@elana๐งprofilefollowI have a doctorate in ancient stuff but these days I'm a full-time plague avoider. Born and raised in Toronto, Canada, now living in Kyoto, Japan.wrote:
Writing.
I want to do it, very badly.
I just don't want to be bad at it.
You have to be bad at it in order to get good at it. But...
Catch 22 isn't it?
I think reading helps quite a bit, especially writing about what youโve read, even to yourself, just so you can go through the process of having to explain something that youโre only recently familiar with. Thereโs something about that process that seems to be helpful, at least to me.
Same.
The way I consume information influences how I think. Short form content makes my thinking fragmented without a clear structure. I feel that fragmented thinking won't lead to convincing writing. And I'm trying to regain some of that lost control.
Very interesting! I think something along those lines might be happening for me too.
Another aspect, always cooking in the back of my mind, "if you can't explain it well, don't explain it at all."
The fear of being bad at it is the only thing holding you back. I experienced something like that with playing music with others. Perhaps you could learn to lay down your critical brain during the creation process and that good/bad isn't a useful distinction until later when you're editing your work.
Coming from HN to post that I've been using Argon all week and it's the first Gemini model I've stuck with for more than a few minutes. It's very good for what I do and it's so nice to read back another style of voice.
I've been working with physical AI - Letting LLMs control robots in a warehouse to collaborate and collectively come up with a physically correct solution to carry out a natural language intent. We had to study the reactive nature of agents, nudges they required, etc.
haha that's a coincidence. Site looks awesome. The moving away from ROS2 part is so true. One of our main potential customers for physical AI solutions (a telecom provider in JP) communicated the need for agents to be able to pick up controls from any vendor
I apologize if that sounded obvious/non-obvious/amateurish haha. I was an intern and worked more on the agentic side of things, it was just a non-primary experiment
Is open source dead?
Everyone has the power now to make their own stacks of things.
The people aspect of open source is dead or dying. I don't think the bunny is going back in the box.
One nice thing about Open Source (at least for me) is different people doing the same thing in different ways and one of those happen to suit my taste perfectly. And in many cases, I won't be able to come with it myself even with Super Intelligence.
Evgeny Morozov has pointed out that using ai through chat interfaces is based on an individualised and privatised model, which (heavily paraphrased) is inherently anti-social.
With an extra layer now of people making small personalised tools, do bigger things still need teams of people who know what theyโre doing?
FWIW Iโd like to see more sharing of vibe coded tools instead of everyone making their own more less the same ones. Mine are here: pauladams.neocities.org
I don't think so. I have been using more open source software than ever before.
You may argue that with LLMs I have the power to make my own stack of things, but this doesn't eliminate that I still need knowledge to properly steer an LLM to create something worthwhile.
If I have no knowledge of what I intend to build, the best thing the LLM will generate is a half-assed copy of something that already exists. Which is what we have been drowning in the past couple of years.
AI can be a superpower in the hands of someone knowledgeable. That makes the people aspect of open source more important, not less.
If you offload your curiosity and desire to learn to AI, you will be a slave to it's limitations instead of being a master of its capabilities.
Of course not. I think it is much more relevant to ask "Is closed source dead?".
Anyone can now replace a proprietary program with a vibeslopped alternative which might suit them just as well. It will have the same level of trust (ie. none) and none of the other drawbacks.
I've written a small persistent agent wrapper for agent.py. I'm impressed with how well it works. I feel like a CEO issuing global memos to all my agents after reading their status reports.
For 30 years, productivity sky-rocketed with the advancement of automation and computing. Meanwhile, a decline in longevity was observed, correlated to sleep depravation. Productivity improvement goes to the owners class' pockets, sleep benefits the workers health.
For 30 years, productivity skyrocketed through automation, yet longevity plummeted as a direct result of sleep deprivation. While these gains accrue to the owner class, sleep remains an undervalued element of the worker's health. We are trading life expectancy for chump change.
That said, in Canada working class started loosing life expentancy for the 3rd year in a row, and WHO figures show a decline in improvement of life expentancy globally. But hey, I am just a no-one on the internet spewing bs :)
a decline in the improvement of life expectancy isn't the same as a decline in life expectancy, it's just that it's improving less fast now, we might be reaching biological limits however?
We tought for years that there was a biological ceiling for sprinters or marathon runners for example, with research backing it, but every one record was shattered. I tend to be cautious of biological explanations.
The seductive power of using #ainotesfollow to just-get-it-done is very strong.
One can't hurry quality. One can iterate quickly though in the hope of arriving somewhere good.
I direct a plan, then break it into phases, and each phase into sections with tasks and sub-tasks. The plan outlines all phases, but only write tasks one phase at a time, since things change. #opinionnotesfollow
That's how I've been working with #ainotesfollow for the past few months to a year. It has been amazingly consistently high quality.
just-get-it-done is for the rough sketch or proof of concepts (I do those too and a lot of them) but never for a quality product. #opinionnotesfollow
@fastidiousprofilefollow"Beware those that would deny you access to information, for in their heart they dream themselves your master." โ Sid Meyerswrote:
The abuse (in my opinion) of the auto-translate feature is super annoying. Thanks goodness for โblockโ.
In the first attempt, Opus 4.7 successfully completed it and proved it. However, the solution is to enumerate all possible space and .filter out anything that doesn't qualify my definition, which is obviously not ideal (if cheating is too harsh).
I can't think of a good way to also prevent that kind of solutions and ended up prompting "please don't enumerate and filter". Opus 4.7 then reimplemented and proved the backtracking algorithm.