Am I crazy or does the skill itself look like slop, complete with disclaimers back to the user embedded into the skill file itself?
"A warning about the original post. It came with an overview graphic of STE. A later fact check found errors in that graphic. For example, "approximately" is an approved word, and the dictionary has no entry for "in order to". This skill follows the rules of Issue 9, not the graphic."
I'm totally fine with that, but I think there are still quality standards that if a bit higher, would let the skill at least be much smaller and thus less token-consuming. One place where I draw the line is when somebody, with minimal effort using an LLM, could have improved the quality of the LLM generated output. Also, I treat skills like programs and skim all external skills to make sure that they're not doing things they shouldn't be doing.
What you linked to is of construction contractors building the plant, not assembly line workers. It's a blurry line and certainly affiliated with BYD, but car assembly itself is moving toward 'dark' factories, populated by robots that are indeed kept in slave-like conditions, unpaid and worked beyond humane shifts.
As a side note, counterintuitively BYD has a lot more workers per car vs a lot of their competitors, but that's because of their extreme vertical integration where most are assembling upstream parts.
BYD pulled off the same thing in EU as well, in Hungary [0].
The dark factory part is simply propaganda, the viral WSJ interview is only about body assembly which is already automated everywhere. The labor intensive part is the general assembly and even they admit this in the video. However here it's like Chinese EV found the silver bullet in car manufacturing.
A simple subway arrival time web page for me and a few of my neighbors showing past headways, predicted next trains based on simple regression, and any alerts. The problem this solves is is we live just 2 stops from a terminus station so that the official subway API predictions start just 2-3 minutes out, when a new train leaves that terminus.
And it also runs a Home Assistant instance inside a docker container, and serves as an MQTT broker for my ESP devices. Everything is through a Cloudflare tunnel.
As for maintenance, your AI harness of choice combined with passwordless sudo (and a good backup system in case things go really south) can make upkeep very pleasant.
From recent ChatGPT (GPT5.6) conversations where I've seen occasional reasoning leaks into the UI, it's clear that something like this is already implemented, and I'd speculate that this is the majority of recent claims of less token usage. Not sure if they are literally prompting for cablese of course.
"Need check output vs prev. Ran script, results fine, need prep next step. Ready? Go."
I suspect you might be right. My conclusions from the digging are that this kind of compression works best with settled instructions/data for machine to machine talk. For something like OpenClaw (which I use a lot), that might mean the AGENTS.md, TOOLS.md, etc. Compression there would free up the context for the agent.
This is their CoT reasoning trace. (Why you see it: Models are supposed to delimit "actual" CoT, their user-visible summary/"cleaned CoT", tool calls, and user output via delimiters, but don't always do it perfectly.)
Rewarding terse CoTs during training seems like a no-brainer, as long as it doesn't impact capabilities, so I suspect the style is completely emergent and probably what you get when you implement a dual goal of terseness and capabilities while still punishing completely non-human-readable CoT. (Failing to do the last part would probably have models speak in ominous Unicode glyphs in no time.)
Reminds me of how we used to search with keywords that ended up sounding like caveman speak, then there was "natural language search", and now the AI is talking to itself in caveman speak again.
And make sure to read the reviews as there are apparently fraudulent versions out there that don't include the 4MB flash chip. That said I took a gamble on a bunch of $2 C3 super minis, and they've had surprisingly adequate signal strength for indoor applications, even in the basement. (-60's dBm)
I've embraced postmodernism and my latest hobby is using LLMs to connect ESP32s to random stuff in my house, so I can unironically WiFi-enable everything from my toaster oven to my 1950s 4-chime doorbell.
I've always thought it would be interesting if the universe also operated on the principles of lazy evaluation, and the deeper you point the telescope, the more is rendered.
"A warning about the original post. It came with an overview graphic of STE. A later fact check found errors in that graphic. For example, "approximately" is an approved word, and the dictionary has no entry for "in order to". This skill follows the rules of Issue 9, not the graphic."
reply