December 23, 2024

Goodbye 2024 with a video & a podcast

Season’s greetings friend,

both from me and temporarily-my dog:

Yes, Barnaby’s back in my life to help me through the dark season and to stop y’all from unsubscribing! And yes, we both needed a haircut, but he can totally pull it off. This is the first time I wish I had bigger ears.

Chatting about Django

I’ve made it on another podcast – this time Django Chat despite having zero Django apps in production! And as it is already custom, they came up empty with how to categorize my opining, so the topic is “Python Tooling”, but we even talk about video production.

Compared to my appearance on Talk Python to Me earlier this year, I think it has a bigger slant towards software engineering. They even allowed me to go on a lengthy rant about design pressures, and if Cthulhu and/or PyCon program committee allow, that may become my 2025 conference talk topic. 🤞

Also, entirely thanks to me being able to retell the riveting story of changing everyone's password in our company due to Django ORM's lack of multi-column primary keys, the 19-year old ticket has been closed!

The reactions to the podcast were universally positive, so why not make it part of your next household chore?

New video!

LOL, remember how the young & naive October-2024-Hynek wrote:

I hope to be able to produce a simpler video by the end of October

Well, I’m sure December 9th was still technically the “end of October” in some airport lounge.

One of the reasons why it took so long, I even elaborate in this video’s epilogue: the second uv video was so wildly successful (over 32k views, as of writing! 🤯), I knew the next one would be a disappointment, and that’s not a great motivator. Little did I know how big of a disappointment: It’s officially my worst-performing video so far.

So, I’ll leave it to you, my newsletter friends, to fix this injustice and invite you to check out:

Master Flaky Systems With Retries in Python 🐍 (avoid self-DDoS)

It introduces the necessity for retries in distributed systems, the dangers, and, if using Python, how to use Tenacity to avoid those dangers. Or, just use my little stamina. 😇

---

Those following me on Mastodon or Bluesky may notice that both the title and the thumbnail have changed since the first release, and I’m afraid that’s something you just have to do, if you want people to watch your stuff on YouTube. #foreshadowing

Interestingly, people who actually watched it seem to like it! It has the best views-to-like ratio and seems to perform well in search, too. Looks like I’m the king of a molehill. To the point that a creator tool suggested that I create more content about that topic because the average watch time (or, as we cool YouTube kids call it, AVD) is off the charts. Thanks, TubeBuddy, I can sabotage myself alone just fine.

Y’all might’ve also noticed I’ve experimented with YouTube Shorts by using clips from the video to drive some more traffic. Not sure much that helped, but I’ve finally had some viewers from the 18–24 age group. 😅

Anyhow, given how timeless the video is, all I can hope for is some slow burn success.

Which takes us directly to…

Reflecting on my first year on YouTube

On December 23rd 2023, I’ve published my first video which was also a camouflaged ad for an open-source project of mine: svcs: Robust & Testable Python Web Applications with Any Framework!

Therefore, today concludes my yearly theme of creation. I’m trying not to be too self-indulgent, but it kinda is a good opportunity to look back.

If you don’t care, that's 100% fair. The important stuff in this newsletter was front-loaded, so thanks for reading and see you in 2025! But there may or may not be another Barnaby waiting for you at the end.

Manual labor

It’s kinda fun to watch my first video today, and realize how much further I’ve come in one year, and six videos.

But that’s mostly about my environment and my camera confidence. Because in some ways, that video had a higher production quality than my latter ones! Everything is hand-animated and took forever.

I’m increasingly spending my energy on streamlining my production pipeline because I can’t stress enough how manual, repetitive, and tedious video post-production is – even if you’re OK with spending money on plugins and templates.

The manual era peaked with Loose Coupling & Dependency Injection the Easy Way!, which is still my most successful non-uv video, the last video I’ve produced with DaVinci Resolve, and it also introduced the jolly tradition of microphone-punching. If you watch it, you’ll notice how I’m experimenting with getting text and source code on screen in a readable way (and not always succeeding). This will be my bane until the retry video where I call it a day and decide to use slides in a symmetrical 1:1 perspective.

Surprise Apple Script!

To make this smoother for me, I have written my first serious amount of Apple Script! I’ve spent a decade conference-speaking, so Keynote is second nature to me and I’m animating as little as possible in Final Cut itself, because, as I wrote, it’s very, very manual and annoying.

So now, I can press one key, and the current slide is exported as a series of images, copied into my current video project and a Finder window opens with the directory selected, so I can drag it into Final Cut. This used to be ~10 clicks. Unfortunately, I haven’t found a way to automatically import them into Final Cut because it doesn’t seem to have a proper Apple Script interface, so if someone has an idea how to streamline this further, please hit the reply button immediately.

As a curious side note, Apple Script is a great example for the limits of LLMs, because even the almighty Anthropic Claude 3.5 Sonnet was completely useless at helping me to write this script. In the end, I ended up old-school copy-pasting some code from Stack Overflow that I don’t fully understand and pester my dear friend Glyph who’s even more old-school than I am.

I suspect it comes down to two reasons:

1. There isn’t that much Apple Script in the wild for chatbot operators to steal train their LLMs on.

2. Apple Script is a terrible, terrible programming language that tries to be human by being highly irregular. It reads like normal sentences, but those sentences have to be in very specific shapes.

So, if Glyph's Theory of Idiosyncratic Tech Stacks holds, this is a great time to get into learning Apple Script.

Audio Quality

This is an embarrassing one. I started my YouTube career with my trusty old Yeti X, and my sound was perfectly fine from my first video right up to my runaway successful second uv video, where I switched to a Sanken COS-11D lavalier microphone (the kind that YouTubers nowadays hold in their hands). If you look around the web, this is pretty much the best lavalier microphone. The problem is that the environment I’m recording my videos in is anything but a studio: I sit in a small room, literally yelling into a bookshelf. What I wanted was more freedom of movement (c.f., microphone-punching) – what I got was reverb from hell.

Long story short, that was a lot of money spent on sounding like garbage (disclaimer: Łukasz, who’s been essential for me to start YouTubing and is my audio mentor, begged me not to spend more money on my audio equipment). I’m not exaggerating when I say that editing the uv video gave me a headache from all the reverb.

Thus, after uploading, and without telling Łukasz, I immediately spent even more money and got a Shure MV7+, which looks like the quintessential podcaster microphone (that would be the SM7B) but is slightly different. Notably, it’s cheaper and comes with USB-C. It’s a really good microphone for my poor conditions and one I’m starting to see all around the web – I just wish I could switch off the childish audio level LEDs on top of it.

---

At this point, I have to thank my GitHub Sponsors, because I would never be able to do such experiments without their backing. So, if you like what I’m doing and where I’m going, please consider joining them: https://github.com/sponsors/hynek/. Special thanks to Klaviyo, who are currently my only The Dedicated Organization sponsor.

Comments

Interestingly, absolutely nobody commented on the bad audio of the uv video. And I’ll have to say something really controversial here: I really enjoy the comments on my videos.

The uv ones come with an asterisk, because since they had such a wide viewership (relatively speaking) and talk about Python packaging, they also attracted people who’ve been burned by Python packaging in the past and came to my comments to air their grievances without even watching the video. Oh well, we all process trauma in our own ways.

But other than that, given my rather quirky style, the comments are mostly fun & friendly and it’s nice to connect to the audience – some of which never heard of me before. I’m curious if I’ll ever get to the point of loathing them, but this is still the parasocial honeymoon period.

I’m sure one factor is that shadow-banning is a first-class concept on YouTube. You may be wondering why you don’t see any egregious packaging dirty laundry on my channel and the reason is that it takes two clicks to silently remove commenters from a channel forever. I don’t love to use it, and ya’ll notice that I tried to talk to people who I thought might be receptive to reason and nuance, but some men you just cannot reach.

Viability

I also have to talk about the viability of my channel. Because as you might’ve read between the lines, the (lack of) resonance to my last video was a heavy gut punch. I didn’t expect anything close to the uv videos, but I also didn’t expect it to perform literally worst:

Funnily, the Internet Ad Complex can smell my fear and loathing, and all my video recommendations and ads are centered around coping with a failing channel now. I shouldn’t be surprised that there’s a whole cottage industry for every stage of having a YouTube channel: how to start, how to go viral, how/why to keep going. The good old shovel-in-a-gold-rush thing. Ironically, I suspect if I had made a video instead of this lengthy newsletter essay, it might’ve performed better than anything I’ve done before – but I didn’t need that kind of stress around Festivus.

But since I’m not doing it for the money, it kinda requires to define what a successful or failing channel means to me. And while I’ve been thinking about that since I started YouTubing, this regression brought the topic to the front, of course.

---

As I wrote in my earlier newsletters, my goal is to find a way to serve the community with my knowledge that wants to leave my head and my… let’s be charitable and call it personality… and the community paying me back with cold hard cash and – nobody is above it – some form of recognition.

I started on YouTube because I wanted to do something new, but also because my old creative outlets, conference speaking and writing, took heavy hits through COVID-19 and whatever Google and Elon are doing to blog traffic. So my hopes were that I could produce something creative again and reach new people who would never find me otherwise. And, if my comments are any indication, the latter part has worked out to a certain degree!

But of course, if I spend a month producing a video, I want more than 1000 people to watch it – otherwise, it’s just not a good use of my time and resources. On the other hand, it would be arrogant to assume I could figure out the demand for my kind of niche content after a year and six videos. After all, “they” say the first 50 are for you to learn to produce videos and to find your niche – which at my pace would take another short 9 years. :) Speaking of coping content, here’s a graph I’ve stumbled over in one of the relevant videos:

So, well, I guess there’s hope? While deflated, I’m also committed to the sunk-cost fallacy, and have to acknowledge one important thing: I’m floored by the amount of encouragement I have received in the past year. I’ve had more people walk up to me at conferences than when I was giving talks!

While the market for introductory videos is infinitely bigger, there is a genuine hunger for advanced content by people who know what they’re talking about. I know I had a hard time finding it, which was also a reason to start¹. I was profoundly disgusted by charlatans and opportunists raking in 10s of thousands of views with overproduced videos teaching how to import a Python module while clearly never having written a line of production code in their life. I won't be able to ever compete with them, but I'm trying to offer an alternative.

I genuinely enjoy creating content I would’ve loved to consume earlier in my career and I don’t feel like I’ve tried enough to find all the Early-Carreer-Hyneks, yet.

As a cherry on top, fellow non-native speakers told me they’re happy to see one of them on the screen, even if my weird English requires subtitles (which I have, thanks to MacWhisper)! It kinda is nice to be an inspiration for something good for a change.

Future

So, as someone who has learned to thrive on spite and defiance, I’m not giving up just yet. I’ve got a few topics lined up, but I will have to play more by YouTube’s rules than I planned until I can go full-on CGP Grey and do whatever the fuck I want. That means I’ve lied to you in my last newsletter: there’s 100% another uv video coming. It also means more shorts; it means video titles; it means thumbs I’m not 100% proud of (but never YouTube face, I swear), and it means I’ll be looking more into the current zeitgeist when picking my topics.

However, doing only popular stuff would burn me out, so I’ll be looking for some kind of balance where I pull in people with hot takes about Python packaging, so I can indoctrinate them with software engineering wisdom later (😵‍💫 coupling is bad, cohesion is good! 😵‍💫).

---

And I think that’s enough indulgent soul-searching for one email. To everybody who had the patience to read all the way to the end, I thank and salute you! 💛 Here's the promised Barnaby:

I have more things to share, but I’ll leave it for next time. For now, I wish you, as the Germans say, a great time between the years, and I look forward to talking to you in 2025.

—h