russ
LightNews LightNews
banner
qqrs.us
russ
@qqrs.us
92 followers 95 following 220 posts
RC W'13
Posts Replies Media Videos
qqrs.us
more doorbot
progress notes
November 11, 2025 at 2:16 AM
qqrs.us
doorbot port feature complete
progress notes
October 24, 2025 at 3:11 AM
qqrs.us
#embedded
winnie the pooh image macro meme template regular bear: "reboot on hard fault" tuxedo bear: "watchdog restore to steady state"
October 24, 2025 at 1:49 AM
qqrs.us
fil-c previous memory models
October 22, 2025 at 2:07 PM
qqrs.us
this week's progress
this week's progress notes
October 15, 2025 at 8:45 PM
qqrs.us
www.bloomberg.com/opinion/news...
Screenshot of Money Stuff newsletter
October 15, 2025 at 6:54 PM
qqrs.us
more doorbot
more progress notes
October 9, 2025 at 9:21 PM
qqrs.us
Working on porting doorbot to the RP2350 microcontroller again
progress notes
October 1, 2025 at 10:32 PM
qqrs.us
September 18, 2025 at 1:21 AM
qqrs.us
September 18, 2025 at 1:17 AM
qqrs.us
did a crossword puzzle for the first time in years

I always felt like... what's the point? I'd rather program a computer to solve it

and now it's exactly the opposite
solved crossword
July 30, 2025 at 10:54 PM
qqrs.us
Big news, Section 174 got fixed

Monumentally stupid it ever came into effect

newsletter.pragmaticengineer.com/p/the-pulse-...
Text of article Text of article
July 17, 2025 at 5:22 PM
qqrs.us
thanks, IKEA!
“This room setup isn’t just for looks! We’re aiming to reflect your lifestyle and provide solutions to enhance your daily life at home.”
July 13, 2025 at 12:50 AM
qqrs.us
IKEA realistic NYC apt layout
IKEA showroom with single bed, dresser, small desk, bare walls
July 13, 2025 at 12:50 AM
qqrs.us
www.jonstokes.com/p/did-claude...
No, the second time around with the Markdown chunker, I did all the AI stuff wrong. I had settled into a comfortable groove with the bot, and I had subtly started changing the way I used it. I wasn’t paying nearly as close attention, and I had the sense in my sessions that the bot was drifting and degrading in performance, but in hindsight, it was I who was doing most of the drifting and degrading. Somewhere in the previous few weeks, I had started to relate to the bot as a junior programmer that I had a sort of mind-meld with, where I could speak to it in shorthand and not spell everything out. And as I began to relate to Claude Code this way, I started to relax my vigilance, take shortcuts, skip steps, and generally abandon the process that had worked so well early on. I was unintentionally giving the bot bigger and bigger bites to chew on, with less and less active direction and focused oversight. And as I was slowly slipping into this mode with Claude, I was getting more and more frustrated with the output. When all of this started to come together for me, I went back to Claude Code for a new feature and fully engaged the earlier process that I had described in my previous posts, i.e. start with exploration and Markdown descriptions of the work, break everything up into small bites, do a little at a time, closely monitor all proposed code changes, manage the context window so that it stays “cool”, etc. Needless to say, the quality of the output jumped all the way back up again. Maybe there really was some model drift in there that degraded the results — I have no way of verifying this. But in hindsight, I’m convinced that 90% of the problem was user error. I had gotten comfortable, lazy, and overconfident in the model’s capabilities, and as I did that, the output quality started to collapse. This combination of narrowing the RAG inputs via tagging and sub-draft-level instructions keeps the context window focused. The idea is to help you with dividing up a large text artifact, like a multi-section newsletter, into smaller bites that are easy for the model to nail on the first try. In general, one of the primary things we’re trying to do at Symbolic is to structure the tools so that they encourage you to do the right thing. We want to nudge you into breaking up the work in a way that will tend to lead to more consistently better results. It’s early days with this kind of work, but it’s going to be needed for the foreseeable future, at least as long as we’re working within the limitations of the current autoregressive LLM paradigm. Commentators and investors who aren’t trying to do real work with these models daily often claim large context windows or some other advancement will eliminate the need to focus the models on bite-sized chunks of work, but these people are all wrong. I think the returns to good UX are massively increasing relative to the returns to effort and quality in most other layers of the software stack. But that’s a topic for another day.
June 9, 2025 at 7:32 AM
qqrs.us
Goodwill find
artwork of the Voyager probe golden record cover
May 31, 2025 at 5:24 PM
qqrs.us
solving advent of code day 1
May 30, 2025 at 10:02 PM
qqrs.us
Borland Turbo C on the T3100e
May 30, 2025 at 10:02 PM
qqrs.us
highlights from article highlights from article highlights from article
May 29, 2025 at 4:01 PM
qqrs.us
my speaker notes
speaker notes from discussion session
May 25, 2025 at 9:20 PM
qqrs.us
I ran an AI & consciousness discussion session at @recursecenter.bsky.social alumni week

very happy with how it turned out, about 20 people came

the questions raised could have easily seeded a dozen more sessions!
Event description AI & Consciousness (NGW Edition) This is a hybrid event. Description A discussion on AI & consciousness. We'll start with a brief presentation so we have common language, then explore a few questions together.
May 25, 2025 at 9:20 PM
qqrs.us
BASIC code from new book The Magic of Code by Samuel Arbesman
one line of QuickBasic code still shot of generated maze
May 23, 2025 at 9:37 PM
qqrs.us
1988-era Toshiba T3100e “luggable” computer
May 23, 2025 at 9:37 PM
qqrs.us
if you see this post your getaway vehicle
train cresting a hill, sunshine looking on, style of R Crumb
May 3, 2025 at 12:33 PM
qqrs.us
scorching 😆
The RIFF specification attempts to be a formal specification, but its formalism lacks the precision seen in other tagged formats.
April 13, 2025 at 9:45 PM