Will everything be faster when Moore's law is over?

Moore’s law is not really a law, it’s just the observation that the number of transistors in computers doubles every two years.

The power of computers has been increasing exponentially for a long time, but we know that due to physical constraints, there’s a limit to how much it can grow, at least in the traditional way of adding more transistors*.

Another law, the Parkinson’s law, is an adage that says: “work expands so as to fill the time available for its completion”. It turns out this law is applicable in many other situations too.

One of the ways it applies, is with computer power: “software expands so as to use all the hardware resources available to it”.

Think about it this way, you code an extremely inefficient piece of code, but it works. It takes about a second to run. Do you bother to fix it immediately? Or do you go and implement new features?

Maybe if you are on your own and you want to do things right you go and improve its efficiency, but if you have a manager to report to, and he is to choose between “spending time on what already works” versus “implement long-awaited feature x”, I think he will decide the latter.

Add this to the fact that we usually build on top of many layers of software, we use library after library, framework after framework, and we don’t have any power on the way they decided to solve the problem. Sometimes a custom solution would do the job, but we don’t always have the confidence, time, effort, and/or expertise required to build it.

Actually, there’s even a law for this one as well, because yes, I’m not the first one to notice it: Wirth’s law: “software is getting slower more rapidly than hardware is becoming faster”. Based on this it seems that we will never enjoy the benefits of more powerful computers.

But my view is that when we finally hit that limit, we are going to start to be much more mindful about the code we depend on, and the code we write, optimizing more and more what’s already written, and software will become much faster.

I have this view from looking back at old devices (or even at current embedded devices). Think about the Game Boy Advance. Maybe the games weren’t as today’s games, but we were able to have extremely nice performance even on much much much weaker hardware.

(*): Maybe we finally develop a reliable quantum computer and we have many more orders of magnitude to spare.