Interesting essay looking at how systems evolve their schemas over time. We're generally ill-equipped to deal with it and this presents options and ideas to that effect. Of course, the more precise you want to be the more complexity you'll have to deal with.

There is some truth to this. Moving some things to data brings interesting properties but it's a two edged sword. Things are simpler to use when kept as code. Maybe code emitting structured data.

Interesting short article. Shows the use of DORA metrics and process behavior charts. This is a good way to test hypothesis and see the impact of processes changes or introduction of new practices. It needs to be done over time and be patient of course.

Pipelines are very widespread nowadays, still I don't see them used much. Having a few refactoring ideas under our belt to replace loops with such pipelines might help.

It's not the only factor leading to troublesome architectures of course. Still, if state and thus data is wrongly handled, you're indeed on the wrong track.

Maybe it's time to stop obsessing about scale and distributed architectures? The hardware has been improved quite a bit at the right places, especially storage.

This is indeed a metaphor which should be more common in enterprise software.

Interesting selection of options to model data structure with some variability in Rust.

A look back at the limitations of deep learning in the context of computer vision. We're better at avoiding over fitting nowadays but the shallowness of the available data is still a problem.

Just looking at averages is indeed quickly hiding patterns. Make sure distributions are visible in some fashion.

Nice piece to give ideas about what type of diagram to consider depending what you're exploring.

Interesting class of data structures with funny properties. Looks like there's a lot to do with them.

This is one of the handful of uses where I'd expect LLMs to shine. It's nice to see some tooling to make it easier.

A nice extension for Postgres allowing to ease the protection of personal information.

OK, the numbers are indeed impressive. And it's API is fully compatible apparently, looks like a good replacement if you got Pandas code around.

It shouldn't be, but it is a big deal. Having such training corpus openly available is one of the big missing pieces to build models.

Interesting dimensions to use when classifying syncing solutions and to see which ones will meet your constraints.

Excellent piece, we're a civilisation whose culture is built on shifting sands and... toy plastics. Guess what will survive us?

More discussion about models collapse. The provenance of data will become a crucial factor to our ability to train further models.

The more releases out there the more vulnerabilities are (and could be) discovered. Some actions are necessary to get things under control properly.