californical
But before LLMs, it didn’t matter because on a good team you’d have a balance where maybe some PRs added complexity and some simplified it. Sometimes people would include a small related cleanup or reactor in a PR for code around what they were doing if they noticed complexity. That has become far less common, I think.
I know it’s not your point, but a 1% decrease 100 times is 0.99^100 = 0.366 which is a 63.4% decrease in understandability. Just throwing that out there
If you want to plan out features then use agents to implement them in small chunks, deepseek has been comparable to Sonnet 5.5.
But this is just early stages for me, I was fully using Claude models for a while and decided to switch after getting annoyed buying extra credits for personal projects.. I was like “eh I’ll just throw a few bucks in openrouter and see how it is this time” and it’s been so much cheaper than buying Claude api credits when my subscription is out
It uses tools in my custom harness about as well as Claude models and makes edits to files doing specific things about as well as I’d want.
You don’t have to ban a thing for 100% of cases, there will always be outliers. My wife and I love driving deep into the wilderness. But still the vast majority of our travel is routes with heavy traffic, let’s replace those trips
But if the pop culture lore has insane depth then it’s probably valuable enough to keep, yeah? If it has a few limited snippets and very few people writing about it as sources, then it’s a lot more questionable
It’s just a rough way to say “mostly this with a decent bit of that” because 20% seems like a medium-small percent that could or could not matter depending on the context
That’s all anyone means when they say 80/20
To your second point, the 80% could be lots of basic things and the 20% could be the important facts, so it’s not always meant as a dismissal that the 20 doesn’t matter
Like sure it’s true that you’re more necessary now, but when you took the job you weren’t expecting to need to be elbow deep in shit every day, and you’d probably prefer to go back to the days of changing lightbulbs and vacuuming hallways
That things which are good for an individual but bad for everyone else deserve scrutiny?
Haha, I don’t doubt it! Pretty damning overall for long-term maintainability of vibe-coded projects, even as the models are great at a lot of individual tasks
With a few thousand lines of code over the course of a week or so of my limited evening energy, I have agentic loops, ide integration, and a limited selection of custom tools which are all very safe (no unintended changes), along with session management and awareness of git.
It works great, is customizable, and allows integration with any provider.
If they need 30k changes per day, it’s not due to the nature of the problem!
Then all of my dotfiles are backed up, with any important other config, in a big gig repo. And all of my documents/photos/etc live on a separate hard drive which gets backed up to the cloud.
All of that to say, with the right setup nix is mostly unnecessary, and honestly the benefit of not using it is that I can wipe my system anytime and restart it with just my usual important programs ready to go. Any mess made along the way gets purged automatically in every fresh install
There is research for a lot of additives, but many have limited evidence or small studies that aren’t replicated. Usually the result is either “it doesn’t affect health either way” or “it is negative to health”. Very rarely is there an ultra processed ingredient with a large body of science that says “we should eat more of this specific thing, it has a positive effect on longevity!”. And if we do, it’s usually medicine rather than a food additive, which due to being heavily researched and tailored to individuals, prescription medications are healthy (obviously). But these aren’t that…
So I think you have to stretch your reasoning that the practice as a whole is probably negative, but I feel like I’ve personally seen enough negatives of certain UPF that I limit my intake whenever reasonably possible.
Obviously having small amounts isn’t going to be horrible but they’re in everything now
I don’t apply the same criteria to things that are intended as non-realistic art, you can have non-slop AI graphics, as it doesn’t matter there. But in photography there’s an expectation that the subject is a real moment in time at a real place.
And at least in America, most commonly-bought peanut butter has additives, and most store-bought bread has 40 different UPF ingredients. Even the jelly will often have “natural flavors” (intentional scare quotes) and pectin.
It’s possible to have a non-UPF peanut butter and jelly sandwich, I might have one today! It’s just that the default options for most Americans contain a ton of crazy ingredients.
Depending on the cheese you buy, many are very simple, and rarely have weird additives. But if you buy a block of cheese or even usually the sandwich-sized slices, it’s not ultra-processed.
But the scale really is completely different. I think the US is still overall good, and we do still have real elections, limits on government power, and genuine turnover in government where power can shift.
And even as the US govt is trying to scare people away from protesting and they’re building a corporate-government surveillance state, personal freedoms are still very high compared to China and Russia.
Their misunderstanding of not being able to hover their cursor over something and interact using the keyboard honestly has me wondering why I thought the real way was intuitive, their logic there makes sense
I feel like I’m going crazy, using all of the best models, spending time to have excellent prompts, configuring tools and skills… and still getting overly complex solutions with mediocre results.
Like it’s still impressive how far we’ve come, and undoubtably cool technology. It’s made a bunch of personal projects possible that I never would’ve started.
But for a business I’m struggling to see the ROI. Sometimes there’s a big benefit and sometimes it’s net negative. Not saying we won’t get there but I’m trying to stay grounded in the reality of today rather than the hopes of where the technology could get to
I had Opus trying to simplify a query for me which was slow - it ran for maybe 30 minutes, including writing and running tests, and came up with a refactor across 9 files with a couple hundred lines changed. I was looking through the output before moving onto the next step, and noticed something a little fishy- I said “why does it do x, isn’t that a more complex y?”
Opus thought for another 20-30 seconds then output “Actually that would make the majority of the diff irrelevant, if we do that change it is just these 4 lines in this single file instead.
So then I had it do that. 5-10 minutes of writing and testing and that was done.
So my company spent $25 in tokens and I spent probably an hour in total for a 4 line change that, in the days before Claude, I probably could have found the correct file and thought through the problem, understood the solution, and written the 4 lines of code myself. Probably in the same amount of time.
So basically there was no benefit at all for my time, an extra cost to the company of $25, and now I understand our codebase a little bit less instead of more if I had done all the work.
As good as Claude is at building greenfield projects it still struggles a lot at complex ones
Isn’t this the issue with using it the way you’re suggesting? At best the model can come up with an after-the-fact rationalization of how to get to the solution, but it doesn’t know what actual path it took to get there - what were interesting traps it fell into, where was a place it was close to the solution but didn’t realize at the time.
Those are things that are valuable to share between humans, those which teach us how to think better, and give us deeper understanding ourselves, and which a model doesn’t have any comprehension of.
It has no memory or experience of working on similar problems. Even if it made one of the foundational libraries that I use in a weather forecasting program, it still has no comprehension of the thought process it takes to understand the problem and build it from zero, and if I’m building on that library it just makes fresh assumptions about how things should work.
It’s not a human with experience or expertise, it’s a computer program that’s really good at turning English descriptions into functioning code
What if you made the data only recorded for flagged license plates, for example? Where all information is discarded at the camera-level (not even sent to remote servers) unless it matches a specific target given by police with an expiration date, with an audit trail for every entry, and manual blind-solution human review to confirm any matches before an event is even recorded
Destroying millions of acres of old growth for cattle farming, which is essentially a luxury for human food, meh.. maybe we’re better off growing cattle in other places that are more suited to it, and eating less beef so that nature in those forests can continue nourishing those parts of the earth.
Some uses of the earth are good for both humans and their environment, and some are good for humans in the short term but at the cost of long-term sustainability.
And if we do things that aren’t sustainable then we use up valuable resources, when doing things in a more balanced way could support and benefit humans indefinitely.