What is abundance? What is lifetime? Well, I use them according to their dictionary definitions, that is, abundance is the prevalence of something and lifetime is simply the interval between that thing’s birth and death. What makes them interesting for computer hardware is that they provide essential insights into the interplay between economics and engineering.

The figure below is a qualitative mapping between abundance and lifetime for some common things. I’ve left out numbers because I’m only trying to create an approximate ranking here. If you dig up the relevant data for these things, then you may very well poke gaping holes in my drawing. Nevertheless, I reckon the rough order that I’ve depicted below is solid enough for the discussion which follows.

abun_vs_life_1

The diagonal in the figure above is partly an artefact of the way I’ve drawn it. In reality you are unlikely to find such a straight line despite the negative correlation between abundance and lifetime. But there’s likely to be a power law relationship between the two axes on a log-log plot, even though I can’t demonstrate that in the absence of numbers. In any case, the key information in the figure above is the presence of an equilibrium that’s remarkably stable - things that are abundant have short lifetimes and things that endure are rather rare.

What keeps this equilibrium going is the cost of repair versus the cost of replacement. As I have noted on the diagram, enduring things are expensive to replace (bottom right-hand corner), while the reverse is true for abundant things (top left-hand corner). It’s also likely that the size of things is a key factor regarding the cost of repair versus the cost of replacement. Big things are rare and long lasting, so making them repairable is a practical necessity, while little things are ubiquitous and short lived, hence their replacement is easier and cheaper. This is a nice parallel with the biological world, where unicellular organisms would lie on the top left-hand corner of this diagram and large animals would lie on the bottom right-hand corner.

How does such an equilibrium emerge? My answer is that the design and manufacture of hardware begin on the bottom left-hand corner of the diagram - this is where the proverbial prototype resides. If a prototype becomes successful in terms of demand, then it begins to move towards the diagonal line before finding an appropriate slot. However, the pace of this movement towards equilibrium is not the same for all cases. For instance, a prototype will sprint towards abundance if it becomes popular, whereas it will walk towards endurance if it is proven reliable. That means, even though the equilibrium appears to be static in this picture, there are fascinating dynamic processes at work here.

One such dynamic process is the position of phones on this qualitative map. The lifetime of a typical phone is two or three years nowadays, since that’s how long people hold onto their phones before upgrading to a new one. If that lifetime elongates further, let’s say it doubles, then phones are guaranteed to become more repairable, similar to fridges and cars. That would of course have consequences for the way they are designed and manufactured in the future. For example, phones may not be hermetically sealed under such a hypothetical scenario, because they would need to be opened up for essential servicing by someone other than the original manufacturer. Also, the price of new phones will adjust as a matter of course, given that supply and demand will match up.

A concluding prediction I will make is that there won’t be anything in the top right-hand corner of the diagram. In other words, nothing can be abundant and enduring at the same time. This is a straightforward implication of the negative correlation between abundance and lifetime, and perhaps also a consequence of the increasing size as you traverse the diagonal downward from left to right. Could we ever see a new hardware design that’s concurrently small, abundant, and enduring? I don’t think so, given the fact that optimising simultaneously across these many dimensions makes our probability of success vanishingly small.