We have so much and we squander it. Here is my diatribe on performance and efficiency
and why engineers (and businesses) should prioritize it.
People use computers to get work done. Today, the internet is virtually ubiquitous and
decently fast. So it doesn't feel painful when MS Teams is a several hundred megabyte
(or worse) download. Perhaps more commonly, users do notice it takes >10 seconds to start
and get into a meeting.
We as a society have normalized software whose resource use is wildly disproportionate to
the job it performs. We have collectively forgotten that we used to be able to download a
20 MB chat/video client which started in one second, and on computers that were literally
100x slower than smartphones of today.
My belief is that many people do not understand actually how powerful computers are today.
If they did, we would not collectively accept such crapware.
I'm pragmatic enough to admit that businesses exist to put products into peoples' hands
which are useful and fast enough to not cause pain. They're not there to ultra-optimize
software and tweak assembly code and produce a demoscene-esque 64KB piece of engineering
porn. However, there is such an enormous gulf between these two extremes. Performance is
not nostalgia, it's not an aesthetic, it's not "Real programmers use C", etc. It's a
discipline of understanding what computers are capable of and removing huge waste wherever
it is found.
Fast software delights people. It feels more fun to use. It feels like people gave a shit
when they made it. It feels like this tool, this product was made with care.
Time is something none of us get back: fast tools respect that. Similarly, when an amazing
program is 2MB, it immediately tells me the creator cared enough to think about this.
Slow tools communicate neglect and thoughtlessness.
I don't think this is malicious. Bloat and inefficiency is rarely one dramatic decision.
It is usually the sediment of a thousand locally reasonable choices that were never
measured together. No one sets out to make slow or bloated software. In my
experience it's either 1) the business doesn't care because their customers have no other
alternatives (and yes, there is some investment to make efficient software), or 2) they
have never seen how fast the computer in front of them really can be. They don't realize
how big a number 4 GhZ is, how much data can actually be stored in 16 GB of memory. These
are insanely huge numbers. My goal is to raise awareness about what computers are capable
of and help people learn more how to measure and improve their software.
Out hardware got exponentially faster, but our expectations did not rise to match it. I
encourage you to measure startup times, network usage, battery impact, dependency weights.
Do these costs look weird? Do these costs justify the value to your user? Will they
delight users?
Let's raise the bar on what people expect from software. Software that uses resources
commensurate with the task at hand, that feels thoughtfully made, that feels like it
respects my time.
