graeber_28927

To me its no surprise that 100% vibed AGENTS.md are unproductive.

To me it is! I thought if an AGENTS.md is researched with much effort once, next time reading the brief conclusion would save on effort and time and cost. Would have seemed logical to me. Like a glossary, looking up a manual, getting briefed before starting.

But I accept the research findings, and I have even had bad experiences with an outdated AGENTS.md derailing the agent.

So now I'm trying to keep to the basics, e.g. just a sentence or two saying "use bun instead of npm" only because I've had to manually prompt that too many times.

It's remarkable how good agents are at exploring, and at treating relatively sparse code reads and get a good picture nevertheless.

I'm guessing it's not a hard coded function, the button invokes. Instead it spawns a claude code session with perhaps some oredefined prompts, maybe attaches logs, and let's claude code "go wild". In that sense the button's effect wouldn't be programmatical, it would be nondeterministic.

Not OP, just guessing.

Electricity analogy is fairplay, but ChatGPT had something like 110% global adoption 5 minutes after its release. The infrastructure and the electrical appliances had to catch up, but the Internet is all built out already.

So I think it's fair to be looking at results a few years in.

Andrey Karpathy famously mentioned in an interview with Dwarkesh Patel [0], that the computer doesn't show up on GDP numbers, there's no noticeable jump or change in slope. Even if Excel is so damn fast, people are likely not drawing its full potential, and institutions are likely actively resisting change anyway.

My take is that the general population hasn't found the productive levers yet, they're at the stage where they're happy to drag down and auto generate the date list in Excel, but don't know to adjust diagrams or read function docs, not to even mention VBS scripting. And the enthusiast (dev) community I'd say is starting adoption with internal tools, and shot-in-the-dark apps, but big successes need time to mature in all the other ways (design, reliability, user feedback, marketing...), which comes back to what you said also, that needs time. Product Market Fit isn't happening automatically by chance or good prompting, I would like to think.

[0] https://youtu.be/lXUZvyajciY?is=CBJI4hIr6w_UHVs9

I thought I was crazy.

Every other month I drive 800km with my Opel Corsa-e, it takes me 10-12 hours and 4 charging stops.

This is 2 hours more than what my friends and relatives manage with their ICE car.

And since I have a child, I've been doing it in 2 days. Allows me to spend half a day packing, spend the other half in the car on the way to a hotel, and the next day I can be with the family for lunch. Kid is also not sitting tied to the seat a full day this way.

But I admit it is getting a bit much, and since I can't charge at home, I'm not getting the financial advantage either, so it's lots of travel time + hotel, and all the gain I get is that I love my car and don't want to switch back :D

Driving 950km with an e-up: I totally believe you that it's fine, but I'm sure I would be eyeing for a slight upgrade.

10% mora range isn't worth double the car price, probably. My personal benchmark is a 200km stretch on my way where I cannot reliably charge, and my current car is about the cheapest fully electric, that can do that. Even if I'm driving 90km/h behind trucks in the winter.

But good in you, nevertheless!

Ice car drivers love to make us feel miserable for the extra time and inconvenience, but it's a package deal, and overall the inconvenience is relatively easy to mitigate by smart planning, thinking outside the box.

And now I want an e-up for my wife :D

Yes, my thoughts exactly. I hate implementing N forms with user state session and navigation, when one big form one the client can hold the state for me, and visual trickery can achieve the same.

Whenever I go debug unnecessary state machines, or have to refactor them (to compress the number if steps), I scratch my head half the time, trying to follow the string of thought that my predecessor felt so smart about.

On one hand I agree. It makes me sad but I'm skeptical they are going to make it.

On the other hand, electric cars seem to be relatively "easy" to build. Sure, Fisker went bankrupt, but Rivian seems to do sort of fine. Xiaomi even managed to build a car, and I actually saw one of them by chance charging next to me today.

Seems to me like a lot more newcomers succeed in getting cars built, than was and is the case with ICE cars.

Whoa, the "japanese" checkbox animation is awesome, so satisfying!

Best part for me is it's model agnostic. I liked Claude Code, it worked better for me than VSCode Copilot Agent, but Claude was too expensive, so I rarely used it, and the price/friction felt bad when I did.

sst/opencode I can use with my existing Copilot subscription, and select Claude Sonnet 4 freely. I never hit the limit before, and all friction is gone! If Google ever builds a better model, I can switch the same day, and keep my workflows, configs, etc.

Also, with Claude COde I always felt a little mistrust, since theoretically they benefit from providing a more expensive service. opencode doesn't have this misaligned incentive.

I still log on from time to time, about once a month, because my family uploads pics in a group after family events, and fb is the only place eveyone is on.

Everytime I open FB, and scroll a little, I feel like an outsider looking into a crazy circus of clown people. I don't get why others don't see what I see.

"I just graduated" - someone whose sibling I met 15 years ago, but whose name I wouldn't recognize without a picture attached

"So proud of my daughter" - a mother who attended a piano concert, couldn't wait to even get home before posting on fb, and will tell her family at dinner how many of her adult friends commented on it, whom none if her family care about.

<vacation selfie> - middle school classmate, or former colleague, who posts swiss climbing and bali surfing images every week, as if they had no life no work just money to burn.

"<long story>, so anyway, that's where I found this purse, please share and repost. You have to prove ownership by reciting the washing machine instructions on the label." - individual who has nothing happening in their life, so they make a blockbuster adventure out of someone's misfortune, rather than handing the purse to the reception in the restaurant where they found it.

Oh, and I always see ten new invitations from people, who would never contact me individually, to events those people know I would never attend, for the nth time despite me never having reacting before.l

Once I caught an old acquaintance of mine recycling their Swiss holiday photos on Instagram a year or two after the fact.

Basically, 2-3 classmates of mine shaped 90% of my view on what a normal family vacation is supposed to look like. Whereas actually, half my class didn't leave the country in any given year, especially if they had renovations or car repairs going on. I later learnt some of the parents were financing even the bus and cinema ticket for one of the girls on a one-afternoon class trip.

I was thinking there will probably be nothing from my home country (HU) since it's a small country, and iPhones aren't as popular anyway. People are comparatively price sensitive.

And then in the 5th video that got recommended to me, the language seemed familiar, and sure enough, it's hungarian. IMG 0397, with 18 views.

The sunroof needs to phone home to work?

IKEA regularly shuffles around it's layout. They won't let you learn shortcuts and exploit experience.

I learnt this the hard way several times in my local IKEA, but I also read an article about how this is absolutely strategic in their thinking. Unfortunately I couldn't find it to back me up.

I love the thought that middle age philanthropy apparently carried potential for making the hackernews frontpage 500 years later.

┳━┳ ノ( ゜-゜ノ)

Problem is, the idiot customer copies and pastes the wrong thing, and then goes on bothering customer support, who go and bother me, the dev, to fix the account. Obviously there are solutions to fix all of this, but that's not how management dreamed up their website, so I'm stuck between supporting idiots and disabling paste.

Sounds really nice, I have to say. Is it the Standard range or Long range model?

Do you make those 800km with 2 or 3 stops? What's your approximate travel time, how many hours do you plan for the trip, assuming there's no serious traffic jam or anything like that?

If you ever drive longer distances with multiple stops, how far do you travel between two charging stops, and what speed do you try to stay on?

E.g.: I drive an Opel Corsa-e, and I usually have cruise control at around 100 km/h in the winter, 110km/h in the summer, and I stop every 150-200km to charge up to ~90%. I fully expect a Tesla to be faster, but I'm curious by how much.

This is an interesting way of describing it! Can you explain why the amount of methane follows the first derivative of the amount of emissions? I understand the math, but I guess I don't know how chemistry leads to that.

The Federal Emergency Management Agency ranks this area in its the highest category for hail risk on the national index.

[Grant Otten, media relations specialist] said the panels are designed to withstand hail, but the size of the hail Friday was exceptional.

I started to worry for the installation on my parents' house, but according to the article, this happened on a high risk location within the American Continent, and was still considered unusual there.

Not sure what exactly you refer to with "debugging", but I'm personally much happier with XDebug in VSCode, spamming my breakpoints and stepping through the code inside of the editor where I edit the code.

I can set breakpoints in my react app too, but it feels weird to inspect code in the browser console. And since modern frameworks get compiled for the browser, it is all even less familiar at a glance.

Reading your message it hit me how trivial it is to not take the phone with me for a minute, and yet how incredibly far I was from thinking of it on my own.

I also work on small apps, and like it there. I needed a day to set up TS the way I wanted, and I don't use strict mode. This way I can have TS check the few places I feel it's most useful, and not bother with the rest.

I feel confident making small components, and functions. But I like adding types to form data and product objects, especially when decorating product objects. Otherwise I'm prone to get confused as to whether the props have been decorated already, or not, maybe check for it? TS just feels really smooth sometimes, but admittedly I am not using it in strict mode.

LLMs,not AGI

Okay, but imagine someone strips ChatGPT of the safeguard layers, asks it to shut down MAERSK operation world wide without leaving tracks,, and connects the outputs to a bash terminal, and the stdout to the chat api.

It is still an LLM, but if it can masquerade as an AGI, is that then not enough to qualify as one? To me, this is what the Chinese Room Experiment [1] is about.

[1] https://en.wikipedia.org/wiki/Chinese_room

To me it's a feature. If I ask any human, often I get "no" as an answer, but with ChatGPT I can be confident that it will produce something, whether it's right or wrong. If it constantly said "I'm nut sure, I'd rather decline this one", I'd be tempted to ask it to go on anyway. If it's wrong, it's wrong, but let me decide. It might be close!

The first sentence seems to be wrong: > A monastic brewery near Munich says...

Klosterbrauerei Neuzelle (15898 Neuzelle) is near Berlin.

Klosterbrauerei Andechs, however, is near Munich, but the powder is made in Neuzelle, judging from the links of other commenters.

95% of the buttons I don't need to be haptic.

But please, let's have climate and volume as 2 physical dials. I promise, I'll be forever grateful, and I won't ever ask for anything more ever again. Just two dials.

Same here. My NUC eats around 18W when idle, which translates to ~58 Euros a year in electricity costs. (serves mostly as a NAS for Plex) 10x that in costs would be quite significant, and my desktop idles easily above 100W. (Ryzen 5 3600 and GTX 1650Super)

When I'm productive, I get to around 0.5-0.8€ a day of electricity costs with the PC, and gaming for just a few hours occasionally puts me above 1€/day.

I don't actually know, but the FW "abuses" splitted thunderbolt connections to get to those features, which I must imagine needs beefier controllers and is more energy intense, than solving e.g. microSD in a more direct/integrated way, without thunderbolt in the middle.