We ran a randomized controlled trial to see how much AI coding tools speed up experienced open-source developers.
The results surprised us: Developers thought they were 20% faster with AI tools, but they were actually 19% slower when they had access to AI than when they didn't.
Pyramid with levels labeled from the bottom up: - not having to use Teams - not having to use Jira - not having to use Salesforce - food shelter water - money love etc.
The real hierarchy of needs
The notion of "relentless refactoring" is an essential part of programming. It's integral--something you do all day, every day, as you work.
1/6
Scale AI’s CEO states DeepSeek rivals or matches the top AI models from American companies. Trained for just $6M (10-100x cheaper than big tech models), it runs on older chips due to chip bans and is open source.
This is a paradigm shift and makes $500B spend on Project Stargate feel misguided.
Love that a lab with a fraction of the GPU power, budget and staff of the big American AI whales has dethroned them and then open sourced their trained model and documented all their secret sauces.
Without any of the bullshit of “we can’t open it, it would become sentient and turn against us”
Layoffs have been going on for years in big tech but it’s still somewhat shocking to see my former employer laying off people with 20+ years of experience at the company as part of performance based cuts with no severance.
This post on Blind is similar to many I’ve seen on LinkedIn & Facebook.
I did the podcast thing! It was great fun chatting with Matthias Endler on his “Rust in Production” podcast. We talked about our little Rust based ECU at Volvo Cars and how it came about. Check it out: corrode.dev/podcast/s03e...
Also, it’s best if the entire team works on one story at a time, moving to the next on one when the current one is done and released to customers. “Little’s Law” tells us that the more things you work on simultaneously, the slower the work will happen.
6/7
People in the Scrum camp sometimes panic when they can’t finish a story within the Sprint, which they classify as a failure. That’s not actually how it works.
1/7
I hope we come up with better standards to protect the kids after this. This is a big one!
Two giant robots. A blue and yellow one with the Python logo on its body, and a green and blue robot with the Clojure logo on its body.
Curious about advanced functional programming techniques? We at @sixtynorth.com have just published a free course on @tubetrain.io, ‘Understanding Transducers Through Python’, which explains a super-interesting technique from the Clojure language by implementing it from scratch, in Python. [1/3]
It took a lot of massacres and killings to "... facilitate the achievement of this object"
en.m.wikipedia.org/wiki/Balfour...
Unfortunately it's been like this for close to a century. It's just being telecast live now.
I was talking about Cassandra in context of ScyllaDB.
CockroachDB is amazing! I hope there'll be something soon under ASF too
I only personally prefer community-led projects if I have the choice. Cassandra is still OSS btw. Not as sexy I guess tho
bsky.app/profile/sysa...
Another one:
www.scylladb.com/2024/12/18/w...
If it's not under the Apache umbrella, it's not worth it!
Chromebooks.
Bought my wife one and that's all she needs to get 100% of her work done!
I'd go as far as to say that Web browsers are the new OS. PWAs are imho 80% of the way there.
What has your experience been? I'm thinking of using Effect TS ..
An absolute must-read from @mipsytipsy.bsky.social, with the single most important piece of founder wisdom that you will ever hear (and most will forget/ignore/be talked out of): "You should ALWAYS have as few employees as possible. Always."
I try to prevent myself from turning my computer on on Saturday. I'm still thinking about stuff of course.
Every software system should be engineered in such a way to prevent errors and breakages, not to report them after the fact.👇
I really wish more people that come up with metrics for measuring team/developer productivity had a chance to work as an engineer in high performing teams.
1/3
Forgot Netanyahu
I ended up working at Intel and met the people I used to interact with on their forums. Many great folks!
I see where you're coming from. I was the founding engineer (first hire?) at IntraNAV (intranav.com) bet their gateways on Edison and Galileo SoCs (ca 2015). Bad decision on my part but we were a software company so we could course-correct fairly cheaply.
Great write-up. Thanks for putting your thoughts together!
👀
