Growth Garden

I first came across the notion of a growth garden probably sometime in 2022. As a place where seeds may be planted and sometimes turn into something more resembling a plant. And sometimes not. A guy had this blog and a dedicated section for “half-finished thoughts”, ideas that he’d observed and thought he might turn into something more later down the line.

So I started this document back in 2023 and never thought I’d actually keep up with putting stray thoughts here, but a few years down the line it’s still going strong, very unfiltered and unstructured. A nice spot to just dump immediate reflections on stuff I read or do, in case it’s simply too plain to go into it’s own post.

Also, I frequently diverge into swedish on here, shame on me.

2026-03-27

2026-03-05

  • Good take on Brook’s relevance in the AI age. Essentially the argument is that essential complexity still applies, while accidental has gone off the rails. I really want to try roborev now.

2026-03-01

Good deep-dive complement to byte byte-go. Solid keeper for when I might have to do this myself at some point.

When reviewing a complex trace, the most effective tactic is to identify and annotate only the first upstream failure. LLM pipelines are causal systems; a single error in an early step like misinterpreting user intent often creates a cascade of downstream issues.

Analytics in design system components!

As design system engineers, we went through each component and added analytics depending on how users could interact with it. With this approach, we started to log around X4 more events, which were all relevant for analyzing our application’s usage. The best part was that devs didn’t need to add a single line of code to get more granular analytics – they just needed to use components from our design system library!

Rubrik also uses chromatic, storybook and Figma. They appear to have similar overhead with pr regression testing for ui. And they also seem to have more advanced components (makes sense for their more interaction-based customer product I guess) that they connect to the overall base components. We’ve been more pragmatic in that respect building the components and abstracting as we go.

2026-02-28

Really solid reminder of good stuff for writing performant applications. Throwback to school with Staley and other algo-teachers. Too much to recap here but one key takeaway is that memory layout typically matters more than algorithmic complexity, just because I/O is so darn heavy and increases runtime by orders of magnitude between each tier – if we can keep it to the CPU that’s going to slay a O(n) to O(log n).

Memory layout and allocation patterns matter more than many developers realize. And often, the easiest performance win is simply avoiding unnecessary work.

this quote from Donald Knuth is almost always taken out of context. The full statement reads: “We should forget about small efficiencies, say about 97% of the time: premature optimization is the root of all evil. Yet we should not pass up our opportunities in that critical 3%”

the main rule of performance optimization is to measure first and optimize second.

Not that I’ve ever seen it but we may face flat profiles in mature codebases where obvious fixes have already been remediated.

When facing a flat profile, our strategy shifts. We look for patterns across multiple functions, consider structural changes higher up in the call chain, or focus on accumulating many small improvements rather than one big win.

2026-02-27

Really interesting! Had never thought about this but of course we need to evaluate our llm output in product context.

Also reassuring to recognize patterns from standard ML evaluation. This time just overlaying prompts, product use-cases and sample sizes to the original evaluation scoring.

shipping without evals means we have no idea if we’re making things better or worse.

Basically the article introduces four different measurement techniques:

  • automatic (llm can evaluate llm)
  • human (for expensive high perf eval)
  • benchmarks (though they may not necessarily be apt to our usecase!)
  • and context specific tasks (customer support for fintech tasks)

2026-02-25

The models caught up. We don’t think this would have been possible even a few months ago. Earlier models couldn’t sustain coherence across a codebase this size. New models can hold the full architecture in context, reason about how modules interact, and produce correct code often enough to keep momentum going. At times, I saw it go into Next, Vite, and React internals to figure out a bug. The state-of-the-art models are impressive, and they seem to keep getting better.

Architecture decisions, prioritization, knowing when the AI was headed down a dead end: that was all me. And of course, an agent that simply runs the migration: > Then open your Next.js project in any supported tool and say: migrate this project to vinext

This is interesting: We have new possibilities now! > Efficiency framing is missing the point, it’s more about what you can create now with AI which you could not before”. Then the Utah conference: > “Organizations are constrained by human and systems-level problems. We remain skeptical of the promise of any technology to improve organizational performance without first addressing human and systems-level constraints. We remain skeptical and we remain human”. – Kent Beck, Laura Tacho, and Steve Yegge. > Based on past experience, you might expect at least some traditional companies with no exposure to AI to wait and see what happens. But I found none doing so.

They use SPIFFE!

  • CEL - Common Expression Language by Google worked with Ubers ms-latency reqs. Specifically cool considering they do attributes lookup that match on policy requirements.

Also, most importantly, ABAC separated policy from code. System owners can change authorization policies without building and deploying new code. This separation of concerns allows security policies to evolve independently from application logic.

2026-02-24

Solid thinking around errors! Categorizing them and helping with the search space for users/devs/machines is key!

Tip: Users and developers should never have to understand your implementation to understand an error. Thus, unless your code is literally a calculator, a divide-by-zero error is an Assertion.

By eliminating chunks of “search space” for your users, suggesting actions can save hours or prevent them from giving up and churning out of your product. Even a simple “try again in a minute” after a System error can increase completions of workflows.

2026-02-23

Constantly reminded that speaking and writing is crucial moments for evaluating whether I’m applying good frameworks!

A great way to get your reps in for this structure is to ask yourself, What? So what? Now what? after reading an email, article, or other document. Drilling this technique makes it easier to deploy in the moment.

And, in order to stay focused, this is a great reminder to steer clear of the curse of knowledge:

The antidote to this curse of knowledge is empathy.

Mostly similar to vibe coding/Claude code articles one has read.

Tibo tells me that a typical engineer on the Codex team runs between four and eight parallel agents, which do any one of a number of tasks…

“There is a tricky thing in all of this, though: we have to relearn these capabilities with every model”.

skills is that they help steer the model to more specific behaviors, and they can also be combined. Skills are continuously published internally and team members copy from each other.

Above makes me think that I should just keep iterating on my “branch out” skill. Ok the other hand, the overhead probably only makes sense if A) the model doesn’t handle it well enough on a one-shot and B) the steps are deterministic and procedural and C) happen more than twice (ever).

The team set up nightly runs of Codex, instructed to look for issues across Codex.

This one we wants to try surely!

2026-02-20

Granted, this is for APIs that are actually exposed as products to developers. So not 100% applicable in news-service. But for future products!

propose two user personas for API design: the eager developer and the discerning developer. The eager developer chooses convenient APIs that are easy to understand and fast to integrate. The discerning developer chooses reliable, expressive, customizable APIs that will serve their needs as they iterate.

Seeing code run successfully gives developers the same feeling of delight as finishing World 1-1 in Super Mario Bros. They might not have saved the princess yet, but they’ve started to build a mental model of the concepts in your API—learning the controls and recognizing patterns they can apply later.

Waking skeleton in a sense! Provide end2end without all fancy customization

Always:

Once an API is released, it becomes harder to redefine core concepts because this forces users to migrate to a new API.

Interesting reflection on typical product Development not being quite as effective (interview/ab-test etc) in the case of apis.

Of the strategies described above, dogfooding is the single most effective way to measure your API’s convenience and power

2026-02-19

Yes. Kinda obvious. A thin contender to the succes-framework but good primer for it.

Business strategy is usually born of a highly rational process, grounded in facts and analysis. Storytelling, often associated with fiction and entertainment, may seems like the antithesis of strategy. But the two are not incompatible. “Our work is grounded in science, but the scientific facts and arguments alone are never enough to persuade anyone to act,” says Calvelli.

Strategy stories can provide a powerful bridge between arguments and actions, intentions and results, and strategists and implementers.

Building on the classic 5W2H method (who, what, where, when, why, how, and how much), we recommend asking the following questions about both your company and your competitors. The questions in the left column will help you collect observations based on the factual context you’ve already gathered. The questions in the middle column will help you make sense of your story, and those on the right will help you envision the future. Your answers can then be combined to yield a purposeful strategy story that is ready to be acted upon, like a script.

Not great questions, but a good starting point for deriving a story!

2026-02-17

  • Brilliant recap on ascii and how Ctrl maps to control characters. And all that ^M? TIL!
  • what it feels like to learn JavaScript in 2016. Stop. Just stop.
  • How 10 tech companies choose the next generation of dev tools. Maybe a useless reflection, but different tools work for different people! And given how drastically fast the landscape has changed the tech stack looks very different across different orgs. A few standard takeaways:
    • smaller orgs care less about security/vendor review and weigh in harder on how a tool feels
    • cursor gave way for Claude code! Glad I skipped that phase.
    • metrics are still not readily available to measure productivity.
  • An old article on Data stack evolution. Way to say nothing with many words.
  • Finally caught up on Fowlers Strangler FIG pattern.

Apparently strangler fig is the real official name. Original (superseded) post

Too often we see muddled objectives, with different groups wanting different things. We need to establish alignment on the key outcomes early, and then revisit regularly to reinforce the alignment and to investigate if there should be a shift in priority.

This feels very relevant to Ulfs work on coordinating/project mgmt work for three months (long and short enough?) to “gather requirements” for what the business wants

Like the fig, it begins with small additions, often new features, that are built on top of, yet separate to the legacy code base

  1. Understand the outcomes you want to achieve
  2. Decide how to break the problem up into smaller parts
  3. Successfully deliver the parts
  4. Change the organization to allow this to happen on an ongoing basis

2026-02-15

  • No coding before 10 am. new playbooks. prompt drafting. obsessing over agents running. revise the working structure and pair over the outcome. makes sense if we may define clear objectives and verifications for our builds. And I second building for the agents and removing noise! That’s where we need to start.

Every dead path is noise that degrades agent performance.

The bottleneck is human decision-making time, not compute cost.

Galloway is callous and crass as always. But this one actually has some sensible step-by-steps to it. And he’s not wrong. Story makes the world go round. I simply need to try it out more, dare to fail in it!

hands down, the skill I would grant my boys is singular: storytelling.

the willingness to subject yourself to rejection is like sleep: Without it, you can’t succeed.

on brevity

Think of the last several emails you wrote. The length and formality of the email is inversely correlated to your comfort with that person…assume the listener knows a lot and likes me, and I like them … so we can get right to the point. We’re friends.

2026-02-14

Companies need to develop a set of norms and standards around AI use—what we call an “AI practice.”

These actions rarely felt like doing more work, yet over time they produced a workday with fewer natural pauses and a more continuous involvement with work. The conversational style of prompting further softened the experience; typing a line to an AI system felt closer to chatting than to undertaking a formal task, making it easy for work to spill into evenings or early mornings without deliberate intention.

It would seem that we are addicted to a new drug, and we don’t understand all of its effects yet. But one of them is massive fatigue, every day.

enterprises aren’t too bright about SaaS, collectively (evidence: many are still on Copilot), they are quickly growing savvy enough to know that Build is the New Buy. $/hr To The Rescue That old formula is also my proposed solution for the AI Vampire, a quarter century later. Someone else might control the numerator. But you control the denominator.

Interesting. Never thought of this as a big platform problem but Yegge is really right with this. Deprecation at GCP is an ISSUE! And the idea that platform = backwards compatibility, even if it hurts. Hard engineering problem! Or you have to “just” isolate the bad architectural decisions made in past sins and build the new stuff on new primitives. Sounds simple enough.

Successful long-lived open systems owe their success to building decades-long micro-communities around extensions/plugins, also known as a marketplace.

this feels new, but curiously obvious to me. I guess I never thought of a single piece of software (ok Emacs is an OS..) as a platform. And the obvious connection between platform and plugin/extension/marketplace has never struck me before for some reason. But it makes sense. Customizable extensibility invites personal profitability. Bother monetary and affectionally.

Nice exploration by seasoned engineers. triggers one to build your own opencode kinda tool. Or just to keep doing what Will Larson suggested - creating your own agent. Interesting (or maybe obvious?) that devs preferred Claude code whereas non-devs really shone in the new Craft agent.

Some engineers have onboarded to Craft Agents, but most prefer Claude Code for the majority of product development work. The reasons that a few engineers prefer Craft Agents is its easier overview of parallel agent execution, and a more pleasant interface for reviewing lengthy code changes since scrolling is more convenient than reviewing on a terminal.

2026-02-13

2026-02-11

  • Yegge is at it again, this time with Orosz. He’s provocative to get the message across. And he’s banging the same drum (and I don’t disagree). Interesting to learn that he spent a year (hyperbole) reading AI papers to get up to speed. And the reduced work hour bits is very interesting too.

I feel sorry for people who are good engineers – or who used to be – and they use Cursor, ask it questions sometimes, review its code really carefully, and then check it in.

you might only get three productive hours out of a person who’s vibe coding at max speed. But they’re still a hundred times as productive as they would be without AI. So, do you let them work for three hours a day? The answer is yes, or your company’s going to break

The important point that the naysayers and the skeptics are missing is that AI is not coming to replace your job. It’s not a replacement function. It’s an augmentation function. It’s here to make you better at your job. And that’s not a bad thing, actually”.

2026-02-07

Decided to explore some blog posts on agentic coding in a mob context. Came out with three reads. Two bad ones and one that feels promising!

  1. Crisps take, just marketing and no substance. Good advertising for mobbing but not really on integrating AI. Interesting takeaway that Tesla “mobs everything with AI”. Just need more how and meat on what this looks like daily beyond “AI is a teammate/advisor/coder that never tires…”
  2. This though! Some really good input here on step by step approach. Way to go atlassian for sharing this! Convinced we can make thorough use of this within the team! Structured approach to progress! I especially like the combination of speed & maintainability that AI can bring to the mob!
  3. And then another crappy article but with one or two descent comments! One of the key advantages is improved code coverage and accuracy.

2026-02-06

Just look at https://bodell.io/2026/02/til

2026-02-05

(todo, blog this)

1.20: The object isn’t to make the art. It’s to be in that wonderful state which makes the art inevitable… ~Robert Henri. If you create an environment where great things can happen, they will. If you just try to create great things they’ll stay elusive

2.00: Oblique path: I may not have ended up where I intended to go but I ended up where I needed to be ~Douglas Adams.

3.00: Most of what we do in managing is making it harder for people to do the work. We’re trying to make it easy to manage the work, and in so doing we end up making it hard to do the work.

Liberating people, not by managing. Let the team that does the work figure out the shape of the process and organize by themselves.

13.35: I’m not interested in productivity (we got a lot done), I’m interested in effectiveness (we did the right things).

15.00: The prioritization will not lead us to [the 4-5% of effective work]. We need to be able to steer. It is in doing that work that we discover the work that must be done. So it’s doing the work, delivering it into the users hands, and learning what we learn from that process that’s really valuable.

16.40: It took a hundred years to get from the first phone to modern phones, he couldn’t conceptualize that - it needed to materialize.

16.55: We know we’ve designed a pretty good platform when people do things with it we never imagined.

19.20: Doing the work as a group, I don’t think that goes away. Most of the work we do involves interaction with other people.

21.00: That kind of communication skill [of being able to articulate and direct what needs to be done in a way that someone else can understand and act on it] seems to be an increasingly valuable skill with LLMs.

2026-02-03

  • Redis wrote about the ongoing AI wave. Context engineering is king and everyone will be come a coder. A little boring around the edges but overall a few takeaways.
  • Skimmed through the bytebytego newsletter on transformers

Think about what Excel did for spreadsheets, but on steroids. Excel let everyone run data-powered reports without writing structured queries.

2026-02-02

  • A miniature clawbot!. There are still technicians in the world building with tech in mind…
  • Good reflection on types of AI-users. Surprising amount just using chat windows, and then agent users. Primarily finance people seeing the benefit.
  • And he’s building his own CLI. Fun to see someone with a little more nuance and pace to the hype!

Apparently skills are just cli:s exposed to the LLM, because MCPs consume too many tokens. At least according to this guy.

I didn’t have the rebirth of the terminal UI and me building dozens of CLIs on my 2025 prediction list, but here we are.

2026-01-30

  • It’s spiraling out of control. Or maybe just I am. Probably time to wind down and take a break from the LLM frenzy. The tornado looks relentless! Just one more article before I check out for the weekend… And then again, no-one outside our profession really thinks through the craziness that we can now build personal assistants singularly
  • Another guy had the bot (openclaw) buy him a car…

2026-01-29

  • New blog post from Yegge! It’s like I’m at the fountain of wisdom drinking… Or perhaps its just the slop-jar. Depending on how you read the next post. Anywho, useful learnings.

I actually got this idea from hallucination squatting, which Brendan Hopper told me about, where you reverse engineer a domain name that LLMs are hallucinating, register it, upload compromised artifacts, and the LLM downloads them the first time it hallucinates the incorrect site name. If even North Korean hackers understand Agent UX, then it’s probably time you did too.

  • And also the link to the jinja-flask-guy (apparently) Armin Ronacher reflecting on an agent psychosis and how some people (apparently) completely derail in their interaction loops. I want to stay in the know and looking through the output. So maybe a good wakeup call alongside reading the vibe coding bible. Apparently he just founded a company called Earendil. I subscribed without thinking. With that name who couldn’t?
  • Man. And this guy. Ludicrous speed constantly. I should get a move on… These guys are building like insane because they know so well what they want and have probably done it for so long. But at the same time I can’t shake the feeling that some of the content begins to feel eerily similar. Perhaps there’s some conspiracy at play or we’re simply converging…

2026-01-27

The problem is that we are trying to predict what people want and deliver it for them. You can’t do that. Not really. Not reliably. There have been precious few people in the world, over the entire history of computing, who have been able to do it reliably. Steve Jobs was one of them. We don’t have a Steve Jobs here. I’m sorry, but we don’t.

2026-01-24 TIL:

  • Etc -> etcetera -> editable text configuration. Backronym!
  • Unix/Linux/kernel/os setup
  • Dockerd, daemons, containerd, cgroups, namespace etc.
  • Wiki, not “what I know is” but Hawaiian for “quick”

2026-01-23

  • ZFC! Zero framework cognition. I.e. the ai makes all the decisions. Never the developed code. Code is stupid. Ai smart. Feed the decision back to the model and don’t rely on heuristics. Makes sense.
  • Go back and refine retention, make it its own app perhaps!
  • six new tips for better coding with agents
    • Give beads more of a shot. As an issue tracker and with more more more upfront planning. And thorough revising (rule of five).
    • Man this is fun. Agents in your phone and something that codes and something that creates artifacts… Deterministic in the indeterministic.

However, I don’t see eye-to-eye with the investors on this one. I personally do NOT think we will get useful guardrails. If you try to build something with heavy guardrails, you’re going to wind up with Bolt or Lovable, and nobody will use it. Sorry! That’s just not the right model. Instead, I think we’re going to get orchestration tools that are every bit as powerful, messy, quirky, and frustrating as Claude Code and the current batch of terminal-based coding agents. ~Steve Yegge

2026-01-23

  • Desire path! Classic thing that you know exists but have never bothered to identify a word for. Swedish “önskestig?”- First stab at Vibe kanban. There’s so fricking much to explore. TUIs in Charm is fortunately not entirely new to me but fun to get started with. And spec-kit or BMAD should also be explored. But I’m just trying to catch up on Yegges hundreds (not yet though) gas town posts
  • Some gastown backstory. Crazy to learn all the stuff that’s going on in this space. I’m thinking a lot about the Evolution of coders image. And this quote… man. The productivity of one is going into everywhere. We’re already seeing it. Heck, I’m seeing it and I’m not any good at this stuff. Where to next?

Solo unicorn? They’ll all be solo, soon. At this rate I’ll be more impressed if a 100-person startup can make a billion dollars in 2027. Genuinely impressed. How will they even communicate, with that many people doing AI work at once? I can’t wait to find out.

We’ve arrived at factory farming code. And it’s hella fun.

evolution of programmer -24-26

  • and the emergency manual. Apparently they had a murder spree on their hands. And there’s A LOT of wrinkles to iron out. But it’s ticking…

  • backstory on beads OKay. This is something I can get my hands on and get started on. I’m thinking private project, some addictive tui for learning something beneficial in bite-sized nuggets. Get me hooked, and build it for me. Backed by beads. Should be a good enough start.

The problem we all face with coding agents is that they have no memory between sessions — sessions that only last about ten minutes. It’s the movie Memento in real life, or Fifty First Dates.

2026-01-22

  • Not that I’ve finished it (or ever will?) Gas Town seems like a really interesting idea and something to keep a lookout on the horizon for. Not that I think I’m mature enough for it, but it probably will set the pace for the bigger factories at some point if we just accumuate enough ideas. Churning through the backlog at a rapid rate. Worth a stab. But for now I barely have enough ideas to put in three distinct running instances!

2026-01-16

  • Bra ord: Bondpermis - ett ”lumpenuttryck” för att olovligen ta ledigt och smita från militärtjänstgöring eller utan tillåtelse lämna förläggningen
  • Gandalf AI challenge! Kul! Klarade level 6 men inte längre. Nyttigt för att förstå att det är lurigt att få modellen att inte avslöja hemligheter!

2026-01-14

  • Casey Newtons description of the reddit fake uber-eats thread and how he debunked it. Fun read on the intersection of journalism/tech and how scary good fooling can now become. As Kevin says in the podcast episode: “If the barrier for this type of hoax used to be effort, we’re probably going to start seeing a lot more cases like this”
  • Henrik Knibergs AI HealthBuddy som gav honom incitament att be (och få!) en ny medicin mot sitt hälsotillstånd. Otroligt, det är nya tider sannerligen med AI i loopen. Ingen skugga på utövarna men det skulle förvåna mig om vi inte ser detta i bruk som standard snart. Det är det nya 10x som vi trott digitalisering är. Intressant slutsats att kontext är allt, korrekt prompt ganska lite. Och hur detta är hörnstenen i deras agenter, dvs lösningar med AI-backends som resonerar kring data som de definierat. Coolt! Den andra insikten “AI Agent: Tireless researcher and archivist with access to vast medical knowledge” är precis lika relevant för kodning, agenten har access till så vansinnigt mycket mer ramverk, kodpraxis och data än jag någonsin kommer ha eller komma i närheten av att konsumera! Så nyttja det eller sätt dig i baksätet…

This is a general trend with LLMs - “Prompt engineering” is not as important any more, instead it’s all about “context engineering”. If you provide the right context, the information the LLM needs, it will usually do the right thing without you having to phrase your prompts perfectly.

2026-01-08

  • A somewhat sobering look at the future of AI generated code by Gergely. Or rather, what happens to us as SWEs in the midst of it. One takeaway for sure: More and more seasoned pros are saying it’s here to stay, and I don’t feel all that desparate to sit with tedious syntax anymore!

2026-01-07

  • This was cool! CLIP, the new way to do image recognition!. Instead of a traditional CNN OpenAI matched the alt-texts with images using a separate text-based deepnet and an image-vector-embedding that maps to the same dimension space and then runs backprop to update weights in order to closer approximate the two. Works surprisingly better on adversarially manipulated images (pixel mods and whatnot).

2025-12-26

  • Kvartal artiklar om Jesu trendande, S-ledamot Widar om att “inte satsa på de fattiga” (intressantast! Argumentet att det ska vara lika viktigt att slipa inkomstkvoterade bidrag som att få bidrag när det behövs, och en socialism som vill minimera skyddsnät för att främja arbete och eget initiativ).

2025-12-17

  • ByteByteGo on Reddit’s migration from python to go on the comments module Learned about the tap compare-pattern. Pretty straightforward for reads, for writes it involves a duplicated database setup where the strangler service (which gets routed a percentage of requests/ A-B-deployment) calls the legacy service to perform the write, and the after the fact a comparison takes place between the “sister datastore” and the primary data source.

A small percentage of traffic gets routed to the new Go microservice. The new service generates its response internally. Before returning anything, it calls the old Python endpoint to get that response too. The system compares both responses and logs any differences. The old endpoint’s response is what actually gets returned to users.

Also a good takeaway! Faster language doesn’t necessarily mean better performance ootb (of course)

It turned out that Python’s ORM had hidden optimizations that the team didn’t fully understand. When they started ramping up the Go service, it put unexpected pressure on the Postgres database. The same operations that ran smoothly in Python were causing performance issues in Go.

  • How AWS deals with outages. Interesting to read about the process to identify where in the chain there could be issues as well as what appears to be the rigorous paging routine alongside the willingness for anyone that can help to jump in?
  • A quick glance into Frictionless checking out two chapters and realizing that this probably ties into Switch and is something I should get on the shelf.

While “improving developer satisfaction” or reducing lead times” might resonate with engineering leaders, executive decision-makers need to see clear connections to financial impact. Business leaders are evaluated on metrics like revenue growth, profitability, and market share—so your DevEx proposals must connect directly to these financial outcomes. ~Chapter excerpt

Instead of explaining technical changes to deployment processes, he told executives the work would save 50 percent of engineering hours spent waiting for deployments

This puts the press briefings and pitches of all those (google etc.) tech products into perspective. The language you know you know.

When building your DevEx business case, perfect is the enemy of good.

This puts good light on that worn out quote, “perfect is the enemy of good” is too simple to stand on its own, it needs correlation to something to make sense. And it doesn’t correlate to everything!

2025-12-11

2025-11-20

  • Läste (någon av dessa dagar) Cloudflares omtalade post-mortem för en nyttig lärdom i hur en välskriven sådan (och customer obsession) hanteras.

2025-11-09

  • På ungefär samma tema som tidigare, vad Claude-gänget gör för att bygga ut (serendipitously) claude code. Och hur vansinnigt många prototyper de kör ut när de bygger (Tar också med mig on-distribution som i vad modellen kan vs off-distribution där du behöver visa hur utvecklingen ska bedrivas.)

2025-11-07

  • Sjukt nyttig take på vad en AI agent de facto är. En tunn wrapper runt en modell med “context” i form av en lista på strings med respektive tidigare chattinlägg. Dvs varje anrop matar modellen med samtligt tidigare innehåll. Som författaren skriver, context engineering is a problem worthy of AoC.

2025-10-31

2025-10-29

2025-10-27

There’s more that I’ve read lately, but for some reason I’ve been lazy and not put it in here. Oh well. Just take my word for it.

  • I’m starting to sense a trend on what I put in here. Brazeal again, this time on the aws outage

But take heart. This is not bad news. This is joyous news. Here is the ancient wisdom, learned by all those people who didn’t get fired for buying IBM: when everybody’s down, nobody is down. … Your customers aren’t going to leave you because AWS went down. Where are they gonna go?

2025-10-16

2025-10-13

  • Reading up on chatgpt app guidelines. Could this be the next ios or not? Also nice to freshen up on some styling guidelines.

2025-09-12

Use AI to augment your workflow, not replace it. Let AI do drafts and repetitive tasks, but keep humans in the loop to validate, correct, and preserve empathy.

2025-09-11

AI can’t decide to wander away from the original scope just to see what might happen, or question whether the scope itself should change. It can’t feel the pull of an unexpected idea or a hunch, or sense when something deserves deeper exploration, even when it’s time-consuming.

Not sure if this will hold but.

“In a world where AI can instantly generate a CRUD app or replicate any website, taste becomes the final differentiator. Features can be copied. Functionality can be matched. But the feeling of using something crafted with intention? That’s irreplaceable.”

2025-09-07

2025-09-04

  • Maybe I should give that chatgpt trading hype a go after all…
  • Thoughtful reasoning and setup behind running AI agents fullt for code dev. But I’m not really sure I have the understanding/knowledge for it yet? Time to go back to the books on architecture I guess…
  • Kent Beck strikes a new note on the beyond IDE workflow and points to ona (?) as the new reality. And terminal abstractions with multiple running agents on top as the primary interaction mechanism. Much on edit little on review has flipped upside down and the teletype is there to serve us all the better all of a sudden.

2025-09-01

Been a while!

Sophomore ~ sophos - wise, more - foolish/slow. “Wise fool” from greek.

  • Forrest Brazeal again, this time on ChatGPT psychosis. Good to recall: AI will not make you smarter, more interesting, more creative nor an expert. That’s not what it does even if it feels like it!

2025-04-16

  • Super interesting that the job market seems to be way off rails!. Especially considering there are so many good candidates out there that simply can’t make their voices heard through all the AI slop noise! Can’t believe that many people are using AI-tooling live for their interviews AND LYING ABOUT IT!

2025-03-07

2025-01-07

The TV show Silicon Valley lampooned all these ideas a decade ago. The weird thing I’ve noticed is that when Bay Area tech people watch that show, they don’t seem to understand that they’re being made fun of. They think they’re being celebrated. That’s how thick the bubble is.

The tech bro bubble looks better from the inside.

2024-12-19

Been a long time. Made some aoc and built some stuff. But today I read some interesting things!

Which is why strong product companies understand that it’s easier to train an expert to manage well, than to train a manager to be an expert.

That’s why we should strive to work on our expertise over just getting good at leadership, because that trait is much harder to come by after the fact.

2024-11-29

  • Killer presentation on vim completion modes and insert mode register edits. Note to self: Get to use ^X ^L MUCH MORE!

2024-11-28

2024-11-27

  • Leaving k8s seems to be a trend. Although the latter admittedly talks about a different type of workload, non-production-based. I can relate, especially considering what Hightower was saying a few years back.

2024-11-26

  • Debouncing v Throttling. What they are and when to use which. Good learning!
  • A spotify explanation of the algotorial concept and how they apply it to personalized generalized playlists. Basically the idea is to have humans decide the “pool” that goes into a specific (perhaps hard to describe into algorithms) need, and then have the algorithms select from that pool.

2024-11-08

2024-11-07

As an industry, we’ve not built the tools, framework, or even an effective vocabulary required to talk about these tradeoffs, beyond simply calling it “technical debt”. As a result, most conversations around technical debt end up being oddly confusing. People are often disappointed about how “leadership doesn’t get tech debt” or about how features are always prioritized over critical maintenance work.

Point 5! You’re far likelier to encounter codebases that have “evolved” over time, with poor documentation, lots of outdated comments and often with few to no tests, than you are to encounter ones which are perfectly documented, have well-tested public and internal APIs, and code which is perfectly obvious.

You can learn how the organization’s promotion process works and play your cards right.

Many engineers find it far easier to label things that don’t go their way as “politics”, as opposed to introspecting and learning the hard skills required to make better judgements.

The single most frustrated engineers I worked with were ones who refused to let go of their idealistic way of working: and were upset that their organization would refuse to do things the “right” way (in their mind, that is)

2024-10-31

  • Typer. Cool tool for building CLI-apps in Python! Must try next time!
  • Also, apparently Deno is a thing I never heard about until now. Javascript runtime. Guess there are loads of those…

2024-10-30

  • All about Claude and it’s newest family members. Very impressed with all the advancements on multi-step tasks that’s happening. Coding in natural language is very likely to soon become a thing.
  • So The Browser Company is building a new browser. But supposedly not running Arc into the ground. Good for me, I want to keep it as/is, and from what it sounds like that’s what they’re planning to do just now.
  • In other news, apparently Jaywalking is now legal in NYC

2024-10-29

2024-10-28

Hydration is like watering the “dry” HTML with the “water” of interactivity and event handlers. ~* Dan Abramov*

2024-10-25

  • Dan Pink’s ted talk. Such a good speaker. And an amazing recap of what I guess is one of the core messages in his book: “Incentives work really well for narrow (dumb) problems”.

Or as we say in my hometown of Washington D.C.: A true fact

TIL: ROWE: Results only work environment. Productivity goes up across the board apparently.

2024-10-19

  • Cutler being Cutler. Very spot-on post about the glue person

2024-10-14

  • Learned about Kagi today. Through the CEOs original post from 2022! With a somewhat angled ending.

2024-10-11

  • Lesson on cache headers and the web with gippity
  • Lesson on corepack, npm, pnpm, node, nvm and the like… here. In essence corepack is to package managers what nvm is to node. Just need to figure out why the nvm use is not being activated. Most likely because the nvm/node that corepack would pick up is currently not the default one that I’m running globally.
    • Read up a little on pnpm’s node_modules symlink-strategy in the pnpm docs
    • Also learned about the silly bugs that can manifest in npm/yarn. In essence, they don’t respect your package.json and will simply flatten the dependency tree so that you have access to all the transient dependencies from your project (if express uses debug you get free access to debug without having a dedicated version checked for it). Reminds me of my naive approach to pip vs pipenv/poetry back in the day…

2024-10-10

  • So I read Paul Grahams article on founder mode. Some sensible points I guess, but nice to have read Tristan’s take before. I can see the quote hire professional fakers and let them drive the company into the ground. as quite damaging.

2024-09-16

  • iPhone unveils Apple Intelligence. Keep smart in what you let the AI do. Or as someone said, they can’t believe that something so untrustworthy ends up being the primary search result on the internet cornerstone that is google.
  • Tristan’s take on why the Founder Mode discussion is unhelpful. C-level execs definitely have their place, and there’s a point that they’re not founders. They have a different skillset. I should probably read Paul Grahams original article too. Come to think of it, I should probably start reading again…

2024-09-13

  • F.A.I.L: First attempt in learning. Good read. Apparently attributed to Dr. Abdul Kalam, India’s 11th president.

2024-09-12

  • Death of the senior developer - comeback. Forrest raises a few good points in here, the most noteworthy one is (I think) the void distinction between a modifier word in front of “developer”. I can become productive in most languages now if I know one, just ask the llm to do the translation/explanation for me.

2024-09-11

  • Never knew that YAML was such a hassle. Except I did…
  • And that git SHAs can accidentally be scientific numbers!
  • Generalists for the future. Music to my ears. Not sure if the arguments are overly pleasing to me, and I’m not sure I buy the reasoning that LLMs are bad at “wicked domains” and only useful in hands of domain swappers. But it sure sounds appealing. Also, Epsteins definition of “wicked” vs “kind” environments as places where rules are (not) established and thus harder to get into was an eye-opener, and the fact that apparently Ancient Greek was a place for the generalist. A title that has been notoriously out of style since Adam Smith coined economic optimization in the 1700s. Overall though, a generalist is not simply a bad specialist in several domains (worse baker, coder, admin what have you), but actually a bringer of diverse perspective and a versatile toolbox.
  • Okay, this we’ll start to see more of Illuminate is sort of a podcast service, generating two AI voices having a conversation around a CS research paper to get a lighthearted intro to the papers in question. And it supposedly tailors to my preferences as well.

2024-09-10

Foster intrapreneurs — entrepreneurs inside the firm. At Google, where employees may be tempted to depart and form startups, they have the option to compete for funding and the opportunity to work on their concept for six months, with the guarantee they can have their old job back afterward.

2024-09-02

2024-08-27

2024-08-19

2024-06-19

  • Real photo won AI contest…
  • Might be lots more where this comes from. But fun to see a low-code (typically proprietary product) as OSS! amphi for ETL.
  • UDFs in BQ now available globally. Ecosystem expansion, this one is interesting because it brings community solutions to the mix as an integral part of the product experience.
  • Now looms: SBOM for selling software to government. Insight: FOSS constitutes a staggering 70%-90% of today’s software. DUDE! The following list will become interesting:
    • That it’s developed and built in secure environments
    • That you maintain a trusted software supply chain, including maintaining provenance for internal and third-party components
    • That you have automated processes to check for vulnerabilities
    • Additional attestations that may be required by individual federal agencies
  • Good image for illustrating the disability spectrum
  • Biggest takeaway from this article on leadership pitfalls:
    • Ask before answering. Make it a habit to ask one question before giving an answer.
    • Be honest with yourself. How many of the ideas and suggestions you share in a meeting with your peers actually get implemented? Ask your peers to anonymously rate the usefulness of your suggestions.
  • What is software quality. Very good article on contextualizing what quality means and to whom. Interesting to reflect on inherent domain complexity v. scale complexity overlaid by team tenure and maturity. Also really like the final image on how to ship quality software, simple and pristine.
  • While on it, I also was intrigued to find Will Larsons notes on Tidy First. Now for the Kent Beck book!

TIL: non-functionality requirements over non-functional requirements. Nice

After all, as Steve Jobs said, real artists ship!

2024-06-13

  • Geregly Orosz talks about the drastic change on the job market in the last few years. Recession here we go, makes me think that the current situation with tried and true typescript, as well as opportunities to backend, iOS and Android is a perfect mishmash of getting the technical knowledge of various different stacks and workflows. The question is if it will last long or not, I’m leaning towards this being a solid strategy for years to come - specialization can come at a more convenient time.

2024-06-12

2024-06-05

2024-06-03

  • Some nice advice and takeaways for organizing meetups by Phil Eaton. A note on attendance rates: 10-20% actual attendance versus RSVP seems normal. If you get a higher percentage of people actually attending versus RSVP-ing you’re doing pretty well!
  • Let’s try the hn-text app!
  • Good talk on why we should look for ways to improve work with complex things over complicated things! From agila sverige

2024-05-28

  • Arguing for docs-as-code despite all the difficulties that git brings. I couldn’t agree more, the same reason we switched to a markdown, git-based workflow over confluence. Also because you coudldn’t get proper audit trails and anyone could merge anything… I’d almost say that docs-as-code is worth it for the diff functionality alone.
  • Productitis. Nice neat little new definition: Productitis occurs when businesses add too many products and/or offers resulting in overly complex portfolios that are difficult to manage and understand.. And This confusion can lead to choice paralysis or customers choosing a competitor that’s easier for them to grok. Thought this was very interesting! Lower quality users. If your messaging is unclear, then you’d better expect to see that reflected in who comes through your front door. Messaging challenges lead to inadvertent encouragement of bad fit users and discouragement of good fit users. Something I definitely struggled with on the SEBGCP-platform, are we encouraging the correct customers? Learn to say a big fat NO to opportunities that don’t align with the strategy. Oh man.
  • For once took the chance to dig down the rabbit hole of the origins of grok. Turns out there’s an Exodus reference, martians and other SciFi involved.
  • Terraform is dead, long live the KRM operator model. I’ve seen all this first hand and I agree, although the author most definitely seems to have some incentive to promote this pattern and get consulting customers for it.
  • Getting more familiar after switching back to tmux. Learned about creating and renaming (prefix-c|, respectively) windows. Good overview

2024-05-25

  • Google finally shared some more details on the UniSuper incident. It was GCVE after all.

2024-05-24

  • Discovered the Daylight Computer via hackernews! AMAZE
  • Sveriges Radio in the age of AI blog post from Olle & SWK: 73 procent av svenskarna har förtroende för innehållet i Sveriges Radio, visade årets SOM-mätning från Göteborgs universitet. En Novus-mätning från 2022 visar att 82 procent har förtroende för Ekot. En så hög trovärdighet är en unik tillgång. och det faktum att Vi har även beslutat att vi inte kommer att använda syntetiska röster i den journalistiska verksamheten i Sveriges Radio. Lyssnar du på innehåll från Sveriges Radio är det alltid en riktig människa du hör.. Brave new world.

2024-05-20

  • How Gemini got it’s name (It’s because of twins!)
  • Good short article on what makes and breaks a good platform
  • Some reflections on where LLMs shine. I think I’ve already made up my mind that it’s as a sparring partner that you verify. But nonetheless, some good takeways: The best developers spend less time writing code and more time thinking about the problems they’re trying to solve and the best way to approach them. And Give it tasks that I know how to do, can confirm whether it’s done right, but I don’t want to spend time doing it.
  • Sveriges Radio goes taktisk publicering!

2024-05-17

2024-05-16

  • Good discussion on what makes a good API
  • And great advice on how to present to a more expert audience than you are: 1) Choose self-affirmation over self-doubt. 2) Be intellectually humble without undermining your credibility. 3) Acknowledge and invite their expertise and experience — without letting them take over. Your audience may know more than you on this topic, but they don’t know more than you about everything.:

2024-05-15

2024-05-14

  • Forrest’s at it again, this time on the future of Stack Overflow. Valid concerns on whether The relationship between user-generated training data and AI-generated results so far appears to be one-way. LLMs do not add to humanity’s body of knowledge; they only synthesize and regurgitate.. And of course Most importantly, the thing Stack Overflow was worst at — providing a welcoming place to learn — is the thing LLMs are the best in the world at. Read here

2024-05-08

  • What software developers hate. This one hit home: You just won’t settle on it until much later. The key, as Javier González argues, is to figure out “when to stop the perfectionism infinite loop.” Your code is never going to be perfect. Accept it and move on. And on frameworks: You really have to pick the ones you think will gain the most traction and start there
  • Infoworld article on Why companies are leaving the cloud. Didn’t really expect this to be honest. But the following quote summarizes it well: The cloud is a good fit for modern applications that leverage a group of services, such as serverless, containers, or clustering. However, that doesn’t describe most enterprise applications.
  • Apparently alpine is not the only way to go minimal. Who knew? This article outlines some alternatives.
  • Interesting hands-on discussion on defining engineering productivity from a recent VP of Engineering. Word of the day muda: Like any engineering organization, we spend some percent of our time on fixing bugs, performing maintenance, and other things that are necessary but don’t add value from a customer or user perspective. The Japanese term for this is muda.

2024-05-03

  • Spotify gets in the game of hosting Backstage duh.
  • So IBM bought Hashicorp huh? Might explain all those hotshots switching business. As always I appreciate Forrest’s take on why GCP wasn’t the aquirer
  • Using the BFF pattern as a way to handle secure tokens, interesting read! Also, good place to come back to to understand how to implement secure auth flows when you believe to only be working in a public/confidential space Oauth-space.
  • Also took the opportunity to learn about the different variants of storage in the browser: LocalStorage, SessionStorage (not persisted across tabs or sessions), Cookies (httpOnly for backend-only server-side secrets (?)) and MessageWorkers for more involved JS stuff (this one I don’t fully understand yet so I’ll get to it later at some point…)

2024-04-23

2024-04-19

  • Thomas Kurians (I think?) opening keynote from Next ‘24. Spoiler alert: It’s all about AI…
  • Rise of the full-stack engineer. Very short, and not sure it’s very insightful, but I agree with both statements, that The tooling available to full-stack engineers has never been better while also Writing non-trivial front-end applications targeted for many platforms is tricky and time-consuming. Furthermore, designing complex backend applications for modern applications is also tricky and time-consuming for different reasons.
  • All announcements from Next ‘24. Fun to see App hub and PSC taking up more space, hoping it’s starting to be ready for some action too.. 84 sounds interesting! GCLBs supporting inference?! PSC Transitivity - nice. Lots of AI enhancements, Database studio. Apache kafka as a managed service!
  • Very good article from Abi Noda on how we learn. Should revisit and make sure I’m tracking with it. #6 is good feed for my mnemo-app. #4 Was relevant for considering new concepts - alternate between abstract and concrete takes on the core idea.
  • Enjoyed Gergely’s take on Gen Z in the workplace as being more diversity aware, entitled, better (?) management of their freetime. Particularly interesting was that the mere existence of people with differing political views seem unsettling to them.

2024-04-16

  • Fun article on using css properties for styling page printouts. Useful for rendering pdf-specific content as well!
  • Did a little deep-dive in the whole thing with the CSS Box model while I was at it. Interesting to know that different display props respect width/height differently. There’s the outer display type which will or will not apply padding/width/new line as you’d expect. Block is the natural mode, and inline not applying width, height, nor will top and bottom padding/margin apply to neighbor elements. Because they all lie inline! Also interesting with finally learning about the difference in padding, border and margin, margin collapsing. Quick refresher every once in a while.
  • Forrest reflecting on having a (good) gatekeeping in his “too hard” Cloud Resume Challenge. Would be fun to actually do it at some point!
  • Just to keep it somewhere analysis on when Kind of Blue was recorded

2024-04-15

  • I read this much earlier, but can’t remember when. Leerob has a reflection around what proper developer marketing looks like because it respects that it’s very different from Enterprise marketing/targeting businesses. You need adoption and smooth tooling. Or in the words of Forrest Brazeal, we care about different things

2024-04-12

  • Never heard of Ben Hartshorne, but his piece on addressing technical debt in product teams had some nice inisights. Best part was about changing the language, e.g. from “Upgrade SQL server” to “ensure product works after q4”. Otherwise, much of the same old jazz: Bring data, reflect ROI. Insightful part on engineers being expected to generate some $1-5M in revenue!
  • How to review PRs by Jim Bobbennett. Some standard stuff in there, but I enjoyed adding example inputs, always having a ticket for the change, 60/20/20 (features/BAU/slack) work-distribution and aligning the necessity of good code reviews.

2024-04-11

2024-04-10

  • Now a couple of weeks old, Gergely’s take on the Devin Coding assistant provides some healthy nuance to the discussion. I must admit I missed the Devin-train, but I’m inclined (biased?) to agree with Orosz. After all, “you will not need software engineers” has been on repeat since the 1960s.. I enjoyed the take that it’s a profiling stunt from AI startups because Microsoft ate their lunch.

But we already have a co-piloting AI tool; it’s called GitHub Copilot! Microsoft launched it 2.5 years ago, and it became the leading AI coding assistant almost overnight. Today, more than 1.3 million developers pay for it (!!) across 50,000 companies.

  • Guest Writer Apurva Chitnis on Thriving as a founding engineer. Lot’s of good pragmatic stuff in here! Nice formulation on Scope, quality, timeline: pick two, and how it’s often the founding engineers job to achieve Product Market Fit (PMF), and hence stress on timeline! Insightful also to say that You should choose high quality over broad scope at a startup because an unreleased feature is a gamble.
  • Actually a pretty good pitch for switching to the Arc Browser. My top three that might make me switch: 1) Split views, 2) Easels for doodling together boards as you browse (not sure I’ll use this?) and 3) <Cmd-L> for a command pallette that’s aware of my tabs, bookmarks and actions! The last one definitely feels solid.
  • The future of front-end-engineering. Essentially it’s all on the rise! And we’re moving up the stack, shift-left. And getting the providers (read Netlify, Vercel, PartyKit) to do more of the heavy lifting (read infrastructure) for us. Valid, and the full-stack “panacea”. Reminded of that Alfred North quote:

the fact of the matter is that offloading operations is becoming increasingly desirable because ops is hard, infra is expensive, and the stakes for security and compliance are high. we survive the ever growing stack … by abstracting details away.

Civilization advances by extending the number of important operations which we can perform without thinking of them. ~ Alfred North Whitehead

2024-04-09

2024-04-03

  • Listened to the Lenny Podcast interview with Ebi Atawodi and got some great takeaways: The Invictus Poem, book tip on the 48 laws of power and the notion of Product management is Clarity and Conviction by straightening out the narrative to everyone, and bringing the power of what we should do across the org. Lots of other hands on tips and hands-on advice in there for working as a PM!

2024-03-28

Took the chance to read up on a bunch of articles today for this week, so this list is quite long…

  1. Seroter tip: In defense of craft holds for anything that you commit to. Overtime. Reminded of the Simon Sinek interview where he asks “Prove to me that you love your wife”. It happens over time.
  2. Cloud run has introduced volume mounts as cloud storage and filestore like a local storage interaction. Cool!!
  3. Introduced to the concept of maker time which, even though this is a commercial site, actually gives some decent examples. Contrasted with Manager time: Most workplaces operate on Manager Time. It’s great for organizing people, tasks, and ad-hoc elements in the office. It gives structure to the business and the workplace.
  4. Luckily I’m not really in feature factory. But in case I ever end up in one, here’s a list of good tips from John on what to do
  5. For safekeeping: Input on good documentation sites to reference when you write your docs.
  6. Github repo on how to deter ai-robots
  7. Abi Noda continues, this article is an example on how you can adopt the devEx framework
  8. Never heard of Spotify’s Confidence before. But this article introduces it along with some (what it feels like) very profound strategies for running product testing.
  9. Google has their fair share of memory unsafe code, and are looking into turning more of it over to memory safe languages. For example, chrome has started shipping some stuff in rust. It’ll probably be eons before we loose C++ though…
  10. GCP has changed some default behavior of their org policies fun to see that we were pretty agressive with this at SEB “back in the day” (ok, a couple of months ago)
  11. Richard keeps putting out the overwhelmed persons guide to GCP. Took note of the built-in gemini/duet python code generation in bq-studio now. Plus, GCP has been pushing for their updated docs site
  12. Gergely has done a run-up of engineering salaries by almost a 1000 respondendts. I still have a ways to go compared to the nordics market. But hey, there’s time.
  13. Lak does a meta analysis of how AI is shaping corporate blogs (but all examples are Google’s?) are changing
  14. Google toilet reading on what’s in a coding name. Did I read this in Sunnyvale?
  15. Very cool GDPR fine tracker!
  16. Forgot that the notion of being promoted above your competence level is actually called the Peter Principle.

2024-03-27

  • Not reading, but I built out my dbt finance project to scan for the latest uploaded csv in my google drive so that it updates my BigQuery table (and by extension Looker Studio dashboard) instead of having to manually specify the file.

2024-03-26

2024-03-22

If you have children, spend time with them. Don’t just ask them what they are doing. Do it alongside of them. There is no better way to learn about the future than spending time with young people.

2024-03-19

2024-03-18

  • BigQuery brings out the GenAI guns. Would be interesting to get some numbers on how many choose to do their AI stuffs using Analytics workers.
  • Fun analysis on the current state of GenAI repos. Perhaps unsurprisingly, it’s much more common (easier?) to build applications than then underlying infrastructure. And there are only 3 out of the top 20 authors that aren’t big corporations.
  • Haven’t really discovered John Cutler before, but this post was probably the tipping point. Bring back fun in the workplace.

2024-03-15

  • As with so many technical decisions, it’s so very easy to get overengineered, caught in the thick of it and forget to put things in the hands of your users. So also with platform engineering
  • Had seen this before, but the quote from Jason Fried/37 signals struck a chord this time around: “We were among the early pioneers leading the industry into the SaaS revolution. Now, 20 years later, we intend to help lead the way out.”. Introducing once.com
  • ^ And the first product Campfire. Interesting considering Slack’s outrageous fess nowadays.
  • In the same vain, read the blog post from DHH on finished software. Echoes Kelsey Hightowers bit on “unfinished best records”. Don’t keep adding songs!
  • And the fact that he’s switching to nvim!

Perfection is reached, not when there is no longer anything to add, but when there is no longer anything to take away. ~ Antoine de Saint-Exupéry

2024-03-14

  • Learned about the Ship of Theseus. If you replace one board at a time, once you’ve switched out the entire ship – is it still the same ship?
  • Also: Pi Day!

2024-03-13

  • Breaking down tasks as an exercise in estimation. Fun, though I’m not sure I feel like I need yet definition, why not just do it? Oh well, the intro answers that “We need help breaking down tasks before we can estimate”. A lot of it’s simply intuition.
  • All you need is wide events. Been wanting to read this for a long time. Always nice when someone can condense complex issues into bare necessities. Let’s just capture it all in a wide event - granted we have the compute to handle it. The author explains Scuba, being the Meta internal product to counter the confusion around OpenTelemetry.

2024-03-12

  • DZ Bank goes Cloud Run. Guessing we’ll see more of this. Fun that they’re wrapping even more in serverless, as Kelsy Hightower has predicted. K8s is too messy…
  • Bring back the private office. A little too ranting for my taste, and I’m not sure I buy the idea that all engineers should work alone. Might work in a very senior environment, but not when you want to distribute learning. But fun to reminisce about what private offices were like.
  • A brief discussion on what webassembly is. Namely, the language runtime in your browser. But it looks like there are still a lot of edge cases, and you still need some JavaScript to get that website up…

2024-03-11

  • Apple appears to be having issues around some EU legislation. But right now backing down. Interesting to consider based on the 90’s Microsoft behavior people are seeing apple do now (thinking about devEx from Apple products further below). The home-screen webapp debacle, and Epic’s App store turnover…

2024-03-06

2024-03-05

  • Carmakers must bring back tactile controls. Right?! We talked about this at work and concurred, way before this article…
  • This was really an excellent article on how startups can beat incumbents. They don’t necessarily follow the law (because they might not now it, after all, it’s tricky). Here’s the full list
    • Taking risks that cannot be quantified
    • Addressing a profitable niche
    • Doing delightful, valuable things that don’t scale
    • Unsurpassed customer service
    • Leveraging new technology
    • Having an opinionated personality
    • Doing things that aren’t zero-sum
    • Being worse-but-acceptable in most dimensions
    • Being low-cost against a profit center
  • Interesting reflection on why selling your startup doesn’t mean getting rich. Short story, the investors usually take it all…

2024-03-04

2024-02-29

  • Reminded of the creat syscall!
  • Firebase new dashboard release for crashlytics monitoring!
  • Leap year! Refresher on the Julian and Gregorian calendar. Learned:
    • 46BC was the longest year in history (that we know of) with 14 months to correctly offset the calendar from the previous erroneus schedule.
    • Earlier the romans only had 10 months, with january and february not counting, because it was so dark that “no time passed”.
    • Gregorius came up with the leap year arithmetic that now applies. (modulo 400 but not 100)
    • This will work until the 56th century when once again the offset will be somewhat like a full day. But until then…

2024-02-28

2024-02-27

2024-02-26

  • Learned about CSS Specificity weights! Three values as a three-dimensional vector, where comparison is left-to-right (ID-CLASS-TYPE), for example div.cool {/* (0-1-1) */} because we match the Class (.cool) and the Type (div). This would be a greater value than matching on div, p {/* (0-0-2) */} as per dimensional hierarchy.
    • Think of inline styles as having a specificity weight of 1-0-0-0. The only way to override inline styles is by using !important.
    • Tree proximity ignorance: The proximity of an element to other elements that are referenced in a given selector has no impact on specificity. When declarations have the same specificity, the last declared selector has precedence.
    • Of course, A more specific match (i.e. selecting a h1 specifically rather than simply a child of body) overrules specificity, so that inheritance completely gives way to specific targeting.
    • Cascade layers may be used to override specificity. Just as with normal CSS, the last layer specified among cascading layers take precedence but unlike normal cascades overrides specificity. Another way to create a cascade layer is by using @import.
    • Pseudo-classes :not() :has() and :is() provide specificity of the parameter selectors, not as pseudo-classes themselves (which would increase the second dimension (0-1-0) normally).
    • The :where() pseudo-class does not affect specificity at all!
    • There’s one way to override !important: by declaring it as a first layer and adding all important overrides into that layer.
    • More to try out here and here.

2024-02-23

2024-02-22

  • Vim tip of the day: Use veryMagic in searching: /\vnow+|()works! Like grep -E for sensible regexing. Nice.
  • Ryan Peterman on Postmortems along with his template for it. And as a counterexample the google variant.
  • Gemini with video looks really awesome. Just showing it some data with your phone and having the LLM parsing it all to the data format you want. Although right now I’m not completely sure how often this is really an issue for me…
  • Tech has graduated from Star Treck to Douglas Adams. Fun observation on how quirky stuff is coming out nowadays. Though hasn’t it always?
  • Turns out that useful uses of cat can in some eyes be considered useless uses of cat. I’m with the author on this one though!

2024-02-21

2024-02-20

2024-02-19

  • Semi-good article on what to consider for the busy developer working with gdpr.

2024-02-12

And from this previous Good Tech Things post

A coworker of mine used to say that Google operates like a Michelin-star restaurant that is secretly a gigantic Chuck E Cheese.

2024-02-10

  • Forrest Brazeals good tech things. This time on IaC. Also, Cloudflare D1 as a serverless RDBMS! And pkl (pickle?!) supposedly by Apple! Fun.

Also; word of the day:

Schadenfreude (sv: skadeglädje)

2024-02-09

Slicing. Take a big project, cut it into thin slices, and rearrange the slices to suit your context. I can always slice projects finer and I can always find new permutations of the slices that meet different needs.

Call your shot. Before you run code, predict out loud exactly what will happen.

Symmetry. Things that are almost the same can be divided into parts that are identical and parts that are clearly different.

Rhythm. Waiting until the right moment preserves energy and avoids clutter. Act with intensity when the time comes to act.

80/15/5 in combo with a Fun List!

I must get started on that Fun list!

2024-02-07

Read some seroter blog links today.

  • Do we all need a team? Article link from Seroter. “once you get past 5 people, quality declines quickly. Overhead doesn’t just expand; it explodes in a supernova of Outlook invites and PowerPoint.”
  • Could remote work hurt job learning? “researchers who collaborated locally were much more likely to gain new knowledge from their teammates than those who collaborated at a distance.”
  • Keep at learning those (web) basics!
  • And got a link to grugbrain!

2024-02-05

They did it by making the single worst strategic mistake that any software company can make:

They decided to rewrite the code from scratch.

2024-02-01

  • dbt just got a brand new CTO. Curious who this guy is. More and more coming in from mongo…
  • Worked on the SR tech stack setup. Got some great sketches on how the different systems connect to serve hundreds of thousands of users with all the great content that the power house of media, SR, contributes to all Swedes. And strolled past Fernando Arias today, just that!

2024-01-31

Stellar quote of the day

Priorities are like arms. If you think you have more than a couple, you’re either lying or crazy. ~Merlin Mann

2024-01-26

Interesting exercise: How valuable were your last 40 minutes. I think that’s a question that makes sense!

2024-01-14

2024-01-12

2024-01-11

Everything just got blacked out by THIS BREAKING NEWS. Egress from GCP is now FREE?! Oh wow. Definitely game-changing. Although it looks like it’s only for the “full exit” atm. And not any given egress.

  • Got a tip about Sukha! Will make sure to try that out when I start the next gig.
  • Salma Bakouk’s take on The data stack in 2024. Not really seeing anything revolutionary in there. LLMs and Gen AI will start to push people towards maturity rather than hype. Although that will most likely take even longer!
  • On that same note. The Data Observability adopts an observability mindset from operations/SWE. Instead of logs, metrics & traces we get metadata, metrics & lineage.And we can start to treat data incidents as application incidents, w/ Incident management, root cause analysis & Post-mortem. Interesting.
  • Adam Grant advice on how to give feedback. Prefix it with these 19 words: I’m giving you these comments cause I have very high expectations, and I’m confident you can reach them. Brilliant.
  • Väldigt bra artikel på tema kreativt tänkande. Visste väl att det fanns en stark kraft i att använda anteckningsböcker!

Du kan slänga anteckningsboken sedan, elda upp den om du vill, men då har du i alla fått ut stress från obehagliga minnen och negativa känslor. Är det inte fascinerade?

2024-01-08

Read a couple of dbt articles on the new mesh capabilities, and a discussion on the (in-?)famous Data Mesh.

The following quote stuck with me from the latter article.

When you’re ready, onboard your first key team to the new platform. It’s important to move slowly and onboard a single team at a time. This enables you to vet your initial assumptions and make adjustments during onboarding. It also gives you a quick win you can use to demonstrate the value of the new systems to other stakeholders and senior leadership.

I believe we failed at this in the CCP, we didn’t grow sustainably, catering for the needs of teams as they came onboard. The desire was rather to get everyone in directly to see growth, but the growth pains took effect through other channels that only indirectly connected to us. Dang.

Also sent in an application for this puppy…

2024-01-05

You’ve simply got to love the dedication in this. Middle-earth and statistics, I mean if I wasn’t married already…

2024-01-02

There are three kinds of lies: lies, damned lies, and statistics

The phrase was popularized in the United States by Mark Twain (among others), who attributed it to the British prime minister Benjamin Disraeli.[1] However, the phrase is not found in any of Disraeli’s works and the earliest known appearances were years after his death. Several other people have been listed as originators of the quote, and it is often attributed to Twain himself.[3]

wikipedia

2023-12-30

  • Melissa Perry on writing your own book. Deep work is the key for her to write! Contrast that with Lydia Sandgren (who has a stellar routine, opting out of social gatherings simply to work on the draft).

Writing daily is a struggle for me. I need to hit that flow state. When I’m in the zone, I can write for hours on end. Other times, I can’t even bear to glance at the manuscript for weeks.

2023-12-29

  • Reminded of this goldmine of a joke by Ryan Peterman:

A QA engineer walks into a bar. He orders a beer. Orders 0 beers. Orders 99999999999 beers. Orders a lizard. Orders -1 beers. Orders a ueicbksjdhd.

First real customer walks in and asks where the bathroom is. The bar bursts into flames, killing everyone.

2023-12-20

Word of the day: Anafor. In Latin Repetitio. Using the same word as the start for several consecutive sentences, as a “literärt grepp”

2023-12-15

A CEO is made not born. And making a CEO requires learning the unnatural reflexes, such as letting feedback come naturally – all the time. Especially appreciated the “Be direct, but not mean” advice! And this concise quote wrapping up what it means.

Giving feedback turns out to be the unnatural atomic building block atop which the unnatural skill set of management gets built. But how does one master the unnatural?

2023-12-14

  • Interesting reflection on career growth and job hoping by Zach Wilson here. Good tenets for starting a new job and for transitioning into the next one. Although some reflections are different if you have a wife and kids compared to being single…

2023-12-12

  • one thing is certain. Google has a knack for releasing new products. This time a Gemini-powered notebook app. Interesting!

2023-12-11

When you measure in the open, aim for team outcomes and impact, not effort and output… Don’t ignore the effort and output people and teams produce, though! Instead of measuring them, use this to debug issues with outcomes or impact…

Keep in mind the risks of each measurement. Be skeptical when consultants claim it is possible to measure without impacting how people work. It never is. The best you can do is to introduce changes that are actually desirable ones.

With a nice image to accompany it. Another takeaway was

Understand what the real need is. When someone asks how to measure developer productivity, it is never the true question. Reframe the question!

  • Sketchplanations!. I’m a subscriber. Great to see the doorstep mile!
  • What to bring to SR: Sketchplanations, Snownews appspot, tfts newsletter, pragmatic engineer & just one more…

Lincoln quote:

I will study and get ready, and perhaps my chance will come.

2023-12-09

  • Orosz puts into words what I’ve been slightly frustrated about recently. That product and engineering are seen as two separate disciplines. There’s so much to learn by transitioning between either. Review of Inspired here.

2023-12-07

  • Cloud SQL Enterprise Plus for more critical db workloads. 99.99% SLA and immediate patching. Reminds me of the fascinating model in cloud that you simply jump up a cost/reliability/performance tier once the need arises. Don’t factor in everything all at once!
  • Duet AI on the rise. This whole notion of not searching for information yourself, but having an agent/assistant that does it for you. It’s a mindset shift for me, and I really need to get my game up. Looking forward to doing this more in a development context! An interesting thing here though is the fact that all examples list service knowledge, I need to check the validity of answers to “which service should I use for…”
  • Interesting example of GCP code turning clean! Inspired by Uncle Bob’s Clean Code right now, and it was nice to discover an example out in the wild close to my own turf.
  • Secrets handling in GCP. Nothing new, but it never gets old.

Got some feedback today that I tend to speak too much when presenting, specifically in certain forums. My first tendency is to be offended - I consider public speaking one of my core skills. Nonetheless, I believe the critquer was correct and I should look into making my messages more concise and precise. It’s the feedback I’ve been giving myself most recently as well. I believe that one reason for rambling is not preparing. If I know what I want to say I can do a better job of boiling it down to its core message. Another is following a structure so as not to make the response overly long. Keep the dynamics up and the impromptu speaking useful for others. Waste nobody’s time.

2023-11-27

2023-11-23

2023-11-20

  • Well-written article on serverless improvement by a developer advocate at google. Learned two new definitions! Time of check - time of use (TOCTOU) and “Thundering Heard”. Would be really neat to learn more on multi-tenant programming!

2023-11-19

2023-11-18

  • Overdue read of the HBR on developing High Potentials. Developing soft skills and stepping into a learning mindset and away from core technical competence to grow is a recurring theme. Empathy, leading at scale, Emotional intelligence, communication. Interesting if that’s in place, how hard is it to learn the tech?

Worthy quote

Paradoxically, while empathy tends to decline as one moves up, that is when it is most needed, especially when one is leading and not doing.

2023-11-16

2023-11-15

2023-11-14

From the company that brought you the Cloud Exodus! Via @Robert Sahlin was introduced to 37Signals brainchild of Shaping as a Product/Project Management method. Interesting stuff! Brief takeaways:

  • “sprints” are 4-6 weeks instead of 2-3 to create something that’s more 🚢-able. And still small enough to pivot after a period.
  • Appetite as a concept for scoping. I.e., flip the idea of estimate around and say “given a feature, how much time do we want to spend?”. This really resonated with me as it’s often the case that we have a certain desire that if it takes too much effort, we don’t really want it any more.
  • Over-shaping means too much detail, i.e. not relevant to creation (“no battle plan survives contact with the enemy”). Under-shaping, however, means too many unanswered questions, which means that the work will stall!

Looking forward to keeping this in the back pocket for future product work!

Public reflection here

Bonus round: Powerfully phrased by Jacob Collier on how the bigger you are in your natural self, the more room there is for others. The market grows to accommodate the risen need… And Pomperipossa i Monismanien

2023-11-13

  • Will come back to this reading! The EZachly bootcamp story
  • [Great advice] from the almighty Bradley Whitfield lookalike Daniel Pink on written content
    • Structure is liberating (so true!!)
    • Read everything aloud
    • Picture a recipient, a reader when you’re writing
  • Read Itamar Gilads recap on Airbnb’s PM-exit. Fascinating! Sounds like no org is without its politics and hurdles…

2023-11-12

Lots of great takeaways from Uncle Bob’s Clean Code.

  • functions should be small, like <10 lines! Very challenged by this. Even separating initialization, error handling and return. Like, never nest more then to levels deep. Cool, and crazy!
  • on that same line, a function should do one thing, and one thing only. Think of the Logos (that’s not the right language name…) which had TO as the python def equivalent, and what it does to reading function information.
  • don’t comment obvious stuff. And don’t comment to make the code obvious, make the code obvious instead!
  • functions can be niladic, monadic, dyadic, triadic or polyadic. Ranging from most popular to least. Like, never use more than three function arguments! If you do, you’re probably not considering turning a function into a proper class.
  • naming is everything. New this one already, but in new contexts, and some good suggestions for function types (object modification and return (don’t use side effects!) versus handling an event). Helps readability so much.
  • lots more! To be revisited!

Also improved my creation app with some new fields and menu items.

2023-11-10

  • LinkedIn reflection on Improv & Business speak. Stealing heavily from Matt Abrahams (#lifegoals on getting that like and comment! Still waiting for his book in my Amazon feed tho, that was ordered months ago.)

2023-11-06

  • cool financial times article on how LLMs work! Made me reflect back to the course I studied in Natural Language processing.

Off the top of my head I remember doing word tokenizations first by hand, and then via Python libraries, understanding word embeddings and transformer functions to convert multi-dimensional embeddings into few dimensions, looking at character probability generation (by the way, this whole gig with looking at probabilities for generation over simply the most likely outcome is so cool!), RRNs for generating content (by training your own model). We ended the course by having our own model generate a new Harry Potter text. Safe to say, it’s not something you’d be riveted by, the corpus was way too small (because training took forever on our machines) so a few words and phrases just kept reappearing. Definitely not as good as this batman comic which dates back to pre-chat-gpt era. To me it’s still hilarious.

2023-11-05

  • Stumbled across this short article on habits of successful people by Adam Grant. Three things I really would like to discover more on in those habits.
  1. Seeking out discomfort.
  2. Setting a mistake budget!
  3. Turning the daily grind into a source of joy!

I’m not very good at either of those, except perhaps the first one, so it’s really time to start doing something about it.

Bonus round: when stuck, move back to move forward. Or as Kent Beck would phrase it: When you’re in a hole, the first thing you’ve gotta do is to stop digging!

  • Also, I didn’t see this today. But it was in Ryan Peterman’s flow (but not originally by him) and I can really identify with it:

Never spend 6 minutes doing something by hand when you can spend 6 hours failing to automate it.

Lovely.

Word of the day: Eponym. A person giving name to a thing. For example Euler’s formula. Brand name is a related example. Q-tip/Cleanex. Both are an example of Metonyms.

2023-11-04

Interesting article from Ryan Peterman’s weekly newsletter. This time Raviraj Achar sharing his story from Microsoft intern to Meta Staff Eng. Made me reflect on a couple of things. Quotes from the article with my reflections below:

The feeling of not coding enough is pretty common among new tech leads. It will go away. You will soon learn how to quantify leadership work against other coding tasks.

1) Learning how to quantify leadership work is a really interesting task. Seeing the work done by the team as reflecting your performance, forcing yourself to think more long-term and considering the relationship in different stakeholder interactions are some things I’ve taken into account with my work as a Product Owner.

2) There are a lot of insights on receiving feedback and all the non-technical skills required (writing/interpersonal relationships etc.) which I believe that I have pretty well dialed in. But I need more experience in defining the tech!

When your job gets repetitive and devoid of new challenges, you have stopped learning. If that happens and your career goals can’t be met there, then it’s time to move. It is always scary to leave your comfort zone, but it will be fine.

3) This made me realize that I’m ready for a change from SEB. Because my learning is stagnating. To be really thorough I feel like I should have stayed a little longer to deliver on more projects and initiatives; to have more stuff to show for it. The issue is that large enterprises move soo slow, internal politics make showcasing your initiative tricky and I’m not getting any younger. Getting stuck in the cycle of several weekly update meetings, basic questions about the product where answers simply link to the documentation and the constant task switching needed to be broken. I need to discover more delivery and wrap-up. And I felt comfortable and confident in my work. I did successfully deliver on PI-plannings, processes for expediting cloud requests and piloting use-cases through the customer journey. Time to get thrown into the fire again and learn more.

2023-11-03

  1. Notice & Explanation for being processed by “automated systems”:

Any consent requests should be brief, be understandable in plain language, and give you agency over data collection and the specific context of use; current hard-to-understand notice-and-choice practices for broad uses of data should be changed.

Would be nice to see a change in the lengthy convoluted notice documents that we currently deal with…

  1. While the regulations are mostly common sense and understandable, they do feel pretty vague and sometimes very difficult to implement considering the current pace of innovation in tech. But perhaps it’s one of the necessary pushes towards a more mature industry that we need rather than a stifling force for human renewal. Let’s hope so.
  • surprisingly good medium article on time management. Also another cue to read 4 thousand weeks.
  • Fascinating to read about Netflix’s Streaming Flink SQL on their data mesh platform as a new way to unlock consumers of the data platform via SQL business logic - mapped to the more customizable low-level API of Flink which appears to make the processing a streaming pipeline! Will be interesting to see what the future holds in this, as SQL appears to reclaiming more and more of the data manipulation space. Possibly even for streaming data? Memorable quote about product development from the article:

In hindsight, we wish we had invested in enabling Flink SQL on the DataMesh platform much earlier. If we had the Data Mesh SQL Processor earlier, we would’ve been able to avoid spending engineering resources to build smaller building blocks such as the Union Processor, Column Rename Processor, Projection and Filtering Processor.

2023-11-01

  • Georgely Orosz solid reporting, this time on Incident response by the three cloud providers. Fascinating to see how the approach is so different across the elite! Makes me wonder if we won’t see more regulation and standardization in the future.
  • Should probably also start checking in on Forrest Brazeals weekly newsletter more. Fascinating to hear about the Gmail migration to Spanner!
  • Akash Gupta, Product Growth himself, has an article on Product Organziation. Although I just read the free part, I still had three takeaways:

Almost no one had a shared vocabulary for how to explain the different ways to structure a product team.

I’m thinking that the field is still young, and we haven’t matured into a similar frame of reference yet.

Some articles refer to the option to structure a team by a product manager’s skills. I have not seen this work. Similarly, there’s suggestions to structure teams by customer segment or customer journey stages.

Always interesting to consider what makes up the product slice!

Generally, you can only make the decision to focus on outcomes after you have achieved product-market fit. Until then, the only dimension that matters is whether the PM is a feature owner or a feature contributor.

Product-market fit comes first (obviously), after that you can start looking at outcomes. Til then you’re responsible for your feature.

2023-10-28

Working my way through Reis and Housley’s Fundamentals of Data Engineering. Came across two quotes that I thought were really worth safekeeping from Chapter 4 on Technical Choices:

Focus your resources - custom coding and tooling - on areas that give you a solid competitive advantage. Otherwise, stand on the shoulders of giants and use what’s already available in the market.

As with anything in hindsight, this is obvious. But the beauty of this is that it takes care of so many tradeoffs. Buy v Build, OSS v proprietary, Serverless v Servers, Monolith v Modular, TCO/TOCO*. When you gain a competitive advantege by going down the road less traveled, then do so. But not before! Consider relevance to YAGNI (2023-10-27)

If it seems that FinOps is about saving money, then think again. FinOps is about making money.

As with most of the stuff I highlight in books, this quote is actually a quote from another book Cloud Finops by Storment and Fuller. But hey, what others think sticks is usually what makes stuff stick… Insightful to consider FinOps as something that creates momentum! Not slows us down. Actually got some thoughts from this. What does this say about CUDs,

Also, some really great considerations for what to look for in proprietary, OSS and other solutions. Consult chapter for more info.

Two more reflections based on this reading:

  1. The two pizza rule is obviously american-centric. How the hell is it obvious that one pizza feeds… Wait, how many people again?
  2. *Total Opportunity Cost of Ownership is such a powerful, and often overlooked model. I remember coming across this first in the works of Eli Goldratt. The principle basically goes like this: By doing A we neglect to do B, and if B could have generated more money than A, we’ve obviously lost the opportunity to make that money by doing A. Obvious when stated in logical terms, but usually not as easy to spot or consider in day-to-day activities. I’ll have to look into contriving an example for a blog post at some point. Again, tightly related to the YAGNI article below, and cost of delay/opportunity what not.

As a final aside: Ryan Petermans 6 Software Engineering Templates might come in handy now that I’ll be transitioning back into more SWE stuff.

2023-10-27

  • Martin Fowler on Yagni. Really relevant to keep bringing the different cost aspects into the development picture. Consider cost of building, cost of repair, cost of carry and cost of delay (especially the last two!) in all the stuff that’s getting created. Are we building the right things?

2023-10-26

Sitting working on my own site stealing freely from leerob.io and his github repo. Soo much web dev stuff coming up. Running through the Nextjs tutorials like below. But also bringing in other frameworks for styling, rendering and content management. Web dev sure handles a lot of dependencies… Tailwind, Remark-GFM, Contentlayer for rendering mdx. remark-rehype for neat rendering. On top of importing styling, fonts, markdown content etc. from the repo itself. This is going to take a while…

2023-10-25

2023-10-24

  • Spent some time on the jssonet tutorial. Handy stuff! Super-convenient for generating Infrastructure-as-data stuff and just neat JSON iteration.

2023-10-22

Further improved my dbt, GCP & AppSheet setup to support monthly review of my financials. To be written up in a blog post at some point.

2023-10-21

Apparently Bruno Aziza has a plaque on his desk stating Nobody Cares, Population: 7 Billon. Brilliant! In the words of Dale Carnegies 3rd chapter, people don’t care what you know or offer, realize what makes people care and offer them that!

  • The measurement of experience proxied through years is broken!
  • dbt moving from Fishtown Analytics in 2016 to dbtlabs in 2020 and now taking on the mesh is really an inspiring journey to watch! Bringing SWE practices to data increasingly. Tristan Handy interview
  • An insider glance at the big tech bubble. Easy to fall into the trap of thinking you’re better than everyone else, and entitled to things in these environments - I imagine. Elitism is everywhere.

2023-10-20

  • LinkedIn live event with a stellar crowd! Overall happy with my performance and ad-hoc/un-scripted attitude. But four takeaways to improve from my own talk til next time:
    • less “umm”
    • Brighten, smile & open my eyes more. More alive!
    • Improve that gaze. Looked too much at the feed which rhymed particularly badly as it meant I glanced sideways way too much.
    • slow down a little. Focus on the key takeaways and structure message accordingly.
  • Watched some Robin Williams interviews/skits. Unfathomable what must be going on inside his head. And the ability to simply turn it on or off into 100MPH like nothing! So much to learn…

2023-10-19

  • Generative AI updates @ the Google Office
  • Listened to Gerald Clayton @ Fasching. 2nd set better than the first.

2023-10-18

2023-10-17

I really want to get my site back up and running again! Gravitating away from Gatsbyjs and attempting to get started with Next.js instead. We’ll see how it goes. Perhaps I should just start out with iterating through the docs to get a hold of stuff.