But don't we, in real times, turn to the people we're comfortable talking to? We are not machines ranking the effectiveness of our peers for best outcome, though some of us surely still weigh that! It would be nice to go the most effective route, but sometimes that brings a lot of pain, judgement, or unforseen externalities.

(and hopefully we pick our close social relationships to make us more effective too!)

> "according to your community norms ..." ... without much real reveal

It puts data to what our understanding of those norms are. I don't know if there's a bigger reveal than "yeah, people generally are pretty comfortable with their spouse"

Tbh I was surprised anyone would/could open up and have an emotional conversation with a neighbor (when looking at risk vs reward). But, there are a lot of types of neighbors too.

> The study included a real-time alert system for intervening if a person indicated a high suicidal intent

Because it sounds a bit dystopian.

And found a little more about it[0]

> For example, in our monitoring studies, we install smartphone apps on participants’ phones. These participants are often recruited from hospitals after presenting with high suicide risk. We tell them explicitly that we monitor responses during certain hours—for example, between 9 a.m. and 9 p.m.—and that we will do our best to respond if we believe they are at imminent risk of harming themselves.

...

> We have real-time alerts when someone crosses a predefined risk threshold, and we follow a very detailed protocol that specifies exactly what actions we take at each level of risk. But we want participants to understand that this is not a replacement for clinical care. We are not a “Minority Report” system that will swoop in and prevent harm in every case.

[0]: https://primr.org/ethical-challenges-in-suicide-research/

The post office already does way more than you think it does. Like identity verification on behalf of the government and tracking which houses are considered "vacant" or not.

It would be nice to focus the USPS, but all these extra services have to be done by _some entity_, otherwise you can't have the final result. In this case, the result is a functional, auditable mail in ballot system

[0]: https://www.federalregister.gov/documents/2026/06/02/2026-10...

Feels like a language limitation to me. It is hard to describe succinctly the complexities of AI use that preserve a human element or recognizes externalities. The wording is not immediately obvious and require subtlety that I rarely see in HN, unless it's about the tech itself and it's minutae.

Curious to see if HN could have any kind of thoughtful conversation on it but so far I haven't seen one.

I found in the current version of Forgejo the most secure way to set it up was just to put the runner on a different host. There's a Docker in Docker setup [2] that was challenging to get working correctly but still didn't quite feel secure...

After setting it all up, I'm convinced GitHub actions as a concept is flawed. I wish the community would make a simpler git-defined CI/CD that didn't have the crazy surface area that actions does... Named shell scripts that can call out to other stuff in the runner Docker image and report it to the UI would be a nice start.

[0]: https://github.com/nektos/act

[1]: https://forgejo.org/docs/v15.0/user/actions/github-actions/#...

[2]: https://forgejo.org/docs/v15.0/admin/actions/docker-access/#...

In the summer example in the hero images, the building placement + small pockets of water on the left looks odd and low attention to detail. A similar poor quality result as if an uncaring human used a scatter brush...

Curious if the examples are cherry-picked and by how much, or if this is one-shoted

> Research on cognitive maps points toward the same conclusion as the study on taxi drivers. In 2023, researchers ran a machine learning model on more than 22,500 people in a national dataset and were able to predict which ZIP codes had higher rates of Alzheimer’s with 84% accuracy based on how complex the environment was. Those living in spatially complex surroundings – the kind that force active map building, such as dense street networks with numerous intersections, diverse points of interest and landmarks, and multiple path options – were less likely to develop Alzheimer’s.

Doesn't that mean guaranteed failure in the market if you use taste as your guiding star? Therefore, there's really no way to enter the market anymore.

I don't know how you then ship software or monetise it...? The market will just clone it.

So... Don't share stuff publicly? Only share with closed groups? Where does this leave small-time software development...?

> I feel a certain lack, an empty space where a particular kind of person should be—all those who used to read comic books

The author describes how there are so few nerds in their ideal, nostalgic time, yet there is somehow a void where there should now be a bountiful amount of nerds in the modern times?

Anecdata tells me the nerds are there. They're just in their own spaces. Out of the public eye.

With the internet the way it is now (bots, scams, AI, overrun of nerd spaces, clamping down of internet speech) I don't blame any of them. You can't be on the public internet anymore and have a good time. The platforms got too big, the rules so impersonal. The interesting people went elsewhere. Left with this hallow "majority". Or as the author says

> it’s problematic that the taste of the majority is swamping out the minority

The "intellectual nerd", the minority, is just elsewhere. If you want to find them, you have to put in the time and the effort to move your eyes or your self there. Cultivate your social surroundings.

Supposedly this is what it does from the opening comment

  # Paxel upload script
  # ===================
  #
  # What this does (up to 17 steps):
  #   On your machine
  #     1. Check Docker is installed and running
  #     2. Sign you in (browser-based device auth)
  #     3. Pull or build the Paxel Docker image
  #
  #   Inside the container — file bodies stay local; only aggregate metrics +
  #   metadata (paths, commit numstat, session events) are uploaded
  #     4. Discover projects and sessions (Claude Code, Codex CLI, Cursor)
  #     5. Read your git history
  #     6. Parse transcripts
  #     7. Summarize each session (cloud Haiku via YC proxy)
  #     8. Group git commits by session
  #     9. Group sessions into multi-day work streams
  #    10. Extract steering traces
  #    11. Extract decision exchanges (cloud Haiku)
  #    12. Redact code before upload (regex pattern redaction)
  #    13. Link decisions to outcomes
  #    14. Analyze code quality (L1 deterministic)
  #    15. Score episodes across 5 axes (cloud Haiku)
  #    16. Assemble your report
  #    17. Upload redacted summaries + scores to the server
  #
  #   Then: opens your results in the browser

Almost as though they want the quality and consistency of hired labor but not the cost

I think that's probably the fourth iteration of it at this point

Leaking is an issue but we're talking about internal services too.