The history of computing confusingly combines shrinking with growth.
As fast as computers increase in power, their size diminishes. I began programming professionally on a DEC System-10 that took up a good-sized room, before moving onto an even bigger IBM 370 mainframe. When the 370 was introduced, its basic form had 64K of memory – 125,000 times smaller than the phone in my pocket. The 370’s basic storage started at 70Mb on slow disk drives. My phone provides 256Gb of fast access solid state storage – over 3.6 million times more.
This positive example of shrinkflation is familiar, even though the scale can still shock. But there is another aspect to the change that has more opportunities to surprise. Those early computers I programmed were monolithic – central mainframes that begrudgingly shared out a portion of their time to various users. With the PC and now the phone, computing has become personal. But the growth of networking, especially wireless networking, also opens opportunities for a new paradigm where small units work together to create a greater whole.
It’s true that we have had instances of parallel processing for decades.
Even a simple laptop passes some of its computing work to specialist chips, for example to deal with graphics. And there have been large computers that distribute work amongst many small processors for decades – but something new is on the horizon. We have the potential to transform the way computing devices interact by learning from social insects such as ants, bees and termites.
Individuals of these species have very limited capabilities. But they come together to form a superorganism, where each insect acts as a component of a greater whole. Termites, for example, are capable of constructing mounds that can be 9 metres tall with complex inner structures. Over a year, in a single colony, they can move around a quarter of a tonne of soil in continued maintenance. They build in air conditioning features to avoid the nest overheating, and they can run farms inside, growing edible fungus.
We are already seeing some interaction between devices, for instance with artificial intelligence, where processing may be partly on a phone, but can also involve interaction with cloud servers.
Increasingly, we will see social hardware where different components with limited abilities – the equivalent of the insects in the superorganism – interact on a large scale to produce capabilities far greater than that of any individual device.
The best example so far is the drone swarms used in lighting displays to produce huge, dynamic illuminated images in the sky. In some cases, drone control is centralised, with a base computer issuing instructions to each drone. But increasingly drones rely on communication with each other to handle positioning in the sky. This approach is similar to a murmuration of starlings, where a huge flock of birds forms a complex shifting shape in the sky as they fly. There is no overall control: all is determined by very basic local interactions between birds.
Drone swarms are just the start, though.
Think, for example, of the science fiction cliché of the humanoid domestic robot, doing chores around the home. The one advantage of the humanoid form is that most objects in a house are designed to be handled by humans. But the human form is extremely difficult to mimic – just standing or walking present a major problem. Future domestic robots are likely to be developed as superorganisms of small devices with different form factors, working together to achieve complex tasks, whether it’s emptying the dishwasher or restocking the fridge (which, of course, would be a component of the superorganism too).
This new approach to automation will require a major overhaul of some aspects of programming. Just as each insect in the termite mound has a very limited ‘program’, so the components of the computing superorganism will not require huge electronic brainpower. But they will need to communicate and interact so that the whole becomes far more than the parts. For coders, this will mean approaching problems in a different way, but one that should enable more creativity in capability once the basic frameworks of interaction are in place, whether the hardware is located in the home or the office.
Forget the futures portrayed in science fiction. It’s time we took a lesson from nature.