Opinion The IT industry is not only full of sharks, it has shark nature itself. It must keep moving forward to survive. Not all sharks are obligate ram ventilators, and not all IT changes all the time, but without innovation the sector would curdle and die.
Venture capitalists and cash-rich tech companies feel this most keenly, which is why they swarm like makos to the bleeding edge. But innovation itself is not enough. Blockchain is very clever, but has made the world a nastier place while losing a lot of people a lot of money. AI’s trajectory from wait-and-see could just as easily go that way as any other. Actual change, innovation that sticks around and becomes part of a better way forward, doesn’t have to and really should not make its inventors billionaires. It is thus often invisible when leadership is measured in dollars. Nonetheless, it matters to billions.
A much more fruitful filter for the future can be stolen from nature. Evolution by selection and descent, the theory of everything for biology, is at once a living demonstration of successful innovation and a morbid library of its failures. It is both tautologically simple — survival of the fittest — and breathtakingly complex in the interplay of environmental factors, selection pressures, and the interplay of change and stasis. It’s not random mutation until something clicks, it’s if and how organisms can change to match a changing environment. In IT, it’s what makes best use of a change in the landscape, which itself is changing as new tech is taken up.
To analyze this, selection pressures are particularly useful. Take a look at what’s going on in and around the peculiar speciation in and around the Linux kernel. There are clearly pressures here from the changing environment outside it. New and updated CPU architectures, evolving models of security, performance expectation in different use cases as sectors rise and fall in importance. Energy efficiency in many different ways. All are pushes for change across the kernel.
Other factors are more subtle and with very different dynamics. The commercial and personal motivations to be involved, the talent pool available to power change vectors, the power exchange between closed and open, the accuracy of perception and the anatomy of persuasion. These are as much social as technical.
Apply this model of selection pressures to our overview of three microkernels in the Linux context, and you can see that viable mutations are generally Rustafarian – personal motivation and skill sets are changing in that direction. Rust has itself evolved in response to those selection pressures, so selection pressure will favor those entities that can evolve to embrace it more easily. Mainstream Linux, with its very high institutional inertia about maintainers, leaves opportunities to the smaller, more nimble. All of the microkernels here have different focuses, different priorities, and the art of predicting their future success is in matching each focus to the imperatives of reality.
What will really make a difference is when an architectural concept offers a path to exploit an environmental aspect that’s unavailable to, or even toxic for, existing creatures. The evolution of photosynthesis produced unprecedented energy for cyanobacteria, while producing atmospheric oxygen that was poison for many other species. The evolution of metabolisms to use that oxygen completed the planetary reset event, and there was all those tasty cyanobacteria to eat.
The equivalent in IT is fixing security, which is currently so toxic that in the UK, there is talk of Covid-level responses to keep entire supply chains alive following a single incident. Information theft, system infiltration, and ransomware are increasingly seen as unavoidable, with corporate fatalism tending towards survivalism. That’s not tenable in an increasingly antagonistic world. Microkernels may not seem to have much to do with ransomware, but take a step back and ask at what point mutation could lead to an inherently resilient data processing environment.
Nature uses diversity and redundancy to enhance resiliency, at all levels from ecosystem to immune system. Imagine those as primary design features of the stack – or stacks. In avionics and other safety critical systems, diversity and redundancy means parallel independent systems performing the same tasks in different ways while cross-checking each other. Building a microkernel that supported the low-level system aspects of that while also doing the primary tasks of functional IT sounds like a very worthwhile experiment.
Doing another that runs on different hardware would also be important. These can start as only open source can, as step modifications of existing code and ideas, building tiny proof of concepts into running systems and encouraging others to start on their own paths.
Nature has created and tested billions of years of evolution, and what magnificence has appeared. But we have goals and a sense of purpose, entirely lacking in natural selection. Artificial selection can achieve in years what may never happen naturally. We can use the same tools as nature but to build towards what we need.
We absolutely do not need $300 billion fantasy AI megaclouds in pursuit of making the world’s first half-trillionaire. We absolutely do not need an IT infrastructure so shaky that a handful of god knows who can wreck sectors at will.
Understanding the big picture behind the little evolutions in open source can get us where we need to be. Evolved to live on dry land, away from the sharks. All of them. ®