ello in the void

Fuck. Reflect Orbital apparently had more time than we thought to respond, and they just did. We have a lot more writing to do. Hopefully some of you who wrote comments will also write and submit replies to Reflect Orbital's FCC response. I'm sure it's total bullshit, and I am not sure when I will have the strength to read it.

Fuck you, Reflect Orbital. I hate absolutely everything you are planning to do. Really tempted to submit THAT to the FCC...but I will rethink when not 100% exhausted.

so I got laid off again, for stupid AI-pilled reasons.

Anybody need a writer or editor with:
- 8+ years of technical writer/content strategist experience, mostly in software
- One published short fiction anthology (as editor)
- A master's degree in sociolinguistics
- A knack for learning new content management systems
- A love of working with invested SMEs
- A basic understanding of web dev
- Strong boundaries
- An aversion to generative AI?

Prioritizing remote work in Canada.
#GetFediHired

Allow me to introduce #MLL coding, the counterpart to #LLM vibe coding. MLL (Manual Labor of Love) coding allows one to spend more time doing a thing, and lets one get better, faster, and 100% understood code.

Blogged "Why I Created phpc.tv"
https://afilina.com/why-phpc-tv

The difference between cucumbers and pickles is jarring.

The famous #Bryggen in #Bergen was traditionally called "Tyskebryggen" – The German Pier – and many street names in the city has influences from the #Hansa times.

After the second world war, there was a mood to change a lot of these, as shown by this sign, meant to supercede the street names of Upper and Lower Hamburger Alley (Hamburgersmauet). Tyskebryggen was also officially renamed to Bryggen, a change that wasn't reversed until 2015.

"Jøssing" was originally used by the Nazi occupiers as a derogatory term for English-friendly Norwegians, but turned into an honorific among the resistance movement – with jøssing jokes being a resistance activity with Nazis as the butt of the joke. The monogram is the one of Haakon VII, the war time king.

Photo by Jacob Blaauw Kooter, via University of Bergen Library: https://marcus.uib.no/instance/photograph/ubb-bs-ok-23408-149.html
#Norway #Norge #NorskPix #Historical #BlackAndWhitePhotography

Expected to find sandy beaches and found large boulders?

#OSIRISREx's encounter with asteroid Bennu resulted in a mystery for scientists who tried to understand why the asteroid didn't behave thermally the way they expected, based on what the spacecraft saw during their encounter.

The rock samples brought back to Terra held the answer to the mystery. Here's a cool graphic as an appetizer for the article.

Credit: #NASA / Scott Eckley

https://science.nasa.gov/missions/osiris-rex/asteroid-bennus-rugged-surface-baffled-nasa-we-finally-know-why/

#solarocks

so elon musk told xai employees that the company needs to build a satellite factory on the moon, then launch those satellites into orbit using a giant electromagnetic catapult. he called the end goal a "sentient sun" of orbital ai data centers

(this is the man who last year said "the moon is a distraction.")

so that's a lot. a whole lot

lets just talk about as a fun little thought experiment what would be involved in a factory on the moon

#kagi, one of the very few internet services that is completely worth the money, has added “LinkedIn Speak” to its translator as a destination language.

Here’s what it did about my sock.

Top shelf internetting, here.

Just laid off along with whole team. I'm in the middle of moving. Movers coming on Saturday. I'm not done packing. Good: time to finish packing. Goodish: cheaper rent at new locale. Bad: older guy computer programmer, but not even in COBOL [C, Perl, Python, PL-SQL, RedHat and Debian Linux. Created a custom RPM just Friday.]

Looking. CT / NYC / remote.

If you want to get into #WomensCycling today is a good day to watch a bike race. It's Trofeo Alfredo Binda. This is the only monument of the women's spring campaign. This race was there before any other, it's over 50 years old and has always been a women's race. It has a finishing circuit with a few nasty Italian climbs in it. It can be a (reduced) bunch sprint, a small group of 2-3 riders going to the line or a solo victory and the odds of each outcome change 4-5 times during the final. This is what Milano - San Remo used to be like.

#Koers #CouchPeloton #TrofeoAlfredoBinda

Lot of LLM users out here thinking they're built different.

The mystery of the fossilised keyboard 📽️

Peeps! Peeps!! FREE showing of The Importance of Being Earnest with Ncuti Gatwa!! 🤩 *swoon*

I am THRILLED. I wanted so much to see it in theatre last year & hoped one day there might be a way to see it!

https://www.youtube.com/watch?v=obX-HGs-PS8

#OscarWilde #NcutiGatwa #Theatre

Happy Stab a Dictator Day to those who celebrate.

@ramsey ugh yes. it's a little bit brain-breaking when i come on here and read (appropriately) negative LLM sentiment about ethics and impact, all of which i share, wrapped around a collossal failure to understand how "ai-native" work environments are actually using these tools.

i'm currently required to use LLMs well if i want to stay employed. here's how engineering at my company works:

1. interactively brainstorm a design spec. the model pulls in whatever it thinks it needs from the codebase, plus supporting docs about your problem from whatever task tracking and doc storage platforms you use, because they've all got MCP servers or CLI tools now. you and the model are gonna go back and forth for a bit as you decide what your solution should look like.
2. spawn a couple agents to review the spec against the codebase and all the supporting context.
3. fix whatever the agents found, then review it yourself. do a couple rounds of this.
4. the model generates a detailed implementation plan from the design, built around a TDD approach and including comprehensive tracing and metrics for your observability platform.
5. spawn a couple more agents to review the plan against the design and the supporting context. sometimes you find design flaws and go back to the design and start over. that's fine.

this is the slowest part of the process. it takes me anywhere from half an hour to a couple days, depending on the size of my project. i'll probably ask my teammates for a design review on the really big stuff, which means they get their agents to review it.

you now have a detailed implementation plan.

1. spawn an agent or six and have them chunk through the plan autonomously. i dump them in a sandboxed environment and take the dog for a walk.
2. use every code reviewing Skill™️ you can find to spawn agents to review the implementation against both the spec and the plan. i have four i use regularly, one of which is a "team" that cross-verifies its findings across different focus areas. have the agents write markdown docs with the issues you think are worth fixing.
3. spawn more agents to review and fix the issues.
4. push a PR for team review, which means (again) the humans on your team get their agents to review the branch and then decide if their agent review has anything worth flagging.
5. point an agent at the review comments. this part might go a couple rounds.

congratulations: assuming you used a frontier model, an opus or a codex, you now have a functioning implementation of your spec.

if you deploy it and it's broken (which obviously happens, but not as often as you'd think), spawn agents to figure out what's wrong with it (which is possible thanks to the observability touchpoints from the implementation plan), fix it, and redeploy it.

anyway, this is how tech companies with money to burn, a high tolerance for risk, and absolutely no ethical awareness or consideration for environmental or social impact are using LLMs. we are very far past the point of "but it doesn't work".