It was somewhat anesthetized compared to running out into an empty lot to play football but yeah better than staring at a screen and calling it networking.
Certainly! I left out the part where I would ride my BMX bike with my friends, without a helmet and without my parents tracking me, 5 miles in an urban environment to get to the arcade, then we'd hike the Hollywood Hills fire trails afterward.
Isn't this suitable for a Bayesian classifier? Label some data (manual + automation using substrings) and use that to train a classifier and then it should be able to predict things for you fairly well.
There's a new feeling that I experience when using LLMs to do work. It's that every run, every commit has a financial cost (tokens). Claude code can write a nice commit message for me but it will cost me money. Alternatively, I can try to understand and write the message myself.
Perhaps the middle ground is to have the LLM write the classifier and then just use that on the exported bank statements.
> Isn't this suitable for a Bayesian classifier? Label some data (manual + automation using substrings) and use that to train a classifier and then it should be able to predict things for you fairly well.
Sure, maybe, if I label enough data? At the number and variety of transactions I do, it wouldn't be be much better than hardcoding.
> It's that every run, every commit has a financial cost (tokens).
Ballpark figure for total cost for the Gemini OCRs for now (for me and a few hundred other people who have downloaded the app), for the past 6 or so months, is a few minutes of my hourly rate.
> "Perhaps the middle ground is to have the LLM write the classifier..."
There was a time when I'd read this comment and then go looking for a tutorial on building a basic "Bayesian classifier". Invariably, I'd find several, which I'd line up like soldiers in tabs, and go through them until I find one that explained the what, why and how of it that spoke to my use (or near enough).
Of course, now ChatAI does all that for you in several Chat sessions. One does wonder though, if Chat is trained on text, and that was the history of what text was available, 10 years from now after everyone stopped writing 10 Blog posts about the same "Bayesian classifier", where's the ChatAI fodder coming from? I don't even know if this would be an outcome of fewer Blog posts [1]. It just strikes me as interesting because that would be _a very slow process_.
[1]: Not that this is necessarily true. People write blogs for all sorts of reasons, and having knockout quality competition from ChatAI does not KO all of them.
Getting LLM to write the classifier should be the way to go.
That’s what I mostly do, I give it some examples ask to write code to handle stuff.
I don’t just dump data into LLM and ask for results, mostly because I don’t want to share all the data and I make up examples. But it also is much cheaper as once I have code to process data I don’t have to pay for processing besides what it costs to run that code on my machine.
I think that's what GnuCash does by default. Even with years of past transaction data it still gets some very obvious matches wrong for me. In my experience it's about 90% accurate for the ones it really should be able to do based on the training data.
> "...it's about 90% accurate for the ones it really should be able to do based on the training data."
What's the pathway for the remaining 10%? Are they simply misclassified, and dropped into a queue for manual labeling? Do the outliers get managed by the GnuCash? Or do they get dumped into a misc 9000 account?
Yeah. I recently stumbled across this in an interesting way. Went down a rabbit hole. I was recreating an old game for my education[1]. Scummvm supports Eye of the Beholder and I used it to take screenshots to compare against my own work. I was doing the intro scenes and noticed that the title screens are 320x200. My monitor is 1920x1200 and so the ratios are the same. It displays properly when I full screen my game and all it good. However, on scummvm, it looked vertically elongated. I did some digging and found this about old monitors and how they displayed. Scummvm has a setting called "aspect ratio correction" which stretches the pixels vertically produces pillarboxing to give you the "original nostalgic feel".
Parent of 3 here. It's not just possible. It's quite normal for young kids to be well behaved in public places without having an electronic baby sitter.
Sure. You’ll notice no libraries, no CI, no issue tracker, written in C, no landing page, no dashboard.
So much of the discussion here is about professional practice around software. You can become an expert in this stuff without actually ever learning to write code. We need to remember that most of these tools are a cost that only benefits for managing collaboration between teams. The smaller the team the less stuff you need.
I also have insights from reading his C style but they may be of less interest.
I think it’s also impressive that he identifies a big and interesting problem to go after that’s usually impactful.
Very much agree. In the rush to "simplify" writing, we've stripped out a lot of the colour in the prose and made it boring. Sentences have a certain rhythm which becomes even more apparent when they're read out loudly or performed by someone with good vocal training.
I can see the appeal in, perhaps, technical writing but even there, I feel that there's room to make the prose more colourful.
I think you've nailed it and it's kind of depressing in a way since there were a lot of things kids were part of just a few decades ago. The local football group, the bunch that gathered together to watch a TV show at one guys house that would air at 7:30 pm on a weekend, the circle of cousins kids would hang out with during a stretch of holidays, the circle of people who they'd meet regularly during a congregational prayers, etc. etc.
These things are as missing as they're necessary in kids' lives and (non-mainstream?) music gives people some semblance of community that has a stabilizing effect.
The Discord "communities" don't have the same effect unless the group actually meets in person regularly.
In around 2002, I got my hands on an old 386 which I was planning to use for teaching myself things. I was able to breathe life into it using MicroLinux. Two superformatted 1.44" floppy disks and the thing booted. Basic kernel, 16 colour X display, C compiler and Editor.
I don't know if there are any other options for older machines other than stripped down Linux distros.
There is something to be said for soldiering through a rough phase. It's not always the right thing to do but below a certain threshold, it's necessary to build some amount of resilience.
Collapsing at the slightest exposure to an uncomfortable situation and having to rely on an extensive support structure that includes a therapist, drugs and other things should not, in my opinion, be the default.
As for Holmes, I read, re-read and practically memorised most of the canon when I was in my late teens and early 20s. Mental health was never one of my take aways. I was fascinated by the intensity of the character and how his work meant so much to him. That the lack of it depressed him might have been something Doyle observed in his patients and decided to use as a foil but I don't think he was "exploring men's mental health" in the stories. He was merely trying to make a believable detective who explains his methods. My feeling is that this is overlaying a 2025 interpretation onto a Victorian tale.
reply