canpan

One big pain point is not seeing the thinking trace. I noticed using pi with a self hosted model that I could spot, stop and correct my prompt much faster. With claude etc I have to first wait for it to think 3 minutes and then notice it got completely off the path I wanted.

I still use a plan, because my local model is not as smart as opus, but I can iterate much faster.

Just for reference our fully electric household (including cooking, water and aircon). Highest usage this month on a very hot day was 25kWh. (Disclaimer: I am not in Australia. Already cover most of it with roof solar. No battery yet)

This would be awesome for OpenMW, too. Morrowind is kind of dated. But I like the original style. So a curated easy install of improvement mods as a service would be great.

I work similar, but still have architecture mds for a few selected cross cutting features. As useful for human readers as for AI.

Normally I do it exactly as you say, point at a few files, but if I know these features are involved I point at the corresponding mds instead. Its a shortcut for me to type less.

Picking max(len(answer)) is the right choice almost every time at the higher level..

Really nice. I build a deck this year too. I would have loved to have a nice blog post like this beforehand. I love his page. Inspires to write. Simple, clean, blog has 2 years break here and there, but writing!

Before doing myself, I did not know just how much work it is. It looked so simple. But it took me many weekends over a few months. Digging holes for concrete feet. Mixing concrete (I found its easy to do in a wheelbarrow) and pouring. Measuring (buy a laser level!). Cutting the posts. Then beams and joists. Noticing the wood I bought was not the same thickness everywhere (I thought my post height was off, but the laser level and my cut was perfect, the beams were not). Lastly just the decking was almost 1000 screws.

All cities have the exact same shopping street somewhere.

Tokyo (Ginza), NYC (5th), Paris, London, Berlin, Sao Paulo..: Starbucks, Gucci, Addidas, Louis Vuitton, Levis, Ferragamo, Apple Store, a little further from there a McDonald's..

When buying fridge, washing machine, oven etc, when I moved, I told the sales person, I like quality, am not price sensitive, but any device that requires an app or has a camera/mic built in is out of question. Some didn't know how to handle that, being used to sell it as a "good thing".

Not only kids. I am an adult, and I also enjoy physical ownership. Physical music (especially LP) sales have been rising yoy since 2020.

I am totally a linux user. But even on the high end, the build quality of laptops more expensive than apple is often worse than apple.

Or it's hit and miss and you need to hope to get the good one. I want a lightweight, non-plastic laptop with good keyboard, solid battery life, and no hinge problems. Apple is consistently delivering it. Good luck finding that outside apple.

I think this is the main point. Most articles conflate consciousness with intelligence or awareness. Without clarifying their definition of it.

To quote wikipedia:

It has been the topic of extensive explanations, analyses, and debate among philosophers, scientists, and theologians for millennia. There is no consensus on what exactly needs to be studied, or whether consciousness can be considered a scientific concept.

Same opinion. Opus is best for coding, but Qwen 3.6 27b Q8 is next, before Sonnet.

Sonnet might have more knowledge and is maybe good for making excel sheets, but it does not write good code and does not follow instructions well.

But 27b Q8 needs a very beefy PC (48GB VRAM or more), so it is not an option many people can use and DS4F is so cheap right now, if you are open to externally hosted models.

A similar thing is happening in video games. Big studios rarely have interesting games. Just GameTitle2026. Reboot of game from the 90s and no one is left of the original staff.

Still a lot of people buy those, so studios continue to make them.

Also indie games are too cheap. I noticed the need to correct my own thinking: Why should the boring game of a big publisher cost more than the great game made by a single guy? And allowed myself to use more money when I want to support smaller studios.

I wonder how they are positioned now in the market.

During covid I wanted a small low power always on server. I thought about Raspi, but at the time it was expensive and I went with an intel nuc, for a similar price.

Now if I wanted to do hobby electronics, I heard I should look into esp32 or stm32..

It might be overcompensation. I think UI, UX and GUIs got better up until the 90s, and early 2000s, but then somewhere GUIs suddenly got a lot worse. So a modern CLI is better and more standardized than a modern GUI.

I don't know about basket weaving, but..

I once had a talk with one of (the?) world's best bonsai gardener in Edogawa, Tokyo. Trees cut by him are worth millions and he has pictures of himself with FANG leaders. This guy wakes up every morning at 5 and works until it's dark outside even though he clearly does not need to work for money, but because he loves it.

How to handle domains? The rest is easy, but your domain registrar blocking you sounds like a pain. My current solution is to use a local small provider, just for the domain. Then if there is a problem with your play account it is out of any blast radius.

I always wanted to get into bsd, especially openbsd. I like the idea of a more cohesive os.

But I don't really know what to use it for to get started. My desktop runs linux with steam for games. My AI server needs rocm drivers so ubuntu-server. My vps runs debian, maybe that one, but there is no DO image for BSD. Open for ideas..

There are! This is outside US (Sorry, page is in Japanese) https://www.ichijo.co.jp/technology/energy/solar/ They are famous for integrating their supply chain, controlling all of the build. If it is possible to make it work financially, they should be able to do it

That is similar to how some of these boats look like, just more modern. Here is an example with pictures https://www.ferry-sunflower.co.jp/en/ (disclaimer, I never went that route, the ones I went with were less glamorous, more modern, but still nice) But yeah, it is mostly one night, because the distance is within Japan.

These places still exist, but you need to look for them. Here in Japan, some remote islands, you can travel overnight boat. I love those. There might be a speed boat or plane, but I love boarding the boat in the evening, everyone feels like having a party, sleeping in a bed and arriving fresh in the morning. (If you are in Tokyo, the nearest is Oshima Island).

There is also slow rail travel, with pretty trains, sleeper car and restaurant. I think Europe has sleeper trains too. I am also interested to go to Europe once by the trans Siberian railway.

It is true, I did learn Basic (QBasic) first, as elementary school student, and C second. What made it particular good was the F1 help. I think it is still one of the best help systems I ever used. Good intro and F1 while pointing on any function and instant help, easy to understand. Additionally no complicated include and compiling. But at the same time, I think I could also teach simple C or JS to an elementary school student. It is just the help and runtime that is bad. Somehow still today there is no major language system that does all of this well. Curious...

I'm In This Photo and I Don't Like It.

Wrote a basic interpreter in C/Wasm last year. Finished it, but in the end never posted it anywhere. While I started my dev journey with Basic as a child, I was quickly reminded why I moved to C as soon as I learned about it. Basic is just not particular useful and has a lot of funny behavior or missing parts for any serious project. I still enjoy C, sometimes ASM, but not basic anymore.

I would recommend to start withQwen 3.6 35B at maybe Q5, it should be fast in that setup. For intelligence Qwen 3.7 27b, is smarter but will run much more slow. Others also mention gemma 4, which might be worth a try.

Llama.cpp with automatic offload to main memory. You can also use Ollama, it is easier, but slower.

Recent models (Qwen 3.6 and Gemma) can really do coding locally. Feels like SOTA from maybe a year ago? But you would want about 32-40GB total memory. 24GB is just a bit short of that. A gaming PC with 16GB graphics card and 32GB RAM brings you very close to a usable coding system.

I am sure ads can work on me, and the HN crowd, if I was targeted. But there are much easier targets that generate more income. So companies spend more on gambling etc ads, where they can make more money.

Ads target a specific sub group. And I think many HNlers are not part of that group. My guess is that ads online target people with more impulsive buying.

Even if you showed me the perfect ad, I probably would not buy it. Because if I need it, I probably already bought it, and if I don't need it, I won't buy it. So there is not much money being made, ergo we get shown ads for the other type of person.

It's not great, but basically if your lock is better than the lock on the bicycle next to yours, they will most likely not steal yours..

I feel bad for the people working on it, but looking at the selected metric "60% faster", I don't think it is enough to turn the tides.

The correct metric would be something like "opens within one frame at 60fps". But that would require product design changes, not a technical problem.