I’m assuming this is an elaborate joke prompted by dieselgate.

Hmm link no worky. Gateways error behind the captcha

The design of git has proven to be a very capable model. I’ve used them all before: RCS, cvs, clearcase!, pvcs. Although it took a bit of adaptation to move from revision numbers to commit IDs it didn’t take long before the first thing I did when faced with another codebase was convert it to a git repo.

It will be fun watching those users who first make the jump to the new project.

Funny 😂

I think the article is over complicating things. I work in a project which is heavily forked for a variety of reasons. While it’s academically interesting to look at the reasons for those downstream forks we have no interest in going to the considerable effort of tracking them all.

If you can take a project and use an LLM to enable your niche use case then more power to you. FLOSS was never about ensuring all patches flow upstream.

It’s not entirely unexpected, all the AI companies have been heavily subsidising inference to get customers.

I don’t use Codex but I’ve been experimenting with ECA and I can track my token API costs across Gemini and Anthropic. I’m mostly using Gemini and a heavy days usage would be £1.50 in API costs and I’m certainly not doing that every day. I have to wonder if these Codex users are conscious of how many tokens they are burning underneath or just YOLOing everything until the computer says no?

ECA allows you to mix and match models to sub-agents and I could certainly see me offloading some tasks like code exploration to a locally hosted models and saving the expensive reasoning tokens for planning.

So algorithms then?

LLMs have some interesting properties and certainly can do a good job sifting through large amounts of raw data. They are however a very brute force approach compared to say a network routing protocol. Sooner or later people will start to realise (again) that engineering is about trade offs and you need to work out what your constraints are and stop trying to solve every problem with massive amounts of multiplication.

It was worse than that. Our understanding of radiation took awhile. While Uranium glass is probably safe I wouldn’t go using it regularly. A lot of women (“radium girls”) suffered from cancers induced by licking their brushes when painting luminescing instruments. This comic looks like 50s era when post the bomb sci-fi was full of “atomics” as the stuff of the future.

It seems the term AI is now synonymous with hallucinating Large Language Models in the general publics head. There is a whole field of machine learning where you can get statistically useful results with various techniques. Alpha Fold is a good example where real progress has been made on finding protein folding solutions that older brute force algorithms are just too inefficient to explore the state space.

This paper is taking about a new ML model for classifying planetary systems that out performs previous data processing pipelines. It’s called statistical validation because it is inherently a numbers game. The paper goes into lots of details about how they calculate false positive rates and compare it to previous approaches. The point is not to definitely identify individual systems but to classify the distribution of system types in the large amounts of data the modern surveys are generating.

What’s your experience of British comedy like? Have you tried The thick of it?

I recently joined the ranks of 3d printer owners. The first thing I printed where a pair of risers for mounting some hall effect sensors on my garage door mechanism. Very simple shapes but super handy.

The other option is to use VirtIO with Native Context support as a software based partitioning scheme that is relatively lightweight compared to the mdev approach.

The kernel on GitHub is just a mirror - the primary source is on kernel.org

I suspect we don’t know enough about the mechanics of consciousness yet to determine what free will really means. We certainly know enough about psychology to understand predispositions to make certain choices and humans as a group are fairly predictable.

I think the most useful thing for this is hosting repos that suffer from constant DMCA takedowns. Emulators, ad-blockers, site revancers etc.

It used to slowly drive haters insane, hence mad haters. There was a theory Napoleon was exposed to excess levels of Arsenic over time although that might have well been background exposure. Nevertheless over time it’s not good for you. Neither is lead.

Can those handle the meta data for the track name, artist and release date. Assuming you want a portable playlist that can then find the track on the recipients preferred platform (streaming provider or self hosting). Given that a lot of tagging is trash maybe also included an audio fingerprint for validation?

You need to pair 1984 with Huxley’s Brave New World to see where we actually ended up.