Reading RPLS is free for the whole profession. Members post, reply, and get the members-only rooms.
I did admitttedly send Wendell a nastygram this morning when I could not read the recent replies. I felt deprived. There was some previous life membership moment in the last incarnation of the Board where I was feeling rich and threw down a certain amount. After an email he honored my previous life member contribution and I was able to catch up. Grateful. Having had an amazing day in many other ways, I scraped up the $79 to become a Founder. Suggest you do the same. Shout out to all my favorite people here.
In other news, I never thought I would apply for a job again. It has been a decade at least since I ever worked for anyone. However, usajobs has openings right now, and if the stars are aligned it might be the best way to share my amassed knowledge for the next 10 years... the essay questions give much food for thought. Go answer them.
@half-bubble No worries. Your email helped me discover a couple other lifetimers that somehow didn’t get migrated properly to the new system. 🙂 And thank you for your support then and now. 🤩
Like what we do here? Start your Membership or Donate
Need a new or refreshed website? Websites for Surveyors
Life revolves around the destinations: Freddys Lowes Lowes Freddys. Repeat. PPI or KR. RE-PC or Value Village or the used bookstore whose name I am drawing a blank on.
RE-PC being the most important.
When I needed a new machine for CAD (specifically BricsCAD) some years back I went into RE-PC and asked for "something good for CAD" & wound up with a machine with 16GB RAM and an RTX2060 with 6G VRAM. Ran it. Made a bunch of drawings.
More recently I am running ollama and one of those famous open source models such as qwen3.5:9B-q8. It was a great discovery along the way via 'watch -n 1 nividia-smi" and "ollama ps" that my entire CAD usage amounts to about 50mb of a 6G VRAM card.
I had a hot rod sitting idle until I started learning about local AI and local models.
I have upgraded yet I am still running 8G VRAM cards used for $200. More focused than any so-called "frontier" model. All these cards have doubled in price since the release of Hermes Agent in February of this year. They only last at the recycled PC store for 24 hours before they are snapped up.
Yes it can write Autolisp. But none of the models have been trained to do so, so you have to train it yourself. The models write python and other languages natively. Grok writes some pretty good Autolisp if you are patient and ask nicely.
I am stacking old books to scan and make available via X402. You should do the same.
We are at a place since February that a 1 bit quantized version of qwen3.5:27b can be run on an 8G VRAM GPU. That is the next already published thing to try locally.
In the meantime, I am running a qwen3.5:9b-q8 on a pair of rtx2070s on ollama in parallel for total 16G VRAM.
Half Bubble giving a multi-year summary. Write if you get a job.