Second order consequences of AI generated code
„Headcount stops measuring capacity and starts measuring how much accountability you can afford.“
karimjedda.com/engineering-...#ai#leadership#engineering#management
Most governments find the loss of entire world regions to uninhabitable climate scenarios and the untold suffering of humans - and animals - easier to imagine, than the transition away from oil and gas.
As Amitav Gosh said, the climate crisis is a crisis of our imagination.
Took a business call this Sunday morning and the caller was impressed how I’m always available for calls and grinding etc.
Actually, I wasn't. I was doing the exact opposite: I was working today because it fit my schedule better, and I can take Monday off
I intentionally work some Weekends so I can take other days off.
This allows me to spend more quality time with my partners as she also works weekends but has weekdays off.
Flexibility isn't about never working weekends. It's about choosing when, and actually protecting the rest.
Work-life balance looks different for everyone.
Some weeks need more work. Some need less. The point is choosing when, and actually protecting the rest.
Find what works for you.
How the tables have turned. Companies trying to “tokenmaxx” a month ago, are now trying minimize the spend and look for ways to efficiently use tokens.
The Germany Economy stands to lose € 112.5bn in the next 4 years due to extreme heat (a direct effect of climate change!)
We need to invest in climate change and climate Adaption now before it’s too late!
www.tagesschau.de/wirtschaft/w... (sorry German only)
#climatechange
A NEW TWIST in Africa's soaring solar story:
It's now manufacturing its own panels. The factories use imported Chinese cells, which are 40% of the value of panels, keeping much of the value domestically.
9 GW of cells+wafers in Mar/Apr alone🧵
Okay, Migration Assistant is interesting. In my head this raises interesting questions regarding software licensing. Assume for example, I migrate a GPL licensed app from iOS to Android.
#GoogleIO
As the new app will most likely be ported from Swift to Kotlin, so most likely not a single line of code made to Android, so would GPL copyleft still apply? Or could I release the app in any license (provided no other trademark is violated)? #GoogleIO
Aha! The #GoogleIO developer demo shines some light on it, as they demo how to create a fine-tuned version of Gemma-4.
I assume they will use multiple fine tuned models for different use cases internally, potentially even on an "per agent" level?
The biggest news for me at #GoogleIO today was TPU 8 with their specialised version for Training and Inference. Making custom silicon is a serious effort, compared to e.g. developing dedicated training and inference cores on the same chip.
The biggest news for me at #GoogleIO today was TPU 8 with their specialised version for Training and Inference. Making custom silicon is a serious effort, compared to e.g. developing dedicated training and inference cores on the same chip.
A bit of searching shows the original TPU 8 announcement is almost a month old now, but it does not go into detail what exactly the TPU 8t models will be doing 🤔 cloud.google.com/blog/product...#GoogleIO
The biggest news for me at #GoogleIO today was TPU 8 with their specialised version for Training and Inference. Making custom silicon is a serious effort, compared to e.g. developing dedicated training and inference cores on the same chip.
So I assume Google must see a future whey they are constantly training models on a massive scale.
I wonder if all this is to develop their next Generation of Gemini Models, of if we soon a much larger model zoo of (personalised) models? #GoogleIO
Wait? That’s it? Nothing on #GoogleBook where they put Gemini right into the OS, or anything for Android besides the ominous „Android Halo“ mention?
#googleio
This whole Shopping Demos are a mess. First they introduce new shopping protocols, now they show a demo that does not use them but instead does AI automation for exiting Android apps. Feels like multiple teams competing with each other or a rushed announcement.
#googleio
So let me get this straight all we get as non-US people is a bigger search box?
Omni, Spark all other new features so far are behind the walled garden of the US border…
#googleio
The reason behind Universal Cart is quite clear: Google fears that agentic shopping is eating up their ad revenues as it happens in the Agent interface and not the browser anymore
#googleio
Valve Corporation has no managers. No hierarchy. No org chart.
Engineers pick their own projects. Teams form themselves and cap at 12.
Need more than 15? They split. 50? They divide. 150? They open a new office.
This is Dunbar's Number applied as company policy.
🧵
A lessen every startup engineering organizations must take to heart.
Yes rewrites sometimes are inevitable but they are also dangerous as they can keep you from adding value.
I balance rewrites by limiting their „life expectancy“ (1/n)
In 1995, Netscape Navigator owned the browser market.
Then they rewrote it.
The rewrite became Mozilla. It took years. Internet Explorer caught up, then passed them.
Netscape never recovered.
🧵
Depending on the clarity of your roadmap the new iteration should add the „Must Haves“ of the next 6-24 month. That cannot be added via „regular refactoring“.
Anything shorter is just regular feature development and refactoring. Anything longer is speculation.
The time allotted for the rewrite itself should be ~20% of live expectancy. (2/n)
Meaning if you are sure that this iteration will last 2 years, you can spend 4-5 months on the rewrite. If you don’t know how the company develops over the next half year. Keep the rework to 1-2 sprints. (3/3)
I learned that you can dynamically set the iPhone wallpaper via Shortcuts. To play around with it, I made a fun clock that shows the time remaining as a bar chart.
While it’s not particularly useful, I was surprised that on May 2nd at noon, already 5% of the month are behind us! 🤯
Every time I see someone talk about Claude I think about how they switched from female to male names as soon as they wanted the robots to seem like experts vs personal assistants.
Josh Johnson has summed up one of the greatest challenges of integrated friend and family groups more efficiently than anyone yet in this little short
telling you that I paid ten times as much for my wedding photographer as for my wedding dress for this exact reason…
m.youtube.com/shorts/Ur-kC...
Most engineers learn software engineering laws the hard way
Your rewrite collapses. You add engineers to a late project and it ships slower. You hit every metric but build nothing anyone uses.
Then you find out a 1975 law predicted it.
Check it: newsletter.techworld-with-milan.com/p/the-20-sof...
𝗪𝗵𝗮𝘁 𝗶𝘀 𝘁𝗵𝗲 𝗱𝗶𝗳𝗳𝗲𝗿𝗲𝗻𝗰𝗲 𝗯𝗲𝘁𝘄𝗲𝗲𝗻 𝗺𝗮𝗻𝘂𝗮𝗹, 𝘃𝗶𝗯𝗲, 𝗮𝗻𝗱 𝗮𝗴𝗲𝗻𝘁𝗶𝗰 𝗰𝗼𝗱𝗶𝗻𝗴?
A long time ago, we coded only manually, but then we started to vibe code. Yet, vibe code is something everyone defines differently.
🧵
joplin-cli 0.3.0 is out!
It’s a lightweight „no MCP server needed“ way to give AI agents access to you Joplin Notes (or just to access them from the CLI yourself)
github.com/vsaw/joplin-...#joplin#mcp#cli#BuildInPublic
Including it into Gemini CLI or other tools is as easy as adding one line to the prompt:
Search my Joplin Notes for Birthday presents. Use the command line tool "joplin-cli" to access my Joplin Notes.
Latest release v0.3.0 primarily focuses on README, and release packaging. So it’s easy to get started as running npm install and setting up your Joplin Token!
𝗟𝗟𝗠𝘀 𝗔𝗿𝗲 𝗡𝗼𝘁 𝗥𝗲𝗮𝗱𝗶𝗻𝗴 𝗬𝗼𝘂𝗿 𝗖𝗼𝗱𝗲
We keep calling LLMs "AI coding assistants." But writing code and understanding code are not the same thing.
Researchers just ran 750,000 debugging experiments across 10 models to determine how well LLMs actually understand code.
🧵
I’ve started a #joplin CLI app.
Compared to the Joplin Terminal App (which is a full blown Joplin with a TUI) I’ve created it as a thin wrapper using the Web Clipper API.
It’s still rough around the edges but if you want to take a look! 👀
github.com/vsaw/joplin-...#joplinapp#BuildInPublic
I use it for
- search/get/update all you Notes and Notebooks from your favorite terminal
- connect to LLMs (as a much leaner alternative to all MCP Servers I’ve found so far)
And potentially other script automations (e.g. I keep READMEs for how I set up all my RPIs at home in Joplin)