https://github.com/anomalyco/opencode/commit/7d2d87fa2c44e32...

The state always thinks of self-preservation. Any bureaucrat is aligned to this goal by getting the benefits from the state. So, the more power it has over its citizens, which is the first threat it, the more safe it is and the less opinions of citizens matter.

Understanding this, every citizen must think carefully about giving away more power to the state.

No, it's just background internet scanning noise

If it was engineered right, it would take:

- transfer model weights from NVMe drive/RAM to GPU via PCIe

- upload tiny precompiled code to GPU

- run it with tiny CPU host code

But what you get instead is gigabytes of PyTorch + Nvidia docker container bloatware (hi Nvidia NeMo) that takes forever to start.

Often bank scams rely on sending money to another account (obviously registered with an ID), and then being drained at ATM. The account is going to be registered on a drop or another victim. Sure, it's burned after that, but as long as it's an insignificant cost, scamming is still profitable.

The same situation with malware, bad actors are incentivized to put effort into bypassing this, so dev accounts will be registered on random homeless people, stolen IDs, or just fake IDs. While normal developers will choose to give away IDs.

And as always, it starts with 'protect the children/elderly/vulnerable', then that authoritarian country requires Google to give away info on every developer to operate legally, then it's UK and other 'democracies', then you can't run your code on your device without the government approval.

With all the labels and disclaimers, there can always be this one person that will get confused. It's unreasonable to demand protection from long tail of accidents that can happen.

Do you:

- Buy open devices?

- Sponsor development of open devices?

- Start open device companies?

- Develop open software that competes with walled gardens in quality and ease of use?

- Sponsor open software?

- Use open software?

- Engage in lobbying?

- Drop exploits (that would be worth a pile of gold) to let people jailbreak devices?

- ...

- Fake-care or real-care?

I have a script that generates an html file, sorting posts to tiers based on domain or keywords. Just as an example the shitty tier posts are now: bbc, wsj, vox, arstechnica; the top tier: itch.io, twitter, arxiv, github. Any posts from rare domains get into top tier, posts from frequent domains to middle tier. So it's mostly sorted to my preferences.

Twitter algorithm is much worse for example and has bugs, you'll need to grind for some time to get rid of popular trash and politics. Any engagement with bad content will only bring you more of it in the feed, and there's tons of baits to engage with something stupid. So mute any accounts and words you don't like, click "not interested" on posts, but it's really slow to update.

If 1/3 of people have a trait, you'll get a good sense it's common after asking 300 people, even if the total population is a million. Surveying more gives diminishing returns as the result quickly converges on the real proportion.

The real problem with all surveys is the sampling isn't truly random, which skews results.