Jan Tschichold is one of the few designers to have radically changed his vision and practice in response to the reality of the war he was experiencing. His personal questioning is quite exemplary and forces us to reflect on our practices. creativepro.com/jan-tschicho...

Disappointed by how many people I respect still post on twitter. By all rights that place should be a ghost town.

sql jail for: sql yelling commas first aliasing tables without as

wow it's like ξενία isn't even a thing anymore

the terran vibes from openai are undeniable

TIL Washington, DC has especially good tap water for brewing coffee. Source: coffeeadastra.com/2018/12/16/w...

@current_problems but for my life

i will watch any movie about a made up hungarian

all i’m saying is that i should get some kind of credit for how good i take care of my plants

!!!

Just an FYI that if you directly own any Apple stock, there is a shareholder proposal up for vote titled "Request to Cease DEI Efforts." You have until February 24 at 8:59pm PT to vote.

Installed PyMC and Blackjax in a Modal container with uv. It worked on the first try. Somewhere, a monkey's paw just curled one finger. #DataBS

me, a duckdb user, whispering in the ear of a summer research assistant at my professor father's villa in northern italy who i've quickly fallen in love with; the late afternoon sun warming the ancient stone walls as cicadas buzz in the apricot trees: "union all by name" #databs

A neat example of why Marr’s levels are useful. Neural networks aren’t blackboxes implementationally (we can see the weights and connections), but they sure are algorithmically (we have a limited understanding of why they yield the behavior they do). People are probably referring to the latter.

Did you know Disney owns the copyright to python's profiling library (profile)? github.com/python/cpyth...

I was literally on Kangaroo Island, one of the subjects of this article, today! It’s incredible how much the island has recovered, but I wish I could have seen it before the fire.

I accidentally followed too many academics and for a while there my feed was largely complaints about reviewers.

I’m currently in Adelaide, which is in the UTC+10:30 time zone. First time being off by half an hour.

It is an underreported fact that Australia is incredible.

1. I have event logs that are being written to a bucket as ndjson. 2. I wrote a thing that gloms those together into larger parquet files. 3. DuckDB reads the parquet and creates some on-disk tables. 4. The tables feed a pipeline that processes them into something useful. Question is… #databs

writing rust to write python to write cuda to write english

an email from a company that i bought a screwdriver from 13 years ago letting me know that they’ve updated their privacy policy

Who is currently winning the vector database race? It’s tough to differentiate between the various offerings right now. Who are folks using?