singingbard

I actually disagree.

People only take your lane if you are in the fastest lane. If you are in any slower lane, people tend to jump in and then leave and I have no problem with people who do that.

You can also keep a gap in the fastest lane but you need to keep track of other cars on the road. You’ll observe that most cars rarely leave their lane. People who tend to leave their lane keep smaller gaps in front of them. Use that knowledge. There are many more factors than just that but if you start observing everyone drive, your little simulation in your head will start putting other drivers into buckets.

To me, they all the same because they are all tools that stand between “my vision” and “it being built.”

e.g. when I built a truck camper, maybe 50% was woodworking but I had to do electrical, plumbing, metalworking, plastic printing, and even networking infra.

The satisfaction was not from using power tools (or hand tools too) — those were chores — it was that I designed the entire thing from scratch by myself, it worked, was reliable through the years, and it looked professional.

LLMs serve the same purpose for me.

I did a fair about of data analysis and deciding when or if my report was correct was a huge adrenaline rush.

A huge test for me was to have people review my analyses and poke holes. You feel good when your last 50 reports didn’t have a single thing anyone could point out.

I’ve been seeing a lot of people try to build analyses with AI who haven’t been burned with the “just because it sounds correct doesn’t mean it’s right” dilemma who haven’t realized what it takes before you can stamp your name on an analysis.

The Internet is to blame!

But you could say we peaked before electricity too:

- It broke our natural circadian rhythm

- It enabled the 24-hr grind

- With radios and televisions, people were now staying home for leisure instead of going to parks and public spaces (TV dinners!)

- It led to immense pollution

Technology really is a double edged sword.

So you’re not missing anything if you use Claude by yourself. You just update your local system prompt.

Instead it’s a problem when you’re part of a team and you’re using skills for standards like code style or architectural patterns. You can’t ask everyone to constantly update their system prompt.

Claude skill adherence is very low.

I think the problem with this hypothetical is that technology was the main constraint back in 1900, not marketing.

Battery technology was significantly much worse. Lithium batteries were only discovered in the ‘70s.

Gas engines were far more polluting but way less complex in 1910.

If only people knew how much of Microsoft Windows has been secretly powered by HTML pages for 20 years…

In Windows 95, Microsoft let you set a HTML file as your wallpaper and let you set up “channels” that were web-based widgets. This was the beginning.

Windows 98 used webpages as core components for Explorer. Literally browsing your files involved J(ava)Script… in 1998.

Windows XP/2000 still had Internet Explorer as a core component. Web tech was involved every time you opened a folder.

Windows Shell using web tech is as on-brand Microsoft as it gets.

“Good” is some mix of taste and skill.

People without taste hide behind skill. They do everything technically correctly and still make something bad. This is the threat of new mediums to them — it takes away their only strength.

But at the same time, something like AI suddenly enables people with neither taste nor skill to produce. I don’t want to see AI art right now — AI art is currently a lot of noise.

The sentiment of photography not being real art hasn’t been a thing for a while now though.

The founders were skeptical of direct democracy because it assumes people have time and expertise they mostly don’t. People should not be voting based on their understanding of tariffs. It’s why we ended up with a republic.

But social media changes the equation entirely. It gives us the speed of direct democracy without any of the structure or responsibility. It pushes people to judge candidates issue-by-issue, often on topics they don’t understand well, while eroding the deliberative layers a republic is supposed to have.

The problem isn’t people or education — America didn’t get this far because Americans are any smarter or dumber than anyone else. It’s the design of the systems. The founding fathers built a system that has so far lasted almost 250 years.

You cannot expect people to change — safety protocols, procedures, govenments — it’s about the systems.

Looks like a Frecciarossa 1000 derailed in 2020 but it was due to a manufacturer defect in a track switch replaced the night before.

The defect was not caught by the manufacturer or the system operator. It was due to two crossed wires in an assembly.

I know a lot more engineering goes into these trains due to the higher stakes. Japan’s high speed rail hasn’t had a fatal accident in 60 years. I’m wondering what the cause of this will turn out to be.

I think the underlying approach seems sensible.

The problem with Gas Town is how it was presented. The heavy metaphor and branding felt distracting.

It’s a bit like reading the Dune book, where you have to learn a whole vocabulary of new terms before you can get to the interesting mechanics, which is a tough ask in an already crowded AI space.

This took me a long time to work through:

1. People’s beliefs are strongly shaped by upbringing and social environment.

2. A belief feeling “natural” or common does not make it correct or benign.

3. What’s most commendable is the effort to examine and revise inherited beliefs, especially when they cause harm.

4. This framework lets me understand how any individual arrived at their views without endorsing those views.

I think this is why responses often split: some treat explanation as endorsement, others don’t. Both reactions are understandable, but the tension disappears once you treat explanation and moral evaluation as separate and compatible steps.

People had already ditched writing HTML for years before Markdown came out.

People were just using other markup languages like rST.

Other attempts had already proven HTML to be a bad language for rough documentation. Someone then just needed to write a spec that was easy to implement and Markdown was that.