Jori Lytter
polarcoder.net
Jori Lytter
@polarcoder.net
Reposted by Jori Lytter
I've been running autonomous agents this whole time.
October 21, 2025 at 8:34 PM
Reposted by Jori Lytter
One rubber duck can speed you up, but two ducks slow you down, because they start talking to each other and distract you.

This is known as the ‘inefficiency pair-o-ducks’.
October 6, 2025 at 3:27 PM
Reposted by Jori Lytter
An event is a fact something happened in your domain.

Named with past tense verbs: `CustomerRegistered` or `OrderPlaced`.

Events are immutable, which makes sense when you think about it because you can't change the past.

This principle is critical to Event Sourcing
September 30, 2025 at 8:24 AM
My current work laptop doesn't have deep sleep (S3) option but only modern standby (s2idle). It consumes over 5% of battery per hour meaning that I can't leave it sleeping on battery power between two workdays because there's no guarantee that it has any power in the morning.
September 30, 2025 at 5:57 AM
Reposted by Jori Lytter
So, wait, like, .... people think it would be good for every developer in their company to produce 20K lines of code per day?

What are they smoking, and how did they smuggle it into the country?
September 15, 2025 at 4:20 PM
Reposted by Jori Lytter
“one of the most devious climate concepts of all times: the personal carbon footprint.” - Benjamin Franta

It’s a distraction invented by BP to deflect blame onto individuals, when the “necessary, and largely sufficient” solution is: stop burning fossil fuels.
September 13, 2025 at 4:47 PM
Reposted by Jori Lytter
They erased the new Banksy mural in London. But the stain remains.
September 10, 2025 at 11:02 AM
Reposted by Jori Lytter
"We use coding agents like junior developers".

Cool. So you get the dual benefits of it being slower than an experienced developer to get a half-decent result *and* it never learns.

Bargain!
August 28, 2025 at 9:31 AM
I've been saying this for years.
1. Build a modular monolith
2. When real life experience with the monolith shows signs that it would bring benefit to extract a module from it then extract it in to it's own service
Microservices were the answer… but in 2025, many teams are questioning whether it's worth the complexity. Are modular monoliths the new microservices? 👀
August 5, 2025 at 6:17 AM
Reposted by Jori Lytter
Trick question, it will never scale perfectly. This is how they tricked us into AWS lambda

mirrors.edge.kernel.org/pub/linux/ke...
August 4, 2025 at 10:35 PM
Reposted by Jori Lytter
I expect that consumer-facing AI programs will continue to improve and they may become much more useful tools for everyday life in the future.

But I think it was a disastrous mistake that today’s models were taught to be convincing before they were taught to be right.
June 19, 2025 at 1:26 PM
Reposted by Jori Lytter
The loud drumbeat that companies should "embrace AI now or be left behind" has more in common with high-pressure sales techniques than biz strategy. And that's fine - tech firms have something to sell. But the rest of us needn’t confuse it with anything more profound. www.ft.com/content/4688...
July 1, 2025 at 11:15 AM
Reposted by Jori Lytter
June 27, 2025 at 8:14 AM
Recently I've come accross mupltiple times a statement that with AI one can generate more code and faster than ever.
To me this seems a reason not to use AI to generate code.
Since when more code has solved anything?
Instead create less code, impactful code, solving real problems.
#AI #code #simple
June 26, 2025 at 3:26 PM
Reposted by Jori Lytter
Time to say it again:

Code is not an asset. If anything, it is debt; every line of code is a potential bug and a potential maintenance problem.
June 2, 2025 at 12:27 PM
Reposted by Jori Lytter
I didn't really think we would see lines of code written return as a measured thing but, thanks to AI, people (not AI vendors) are now citing it in quarterly earnings reports as though it matters?
June 2, 2025 at 11:37 AM
Amazed of proprietary corporate VPN's and how badly they're still implemented for linux at 2025.
And since they're proprietary reverse engineered openconnect solutions have a 50/50 change on working. Does not work in my current situation.
May 18, 2025 at 6:57 AM
I like the idea that Christian Ekrem is proposing on when and how to use AI in the process of software development.

Give it a good 6 minutes of read to see if it makes sense to you.

#AI #Coding #Craft #Programming

cekrem.github.io/posts/coding...
Coding as Craft: Going Back to the Old Gym
Why 'reflexive AI usage' sounds like a diagnosis I don't want next to my name
cekrem.github.io
April 22, 2025 at 12:41 PM
Reposted by Jori Lytter
I do not believe anyone is actually doing this.
March 13, 2025 at 7:03 PM
Reposted by Jori Lytter
Most of what we call "tech" is about getting the right information into a database and trying to prevent the wrong person from reading or updating it.
March 3, 2025 at 10:10 PM
@tastapod.com is on to something with the idea of "best simple system now". It resonates with me. I at least like to think that this is the way I work during my better days every then and now.

Definitely worth a read dannorth.net/best-simple-...
Best Simple System for Now
You can have your cake and eat it, as long as you bake it carefully. ‘We can do this the quick way and pay later, or the thorough way and pay now.’ This seems to be a fundamental dichotomy in software...
dannorth.net
February 8, 2025 at 4:21 PM
Reposted by Jori Lytter
Do you start each programming task/job/ticket with a clean start? www.industriallogic.com/blog/clean-s...
The Clean Start Protocol
If you want to program, begin with a clean start.
www.industriallogic.com
January 14, 2025 at 11:21 AM
Reposted by Jori Lytter
Imagine applying for engineering jobs with recent work experience at Meta on your CV
Holy shit this filing update is wild! It states that:

1. Zuckerberg gave the ok to train off copyrighted works

2. Engineers wrote scripts to remove copyright info

3. Meta downloaded illegal torrents to train models of

GenAi is a house of cards, I swear.

techcrunch.com/2025/01/09/m...
Mark Zuckerberg gave Meta's Llama team the OK to train on copyrighted works, filing claims | TechCrunch
Meta CEO Mark Zuckerberg gave Meta's Llama team approval to train on copyrighted documents, according to a new court filing.
techcrunch.com
January 10, 2025 at 8:03 AM