On Starting

Or: “Fuck It!” and the duality of applying this mentality to only certain facets of life.

When: July 7, 2023

Where: Saysutshun (Newcastle Island) Marine Park, Nanaimo, BC, Canada

Well would you look at that: a blog post. However, if you’ll look at the source code for this site, at least at the time of writing, you’ll find no blog-specific tooling here to help me. In fact, you’ll find only a fairly small wrapper script (helpfully sourced from elsewhere) around lowdown(1) and a Makefile to help shrink images and so forth. Otherwise, there’s nothing to help tag these articles into categories, there’s nothing to auto-generate an Atom feed, there’s nothing to deploy this automatically. And yet, here I write.

This is long overdue: I’ve been talking about wanting to take part in things like #100DaysToOffload for months, and about wanting to “get a blog going” for literally years. Actually, we might be coming up to closer to a decade of me half-assedly murmuring something to the effect of “I’ll get a blog going about <thing>” with about the same intentionality as someone telling their partner they’ll “totally” get out of their warm and cozy bed in the morning to help with whatever the chore du jour is, while already drifting back off to a peaceful slumber. My “peaceful” “slumber”, I suppose, was probably microblogging, or other dopamine-driven social media over the years. Case in point, one of my many TODOs for this site is to backport a ton of longer-form Facebook posts to here and intersperse the images. I probably have a few dozen such pages to get to.

I’m not going to blame this all on like-button-fueled dopamine rushes, though, because I don’t think that’s actually the entire situation. I’ve been slowly pulling away from traditional1 social media anyway for a while. I think there’s a combination of UX and psychological factors at play in this as well, and that brings us to today’s real topic: self-induced startup costs and the empty canvas effect.

I find that hobbyist software folks and artists2 I’ve met often fall into one of two categories:

I, alas, fall into the latter camp for hobby work. It’s been a lifelong curse: outside of work and volunteer spaces, overcoming startup inertia is hard. One good look at my GitHub repositories list can clue one into this: this list largely resembles a graveyard of ideas past that typically never got off the ground for one reason or another. Maybe it was trying to use a new language for the sake of learning it and instead getting neither a (significantly) deeper understanding of Haskell or a music library manager out of it. Maybe it was getting partway through the initial prototype and then suddenly changing everything, including the implementation language, before the MVP had ever published a public artifact 3. Or maybe it was one of the countless other half-baked ideas that never made it to a text editor at all: a repo made, high-level ideas bounced around, and then… what? And then what?

There’s a well-known saying in English (and a few other languages): “perfect is the enemy of good”. It’s a saying I like to say to other people all the time, like when encouraging someone else to get their own cool thing off the ground, or to get a project underway at work. “Yeah man, whatever, we can refactor it down later, let’s get some kludgey shit thrown together for now to get something going!” I’ve learned, however, that, as with many adages I offer to folk, I’m actually pretty awful at following the advice myself4, and I’m especially bad at following it when left to my own devices. I’ve been working to change this, and as part of working to change it, I’ve had to give some thought to why I am this way. My theory on being such a perfectionist in my personal projects is somewhat twofold:

First, anxiety of publishing things in the public realm. “Nononono dude, I don’t want to have horrible code up on my GitHub, I’d rather have no code on my GitHub, at least then employers will assume I don’t write code at all in my spare time, rather than writing shitty code”. I’ve found this side of the reasoning has died down significantly in more recent years as a function of age and career growth: I no longer (as badly) feel that I have something to prove to the industry. I’m here, I’ve been doing this thing for a good while now, and I’m decent at it. Cool. If I have some shitty code on my GitHub or my SourceHut or whatever, who cares?! It lives alongside other actually decent code I’ve published, either myself, or on behalf of companies I’ve worked for.

Second, a sense of agency - agency I longed for in day-to-day work. In industry, typically one will be using whatever the flavor of the week toolchain and language is, or whatever the team agreed on a gazillion years before you showed up, or whatever some manager thinks they can easily hire for, and depending on your creative interests as a developer, this may or may not fulfill you. I know for me, it often did not: reading about languages like Haskell and Rust pushed my brain to think in ways I previously hadn’t, and then going back to whatever my dayjob of the time was to write, say, RESTful CRUD endpoints in Python, was a touch depressing. So in my free time, where I got to write all the rules about what got used and why, I lunged at the opportunity to try all the things, all the time, and I wanted to see how to use those things to write software better. This now meant I was, and little did I realize the gravity of this until I started doing all this reflection, fighting a war on two fronts of sorts: not only did I want to ship a “product” (or a toy, or a whatever), but I also needed to climb this giant mountain of learning a new tool… and learning that new tool well enough to do things “right”… whatever that meant.

Like many5 folks in software, I have a complex and complicated life outside of software and its development. Some might say I wear lots of hats or have gotten to live lots of chapters in a very short time. A saying that was for a long time (and in some circles still is) associated with me is “strong fuck-it reflex”, referring to the long list of things I’ll say “oh fuck it, why not” to. This list includes, but is not limited to, some major items:

All of these share a common trait: there comes a day when you’ve just gotta decide. Launch or no launch? Stay where it’s safe, or jump into the unknown? Consistently in my meatspace life, I find that I opt for “jump into the unknown” - even on less life-upending things like deciding to go on dates with someone, I find that, given a sliver of interest to begin with, my answer is generally likely to be “yes” until proven it should be “no”. And so I jump in head-first. And regardless of how any of these major decisions (or the more minor ones!) have turned out in the long run, I consider none of them failures in the grand scheme. Maybe the objectives I set out to achieve by jumping in (if there were any beyond “fuck it” to begin with, which isn’t always guaranteed) weren’t met, but I can consistently say that I’ve learned something from having this reflex (perhaps more politely called a “just say yes” reflex, which I vaguely recall is a concept some business book teaches to folks to whom this doesn’t come naturally).

Great - so… if this is how I am away from the screens, why not be this way at the screen, too? Why can’t I be that first type of hobbyist dev I alluded to, rather than the second?

Writing is one of these aforementioned cursed “blank canvas problems”. Every blog post, every book, every newspaper column, every technical manual, every copypasta shitpost, they all start with something vaguely approximating a blank white page onto which a series of characters can eventually be spewn. This is, to put it gently, an intimidating process. There’s ways to make it less painful, like doing an upfront outline before writing any of the prose7, but even that outline has to start somewhere!

Getting the infrastructure for a blog or personal website set up is another “blank canvas problem”. You have to somehow traverse the void between an empty directory on a filesystem and having some documents written, and between an empty body in an HTML document and actual readable pixels fit for human consumption8.

Solving both of these problems at the same time is the herculean task I’ve been up against for years, with the added self-own of explicitly not wanting this site to look like a templated blog site. Because, you know, if you’re already solving two hard problems, why not make it a billion times harder on yourself and willfully exclude all the wonderful options that already exist out there with the goal of trying to bridge those two voids for you? Genius.

Thus, for quite a long time, I took the path that solved at least one of these problems in the worst way: “hey look! an input box just sitting there waiting for me to type in, thanks Facebook or Mastodon or whatever! I can put off this blog thing yet another day”. And procrastinate I certainly did, much like how I’ll also procrastinate cooking if there’s anything prepared laying around, or will procrastinate doing the dishes if there’s even a single fork anywhere that’s still clean. To some degree, I think this is human nature: we’ve evolved to take paths of less resistance and less risk and less energy expenditure, often for good reason. But I’ll still call it for what it is: laziness combined with a fatal hobby-perfectionist streak.

Finally, several months ago, I did something very important: I decided to sit perfectionism aside. Who cares if the CSS isn’t rendering quite right on all devices? Who cares if the Markdown parser has opinions slightly different than mine? Who cares if I don’t have a fancy build script and deployment setup to get these HTML files from my disk to the Internet? Who cares if everyone looks at the code that powers this stuff and laughs their asses off? Just Do It.9 It’s taken a while, but this place finally has a few pages worth actually reading, and the startup cost of making new ones is ever-decreasing.

I’m missing some admittedly really nice to have features I’d love to some day backfill: there’s no way to link to other pages in a way that isn’t liable to break if a page gets renamed. There’s no automated RSS/Atom feed of site updates (short of following the Git repo, I guess). The images of the site are all stored bare in the Git repo rather than being in LFS or similar, where they belong. The deploy process is literally a manual rsync(1) incantation I keep in my shell history - it’s not even in a shell script or Makefile yet.

And yet… it’s mine. It’s unique. I achieved that “it’s not off-the-shelf” goal, and I have something tangible on the internet. Unlike, say, aforementioned gawsh, any rewrites or refactors I do now, are in the context of already having a working thing to enjoy in the meantime. The urgency of the refactor is entirely my own decision now, and it doesn’t impact having the damn thing shipped already.

Months have gone by with the site existing, but I had still delayed writing any long-form prose, in part because of how relentlessly busy I’ve been, but in part because I also didn’t know what I wanted to write about to inaugurate the blog section. Here we go again with blank canvas problems, hey? Sure, there’s a navbar now, but the articles section is empty - what to fill it with?! So tonight I decided to apply the same lesson I’ve been trying to pound back into my head in hobby software development over the past year or so. I decided to sit down and write this article in a single take, top to bottom. I’ve made only minor edits to the opening paragraph, and a proofreading pass over the whole document, and I’m about to commit this Markdown file to the Git repo, run the Makefile, and push this thing - whether it’s a work of art, a total shitshow I’ll wince at in a year, or (most likely) something in between - up to the public web.

That lesson? Fuck it. I’m starting.

  1. Can we call any of this mess “traditional”? My sense of time is completely screwed I guess: I’m somehow from that bizarre cusp set of years where I understand why pencils and cassette tapes go together (and remember having to reverse VCR movies before returning them to the honest-to-god-it-existed-kids-I-promise Blockbuster), have watched the entire rise and fall of Google+ and Twitter (too soon?), and have been terminally online for longer than many of the robotics students I volunteer with have been alive. Facebook feels “traditional” in that sense: I recall using netbooks at the Boy Scout Jamboree when I was like 13 years old to access Facebook and its chat (which was still the era when it was XMPP-compatible, RIP to those days). That was - literally - half my life ago. “Traditional” seems to fit. ↩

  2. I’m intentionally scoping this to hobbyist works because, with some exceptions, most hobby work is done without a financial incentive or imperative, and further, without external managers imparting milestones and deadlines upon us that help shape the early days of a project. At work, I’m generally well-known for being the kind of guy who can get a greenfield project off the ground quickly, but this discounts both the sense of urgency that comes from knowing I have to build this thing to stay housed and fed, and the upfront, often team-scale planning that goes into brainstorming up such tools - a preemptive brainstorming that is often lacking from my hobby work for one reason or another. Diving into this could easily become its own article, so I’ll stop here. ↩

  3. It actually gets worse from here: after moving from the Rust monolith to the Makefile+Zig architecture that currently exists in that otheridea branch (and I do hold that this migration was the technically-better one: the perf numbers speak for themselves), I finally got frustrated with Zig enough to go down what has now been a two year rabbit hole of programming language design in trying to find the ideal balance of strong typing, correctness checking, speed of implementation, usable syntax, and understandable internals, ultimately leading to what is now slowly shaping up as Gale. In many ways I don’t regret it, but in other ways, I have to laugh that I somehow have taken two years to build something like stagit by way of building a whole programming language with which to eventually built that Git tree renderer along the way. ↩

  4. Relatedly, we’ll all (it’s all, right? it’s not just me…) tell our friends what red flags to look out for when job hunting or dating, and yet will totally wind up taking the wrong job or dating the wrong person in spite of that advice we dole out. ↩

  5. But unfortunately not all… ↩

  6. I did eventually go visit the place 6 days before I loaded my car up and drove north-northwest from Dayton, at least. And I did love my years there: this was a decision that paid off, big time. ↩

  7. Guess which article had no such outline earlier this evening? ↩

  8. Or at least fit for ChatGPT to gobble up and regurgitate into some nonsense that someone will wind up citing in a paper with credit to the AI rather than to the original author. ↩

  9. Obligatory Shia LaBeouf ↩