How this started
As much as I enjoyed hosting the first 90 Ship It episodes, from May 2021 to February 2023, I missed having time to experiment. More than that, I wanted to invest in a few small bets and see what happens. One thing was clear: I had to take a break from the weekly podcasting grind.
For most of 2023, I spent time experimenting with Talos Linux, the Kubernetes OS. I started from scratch, with an 11-year old SuperMicro motherboard, and a newbie attitude. This was the starting point for rediscovering the joy of playing with tech consistently, over many months, just for the fun of it:
By the end of 2023, all my infrastructure was running on Talos, including all the production workloads that I have been hosting since 2003. That's right: 20-year old PHP & MySQL websites that went through many changes, and finally made it onto my production Kubernetes setup!
My Talos-powered fleet of clusters expanded to include a Raspberry Pi 5, a LattePanda Sigma, a custom built fanless AMD Ryzen 7, as well as bare metal hosts in Paris & Amsterdam (a.k.a. production). A story for another time:
The Joy of Making it Work
In December 2023, I started recording my first content after Ship It. The focus was on screen sharing with a hint of pair-programming. Having honed this skill over many years at Pivotal, it felt second nature. While I like talking, I enjoy doing more. It feels more concrete, more useful & I personally learn more from experiencing the practical side of ideas.
For me, the best part is concrete examples of dealing with unexpected situations. Stuff always fails, especially during demos! This cannot be planned or scripted. Capturing those authentic moments, the magic that comes from figuring it out & then making it work is what I enjoy doing the most. That's what I decided to start investing in 2024. I capture it best in this 2 mins segment from Changelog & Friends #26
The Hardest Part
The most challenging part of the new Make it Work content space was figuring out the editing process. Finding the tools that work for me was particularly difficult.
While I started out with Descript & Squadcast, I kept hitting various issues:
- Audio micro-corruptions (here are a few examples which I couldn't cut out - many more which I did);
- UI lag when editing 4k video made that workflow impossible;
- Missing screen shares after recording 9 in a single session; cloud backups were low quality (a low resolution screen share was the end-result);
- Audio recording with no sound after input device changed (luckily this happened in a trial run).
Even though I reported a few of these issues and engaged with support, we ended up working around problems. I didn't like this. I knew that I had to do something different.
After several weeks of research and trying out various other editing apps, I settled on DaVinci Resolve. The more I use it, the faster & smoother my editing gets. Also:
- Audio mastering (a.k.a. Fairlight) makes even the most challenging recordings sound podcast quality (especially with DaVinci Resolve Studio);
- Special effects (a.k.a. Fusion), in particular Magic Mask, work best with in an NVIDIA GPU;
- Colour grading is so comprehensive that I can make the videos look like Hollywood movies (with enough time & practice);
- The export process makes the final audio & video be exactly as I imagine it: pro grade.
Six weeks later, I got the hang of DaVinci Resolve Studio. I am editing on a Linux desktop running Pop!_OS 22.04, and an NVIDIA RTX 4080. After decades of trusting all my production systems with Linux, I can finally trust my dev too! 2024 looks like the year of the Linux desktop for me 👍 It seems that I am not the only one.
I also tried Canva for both video & podcast thumbnails. I can honestly say that the Pro subscription is worth every cent! The UX is really good, and creating thumbnails is finally enjoyable.
I cannot emphasise enough the importance of good thumbnails. They will determine if someone will click on that video or not. Let's try it! Which thumbnail are you more likely to click on?
Thank you Robert for your comprehensive feedback! It had a meaningful contribution to the time & attention that I give to thumbnails 🤓
Why not outsource editing?
Even though I engaged with several editors, none worked out. After a few weeks of messages on Upwork, I finally found someone that was ready & willing to take on a trial edit: 20 minutes of raw audio & video that needed to be edited into a 17-18 minutes long YouTube video.
Out of these two final videos, which one would you say was professionally edited?
- 🎬 Works on my Computer - Docker @ KubeCon EU 2024
- 🎬 GitOps & ClickOps beyond Kubernetes - Akuity @ KubeCon EU 2024
The learnings from this experiment were:
- It didn't save me time. I had to explain, review & give feedback, then do the final review;
- It also cost money on top of roughly the same amount of time that it would have taken me to edit;
- The end-result was not that much different from what I was already doing (unbiased user feedback);
And this was the most straightforward edit! The other ones start as 60+ minutes highly technical conversations that need editing into 18-25 minutes videos. Sharing which parts should be removed simply takes too long to be practical. Maybe if I did the first edit, and left the finer edit to a professional video editor, this might just work. For this round of edits, I ran out of time, patience & money to try this out. Maybe I'll try this approach out in the next three months.
After all, the focus is on shipping great content. Outsourcing video editing is a means to an end. While I expect to get there eventually, 👋 Ali Abdaal, it doesn't seem to be straightforward for my current context.
The Best Part
After the first three months of Make it Work, the best part was being able to record, edit & publish a talk that did not make it through the CFP stage for KubeCon EU 2024: 🎬 The Square Hole
Since I had my recording & editing process mostly figured out, and since I was comfortable doing everything from start to finish, this went smoothly.
By the way, can you tell from the above thumbnail whether I was using Canva at that point?
I am really glad that I went ahead with The Square Hole as an online exclusive! I am excited about the idea of following up with The Pentagon, The Hexagon, etc. Producing content like this every KubeCon makes sense to me and, who knows, maybe one day I will give one of these talks live at KubeCon.
The second best thing that I was able to do is record a handful of conversations at KubeCon EU 2024 and share them both as 🎧 audio (special episode), and also 🎬 multiple videos
Recording in the showcase booths, with 360˚ video, allowed me to capture the atmosphere as it happened. Even though at the time I did not have the video editing app that would allow me to combine the good audio with the 360˚ video, I am happy with the outcome. Now that I have DaVinci Resolve in my tool belt, I am confident that I can improve on this further 💪
For what it's worth, if I had more time, I would have edited the other 60% of the videos that I recorded:
gerhard 🌐 h22 /Videos/KCEU-2024.03
❯ du -kh
126G ./unedited
75G ./edited
201G .The Unexpected Part
It was surprising & entirely unexpected for the last published video - 🎬 Let's build a CDN - to be viewed & liked more than all of the other videos before it combined! And when you think that we didn't make it work, it makes me wonder whether I should pivot to This Doesn't Work (from Make it Work) 😆
Even more unexpected is that I still haven't edited & published some of the conversations that I recorded in January 2024! That makes my current turnaround 16 weeks from recording to publishing. Getting this reduced by half is at the top of my list of things to improve in the next three months.
By the time I have the entire process dialled in, the record-to-publish timeframe SLO that I am aiming for is 2-3 weeks. This is what the record-to-publish timeframe looks like for the first 5 episodes:
E1 - 8 weeks 🗓️🗓️🗓️🗓️🗓️🗓️🗓️🗓️
E2 - 7 weeks 🗓️🗓️🗓️🗓️🗓️🗓️🗓️
E3 - 9 weeks 🗓️🗓️🗓️🗓️🗓️🗓️🗓️🗓️🗓️
E4 - 5 weeks 🗓️🗓️🗓️🗓️🗓️
E5 - 16 weeks 🗓️🗓️🗓️🗓️🗓️🗓️🗓️🗓️🗓️🗓️🗓️🗓️🗓️🗓️🗓️🗓️The Stats
- 12 YouTube videos published
- 9 conversations edited & published
- 4 conversations waiting to be edited & published
- 5 podcast episodes published
- 1 special episode 🎧 KubeCon EU 2024
- 200 stickers shared at KubeCon EU 2024
- 1 newsletter sent - this one!
- First direct feedback received, and 90% implemented 👋 Robert
Key Takeaways
- Discovering the editing setup that works will take longer than expected.
- Write the process down. It's long, and consistency combined with an attention to detail make all the difference.
- Good thumbnails are essential. Take the time, iterate through at least 3 versions, otherwise the rest of the effort will have been in vain.
- Do not publish an episode on Transistor.fm until after the final thumbnail is in place, otherwise the likes of Overcast will remain stuck on the old thumbnail.
- It's impossible to know which content will be popular. Do your best every single time.
- Editing the first pass of highly technical content is not worth delegating.
- Being self-sufficient brings special joy & highest personal satisfaction.
- Make every episode better than the previous one. Improvements compound.
- Start writing the newsletter (this one!) in small fragments. Observe how it grows & improves. Finish it on holiday.
- If you believe in something - i.e. The Square Hole - find a way to make it happen. It may be the best content - see point 5.
Next Three Months
These are the things that I would like to accomplish in the next three months:
- Reduce the maximum record-to-publish timeframe to 8 weeks (currently at 16 weeks)
- Publish 7 new podcast episodes
- Publish 8 new videos
- Including a Ship Fast special episode
- ✅ Join 1 live stream 🎬 Let's Profile Dagger
- Enable listeners to own this content (I expand on this below)
Stretch goals which are beyond my control:
- 1 review on Apple Podcasts - Can you help?
- 1 review on Spotify - Can you help?
Lastly, is there anything that you would like to see in the next three months?
Own This Content
My last goal for the next three months - enable listeners to own this content - deserves unpacking.
Make it Work is not sponsored in any shape or form - not even free stuff. I really like this status quo since my only focus is on producing honest, great content. The content is the product.
I never liked ads. I hardly used YouTube before I started paying for Premium & experienced it ad-free. I tend to skip sponsored videos, and many times I stop watching videos as soon as the sponsored segment starts. I rarely skip over the sponsored segment - maybe 1 in 20 videos. The bottom line is: I really don't like ads.
I am a strong believer in owning content, forever. Don't you hate it when one of your favourite TV shows just disappears from INSERT-STREAMING-SERVICE-NAME-HERE? I think that it's valuable not having any big corp in between, just a simple download link from a CDN. A video player to stream directly from this CDN is nice to have (a form of convenience, a backup to streaming from your LAN), but the important part is content that you pay for once, then own in perpetuity. The key take away here is that you are not the product (i.e. ads). The content is the product.
There is one more point worth expanding: pay once. Subscriptions are like rent, and I don't think that they work well in this context. You shouldn't stop having access to content just because you stop the subscription. Instead, I see it as an investment that should keep on growing, and becoming better over time, not stop because your circumstances changed. Think compounding long-term value in the form of being part of something that you enjoy.
If the above resonates with you, subscribe to makeitwork.gerhard.io to see how this unfolds. If you are already a subscriber, I expect to have this option in place within the next three months. As for everyone that I recorded with, they have a 100% discount code secured. I will be sending them out as soon as the new setup is in place.
In summary:
- Own the content that you enjoy watching;
- Pay for it once.
Let's see how well this small bet works out in practice 👋 Until next time!
