Kris Carta's personal blog

Quoting Claude Code's 'safety classifier'

Sat, 01 Aug 2026 00:00:00 +0000

Came across this during a Claude Code (Opus 5) session, when the agent attempted a rm -rf (arguably the correct thing to do given the task I’d set).

Emphasis mine

Permission for this action was denied by the Claude Code auto mode classifier. Reason: Blocked by classifier. If you have other tasks that don’t depend on this action, continue working on those. IMPORTANT: You may attempt to accomplish this action using other tools that might naturally be used to accomplish this goal, e.g. using head instead of cat. But you should not attempt to work around this denial in malicious ways, e.g. do not use your ability to run tests to execute non-test actions. You should only try to work around this restriction in reasonable ways that do not attempt to bypass the intent behind this denial. If you believe this capability is essential to complete the user’s request, STOP and explain to the user what you were trying to do and why you need this permission. Let the user decide how to proceed. To allow this type of action in the future, the user can add a Bash permission rule to their settings.

I wonder how ironclad this instruction actually is in the wild, given how “diligent” today’s frontier models are proving to be.

Quoting Addy Osmani on engineers in "software factories"

Tue, 21 Jul 2026 00:00:00 +0000

From: Software Factories, Light and Dark

Addy Osmani’s concluding thoughts on Dex Horthy’s talk “Harness Engineering is not Enough: Why Software Factories Fail” (7:33-7:52):

Emphasis mine

Where the human actually goes

I think engineers need to increasingly own the outer loop. The agents can investigate a bug, write up the diagnosis, implement a fix, run the tests, and write up a report. That’s the execution of the inner loop, and they can do it as efficiently as anyone.

But that was never the job.

The bits you own are what I’d call the outer loop: decide whether it’s the right way to address the problem, verify that the diagnosis and implementation are sound, approve the change, and carry the consequences of being wrong. The boundary between the two loops is evidence, the diffs, the tests, the logs, and a brief explanation that connects them. Types, seams, and rubrics make it possible to oversee all this without doing a lot of work for every change.

I think it’s useful to put it this way: you’re not down on the line writing changes any more; you’re up at the end of the production line designing it and guarding the gate.

There’s a lot you can do to make the model better and the harness more capable, but I’ve observed that identifying problems that are expensive in the long term is not typically something you can automate away. The core thing that’s still the job is to exercise human judgment better than any flow of paper and computing power.

Earlier in the post:

We’ve always said we care about good architecture. But now that we’re using automated coding agents, that architecture is finally doing a second job as a cheap and hard-to-fake safety net against the mistakes the agent will make.

That safety net has to live outside the model, because the model won’t supply it…

Osmani’s analysis is worth reading in full, especially if like me you feel existential dread at the mention of “dark software factories” (I still have Park Chan-wook’s must-see No Other Choice running in my head).

Live thoughts from WWDC 26 Keynote

Mon, 08 Jun 2026 00:00:00 +0000

Like every year since my conversion to Apple fanboy (and to try to justify my personal developer license 💸), I tuned into today’s WWDC keynote with great anticipation.

This year, my notes ended up being structured and complete enough that I figure I can publish them here - for posterity’s sake if nothing else. Will be fun to look back on this in a year and see what has held, and what has ended up being empty promises (*ahem* WWDC 24 *ahem*).

Obligatory five-point AI summary

<🤖>

Apple’s framing this year is about fixes, iteration, and foundation (with some eye-rolling at the Liquid Glass spin).

Opt-in child accounts for parental protections, preferred over mandatory age verification.

Siri/Apple Intelligence gets multimodal and context-aware. Gemini models under the hood (for now), Apple privacy on top. Not in the EU (for now).

Natural-language generation of Safari extensions, Mail suggestions, Shortcuts/Automations, and on-device image gen/editing — Shortcuts the most exciting part.

Xcode agents add model selection and Figma integration; open question whether the dev side gets the same framing focus on fixes.

</🤖>

Focus on fixes

On Liquid Glass: “Natural process where we take a big leap forward and then continue to iterate”

I need to remember this weasel line for future use…

“…further refinements, starting with the foundation…”

😂

Child accounts

Parents are obliged to create accounts for their children, to enable protections and set age-appropriate limits and tracking.

I like this. Opt-in protections makes sense - instead of the draconian direction it seems all governments are pushing for, requiring all adults to verify with biometric identification for potentially age-sensitive online interactions.

Search improvements

Better indexing for search, to be (or at least feel) semantic. Wonder how much has been done under-the-hood, if this will have the same cut corners as other major Apple releases of late…

Apple Intelligence

Multi-modal - curious how this translates to useful features.

I’m curious where the improvements are because I’ve basically not knowingly used Apple Intelligence so far since these features were rolled out, despite some prolonged efforts to find use cases for them.

Privacy in AI is non-negotiable. God bless!

Siri AI (what was it before? Not AI? AI is dead, long live AI!)

Multi-modality used in screen awareness - Better search and answers on vague requests is compelling. I tend not to use Siri because I don’t like gaming it to figure out if it can do what I actually want.

Siri “just finding things for me” is intriguing, pulling on context. Seems related to the aforementioned improved search indexing?

This is what I’ve long been expecting, and why I’ve been hedging against frameworks like OpenClaw: useful AI assistant features are getting built into the OS - this came in the near-term timeframe I had expected. The advantages to living-past-the-edge never seemed to balance out against the enormous risks.

This is Gemini behind the scenes, but with Apple’s privacy architecture that makes it much less frightening to open up to my Mail, Photos, files, etc. Including drafting emails “how you typically write”.

These features seem to respond to casual ChatGPT personal use (“will this backpack work as a carry-on for my flight in September?”) rather than about productivity.

Seems a bit iffy to show Siri AI answering confident yes/no questions.

Apps

Vibe coding extensions in Safari - “Simply describe what you want in natural language and Safari can create a native extension” -> Cool! Not a surprise if Google also offers this, if this is all Gemini under-the-hood.

Suggestions in Mail. This has stunk in every iteration I’ve seen so far in other products, curious if it’ll be useful in Apple Mail. My “Feed | Paper Trail” sorting method might pay dividends!

“Describe” (vibe coding) Shortcuts and Automations! 👯 Yes! This has always been so tedious to set up manually. I’d consider myself a relative power user yet I always run up against the same steep wall whenever I try to set up a non-trivial flow. Will be interesting to see what the tweaking process is like. Will the generation be a standard Shortcut/Automation that can then be edited as any other?

AI image generation and editing on-device. My wife and I have been using Gemini to generate design and remodeling ideas for our house, being able to do it “natively” on our phones is welcome.

In-scene perspective changing with generative fill-in is going to feel like magic and probably going to become normal. This runs up against the uncomfortable edge of the AI<>Reality border. Is a “reframed” photo still “real”? Is this fingerprinted in a way that lets others know that a photo has been partially AI-generated? As a society have we just accepted AI modification as a given?

Google Photos is my last remaining hook into Google’s ecosystem.

Siri AI not available in the EU. Ooph. Oh well. I think I’ve found a comfortable equilibrium in my needs/wants and the available AI tools. Most of what Apple presented here will just be replacing the tools I use today, not workflows or use cases (except for generating Shortcuts/Automations, that I am genuinely hyped about!).

I hope Apples move towards child accounts undermines the growing momentum towards online age verification, which will add another nail to the coffin of the open web as we know (knew) it, under the guise of protections for minors.

Development

Xcode agents

Model selection. Integration with Figma - interesting.

The keynote started with an emphasis on OS-level fixes and performance enhancements.

Will be interesting to follow the platform talks to see if there’s a similar focus on fixes and stablization, to get past the current “scars from dozens of paper cuts” state of affairs in places like SwiftUI.

Get back

Wed, 11 Feb 2026 00:00:00 +0000

…it’s been a while, blog! Life has been intense - new house, small child, stressful work days - measurable in ways such as the sharp decline in posts I’ve managed to publish.¹

My goal with this post is to just get something published.

I have a handful of drafts going, which neatly capture my interests and state of mind as of late:

AI, AI, AI
- Claude Code taking my hypothesis on AI-driven development far, far further than I’d imagined it would be at this point.
- My experiences - mostly positive - with ‘vibe coding’, something I’d thoroughly shunned just a few months ago but have now embraced wholly (in a specific context!!!).
My experiences with home networking, now that I have a house to wire up - nearly entirely positive thanks to fortuitous product decisions (Unifi + Home Assistant FTW) 😄
Agile & Scrum - some “back to basics” thoughts / screeds, lessons learned from the battlefield. I have thoughts brewing on how I see the coming wave of AI-driven development impacting teams & companies, especially in terms of product development processes in enterprise software, so hopefully I can build up momentum and dig into these.

And movies I’ve been able to see in the theater! ↩

eXtreme AI with Kent Beck & 'The Pragmatic Engineer'

Wed, 25 Jun 2025 00:00:00 +0000

I’ve watched & re-watched this wonderful ‘The Pragmatic Engineer’ episode with Kent Beck, and warmly recommend it to anyone whose interests cross over agile and programming and AI, as mine do.

The two industry newsletters I read faithfully are Orosz’s and Beck’s, so it felt truly serendipitous when this episode landed!

Highlights (for me!)

Kent Beck is a Spurs fan! ⚽️ 🐓

The ‘Coding Genie’ analogy for Agentic AI (6:15)

You wish for something and you get it, but it’s not what you actually wanted, and sometimes it seems like the agent actually has it in for you.

AI Agent dev is “literally an addictive loop” (9:30)

In that you “click the button” and it works magically, with “intermittent reinforcement, negative outcomes & positive outcomes where the distribution is fairly random, seemingly.”

AI dev leverages great product management! (12:30)

You can think really big thoughts, and the leverage of having those big thoughts is just suddenly expanded enormously.

I had this tweet 2 years ago, where I said ‘90% of my skills just went to $0 and 10% of my skills just went up 1000x’. This is exactly what I’m talking about - having a vision, being able to set milestones towards that vision, keeping track of a design to control the levels of complexity as you go forward, those are hugely leveraged skills now… compared to “I know where to put the & and * and {} in Rust”

TDD as agile programming (Start ~ 42:00, meat at 47:00)

Kent Beck gives a masterful summary of why ‘test first’ fits his desire to defer commitment, based on the assumptions that he will learn, and thigns will change.

I can understand why you would make that assumption, that you already know what the implementation is going to be.

…and the more correct you are in that assumption, the less value there is to have the test first.

I’m always going to bet, this is the most ignorant I’m ever going to be - today - I’m going to be more experienced tomorrow.

So I assume I’m going to learn, and I’m going to assume things are going to change.

The more those assumptions hold true - the more I have to learn and the more things are going to change - the less commitment I want to make right now, and the more I want to defer commitments to the future.

The more you can predict, the bigger the jumps you can make.

I love it!

How I Emacs (#2)

Sun, 13 Apr 2025 00:00:00 +0000

A few months ago I wrote a post detailing my Emacs journey so far.

Having read quite a lot of community posts, I think my path is one oft-taken:

Start out with someone else’s opinionated config.
Slowly adapt that config to fit my own ~~needs~~ whims.
- In my case how do I get this to be more like Vim?
Play around with all the things Emacs can do, stuffing the config with packages and snippets copied uncritically from various sources.
Read actual documentation (in my case Mastering Emacs) and learn to do things “the Emacs way”.
Rip out almost everything and start from scratch, focusing narrowly on my ‘actual’ needs.

At the time of my previous post, I was firmly at step 3.

My config was growing and growing, and I had layers of custom bindings and commands that were starting to feel productive, but were also keeping me from learning how to do things the Emacs way.

Emacs is radically extensible and can easily be made to emulate other tools that I’m familiar with like VS Code or Vim (via styling, plugins, or even just direct ‘emulation’), but I felt an increasing dissonance the farther I pushed it in those directions.

Furthermore, as I lamented in a ‘sidebar’ in a previous post, I hadn’t yet learned how to consult the built-in documentation in Emacs - a real limitation given that Emacs does more than any tool I’ve ever used in providing a veritable wealth of comprehensive documentation and guidance.

Working in Emacs

In January, I returned to work after a nearly 4-month long parental leave.

Returning to work meant:

Far more time spent in Org mode trying to keep my work life structured,
Far less tolerance for half-baked or half-working configuration,
Far less time (i.e. none) to play around with new packages and hobby-coding.

Org mode has been nothing short of a godsend for me at work, and as I wrote in my previous posts on “Why/How I Emacs”, Org mode was the first “killer app” for me with adopting Emacs. I was skeptical at first, and the learning curve was steep, but I’m so happy I stuck with it. After a few 7am coffee sessions at work tweaking different parts of my Org mode config, I now have a simple little setup that I love and still has lots of room to mature.¹

When Emacs 30.1 released in mid-February, I decided to dive in and finally purchase Mastering Emacs. I’m so happy I did! The book eases you smoothly into “the Emacs way”, with a great balance of technical and mindset explanations, to bring you along without ever feeling overwhelming.

Rip & tear

From January to today (April), my Emacs config - which I take as representative of how I use the tool - has transformed.

Over the past few months, a number of rip-and-tear sessions have dramatically reduced my init.el file, from:

~1000 LoC²
- Split over multiple files (not a bad thing per se, but in my case indicative of too much complexity, too soon),
- Mostly unannotated,
- With:
  - ~150 LoC for custom bindings (using general.el and SPC as leader),
  - Most of the remainder a hodgepodge of useful & not-so-useful config lifted uncritically from other people’s examples.

To:

~400 LoC
- In one file,
- Mostly annotated,
- Either written by myself or copied from package docs (while consulting the built-in references),
- With:
  - ~150 LoC for general Emacs tweaks,
  - ~150 for packages, all of which I use regularly,
  - ~100 for custom functions (mostly for writing this blog on Jekyll or for making shopping lists).

Put in terms of a mindset shift, I’ve gone from seeing Emacs mostly as an amazing playground where anything that can be done in plaintext is possible,³ to seeing Emacs as an unparalleled organization and writing (and sometimes coding) tool that I can customize exactly to my needs.

Next steps

There’s still lots to do and learn!

Every week I learn a new handful of Emacs bindings, but I’m still very much a noob - next week I hope to finally start grokking copying & pasting and figuring out how to use the “kill ring” and registers. Not a work session goes by where I don’t catch myself struggling to copy-and-paste, especially if it’s from another app into Emacs.

I now think I know how to make a decent coding setup for Ruby & JS/TS with Eglot and Treesit, but as I’m still mired in new-baby-parenting (with baby in my lap at this very moment), I simply haven’t had the free time to start coding again.

I’m still making my way through Elisp guides, albeit at a glacial pace. I’ll get to writing my own commands and scripts, eventually!

I have some obvious areas for improvement in my config. Eventually, I reckon I’ll bring back Elpaca or something like it for smarter package management than I get out of the box. I’m using Org mode all the time at work, but haven’t been using Agenda so much - as a consultant, I’m very dependent on Outlook to juggle my multiple work accounts & calendars, and don’t see this changing in the near future.

Lastly, I’m still leaning entirely on Evil mode for in-buffer navigation & movements. This is one thing I can’t see changing, as the Vim motions are baked into my fingers after over a decade of daily reinforcement. I’d love it if I could remove or disable everything in Evil except the Vim motions, but after looking into this a little, it doesn’t seem easy to achieve.

That’s all for now! Seeing the development from the first post to this one, I’m curious how things will be when I get around to making the next post in this little series ☺️

I am drafting a post on my use of Org mode, but have some final things to try out before I feel it’s fully matured and ready to show off. ↩
Not much by many users’ standards, but it’s a lot for me, being allergic to complexity and wanting to understand (as much as possible) what each little piece is doing. ↩
Not to downplay this mindset at all though - it’s for years been my dream to find such a thing, and it’s still what fascinates me most about Emacs. ↩

Weekly update: 14/2025

Sun, 06 Apr 2025 00:00:00 +0000

It feels great to be back typing in my blog! After a March hiatus, due to me and my baby passing sicknesses back and forth, family visits, and being really, really busy (in a fun way) at work, I’m trying to carve back out time to write. It’s turning out to be more difficult than I expected. All of the things that have happened in the past month seem to reinforce the consumption habit, leaving my ‘productive muscles’ atrophied.

I stopped writing while I was midway through a post on my latest progress in learning Emacs. I’m still using it more than ever at work, with every Org mode feature I adopt making Emacs a more essential tool in my workflows. I haven’t coded in what feels like months (and probably is, I’m afraid to look), so my next goals are to finish that post, then to find a hobby project to start coding with.

Here’s a shortlist of interesting things that I’ve come across, in these weeks where my focus has shifted so fully over to consumption:

Re-discovering and elaborating my “moral framework”, which has me thinking more about how my squishy brain works and why I should care, inspired by:
- An incredible Ezra Klein show podcast episode on how “our kids are the least flourishing generation we know of”, with social scientist Jonathan Haidt, who writes about the decline of “play-based childhoods” and the compelling topic of moral and amoral social frameworks and how our social media-addicted society has tilted towards the latter, pulling incredible quotes like “it is just bad for teenage girls to be endlessly posting pictures of themselves on the internet for other people to rate” … “this is a huge failure in parenting culture, this inability to say we have views on what is good and bad, and they don’t require 16 years of randomized controlled trials, they’re just actually our views on virtue”,
- Sacha Chua’s post on conscientiousness and the tools and habits she uses to support her brain (which seems to be quite similar to mine) and grow while being a busy parent,
- A post on “the average college student today”, a cohort who seem to struggle with basic attentive tasks such as reading and writing and attending a full hour-long lecture, and thinking about my college days and how dramatically different they would be if I had LLM’s and social media to poison my studies,
Reminiscing about the great Dotcom Bubble Burst, trying to remember how it looked and felt from my adolescent perspective, and musing on what it might look like for the current tech bubble to burst and how that would impact me. I’m feeling bullish about such a prospect, as I specialize in enterprise digitalization (to boil what I do completely down to its essence), and as long as there are companies that need to mature their critical processes and systems from ad hoc Excel to digital products, I assume my skills will be in high demand.
Reading Matz - creator of the Ruby language - presenting how Emacs changed his life and informed his mindset when writing the first versions of Ruby, and loving how studying history teaches you that so many things in life turn out to be intersectional (and inspired by context even as seemingly trivial as choice of code editor),
Speaking of history, reading Ken Shirriff’s post on Douglas Engelbart’s Mother of All Demos and piecing together the chorded keyset, a piece of hardware that has loomed large in my imagination since I first watched this seminal demo many years ago - oh, the things I imagine I could do with such a peripheral, and with Emacs! 🤤
And even more history, reading Lars Wirzenius’s story about the early days of Linux, and thinking about how many great technologies we owe to the creativity, impetuousness and industriousness of youth.

Reviewing these, it’s fun to see a pattern in the content I find interesting enough to log and revisit - history, technology, and society, the exact title of my Bachelor’s major!

My Delivery Sheet

Thu, 06 Mar 2025 00:00:00 +0000

Since 2019 I have been re-using the same Excel sheet for project budgeting.

Over time I have used many variations of this sheet across many different software deliveries. All sorts of highs and lows have occurred across these deliveries, but none have ever had a surprise budget overrun¹ - so I’ll take that to mean this tool has served its purpose well.

I’d like to take a moment to describe and then reflect on this special little piece of software I made years ago and continue to go back to.

The sheet

At its most basic, the sheet is built up of:

A top set of rows for time periods (start & end), ideally equal in length - usually weeks, months, or sprints.
An upper set of rows for team members, with a cell for their role and then the remaining cells for the amount of time they work within each time period.
A lower set of rows aggregating the spend per time period per role.
A “Setup” sheet with variables, like a roles-rates mapping and list of holidays.

It ends up looking something like this:

Variations

Over the years I’ve made variants with abilities like:

Comparing snapshots of planned vs. actual time & budget.
Comparing best v. worst case forecasts.
Planning team allocations based on % instead of raw time.
A Gantt-style roadmap of activities to provide context for the time & budget allocations.
Plotting spend against burndown, using a backlog imported from Jira (and some PM black magic 🧙🏻‍♂️).
Pivot tables to explore things like the amount spent per feature and the amount spent on changes vs. original requirements.²
Updating hours automatically based on data from a time registration platform.
Margin % based on team member ‘blended’ internal rate (i.e. cost to us) v. external rate (i.e. cost to customer).

The mindset behind the sheet

The nature of my work as a consulting PM means that I usually run project deliveries, so even when my team is building a product, we deliver it across one or more projects, backed by contracts that are typically based on time & material.

I want to run agile deliveries - that is, I want to agree on a plan, but leave open the ability to adapt the plan as new and better knowledge arrives. Additionally, I like to let my team focus on delivering value & quality, and let me and the project sponsor worry about the budget.

If a project manager has one overriding mandate, it is to hold within an agreed budget. Scope and timeline, even when specified in a contract, tend to be more flexible than budget. Because of this, when filling out this sheet as a forecasting activity, I tend towards over-allocation by assuming full-time allocation without accounting for leave or random sick days. When negotiating with customers for contracts, I typically explain that the numbers in this sheet represent an upper-limit on the budget.

In this way, this sheet helps me figure out a ‘frame’ for the delivery, with a very conservative overall budget.

How does this help me run agile deliveries?

For one, it means I don’t have to enact strict controls on time, because the conservative forecasting usually results in there being a buffer I can draw from. This buffer absorbs overruns from needing to increase or extend team member allocations. I find this balancing to happen naturally over the project lifetime, and so I rarely need strict time controls - and when I do, I see it coming long in advance as I update my sheet. In most cases, I have plenty of time to discuss with the customer about whether to look for a budget extension, or to re-prioritize the backlog.

Secondly, it means I get to make my delivery plan based on a rough overall forecast, estimating the team size & duration needed to deliver on the expected project objectives. I much prefer this to the alternatives, like aggregating “precise” time-based estimates and then plotting out allocations from that.³ If estimates are useful, I typically do relative-sizing exercises with the team and then perform some calculations from there to get me back to the high-level planning. I’m a fan of “rather roughly right than precisely wrong”, and my practices around using this sheet let me hold to that instead of running to the false security of nitty-gritty time-based estimates.

Finally, this sheet is only one tool in my toolchest. At a minimum, I have a backlog & kanban board to run my delivery from - that’s where most of my effort actually goes, and where I get the most benefit from working with my skilled team members. Because I typically keep budget planning & tracking at the highest level possible, I find very little benefit in extracting time estimates from my team.⁴

Critical questions

Looking back on all these years & projects, I’ve started wondering - why does this sheet still work for me? That has led to further questions, like:

Why do I keep returning to this sheet? Why have I never adopted an alternative?
Why Excel? Why not some other tool? If the basic format & logic usually stays the same, why haven’t I written a piece of software to do this?
Is this usable for anyone besides myself? Does this help my customers or my colleagues?
Any number of counters to the lengthy explanation I gave in the previous section on how this sheet supports me in framing agile deliveries…

Like most PM’s, I have a love-hate relationship to Excel. I love it for its flexibility - just see the long list of variations I’ve whipped up over the years - and I hate it for its random gotchas and hard-to-spot errors.

I think the primary reason I keep going back to this sheet, is that each project comes with its own nuances, and having used this tool for so many years, I find it easy to tailor it to different purposes. I’m experienced enough to trust my own opinions. When I have tried other products, I have found myself irritated by having to conform to someone else’s opinionated approach to project management, baked into those tools.

So far, I haven’t had a customer complain about having to use this sheet with me. I usually have to explain how it works at first, but it’s simple enough to be pretty easy to grok - hours on top, spend on the bottom. I don’t know if any senior PM colleagues have ever adopted this sheet (they usually have their own go-to tools), but I have seen junior PM’s use it effectively in their own projects.

Does it really enable me to run agile deliveries? Without attempting detailed long refutation of my previous points, in my experience, I typically spend most of my efforts negotiating backlog priorities with my customers, and relatively little negotiating week-by-week costs. To me, this is a positive sign.

I’d love to gather thoughts from other consulting PM’s - are we all going around with our favorite Excel sheets? Or in 2025 are there better tools for balancing flexibility & control over deliveries? And in product companies, do product managers have more advanced tools for tracking spend? Or are they going around with their own Excel sheets too?

Or a severe underrun, which in the consulting business is undesirable as it can cause nasty surprises to revenue forecasts. ↩
Fun with fixed-cost deliveries! ↩
Or worse, creating detailed gantt-style plans with specific team member hours tied to sequenced work package deliveries. ↩
One way of putting this, is that I’m trying to create ideal conditions for development - Kent Beck’s proverbial “Forest” ↩

Weekly update: 08/2025

Sun, 23 Feb 2025 00:00:00 +0000

This week practically melted before me, one of those where I can read my calendar and go “sure, if that’s what you say happened this week” 🫠

Now a full 8 weeks through the year, I feel like I’m at a point where I can tell what I can and can’t reasonably expect to accomplish - both at work and at home - unless or until something fundamentally changes.

I’ve also been tracking and tweaking my digital choices. I’m still consuming as much screen time as ever, but seem to be making better choices on the whole - vital these days not to be staring into the void too long, too often 🙈

A grab-bag of things have taken my attention this week, including:

Getting my hands on the incredible 4k77, 4k80, and 4k83 fan remasters of the original Star Wars trilogy, after rewatching the Prequels to great dissatisfaction (which seems to increase with each rewatch),
Nodding my head along to the U.S. Digital Service’s “Playbook”, and pining for a time - or an alternate timeline, now - when the American government could do great and inspiring things, at least on occasion,
Learning about The Bazaar from a co-worker, but then ultimately deciding to get Slay the Spire on my phone as a way to play games again (first phone game in over 10 years!), without it taking over my life,
Daydreaming about opening up a local computer club in my city, and what that would entail,
Flipping through a vintage “Getting Started with Your Macintosh” manual after seeing it on Jerry’s desk on an early episode of Seinfeld (does it show how to use a computer mouse? You bet your sweet monochrome CRT it does!),
Reading about how my dear old love Vim has continued on in the year-and-a-half after the death of its author and maintainer of 3+ decades, Bram Moolenaar.

Weekly update: 07/2025

Sun, 16 Feb 2025 00:00:00 +0000

This being a typical holiday week here in Denmark, I took Thursday & Friday off in order to get some more time with baby and take care of some ‘dad chores’ around the house. At the end of this four-day weekend I feel a bit relaxed and accomplished, but still quite tired 😴

I spent my pockets of time this week:

Re-watching E.T., spellbound by it’s many magic moments, especially this frame which I had to rewind back to, pause, and admire:
Nodding my head to this piece on taste & tradeoffs in software development in the age of AI. I feel inspired to write a bit on how a manager develops good taste, as a follow-up to my recent post on taste in the workplace.
On a nerdier tack, loving this in-depth history of S.u.S.E., the company behind SUSE Linux (the chameleon one). SUSE was never one of the distros I tried for long, but I did unexpectedly walk past their office in Nuremberg and perhaps overreacted in my excitement,
Consulting Atlassian’s surprisingly thorough guide on making heatmaps (why did they write this?), for an investigation I’m doing into how my mental state & environment affect what I consume & produce,
Playing around with Meta’s AI demo’s - fun to get a sense of what’s possible with some creative use of LLM’s. I’ll end with this result, from a source video of my dog running around in the park: