Oh No, I Think I Might Have to Move to Home Assistant
And Other Major Life Changes
The last few months have been pretty wild… In April, my wife and I moved out of our tiny Oslo flat (41 m²) into her childhood home. This is a large (for us) 290 m² house in a smaller town. On the first night in the new house, with things pretty up in the air, our first baby decided he wanted to be born 3.5 weeks ahead of schedule. He’s a lovely guy! But it’s fair to say moving, renovating, selling a flat, and taking care of a bundle like that, is pretty intense — hence why I haven’t written the last few months.
I do want to refrain from posting a lot of our son online — but hopefully saying that his name is Alfred, and posting the pretty anonymised photo below is OK ☺️:
I got to make the house smart 🫶🏻
As part of the deal of us moving to «my wife’s» house and hometown, we agreed that I was allowed to make the house smart. Not that she was that difficult to convince, as she doesn’t find a smart home annoying. Especially if I manage to follow my own principle: Smartness should always be in addition to regular functions. There should always be a button to toggle a light — and then you can add smarts on top.
The current plan is to have the following be smart:
- Lights
- Blinds
- Heating
- Main door lock
- Garage door
Playing With Icon Composer and the Bike Icon
One of my favourite apps to use, is Bike Outliner by Jesse Grosjean/Hogbay Software. It’s simply delightful through-and-through.
He’s hard at work on a really promising 2.0, and I wanted to play with Apple’s new Icon Composer. So I wanted to see if I could contribute and adapt Bike’s icon to the new style.
Not optimal
The original Bike icon (which I don’t love as much as the app itself) isn’t optimal for this style. If you look at Apple’s documentation, it’s especially clear to see that it’s easier to get a nice result with few largish objects you can layer.
Also, I’m absolutely not a designer. But I did my best to create:
A first draft
One thing I’ve thought about the original icon, is that the spokes on the wheels are many and small. On the sizes where the icon is actually used, this makes them blur into each other. However, this isn’t necessarily something negative. (And I know Jesse likes it.) To me, it gives an impression of the bike being in motion.
Here’s what I did:
- I redrew all the elements (except the head) to make them thick enough so that the glass effect in Icon Composer would apply.
- Then I separated them into the maximum of four layers Apple allows:
- The spokes and wheels at the bottom,
- then a layer of transparent circles that blur the spokes,
- then the frame,
- and lastly, the seat and handlebar on top.
Quick Recommendation #11: StarCraft 2 (Video Game)
When Blizzard Was the Best in the Business
Growing up, I played a lot of the original StarCraft. (Even more than Broodwar.) So I was obviously hyped when StarCraft 2 launched 12 years later, in 2010.
This was back when Blizzard Entertainment only made great games (even though I’ve never been into Word of Warcraft), and hadn’t discovered microtransactions and undercooked remasters. I both played through the campaign and was mediocre on the ranked ladder,1 and it was great!
Even though the game is 13 years old, there’s still an active pro scene and community. And recently, the YouTube algorithm decided to serve me a really charming channel: WinterStarcraft. His bread-and-butter is him casting pro matches, and I just love it. (To actually give the videos a chance to reach new viewers, the titles and thumbnails are really click-bait-y. But the content is good!)
For a place to start, I can recommend this video and match. (The third game is especially great.)
I highly recommend playing the Starcraft 2 campaign in 2025.
The game went free-to-play, via Battle.net, in 2017 – and this also includes the first part of the campaign: Wings of Liberty. That’s 20+ hours of gameplay right there – and if you’d like, you can then pay to unlock the rest of the single player content (Heart of the Swarm and Legacy of the Void + Nova Covert Ops). This is also included in Xbox Game Pass.
Even though real-time strategy can be a bit daunting, the Starcraft 2 campaign is really accessible. It teaches you the game in a great way, feeds you mechanics bit-by-bit, and has a lot of challenge adjustability with difficulty levels and optional objectives. I just really like the world and story. (The Zerg, admittedly quite inspired by Xenomorphs, are some of the coolest things created.) And the game is just so polished.
This was also when Blizzard ported their games to Mac.
So even though it isn’t very optimised for Apple Silicon, it still works great on my M1 Pro. However, I did have an issue with the sound – but it’s fixable:
For some reason, the sound output would crackle if the microphone is active.2 However, barring StarCraft 2 access to the microphone, through Privacy and Security in System Settings, fixed it. (I also have to close SoundSource.)
GL HF!
Quick Recommendation #10: Hemispheric Views (Podcast)
Episode 137 Is the Perfect Place to Get Into the Most Charming Tech Podcast
Some reasons why Hemispheric Views is one of my favourite tech podcasts:
- Three sympathetic hosts, with great chemistry.
- Some Australian perspective in my life.
- The goal of the episodes being “a tight 45” (minutes).
- Great bits and running jokes. (For instance, their member program is called One Prime Plus. 😁)
I can recommend last week’s episode, 137: I Had a Pi in the Drawer, as a good place to start! It’s both accessible and gives a good impression of the show. And as I’m a bit late to posting this, you’ll then also have this week’s episode ready if you want another one immediately.
I also recommend following the hosts:
Don't Chop Up Your Apple Trees to Make Barrels for Your Apples
The CEO of The Browser Company, the company behind the Arc browser, recently posted a lengthy Reddit post. He explained why they’ve abandoned Arc, and discussed a bit about their plans going forward.
They’ve made a huge pivot to an upcoming AI browser they’re calling Dia. But I just wanted to comment on this, very stupid, approach, from the portion about what makes an “AI browser” different:
1. Webpages won’t be the primary interface anymore. Traditional browsers were built to load webpages. But increasingly, webpages — apps, articles, and files — will become tool calls with AI chat interfaces.
They really don’t see the major flaw in this approach?
The AI models are completely worthless without training data,1 which they scrape from webpages. If people don’t visit webpages, the incentives to create the content disappears. What do they think this will do to the quality of the “tool calls” over time?
“Hmm, we need some barrels for all of these apples we just picked from our orchard… I got a great idea! The trees are worthless now, without apples – so let’s just chop them down and use the wood to make the barrels. I’m sure this won’t have any ramifications for the future.”
The Single Piece of AI Legislation I'd Start With
One Simple Rule
I know that creating laws is very hard — especially at a global scale.1 And figuring out what to do about the rise of AI is the same. But I have an idea about where I’d start, that I would love to spitball.
My opinion regarding generative AI is currently something like this:
- It’s not as useful and amazing as the salesmen claim it is. And overestimating it has its dangers.
- At the same time, it also does have plenty of very useful use-cases — and more to come.
- However, it being useful isn’t the same as it being a net-good, or that there aren’t very problematic consequences that need to be dealt with. (I wrote more about this here.)
My suggestion…
… for the first rule is quite simple (in concept):
It must be easy to find out if a piece of content is part of a model’s training data or not.