Not All Software is Legacy

There is a perverse and prevalent notion perpetuated ad nauseum through the software development industry that all software is legacy the moment it is written. This cynical, seemingly unscripted, and suspiciously unanimous consensus betrays a cultural shift so diametrically opposed to the programmer/hacker sub-culture of the 90s and early 2000s that I refuse to acknowledge it to be derivative.

First though - as befits the current social climate - a disclaimer; I must warn the weary reader that this is not a feel-good article. Reading on will not provide you the dopamine hit you have come to expect and demand from the media you consume. Go back to your Instagram Reels, your TikToks and YouTube Shorts, this is not for you. Nor will I be dumbing this down to appeal to the broadest possible audience with the vain hope of earning some measly ad revenue. Not to be found here is any form of encouragement and appeasement in the face of your eccentricities, your ignorances and your imposter syndromes. I’ve already lost one job prospect because of the contents of this blog, and so I now gleefully remove the proverbial gloves. You have been warned.

What Software Was

Like everyone before me I found myself in the middle of something new. This new thing, these thinking machines, were thought to have the potential to change the world. No one knew for sure though. I wrote my first line of code in 1996 using Borland Turbo C++, and later that year I made my first socket(AF_INET, SOCK_STREAM, IPPROTO_TCP) API call. I saw my first URL in the real world on a newspaper my father brought back from a business trip to Seoul around 1998. The HTML tutorials in books showcased checkboxes and radio buttons on - at the time - hilariously futuristic pizza delivery pages. We watched Captain Kirk use a comically small device to communicate with his space ship on Star Trek. People joked NetBSD could run on toasters. There were stories of Ken Thompson writing an Operating System, a shell, an editor and an assembler in a month.

The folks I hung out with on IRC really bought into this whole new computers and Internet thing. These people were curious. Not curious in the sense of having a singular fleeting non-trivial thought on a topic. Not curious in the way a Chess Master may look for 40 minutes at a novelty on move seven of the Sicilian Najdorf. These people wanted to learn deeply how things worked. Hours a day, for days, months and years. Not just technology, but the world around them - of which technology was a small part. It was these people with their wide range of knowledge that took lessons from - for example - Biology and applied them quite successfully to Software (“Object Oriented Programming” and the “perceptrons” for Machine Learning come to mind). They knew how locks and keys worked. They wondered how traffic lights worked. They took apart things, figured out how they worked, and put them back together. Social engineering became a thing. They had an intense abhorrence for authority and viewed the coming of The Information Age as a catalyst for bringing equality of opportunity to the world, and destroying centralized structures of authority and despotic governments. All of what we have from Software and the Internet is thanks to their inquisitive nature and hard work.

What Software Is

In 2014 I sat in a room full of solemn and serious engineers at arguably the most successful eCommerce company of all time. With a completely straight face they’re discussing a document that estimates the time it will take to change the color of a button on a website; 2 years of development effort. WhatsApp leverages FreeBSD’s great TCP/IP stack and Erlang to run 3 million concurrent connections on a single box. Later I see an I/O bound Java web application running 2,000+ threads on a quad-core machine. People calmly discuss how their code that stitches strings into a 60kB HTML page takes 400 milliseconds to run on a 2.8GHz processor with a 22 nanometer feature size. Someone posts the Apollo Guidance Computer source code to GitHub - it got humanity to the Moon and didn’t even have a call stack. For the uninitiated; these are wildly contrasting, alternating, interleaved worlds.

These are two parallel realities that exist in this industry. One is populated by a group of people who know what they’re doing. The other is filled by two generations worth of people who picked a career they thought would be lucrative and have no interest in understanding anything other than office politics and not-getting-fired. I have new hires that don’t know what OOP is for. I patiently listen to Principal Engineers talk about concurrency who have no fucking clue what epoll is. A Java-loving L8+ IC at AWS talks shit about Erlang and the Actor Model, and then proceeds to say the greatest model for concurrency is an isolated, stateless green-thread only allowed to communicate by means of receiving input state one at a time on a bounded queue such that it can be reasoned about as if it were single threaded. This is in fact, the exact description of an Actor Model ideated and published before this guy was even born - around 1973. But sure, let’s learn about concurrency from the Java runtime.

Legacy Code and Tech Debt

So no, not all code is legacy the moment its written. Tech Debt is not inevitable. It’s just that our developers have been taught the wrong things. Not enough people question writing shitty code. Too many people believe the myth of software estimation and are pushed into unrealistic deadlines. Too many new developers are mentored and managed by people who have no idea what they’re doing. Too many companies have become successful despite writing shitty code, so the evolutionary forces are against (temporarily) against the perpetuation of craft and skill.

Takeaway: Advice for New Developers

Dive deep. Take anything and pick it apart, understanding it to its very core. Then learn to explain it to someone - anyone. The more different kinds of people you can explain it to, the better you become at understanding it yourself, and the more valuable become as a problem solver to those around you and the society you live in. Reject the notion that no one knows what they’re talking about, and embrace building expertise through slow, deliberate and consistent effort.