Robert D. French

Our Industry is Young

My parents are older than all software. We are at most three generations out from the advent of our entire craft. Human civilization is what, ten thousand years old? My friends, we need some goddamn perspective.

The personal automobile is roughly twice as old as software. Vehicles that continue to function after 25 years qualify as “antiques”. Their value rises. People get special license plates and emissions exemptions for them. Politicians pay to ride around in them during parades. We treat these vehicles with respect because they have stood the test of time. Why should we denigrate software that has done the same?

We are literally just getting started with this shit. We don't know yet what works well and what doesn't. Personally, I believe that type safety, static analysis, and unit tests make software more robust. But what about that old mess of Perl code that's been silently chugging away, handling your company's payroll for 20 years? If anyone wanted to mount an argument against the value of type safety, that would be a strong piece of evidence.

What does 'Legacy' even mean?

That 20-year-old Perl codebase is definitely 'legacy', right? What about C89? In the Year of Our Lord 2025, the 1989 form the C programming language is 34 years old. It's 14 years older than this hypothetical Perl program. Yet Linus Torvalds requires all new Linux code to be written in C89. Is C89 not 'legacy'? Wouldn't that mean Linux is also 'legacy'?

On close inspection, the word fails to have objective meaning. And that brings us to the real point: “legacy” has never about about age or maintainability, it's always been about fashion. Linux is still cool, so it gets a pass despite constant CVEs. Git is definitely cool, right? Even though it relies on a 30-year-old, broken hash algorithm? Why is CGI legacy, but “serverless” isn't? It's exactly the same execution model, but Amazon has made serverless seem new and cool and different.

Show Some Respect

That old Perl code may be ugly. It may be scary to change, hard to debug, and riddled with security bugs. But it deserves your respect.

Unless you wrote it, it has probably been working since before you even joined the company, and it is still adding value. Have you added the same amount of value? Which is more indispensible, you or that Perl program? How many of your co-workers have lasted as long as that software? Think for a moment about what would happen to the company if you quit, and compare that to what would happen if that Perl program suddenly evaporated.

We have a lot to learn about software still. Humanity is new at this. Any software that has continued to work reliably for more than a few years has, proportionally, survived a substantial chunk of the history of computing. That is something that should spark pride, awe, wonder, and admiration. That is a feat to be envied, and it is something to which we should all aspire.

Shout out to Caleb Wherry for helping me collect my thoughts on this one.