Henning
@radbrt.bsky.social
free-range data clown. Professional YAML indenteur. Writing python & sql, dreaming of Rstats and a turtle. Recovering economist.
Writing for humans at radbrt.com, and for machines at github.com/radbrt.
📍Norway
Writing for humans at radbrt.com, and for machines at github.com/radbrt.
📍Norway
Every once in a while I translate IT terms to Norwegian and I’m like "wait, what? This is what programming sound like to Americans?"
November 10, 2025 at 5:53 PM
Every once in a while I translate IT terms to Norwegian and I’m like "wait, what? This is what programming sound like to Americans?"
Excel: what are you doing? Saving as CSV will cause you to lose functionality!
Me: That is the entire point of it!
Me: That is the entire point of it!
November 7, 2025 at 8:23 AM
Excel: what are you doing? Saving as CSV will cause you to lose functionality!
Me: That is the entire point of it!
Me: That is the entire point of it!
The tensorlake playground was, unlike AWS textract and every other tool I have tried, able to parse my angled, low-quality scan of Norwegian pay statistics from 1926.
Not that 1926 Norwegian statistical tables is a generally useful benchmark…
Not that 1926 Norwegian statistical tables is a generally useful benchmark…
Document parsing benchmarks have been measuring the wrong thing.
We tested every major parser on real enterprise documents.
The results will change how you think about OCR accuracy 🧵
We tested every major parser on real enterprise documents.
The results will change how you think about OCR accuracy 🧵
November 5, 2025 at 7:52 PM
The tensorlake playground was, unlike AWS textract and every other tool I have tried, able to parse my angled, low-quality scan of Norwegian pay statistics from 1926.
Not that 1926 Norwegian statistical tables is a generally useful benchmark…
Not that 1926 Norwegian statistical tables is a generally useful benchmark…
On the bright side, humanity will finally go extinct.
New survey study of Austrian youth aged 15–25 finds that:
• 75.4% follow influencers
• 77.6% trust influencers they follow for health information
• 30% have already bought a health product on the recommendation of an influencer
www.jahonline.org/article/S105...
• 75.4% follow influencers
• 77.6% trust influencers they follow for health information
• 30% have already bought a health product on the recommendation of an influencer
www.jahonline.org/article/S105...
Engagement With Influencers as Sources of Health Information and Product Promotions: A Cross-Sectional Survey of Austrian Youth Aged 15–25 Years
Social media influencers have become a key source of health information for young
people, despite often lacking medical expertise and being driven by commercial interests.
This study examines influenc...
www.jahonline.org
November 5, 2025 at 1:52 PM
On the bright side, humanity will finally go extinct.
Norwegian aid to Ukraine is predicated on spending the money abroad so as to not overheat the Norwegian economy.
Very unsolidaric to push these externalities onto other countries.
Very unsolidaric to push these externalities onto other countries.
November 5, 2025 at 12:44 PM
Norwegian aid to Ukraine is predicated on spending the money abroad so as to not overheat the Norwegian economy.
Very unsolidaric to push these externalities onto other countries.
Very unsolidaric to push these externalities onto other countries.
I get a childish joy whenever I’m able to remove modals in css in order to access content.
This time it was a spammy "please give us your email" modal, I’m not depriving journalists of dinner or anything.
This time it was a spammy "please give us your email" modal, I’m not depriving journalists of dinner or anything.
November 4, 2025 at 1:27 PM
I get a childish joy whenever I’m able to remove modals in css in order to access content.
This time it was a spammy "please give us your email" modal, I’m not depriving journalists of dinner or anything.
This time it was a spammy "please give us your email" modal, I’m not depriving journalists of dinner or anything.
Today in cursed autocompletes.
November 3, 2025 at 7:36 AM
Today in cursed autocompletes.
New contender in "how to Y-axis" just dropped.
November 2, 2025 at 6:04 PM
New contender in "how to Y-axis" just dropped.
I have tried it out a little, but as much as I respect dlt’s hustle, my initial reaction of "what kind of YOLO-shit is this" is still valid. I would prefer Meltano in a generic production environment (of course, the choice depends on the context).
November 1, 2025 at 6:05 PM
I have tried it out a little, but as much as I respect dlt’s hustle, my initial reaction of "what kind of YOLO-shit is this" is still valid. I would prefer Meltano in a generic production environment (of course, the choice depends on the context).
Interesting eulogy for dbt: timodechau.com/dbt-blues/
dbt blues
When dbt Labs announced their merger with Fivetran, the data community had feelings. A lot of them. The dbt fear index—Oliver from Lightdash's brilliant term for the spike in repository forks—told the...
timodechau.com
October 23, 2025 at 5:29 AM
Interesting eulogy for dbt: timodechau.com/dbt-blues/
So Americans work 12-hour days but does that include the time in golf simulators?
Landlords are seducing remote workers with golf simulators and rooftop cabanas
New “amenity buildings” are luring Americans back to the office
Landlords are seducing remote workers with golf simulators and rooftop cabanas
econ.st
October 21, 2025 at 9:33 PM
So Americans work 12-hour days but does that include the time in golf simulators?
Sixtran shows that we need low interest rates again. Everything else shows that we don’t.
October 13, 2025 at 6:30 PM
Sixtran shows that we need low interest rates again. Everything else shows that we don’t.
I am simultaneously worried by the dbt/fivetran merger, and quizzical why we, as paid professionals in profit-seeking entities, should have parasocial relationships with open source python packages.
October 5, 2025 at 2:25 PM
I am simultaneously worried by the dbt/fivetran merger, and quizzical why we, as paid professionals in profit-seeking entities, should have parasocial relationships with open source python packages.
I am still convinced the only thing European clouds need in order to be competitive is an actual IAM solution with passwordless M2M authentication.
Stuff like Azure Managed Identity and IMDS. Please throw in Oauth CLI authentication while you’re at it.
Stuff like Azure Managed Identity and IMDS. Please throw in Oauth CLI authentication while you’re at it.
September 15, 2025 at 10:22 AM
I am still convinced the only thing European clouds need in order to be competitive is an actual IAM solution with passwordless M2M authentication.
Stuff like Azure Managed Identity and IMDS. Please throw in Oauth CLI authentication while you’re at it.
Stuff like Azure Managed Identity and IMDS. Please throw in Oauth CLI authentication while you’re at it.
Finished a workshop that seemed to have the mantra "why don’t people realize they can use this awesome technology that they don’t know about to solve this problem they don’t know they have?".
I guess there’s a lot of money in selling people both the problem and the solution.
I guess there’s a lot of money in selling people both the problem and the solution.
September 1, 2025 at 7:42 AM
Finished a workshop that seemed to have the mantra "why don’t people realize they can use this awesome technology that they don’t know about to solve this problem they don’t know they have?".
I guess there’s a lot of money in selling people both the problem and the solution.
I guess there’s a lot of money in selling people both the problem and the solution.
I’m ready to call it. Databricks governance is a fucking dumpster fire.
August 28, 2025 at 2:15 PM
I’m ready to call it. Databricks governance is a fucking dumpster fire.
Databricks copilot is amazingly bad. Run bad command > tells me the syntax is X > Get error > tells me the syntax is Y > Get error > tells me the syntax is Z.
August 25, 2025 at 7:54 AM
Databricks copilot is amazingly bad. Run bad command > tells me the syntax is X > Get error > tells me the syntax is Y > Get error > tells me the syntax is Z.
Absolutely bizarre that Databricks on Azure (and GCP) only lets you have one unity catalog per region - basically forcing you to develop in production.
August 22, 2025 at 8:55 AM
Absolutely bizarre that Databricks on Azure (and GCP) only lets you have one unity catalog per region - basically forcing you to develop in production.
Wrote up how to get GitHub Actions with OIDC to Azure working. Don’t know what that is? It is basically a way for your GitHub actions to log in to Azure without using a password. And Microsoft is making it increasingly easy to set up. radbrt.com/posts/github...
Github Actions OIDC | Radbrt
GPT came very close to giving a complete working tutorial on setting up OpenID Connect federated credentials that lets your Github Actions authenticate to Azure. This means no passwords, exceptionally...
radbrt.com
August 22, 2025 at 7:37 AM
Wrote up how to get GitHub Actions with OIDC to Azure working. Don’t know what that is? It is basically a way for your GitHub actions to log in to Azure without using a password. And Microsoft is making it increasingly easy to set up. radbrt.com/posts/github...
Really want @markzandi.bsky.social and the Inside Economics crew to comment on the idea that was floated to only publish employment numbers quarterly.
I get the need to have new data ahead of fed meetings, but other than that a lot feels like impatience in a trench coat.
I get the need to have new data ahead of fed meetings, but other than that a lot feels like impatience in a trench coat.
August 13, 2025 at 9:42 AM
Really want @markzandi.bsky.social and the Inside Economics crew to comment on the idea that was floated to only publish employment numbers quarterly.
I get the need to have new data ahead of fed meetings, but other than that a lot feels like impatience in a trench coat.
I get the need to have new data ahead of fed meetings, but other than that a lot feels like impatience in a trench coat.
The difficulty of creating a grouped aggregation with a weighted average in Pandas tells you a lot about the users.
With R, this is built-in.
I wish the pandas aggregations were pluggable and we would be able to create things like weighted mean, gini etc.
With R, this is built-in.
I wish the pandas aggregations were pluggable and we would be able to create things like weighted mean, gini etc.
July 31, 2025 at 8:37 AM
The difficulty of creating a grouped aggregation with a weighted average in Pandas tells you a lot about the users.
With R, this is built-in.
I wish the pandas aggregations were pluggable and we would be able to create things like weighted mean, gini etc.
With R, this is built-in.
I wish the pandas aggregations were pluggable and we would be able to create things like weighted mean, gini etc.
💯don’t demo with fake data to the business. Or, actually, either demo with very correct numbers or so obviously fake numbers that everybody gets it.
July 30, 2025 at 5:28 PM
💯don’t demo with fake data to the business. Or, actually, either demo with very correct numbers or so obviously fake numbers that everybody gets it.
So, my iPhone lock screen can’t render the letter "a"?
July 27, 2025 at 7:29 AM
So, my iPhone lock screen can’t render the letter "a"?
Read @edzitron.com’s guest post (the unpaywalled part): www.wheresyoured.at/the-remarkab...
The article is about why companies buy software they don’t need. I have argued almost the exact opposite - that they build unnecessarily complex software themselves. 🧵
1/
The article is about why companies buy software they don’t need. I have argued almost the exact opposite - that they build unnecessarily complex software themselves. 🧵
1/
The Remarkable Incompetence At The Heart Of Tech
Hello premium subscribers! Today I have the first guest post I've ever commissioned (read: paid) on Where's Your Ed At - Nik Suresh, one of the greatest living business and tech writers, best-known fo...
www.wheresyoured.at
July 19, 2025 at 5:41 PM
Read @edzitron.com’s guest post (the unpaywalled part): www.wheresyoured.at/the-remarkab...
The article is about why companies buy software they don’t need. I have argued almost the exact opposite - that they build unnecessarily complex software themselves. 🧵
1/
The article is about why companies buy software they don’t need. I have argued almost the exact opposite - that they build unnecessarily complex software themselves. 🧵
1/
This morning I read blog posts that convinced me that there are only two problems in software: that we buy software, and that we build software.
I guess there is only one problem in software: software.
I guess there is only one problem in software: software.
July 19, 2025 at 10:04 AM
This morning I read blog posts that convinced me that there are only two problems in software: that we buy software, and that we build software.
I guess there is only one problem in software: software.
I guess there is only one problem in software: software.