overlooked potential at the lower end of the computing spectrum. There remains substantial room for improvement and innovation by revisiting architectures through a lens of strict resource scarcity. By optimizing systems for extreme efficiency rather than scale, it is possible to develop highly resilient computing platforms that do not rely on massive infrastructure.

DARPA is looking for responses that address at least one of the physical resources constraints: low power, low memory, low reliability, or low technological level, and may also address one or more logical resource constraints in addition: low trust, low privilege, self-hosting, or low complexity of user experience.

As Colton Jones described the RFI in Defense Blog:

The Pentagon’s most ambitious research arm wants to build computers that can think in the dark, operate on almost no power, and keep working even when their own hardware is failing, and it is now asking the technology world to help figure out how.

Sure, why not?

I can’t help thinking of the work Leidy Klotz and colleagues have done, as described in his book Subtract. Basically, they argue that our design bias is always to add, never to subtract, and that bias causes us to overlook many opportunities. The proponents of frugal innovation would understand. The RFI suggests that it’s time to strip computing down to its essentials, and see if we can rebuild differently.

My first thought, right or wrong, in seeing the RFI was fungal computing, which I wrote about a couple years ago. People like Andrew Adamatzy, Professor of Unconventional Computing at the University of the West of England, or Robert Shepherd, leader of the Organics Robotics Lab at Cornell have been on this track for years.

Professor Adamatzy is also the Editor-in-Chief of The Journal of Unconventional Computing, which publishes research on varied topics like chemical computing, biomolecular computing cellular automata, or logics of unconventional computing. Professor Shepherd is more focused “on using organic chemistry of soft material composites for new capabilities in robots,” but has used fungal computing to guide them. Neither may be the kind of person who will start the next Apple or the next Microsoft, but both seem like the kind of people who see the future differently.

Exactly the kind of people DARPA is hoping to hear from (although, of course, Professor Adamatzy is English, and neither may be interested in helping DARPA or the Defense Department).

If you think that fungi would be an odd choice to power computing, well, you must not have been paying enough attention to them lately. Their networks are everywhere and transmit startling amounts of data, as a recent paper illustrated. As the paper’s lead author, Dr. Toby Kiers, told Alan Burdick of The New York Times: “Are fungi capitalists? No. They’ve developed a system that is much more sophisticated than the economic system humans use.”

Silicon, you’ve been a good friend and ally for many decades, but perhaps it is time we revisited some old friends.

I’ll give another example of out-there technology the DARPA RFI made me think of: 3D printing of batteries, as Christopher Mims recently profiled in The Wall Street Journal. Batteries have long been a pain point for all of our devices, from watches to smartphones to laptops to electric cars, but some innovators don’t think they have to be. Mr. Mims writes:

There’s a revolution in battery technology hiding in plain sight: The 3-D printing of batteries has the potential to put energy storage inside any device. This will enable lightweight and long-lasting consumer gadgets, long-range military drones and even nanoscale robots.

 He goes on to posit:

The promise of battery-tech 3-D printing (aka additive manufacturing) is simple: What if batteries could fill any available space, even structural elements of our gadgets, rather than always taking a rigid shape like a pouch or cylinder?

What if, indeed. What we think batteries are and have to be, may not be. How many other components of our existing computing platforms is that true of?

Responses to the DARPA RFI are due July 17 (which doesn’t seem like very long), and DARPA will hold an in-person, invitation-only workshop in August. I don’t know what kind of responses DARPA will get or who will be attending the workshop, but if I were a venture capitalist looking for the Next Big Thing, I’d sure be paying attention.