HNHacker News
TopNewBestAskShowJobs

flexagoon

2,777 karma · joined December 29, 2021

meet.hn/city/50.8512438,5.6909768/Maastricht

Socials: - github.com/flexagoon - t.me/flexagoon

Interests: Biotech, DevOps, Open Source, Philosophy, Privacy, Programming, Research, Science, Technology

---

submissionscomments
flexagoon··on Ask HN: Does anyone let AI agents play games just for fun?
I was just using the terminology of the person I was replying to
flexagoon··on Grok CLI uploaded the whole home directory to GCS
Not gonna argue about the utility of AI, but isn't the statement "AI is not here to help people" completely meaningless? AI itself is not "here" for anything; the problem is big tech doing big tech shit as always, not the current technology they're doing it with.
flexagoon··on Ask HN: Does anyone let AI agents play games just for fun?
The huge energy waste is from model training, not inference
flexagoon··on Since Chromium 148, Math.tanh is now fingerprintable to link underlying OS
You can just use one line of JS instead

https://fxgn.dev/blog/anubis

flexagoon··on Show HN: Learn by rebuilding Redis, Git, a database from scratch
> you need to run it on your own machine

Isn't that the best part of codecrafters? Being able to just use your own text editor instead of a janky web based one which loses everything you typed on page refresh?

flexagoon··on Ask HN: Why isn't Google indexing information about the AT Protocol?
All the results for "e os" are about /e/OS, and that's its full name, so you wouldn't just search for "e". D language is a better example — its just called "D", but searching for that doesn't give results about the language. On the other hand, for "C" and "R" the top results are all about the languages. "Go" is also a pretty bad one, they insist it's not called "golang" but you basically have to search for that to get any useful information.
flexagoon··on Apple sues OpenAI, accuses ex-employees of stealing trade secrets
No no you just don't get it, they haven't released one because it's too dangerous and a you're not allowed to have access to it for your own good!
flexagoon··on Ask HN: Are systems ready for the first negative leap second?
If you're calculating time that a task took by subtracting two UTC datetimes instead of just using a monotonic timer or at least unix timestamps, you're doing it wrong either way
flexagoon··on No leap second will be introduced at the end of December 2026
The only better thing is the organization being called "International Earth Rotation Service"
flexagoon··on Top researchers leave USA for the Netherlands (in Dutch)
Independence from a country doesn't require excommunication of all citizens of that country
flexagoon··on Top researchers leave USA for the Netherlands (in Dutch)
Dutch researcher salaries are also pretty high (compared both to median Dutch salaries and academic salaries around the world), so it's still not like you're leaving USA to live in poverty
flexagoon··on OpenPrinter
Unlike now, where instead of saying your hypothesis you're just letting every reader come up with what they think is a crazy conspiracy theory and ascribe it to you?
flexagoon··on Organic Maps
There is a tag for ALPRs on OpenStreetMap:

https://wiki.openstreetmap.org/wiki/Tag:surveillance:type%3D...

So you can just add them using a normal OSM editor like EveryDoor or Vespucci

flexagoon··on Google Books (or similar) all book scans – $200k bounty (2025)
No specific figures, but see, for example:

https://annas-archive.gl/blog/ai-copyright.html

> Virtually all major companies building LLMs contacted us to train on our data. Most (but not all!) US-based companies reconsidered once they realized the illegal nature of our work. By contrast, Chinese firms have enthusiastically embraced our collection, apparently untroubled by its legality.

> We have given high-speed access to about 30 companies. Most of them are LLM companies, and some are data brokers, who will resell our collection. Most are Chinese, though we’ve also worked with companies from the US, Europe, Russia, South Korea, and Japan. DeepSeek admitted that an earlier version was trained on part of our collection, though they’re tight-lipped about their latest model (probably also trained on our data though).

It's at least 30 companies, each of which paid hundreds of thousands of dollars.

flexagoon··on Leaking YouTube creators' private videos
I don't think it is. It reads exactly the way I would write it myself.
flexagoon··on Google Books (or similar) all book scans – $200k bounty (2025)
No, but they openly make a lot of money from selling their library to AI companies. Fast enterprise access to Anna's Archive starts at $100.000
flexagoon··on Learn Vim motions with an ice-cream van
Idk, I get what you mean but I'd personally rather not use my pinky finger for navigation, sounds really uncomfortable.
flexagoon··on Google loses fight over record $4.7B EU antitrust fine
I don't believe that you actually see no difference between this and the case in the lawsuit.
flexagoon··on Learn Vim motions with an ice-cream van
How are they "not that touch-typing-friendly" when they're literally the keys that are always under your fingers when touch typing?
flexagoon··on Learn Vim motions with an ice-cream van
I just have a layer on my keyboard that puts arrow keys where hjkl are on qwerty when I hold the tab key. Since tab is right next to my pinky finger on my laptop, it easily became muscle memory, and allows to use home row navigation in all apps, not just the ones that support hjkl.
flexagoon··on Oomwoo, an open-source robot vacuum you build yourself
If someone has the time and opportunity to create a robot vacuum from scratch, they probably also have the time to write a short introduction for it
flexagoon··on The first early human eggs from stem cells
Child mortality rates are rapidly dropping globally as well

https://ourworldindata.org/child-mortality

https://ourworldindata.org/children-saved-global-health

flexagoon··on Godot will no longer accept AI-authored code contributions
> Submitters just add stylistic markers to make their accounts and output seem human-generated

https://xkcd.com/810/

flexagoon··on Type-checked non-empty strings
I've had a Bellroy bag, they're not the most fashionable but super high quality and well thought-out. Just like Haskell code—maybe that's why they like it.
flexagoon··on Show HN: Adrafinil – keep a lid-closed Mac awake only while agents work
Everyone dreams about making the next cool AI tool and doesn't bother checking if someone has already made the same thing years ago without the word "AI" in the description
flexagoon··on Previewing GPT‑5.6 Sol: a next-generation model
For even more of a challenge, try replicating Richard Stallman's personal setup:

https://stallman.org/stallman-computing.html

flexagoon··on Medical students are using popular research tool to pump out misleading studies
"Don't believe random quotes on the internet"

- Albert Einstein

flexagoon··on FUTO Swipe – A new swipe typing model
Heliboard does have swiping! You just have to manually download the GBoard swiping library and upload it in the app.
flexagoon··on Polymarket has flooded social media with deceptive videos by paid creators
> Exactly same reason why porn sites use their own specialist payment processors

They're called high-risk payment processors and they usually handle all sorts of "legal but sketchy" sectors like porn, gambling, crypto, legalized drugs/vapes, MLMs, etc. It's not even some shadowy darknet industry, if you just search for "high risk payment processor" there is plenty of them which look very professional and even provide official Shopify integrations and stuff.

Interesting that if you go to one of their websites they also handle some pretty normal industries like "furniture" or "travel" just because of the high chargeback rate in them.

flexagoon··on There is minimal downside to switching to open models
Because it offers cheap open source models, not GPT and Claude. I mentioned it as an alternative to Z.ai's subscription in OP's comment, not to Codex.
← PreviousPage 7 of 25Next →