Rick Ballard

@rballard.bsky.social

DevTools engineer, @ Apple since '05. Cheerful Californian. World traveller. Food geek. I'm not a snob, I'm an enthusiast. (He/him) 📍 San Mateo, CA Also: 🐘 https://mastodon.social/@rballard

Rule of thumb: AI writing is for other AIs, humans deserve human writing. Maybe not an *absolute* rule, but if something isn't worth the time to write yourself, at least clearly mark what part is AI and include the human touch above; and don't assume the AI part will even get read.

i'm excited for newer models like Astra that are trained to subagent well imo a large portion of the "AI did dumb thing" complaints are solved by looping with a cheap verifier subagent to check results but teaching people how/when to do this will be quite hard, needs to be trained into the models

As most engineers have adopted AI now, it’s helpful to realize that this is the worst that AI engineering will ever be. Models & harnesses will get better. Engineers will get more practice. Scaffolding & context eng will improve. Organizational culture & process will adapt. This moment is the nadir.

I think of negative LLM impacts by “default outcomes”. A default outcome is high-velocity low-quality PRs. A default outcome is skill atrophy. Etc. I want more people to see that these are *defaults*, not inevitable. If we approach each of these as a problem to be solved, we can steer the outcomes!

It is wild to me to still see popular takes calling AI coding snake oil. Most people haven’t built the expertise to get the massive benefits (and going beyond vibe-coding takes expertise!), but it’s real. The longer folks are in denial, the later we’ll start grappling with how to mitigate the harms.

The fun thing about agent harness engineering is that it forces you to do meta-cognition. How *do* I solve problems? Do I have a process? Where do I switch strategies? When I work, much of this is tacit. When building an agent factory I have to reflect and articulate it.

I don't really feel comfortable with MCPs. Like, one of them literally digitized Flynn and trapped him in the grid, but we're supposed to just trust them now?

Despite getting somewhat inured to the unending stream of terrible things from this administration, today's threats are a deeply upsetting new low. TBH having some trouble functioning this morning. Also thinking of my HS-teacher spouse, who needs to go explain this to kids today.

It's really remarkable how dramatically the way I work has changed in just a few short months. The parts of my job in front of a computer are almost unrecognizable vs. before. I'm lucky that it clicks so well for me, but I can completely understand why so many engineers are deeply disconcerted.

And what is the emergent shape of the system in which many more people are doing hyper personalized software that has the ability to interact with standardized software, I don't think we've ever REALLY been in that world

I love this -- a really fantastic start at turning LLM-assisted coding from something folks fear will sap their skills into a great opportunity for clear-headed learning and continuing education. And instead of making those separate activities, why not both at once? Try it out!

Cat Hicks@grimalkina.bsky.social · 6mo ago

Key to efficient learning is realizing how we ACTUALLY learn, not just what FEELS like learning. I wrote a Claude Skill for some friends to help them think about this and they've liked it -- see Principles for some directions you could explore github.com/DrCatHicks/l...

For all the talk about AI coding slop, the right workflows can go the other way. I pride myself on writing meticulous code & building deep understanding, and can change fragile code without regressions; but running my commits through a good AI review often still catches things I would have missed.

Manifesto: 1/2 Early AI boosters (the type that pivoted from NFTs/metaverse shit) are basically a subspecies of semisentient slime mold. These are the people that made genai image/video gen the visual shorthand of fascism. They deserve no credit for accidentally guessing right.

I haven't been posting about AI because people get dogpiled for it. And it is a societally complicated topic. But: I have been having more fun lately than at any point in my career. (And I've had a fun career 😄). Agentic programming has gotten *really* good. Especially with a thoughtful operator.

Cooking turkey on a charcoal grill once again has proven to be the best. Break down a turkey into parts, like you might a chicken. Kosher salt & leave in the fridge for a couple days. Grill; take each piece off at the right temp. Gives you moist, delicious meat with just enough smoke to elevate it.

I am well aware that they are and will continue to do massive damage to this country but does it feel like they’re on a glide path to successful authoritarian consolidation anymore? I submit that it does not

Addressing the housing crisis requires lowering costs, not just legalizing housing. One great proposed reform is allowing single-stair building design up to 6 stories, instead of the dual-stair design current CA fire codes require. Easy link to send a letter here: actionnetwork.org/letters/make...

Tell the State Fire Marshal: Legalize Single-Stair Buildings in California!

Improve California's building code so that more families can afford homes in our cities.

actionnetwork.org

My new cooktop is induction, and induction is *so* nice. Heats up startlingly fast, and then easy to control. Water boils shockingly fast. I’ve zero regrets about saying goodbye to gas.

I hate how the modern answer to "what file format should I use for this config file that should be both human and machine editable" is basically all bad options. Everything out there has serious limitations, or isn't widely supported. For all its verbosity, I kinda miss XML?

The passage of SB79 is a huge achievement. (No thanks to my representitives on the penninsula). It really feels like there's been a sea change in how housing policy is seen — albeit a change that has had to be fought for tooth and nail.