Week 13, 2024 - On Newsletters and the next Tech Radar

Happy Easter if you celebrate it — otherwise, enjoy a nice spring weekend. This issue got a bit long, I almost moved my thoughts about the new Tech Radar to a separate article to keep the size down. Eventually, I decided to keep it here: there’s a structure emerging in how I organize my articles: I create separate posts during the week for bigger topics that can be relevant regardless of time — and keep commentary, ephemeral thoughts, and other ramblings for the Friday newsletter. Once this crystallizes, I’ll also update the two mailing list descriptions to reflect this duality, so people can better decide what content they are interested in. Until then, thanks for bearing with me, it means a lot.

🌩️ What I learned this week

I finished the longest article on the site yet, about How To Represent Decisions You Disagree With. There are just so many aspects to a difficult situation like this, so I feel the depth justified, though the longer the text is, the more there’s a need to structure better and have summaries / conclusions to help process it. I should improve on that. Otherwise, I’m happy with the result, and this article will be a good reminder to go back to the next time I’m faced with having to navigate my teams through turbulent times.

On the home-lab front, I mostly struggled with sharing a USB-attached HFS+ drive with VMs and containers. I had to settle with SMB for now, NFS was failing for some reason. Good enough for now. Not much progress on the public services front yet, though I did some rough firewall tuning and network planning, and have a good Docker-hosting VM template to host future containers on. Also, backup, while on-site only yet, is sorted out nicely.

📚 What I want to try next week

There’s an async book club being organized in one of the leadership communities I’m part of, reading Cooperative Software Development from Amy J. Ko. I’m equally excited about the format and the content, and will probably report back here in 2-3 weeks when we’re done.

Unless something makes me change my mind, I plan to write about things an aspiring IC can do to grow and practice management / leadership skills in their current role. (I should also make a better distinction between “leadership” and “management”: a comment on last week’s newsletter made me realize that I sometimes use these words interchangeably, though when asked, I see a lot of obvious differences between the two.)

If I have some time, I’ll add off-site backup and maybe the analytics for this site to the home setup too. It’s so easy to get lost in tinkering, even the frustrating periods are fun. I should watch my time better.

🤔 Articles that made me think

Why I Write a Newsletter Every Week

Jordan Cutler’s amazing and honest share on the first anniversary of his newsletter made me think about revisiting my original reasons in light of the 2 months that passed since. Here were my reasons back then:

🧠 ”Retain the lessons from my engineering leadership challenges”: This is already so much more than just a track record of events I went through. Writing about a topic forces me to organize, revisit, challenge and solidify my thoughts on the subject. Seeing the Learning Pyramid linked from Jordan’s article made me realize why I feel this is the biggest value I get out of my blog so far.

📅 ”Holding myself accountable”: The weekly cadence of sending out something on a Friday worked so far, but I know I’ll face harder challenges eventually. I’m hoping I’ll be able to timebox and cut scope when that happens, to avoid delays.

🧑‍🏫 ”Helping others”: This starts to pick up slowly if I look at subscriber numbers, where my content is shared, and what feedback I receive. It’s great to feel helpful, but I’m still focusing on the above two points because I have more control over those areas. Regardless, if you find this content useful, share it with someone who you think could benefit from it. Thank you!

Things that were missing 2 months ago but now I see them as goals:

🧑‍🎓 Pushing myself to learn new things: Initially I only focused on “sharing my lessons learned”, but I quickly realized I could and should also do this with the new stuff I’m just learning about. During the preparation of a topic, I always do research, and discover great ideas that I can incorporate. Also, writing this newsletter and blog helps me be more conscious of how I spend my time on experiments and new stuff. Without the public eye, I can tend to abandon something after a while, move on without concluding, etc. Knowing that I’ll need to write about it helps me to be a bit more disciplined and organized.

👯 Networking: I had a few discussions around the content of this blog with people I wouldn’t have met otherwise, which was a great feeling (both building new connections and helping others). I’ll keep this as a focus point going forward.

I’m thankful to Jordan for sharing his journey, similar to the second part of Nicola Ballotta’s Ultimate Guide to Writing a Newsletter, which was another great read this week on the topic.

Thoughtworks Technology Radar #30 Preview Webinar

If you’re not familiar with the Tech Radar, in short, it’s a regularly updated visualization of techniques, tools, platforms, and frameworks / languages in software development, putting these “blips” on the radar in groups of hold → assess → trial → adopt. It’s a great way to get an overview of the tech zeitgeist, and to have a more or less objective validation of current trends (even if they are explicitly “opinionated”).

Every 6 months Thoughtworks publishes an update, and a week before it’s live they hold a webinar discussing the most interesting findings. This time, not surprisingly, AI was everywhere. They mentioned that the finished report has 40 AI-related “blips”, and hinted at possibly starting a separate quadrant or even a new publication, something along the lines of “state of AI in software development”.

Side note: I have a feeling akin to the late nineties, when I had the chance to watch the dotcom boom from close, starting my career as a leader of the newly formed web development team of a big Hungarian ISP. The parallels are strong: it’s the same excitement of suddenly discovering new possibilities — and the same overinflated hype of exaggerated predictions. Day by day, revolutionary stuff gets announced, and it’s all there to play with. Similarly, I still remember how surreal it felt to order pizza online for the first time in 1999 - I hadn’t left my desk or talked with a human being, and in an hour or so, reception called that there was someone with a pizza for me. The feeling I have playing with some of the new AI tools is very similar to the amazement of a younger me.

For the flip side, as a light experiment, I put together a chart overlapping the time-shifted stock price of Cisco and Nvidia. I believe both companies are great representatives of their periods, selling the hardware that powered the exponential growth of the internet and AI respectively. To make them overlap on one chart, I added exactly 24 years to Cisco’s data, making the dotcom burst to align with today, which is obviously opinionated, but that's exactly the point I'm trying to make on where this excitement could lead to.

I hate serious charts with a hidden agenda, so let me be clear: it’s not serious, and I don’t try to hide my agenda. But I also don’t think I’m alone feeling that there's a bubble to burst. Cisco was far from being the only one benefiting from the internet, and similarly, the moment tech giants move on by creating their own GPUs, the days of free money will be over for Nvidia.

Anyway, here are my random notes from the Tech Radar webinar:

🏦 A fun case study: an Italian firm modernizing a legacy codebase consisting of 250 MLOC (that’s a million!) of Cobol, using Generative AI to understand and identify possible entry points for isolating features to start the migration with. Interesting products to watch in this area: Bloop, Driver AI.

🧑‍💻 A lot of discussion about how (if) AI will replace developers. I agree with the consensus: writing lines of code is not the hard part of software development. Time and time again new technologies and tools elevate the work to a higher abstraction level. The majority of programmers are not familiar with machine language anymore, let alone logical gates in a CPU, both of which were day-to-day parts of the job of their ancestors. Similarly, we can expect a shift from coder to overseer or conductor of AI agents. To thrive in this era, developers need to better understand the problem they are out to solve, the product they work on, and the users of it. It’s funny that these are the exact skills that members of an efficient software development team need to have already, without anything to do with AI. Effective work requires close collaboration within and across different functions of the organization, and it's exciting to see how AI can eliminate the boring parts of this process.

🤖 On this topic, AI Team assistants are being assessed. These are tools that help software development teams with story writing, creating acceptance criteria, helping automated tests, etc. There was a mention of the “reusable prompt” concept, storing team knowledge and making it available to the members via AI agents. Thoguhtworks, the writers of the Tech Radar are actually working on a solution (codenamed “Team AI”) that they’re planning to open source when ready.

💻 Warp adds AI to your terminal to help understand error messages, access extended autocompletion, and get things done plus learn in general. After using it for a few hours, I think this approach better suits my workflow than the way Open Interpreter which I wrote about two weeks ago works. I feel more in control of what’s happening, and this way of interacting with an AI feels closer to traditional work in a shell. That being said, Open Interpreter aims to be a much more generic tool than Warp.

📝 Cursor seems like a cool AI-first Visual Studio Code fork. All my extensions and settings from VSCode were picked up nice and clean, but I need to play with it a bit more to see if it can really be a viable replacement.

©️ AI and Intellectual Property came up too (something I scratched the surface of last week). Some imagine this will be handled similarly to how AirBnB and Uber’s case was: they did something arguably illegal (operating unlicensed hotels and taxis), but by the time legislation caught up, they were too big to kill. Companies are betting on something like this: Microsoft pledges to provide indemnification if you work with their Copilot tools and get sued for using someone else’s intellectual property that their models were trained on.

🌐 Infrastructure as Code is seriously maturing. This was the only non-AI topic on the call. Among the mentions, there were Terraform alternatives and forks (Pulumi, OpenTofu); and two interesting visual products: Winglang combining infra with runtime in a new language, and System Initiative offering a collaborative visual tool to manage infrastructure (imagine Miro for Terraform?).

Volume 30 of the Technology Radar will be published on Wednesday the 3rd of April here, but you can already watch the recording of the above webinar.

That’s it for this week, no separate “Something Cool” as the newsletter is already pretty long, and there was a lot of cool stuff to play with in the section above anyway.

Have a weekend full of sunshine and springtime winds,

Péter