What's New and What's True

Self-Indulgent Wastes of Time

Old-timers and experienced web developers often complain that each new framework is just hype, just recycled ideas from the past couple of decades. They add nothing, and there’s no reason to continue learning new frameworks except to speak the same language as their less-reflective colleagues.

This criticism isn’t entirely off the mark. We programmers love tweaking and rewriting things to fit our own computational aesthetics. Especially in a business environment, we often have to be reminded not to re-write the entire damn codebase. That code works, and refactoring definitely will take longer than you think.

As Rich Hickey notes, programmers know the benefits of everything and the costs of nothing. Just how large are the benefits of this new language, this new framework, this new library? Do they justify the costs of setting it up, of integrating with the surrounding ecosystem, of learning it?

A Sigh of Resignation

One response to this criticism can be a resigned eye-roll. Paul Graham notes that it’s very hard to tell who’s a good programmer, but one sign is that they’re finicky about their tools. They just don’t like to work with inferior languages or frameworks or editors, just as a good chef can work with knife but might insist on Wusthof knives.

Given that fact, we might accept this ceaseless experimentation as the price we pay for good engineers.

Nevertheless, it’s important to emphasize that in the larger picture, all this frenetic writing is not a waste of time.

The Good Old Days

Let’s remember that not so long ago the front-end and back-end folks sat in different rooms, as it were, leading to the traditional diagram “client – business logic – database” or Model-View-Controller paradigm.

And not too long ago the front-end was seen as “not real programming,” with all the heavy-duty stuff taking place in the back end. You needed designers and front-end specialissts, and then you needed sysadmins and you needed back-end programmers.

Today more and more of the logic lives in the front end - I’ve had backends that literally just stored the user’s information in Redis and then returned it on login. All the computations, logic, navigation, etc., took place in the browser. Stupid backend engineers, what do they know?

These days full-stack web developers are more the norm than the exception. Of course any larger company or well-established startup will have front- and back-end specialists, but the distinction isn’t nearly as sharp anymore.

Open Source

Part of this story, by the way, goes beyond the specific languages and frameworks and libraries that actually see a lot of usage, to individual projects and tutorials which individual developers create for fun and goodwill.

This massive community effort, much of it enabled by GitHub, allows newbies to learn and get up to speed through examples and instruction. The modern open source movement continue to create astounding wealth by opening the craft of programming to those without the luxury of learning on the job.

Sysadmin, and Emacs

Just as remarkably, with Docker and Dokku we are starting to see the full-stack developer extend to the operations side, to server management and system administration.

Before Docker and Dokku, services such as Heroku enabled solo developers by taking all of that off their hands. I take the increasing popularity of “DevOps culture” as another sign that the solo developer will run servers as well.

The same multiplying effect can be seen in tools like Emacs and Org-Mode. I can efficiently plan and track projects, administer servers, produce reproducible research, perform polyglot literate programming as in Jupyter notebook, organize research and writing, format mathematical writing, publish to a blog, and code in a personally-suited IDE in every language. Of course there are other goodies like listening to music, because Emacs, but the point is that without this tool I’d need two of me to do the same things.

The Bigger Picture

Due to the constant rewriting of languages and libraries, today one programmer can accomplish the work of an Army company of programmers. Famously, WhatsApp, Mailbox, and Instagram each had astoundingly few employees. Instagram exited with 13 employees all told, at a market valuation of $100 million per employee.

I’m by no means a demigod or even lesser hero of programming, but by myself, I can do what 20 programmers would have accomplished about the year 2000. This churn of technologies enables individual programmers, and thereby enables new businesses with their own societal innovations to get started.

As many other have noted, just starting a software-based business, which doesn’t even have physical production expenses, cost millions only a couple of decades ago. Then it cost hundreds of thousands. Now if you give me $10k, I and a co-founder can live on that for 4 months and start a company - and we can do that because the programming is easier.

Constantly rewriting software to your own specifications may look incredibly finicky and self-indulgent, and maybe sometimes it is. But it’s also the source of great progress in programming, and therefore also the world.