What is computation?

_See Pieter Hintjens Our decentralized future _

For our purposes let us refer to computation as the sum of hardware and software in our lives. We tend to think of these two concepts as discrete, and for many intents and purposes it is important to remember that hardware sits underneath computation, in that hardware permits computation, and of course there is a hardware and software revolution - but it’s also useful to think of the sum of both of these phenomena together briefly.

If this is a macro-trend what exactly is it doing for us? What are some of the possible pressures that are leading to this computational revolution? Basically when we use the term ‘computation’ what do we mean exactly?

I’ll break computation down into a few important pieces and make analogies to ordinary (pre-industrial) human behavior:

1. “Pure reasoning” is a classic image of computation. We imagine computers as big brains that think very hard on problems. A good example is a computer vision algorithm examining an image and performing image segmentation on the image. In this case there’s very little data going in or out, and the computation is simply evaluating or inspecting the image based on its own algorithms (say a deep learning algorithm or even just a simple edge detection algorithm). For human minds even doing pure reasoning is expensive calorically. To identify a certain kind of bird, say as part of an endangered wildlife field count, or to have a security guard watch a construction site all night long requires significant attention and focus. There is huge incentive to build software agents that can perform these kinds of roles. Another example would be computing the forces required to produce an optimal flight path for a SpaceX bellyflop.

2. “New sensing”. There is a class of computation coupled with new sensing capabilities, such as peering into the infrared or the ultraviolet spectrum. Another example is new odor sensing devices to identify smells such as gas leaks, or perhaps some day even the composition or even terrior of a wine. In this case electrochemical sensors react to volatile compounds ultimately producing a digital value.

3. “Sense making”. Putting information into different buckets is similar to feature extraction but at scale, organizationally, it becomes a human scale challenge. Sentiment analysis, contextual network graph traversal, topic clustering and many other services are capabilities we take for granted today when using Google Search. Soon we may also see tools like this to help us organize our own lives and our own data.

4. “Robotics”. Robotics is in some ways just a variation of the other themes - but with fine grained digital control of the real world we start to automate traditional human roles (for better or worse). What’s occurring here is a bit subtle. Rather than having to use “atoms” to make a stable airplane wing, one can make a much lighter but wildly impossible-for-a-human-to-control airplane wing and then use an even faster computational engine to solve for that wings desired trajectory in real time. Control is a way to reduce compensatory bulk, buffers, guardrails and other patterns we’ve used previously in manufacturing. So much of human value is embodied in the idea of labor, and so far this has been one of the last territories for machines to take over. Human landscapes are highly suited to humans and are in a sense protected and vulnerable spaces. Machines, such as self driving cars, have a higher barrier of performance, deftness, contextual awareness to overcome than in other domains.

5. “Artifact virtualization”. Many of our physical artifacts are in fact purely information objects that happen to be co-embodied with a physical transport. Books are a good example. Art is often arguably another example. We’re extremely used to this physicality and we struggle to understand the separation of the two realms. There’s been a recent discourse over NFT artifacts for this reason. Nicholas Negroponte coined the phrase “from atoms to bits” to describe this. Atoms have mass, require transportation. By comparison bits can be sent around the world instantly and take up less room to manage, store, manipulate in general.

6. “Speed”. The crowning achievement of the computational age is a kind of battle with time. As humans we live in a time constricted domain; we simply cannot think hard enough or fast enough in our heads to navigate some of the larger harder issues that we face. Our brains only react at 200 miles per hour. There is phenomena in our world that vastly exceeds our ability to respond or even perceive it before it kills us. It’s useful to identify a flower, or to organize our photos, or self-drive a car, or organize our record collection - but being able to respond to stock market fluctuations, or a pressure build up in a nuclear reactor is where utility is created.