Fears about too much driving is worth addressing, but we need to reduce VMT regardless if the driver is autonomous or human. It's not helpful to somehow pick on AV VMT. So is excessive gas-powered, single-occupancy joy-riding in fart cars okay now if it's human? Pollution is pollution, tax it all...
Waymos are far safer and many skeptic pieces have been double counting when comparing with human drivers. I haven't had time to write about it, but I'll say it's a warning sign that skeptics have been unable to identify this obvious numerical issue.
Also trying to use Tesla to tar Waymo. /
The verdicts against Meta confirm what we’ve known all along: they are harming our kids.
As a parent, I am committed to holding Meta accountable - and we’ll do just that when we go to trial later this year. Our kids deserve better.
www.nbcnews.com/tech/tech-ne...
Bluesky is still too much of a mirror of Twitter. "since:" for search makes Top a lot more useful and is apparently supported but just as hidden as it is on Twitter: bsky.app/search?q=%22...
Ah yeah the permissions layers can be really annoying when you don't want them. And across installs, the built-in usernames and passwords even vary which is very confusing.
It would be nice if there was an easy to use standard Docker image that makes it more like sqlite by default.
Uh oh 😅. How did PostgreSQL hurt you?
So even if AVs were the silver bullet, you'd have to have an answer for why people seem to tolerate such deadly devices at such high cost today while good alternatives already exist. Anyways. It's not just about knowing what the right answers is my point. Something we should all thinking about more.
Because there is a huge body of work around Vision Zero and safer streets about what works. And it's all very cheap and will save both lives and money in the long-run. Yet American society has not rushed to adopt proven life-saving safe streets measures. /
There are related lessons here to think about for people frustrated with adoption curves of other ideas. Housing, safe streets, etc. For example, if you accept AVs are safer, "Why don't we just adopt autonomous vehicles to save lives?" is actually a very weird question. Why? /
On a variety of subjects, the Discourse tends to demonize people with bad ideas as uneducated & irrational or comically self-interested villains. But the educational establishment does not seem to be obviously either of these things. There is something else with adoption of ideas that goes beyond /
The ideology tried to have kids just guess using context and memorize sight words rather than show kids the phonetic structure of written words. It never worked but was a national phenomenon. The push back is many years long and even with a new good policy direction in NJ, reversal is uncertain /
Thinking about technology adoption: For good ideas, a lot of times, it's not the research or data that is lacking. Proven ideas are often well-known.
A woo woo educational fad called "whole language" and "balanced literacy" moved away from phonics and destroyed reading for a generation. /
I just signed up for Claude as I respect Anthropic for putting responsible use above profit. Hegseth refuses to agree that AI shouldn’t be used for mass surveillance of Americans and for creating robots authorized to kill without a human decision-maker. (1/2)
I think also worth contemplating that machine translation of human languages was really good even before this era of LLMs. It's an easier task. Moving between programming languages or frameworks is often very valuable as it is here to prevent vendor lock-in and lower compute costs.
AI Software Translation: People have been bouncing between bullish and skeptical about AI coding. This is a good example where if you have something that is very well-defined, AI is the clear catalyst that enables a little bit of human and a little bit of compute combined to go a really long way.
Disturbing hypocrisy – reports on the other site saying that Anthropic appears to have distilled *from* Chinese model DeepSeek. I was also able to personally reproduce.
Much has been made of the AI-driven SaaS-pocalypse's effect on employment and the broader economy. But on a lighter note, I wonder if there is a detectable downturn in asking friends simple questions.
Every few months, I write an updated, idiosyncratic guide on which AIs to use right now.
My new version has the most changes ever, since AI is no longer just about chatbots. To use AI you need to understand how to think about models, apps, and harnesses. open.substack.com/pub/oneusefu...
Oof. So far there doesn't seem to be one model to rule them all on everything yet.
Saw this today, seems pretty interesting. Claim is changing how diffs work / context can improve Gemini for coding by 21%: blog.can.ac/2026/02/12/t...
😂 Uh oh. Net positive though? Or Claude Code or bust?
Clever. They've protected *you* from the bloatware pit that is Confluence.
The incentives for implementing a new corporate pseudo-religion are so high. Oh look, Atlassian has a translation page for ITIL to Agile. Of course they do 🤣: www.atlassian.com/whitepapers/...
No other software came with trainings that turned methodologies destructive to positive team dynamics, creativity, and flexibility into a pseudo-religion called capital-A AGILE. No other software religion protested about something so deeply – waterfall – that they most engaged in.
JIRA bravely asks, what if we created a system to catalog all the things that don't matter, then reward teams for maximum distraction aka "velocity"?
Confluence bravely asks, what if we were Google Docs but much worse and after you write everything down, you're never able to find it again?
This is very 2010s tech trauma and spiteful, but I don't mind that $TEAM is in decline due to fears of being replaced by AI. No other company did more to create little dictatorships within the industry with more annoying software. Go ahead and drag JIRA and Confluence to the trash bin please.
Just an utter catastrophe for the industry and for democracy.
This chart. Wow. reportearth.substack.com/p/the-washin... via @chriscmooney.bsky.social
There is so much hype about Anthropic because of Claude Code and business success, but for Q&A, I still find if I really want to check something, GPT-5.2 is worth waiting for and has much stronger logic. If I want conventional wisdom / pretty good + speed, I find myself going to Gemini.
