2020 is almost over! Thank all the supernatural beings in whatever pantheon you wish. I wish you all, readers, a 2021 that is orders of magnitude better than 2020.
Despite the crazy lockdowns outside, somehow life is even more hectic this “slow season” than normal — probably because of a rampaging toddler in the house. So while I had a great idea in my head about playing with eclipses for a newsletter, I didn’t have time to do the work for it, so onto the back burner it goes.
So please forgive me if I take the somewhat easy way out and do a data analyst take on the “year in review” type of post. This newsletter started in much more innocent times, Jan 24th, 2020, and pegging things to the calendar year seems much more convenient going forward.
Maybe there’s some interesting information that someone interested in getting into writing about data can find useful.
Slow start because what is marketing?
I’m something of an accidental blogger. I only started because Sean Taylor had tweeted asking about data science in small startups/organizations and I was inspired to slap together a Medium article (reposted on substack w/o paywall) about the topic in 2019. It surprisingly went viral and I started writing semi-regularly, about one or two pieces a month when the mood struck me.
Substack was the point where I put myself on a rigid “publish something every week” schedule. Occasionally I’ve accidentally published articles a day early because I seemingly can’t read a calendar properly, but I’ve thus far never missed a post. This is primarily because I know myself and if I let myself slack just once, I might never return.
As Medium slowly descended into its own weird brand of semi-decay, I followed a bunch of other bloggers and moved things to Substack.
So, I technically did start this Substack with some semblance of an audience, not a complete blank slate. At the least, some nice people on twitter also signed up early on. But on average, because Medium feels quite bad at maintaining continuity of readership, I’m not sure how much carryover came through.
The biggest takeaway from the growth chart above is things grew at a very slow and steady pace. Every so often, completely unpredictably, I’d get a spike of activity — either a shoutout from someone more well known, or having a thread or article go viral. I didn’t have an actual goal in mind for growth, so I largely just ignored the grind, but looking back at it in hindsight seems interesting.
Publishing date doesn’t seem to be related to anything
With very minor exceptions, I schedule my posts to publish on Tuesday morning, 8:05AM, Easter Time. The only thought process to go into it was “8AM is good enough for a morning mailbox appearance” and “Tuesdays aren’t usually holidays”.
In the end, WHEN something is published doesn’t seem to have a strong effect on things. If anything, Wednesdays, the day after a post are surprisingly low in terms of signups for some reason. Maybe if I controlled for viral explosions a better pattern might emerge, but I’m skeptical.
Hopefully this information is a bit of a relief for anyone who wants to publish stuff to the internets. It seems that content will find its way around over time and micro-optimizations are probably not worth it. Just throw stuff out there, it’ll sort itself out. There’s really no predicting what will or will not spread, so steadily grinding away at publishing new content.
That said, macro-optimizations in timing can be useful. For example, riding on a current events wave is a relatively cheap way to “be relevant”. At the same time, it is exhausting trying to keep any pace with the modern news and meme cycle. If you have the energy for it, more power to you.
You can see snowballing effects
Network effects are super important when your entire growth strategy is effectively word of mouth. So you can see in the signup growth chart that the slope of the curve has been slowly increasing over time. Network effects are starting to kick in and the rich getting richer effect is apparently.
It makes a fair amount of sense why this would happen. If you assume that everyone has an equally low likelihood to share an article with someone (and thus potentially get a new signup), then the more people with access to the initial post will proportionately increase the growth rate.
Being laser focused is tiring, so… don’t
Ostensibly this newsletter is about data and tech related topics. Ostensibly. As you all know, I wander pretty far afield at times, though I usually find sneaky ways to tie seemingly unrelated things together.
This was a deliberate choice on my part because I don’t have the ability to constantly write about the same topic, or even the same field or family of fields, for any sustained amount of time. Writing about only data collection, cleaning and database management for 52 weeks straight would…. drive me even more nuts than I am now. Some people are somehow able to sustain that focus, but I certainly can’t.
Instead, I’ve decided to write about things that my past self would’ve been interested in reading. Primarily it winds up being data things because I want to maintain a bit of a theme thing. But I have given myself permission to ramble on about side projects and myriad hobbies. This way, I can keep writing from week to week, even if my brain is completely obsessed with some new fancy thing on the side.
I think it’s too tiring to constantly maintain a projected facade of being “the expert on this one tiny thing” for a sustained weekly blog. At some point, being a well-rounded human with other thins needs to show through just as a break.
Notes are important!
I’m horrible at keeping notes, leaving things scattered around my desk or lurking in my brain. But the more I’ve needed to write at a regular cadence, the more I had to keep track of the little rabbit holes I’d fall into during the week.
Right now I’ve got a stack of links about eclipses, a couple of links to interesting tweets I might want to write about, and a pile of open tabs on various topics. One of those might be good enough to work with. I hope.
I’m a long ways off from other people I know who write for a living and thus keep entire journals of ideas and things, but it’s the best I’ve got.
Research rabbit holes have no end…
Fairly often, I start writing about a topic and wind up doing background research about it. This is especially true if I’m explaining/teaching about stuff. It’s really true how much you have to learn in order to teach something. So writing is definitely a good way to become an expert.
Since I very often take a view that tech decisions are embedded in a historical context, it winds up trying to hunt down original RFCs and sources which are often multi-layered and peculiar, assuming they’re not paywalled or locked away in a dead tree book somewhere…
But due to (or in many cases, thanks to) deadlines, at some point you just have to draw a cut line and stop trying to trace the interweaving threads of history about a topic, like Time Zones… This luckily puts a hard stop on how far I can possibly spend running down a weird garden path. But sometimes, it means I don’t get the whole picture, and it annoys the heck out of me when that happens.
Titles, Intros and Outros are oddly tricky
Even without trying very systematically, it appears to me that having a good catchy title seems to have an effect on initial interest levels. The problem is writing good ones. It always feels like it takes more time than it should.
But beyond the obvious title, I find I have the most trouble writing the introduction and concluding sections of every single post. Even if I save them to do last after everything else has been written, they’re STILL difficult. There’s something about doing a cold start and leading into a long article, and that same long article off at a stop point and saying “and we’re done here” that I find very hard to pull off naturally.
If anyone has good tips for that, let me know because then I can lower this burden upon my cluttered brain.
Interaction is hard =\
The one thing I’d like to get more of, but is apparently very hard to do, is find a way to get more discussion and conversation going on. This is primarily for my benefit, in that it’s more fun to talk WITH someone than to talk AT someone.
The problem is of course that newsletter by their very nature are still very much a push medium. Sure there’s a like button, a comment functionality that sends pings and data back to the central server, email replies even go back to the author. But it’s quite bolted onto the whole archaic email infrastructure.
Twitter is of course much better suited for back-and-forth style conversations where others can chime in, but I haven’t had much luck bridging the two media. The paradoxical thing about the situation is that I probably wouldn’t even be writing, or have any readers if it weren’t for having lots of fun interactions online with other people.
I guess if I really want to make something happen I’d have introduce a forcing function of some kind, like an event.