mglvsky

finally: unshittification

Imagine SRE that very quickly triaged the incident, gave solution and it turns out that solution made system ever worse, and after that he/she said: "You're absolutely, right!"

can I choose Lambo as tractor?

This would be true when software factories succeed. For now, AI needs initial input from a human

can I use this harness with subcription plans without being afraid to banned from violating some ToS?

Why they didn't prompt something like "please make whole system more reliable, don't make mistakes"?

are there clean-room games?

eternal hacktober

My current take: at the moment, vibecoding is much fun as being IT manager.

YMMV, not so fun for me being human/robot proxy constantly, there is "review dilemma" with AI tools:

review everything AI does - getting burned out quickly

skip review what AI does - getting wake up call in a weekends' nights and getting burned out quickly as well

Maybe I'm in the bubble, but I don't see a lot of "software factories bros" here, there are plenty of "coding is solved"-bros (they're also annoying btw)

edit: grammar

impressive, but brand websites mostly "eaten" by curated accounts in social media

I've lurked about PKPASS[1] file format a little bit, sounds like any vibecoder can make editor of those files, but maybe I am missing something (verification?)

[1]: https://en.wikipedia.org/wiki/PKPASS

Call me introvert or not, before doing the exercise I feel a little resistance to do it, but after - feel very energized. For parties: it's diametrically different - feel big boost of dopamine before party, but after - I feel some regrets, so I'm just trying to not think about later.

...that uses SQLite as its storage engine

but where are new shiny apps from whom who uses $RECENT sota models?

Here is my approach:

"coding is solved" == "gastown-like systems give a brand-new and useful software"

I don't recall whether GasTown succeeded...

From my experience, the game crashed sometimes on Windows 7 (on old HP ProBook)

I'd play Messiah[1] - such a hidden gem

[1]: https://en.wikipedia.org/wiki/Messiah_(video_game)

that could be "new" technique to use agents in software development: instead of tiresome prompting/reviewing large swathes of text - pointing to specific area in a map, just like in a strategic videogame

A good quote: "Sell people a worldview, and they will advertise it for free because criticism of the product now feels like criticism of the self"

> We just need better, efficient and faster GUIs to put these TUIs to an end.

I think, firstly, we need less meaningless red-tape shenanigans from MS/Apple to publish GUI apps

For me as introvert, it looks like naive dream about getting big income solo, without dealing with other people and politics and all "jazz". But it turns out there are clients to engage with. Oops. So why not doing business with a few other pals?

I'd go further - just leave social media

> If they give that up, they are pushed into insecure positions where they cannot be replaced by AI and machines

Is there typo? Great passage anyway, but I think the virality still isn't only venue to "enough money" for poor. For "shitload of ton of money" amount - perhaps

That combination of soundtrack and graphics reminds me of [1]

[1]: https://en.wikipedia.org/wiki/Rock_n%27_Roll_Racing

I think, considering length of "self-help content", there is spectrum, and in the both ends of such spectrum there is conflict of interests - audience vs authors.

In the first end - short-form content: yt shorts, tiktoks, tweets, ig posts - "authors" have interested to just have more engagement, so we end up with ragebait, fringe ideas, etc - everything for just viewers count go up, the substance doesn't matter at all.

In the other end - "long-form" content: books (usually thick ones), writers for some reason need to fill extra content alongside the substance, some off-topic stuff, I'm not professional writer/editor, but I suppose it's protection from some nitpickers, adding some scientific basis to look more solid (don't know why: author can pick any research that support's book idea, even already debunked), maybe there is requierement from publishers, editors, etc.

The sweet spot here are thin books here, but even that case they cannot help much, because even a therapy sometimes cannot help for someone (heck, there is a real person trying help, not just text with ideas!)

I prefer this policy: https://github.com/ghostty-org/ghostty/blob/main/AI_POLICY.m...

> If you can't explain what your changes do and how they interact with the greater system without the aid of AI tools, do not contribute to this project.

edit: added that quote

what about barebone/starter desktop app that can be modified itself by prompts?

that's would be Electron app, but without unneeded bloat

so is harnessing tmux/tailscale new "rsync/FTP is enough" thing nowadays?

thanks, will check the video