Doing this on a global scale requires "CFC-ban"-levels of global coordination which I cannot see happening in the world we live in today. Just look at how global CO2 reduction and climate change is being handled today at the global scale.

How I manage this is that every time I want to open-source a previously private feature, I take the changeset diff and apply that to the files in the public repository. Same features, but plausibly different hash.

Comments URL: https://news.ycombinator.com/item?id=41047950

Points: 1

# Comments: 1

At first, I thought it was just to reduce the complexity of managing hardening rules for several OS and OS versions.

Industrialization as we know it would have never happened if we artificially limit progress, just so that people could still have jobs. I guess you could hold the same kind of argument for the copists, when printing became widespread; for horses before the automobile; or telephone operators before switches got automated. Guess what they have become now. Art made by humans can still exist although its output will be marginal compared to AI-generated art.

LLMs are not humans but are used by humans. In the end the beneficiary is still a human.

Also, using state of the art IDEs to be able to navigate between my own and third-party code removed pretty much all of my anxiety when dealing with code in general.

There should be a way to make the Tails OS in such a way that each time the USB is plugged in, the contents of the flash drive are moved to your computer RAM, so that if you plug it off without running `shutdown` explicitly, not only you'll shut the computer down in 2 seconds, but your USB is also clean. The `shutdown` command being run with the USB still plugged in then copies the image from RAM back to your thumb drive. Your non-destructive shutdown routine will obviously take a lot longer though.

People convicted for Jan 6 Capitol attacks are looking at 18 years[1] at least

[1]: https://www.pbs.org/newshour/politics/oath-keepers-founder-s...

The "proofs" for conspiracy were apparently very light and the fact that the accused had used privacy protection measures, like instructions for a de-Google'd smartphone, Tails, TOR and encryption, were apparently enough to sue, detain and to have the DGSI investigate... I wonder what would happen if a similar case happens in the U.S.: they would have perhaps been suggested to have a plea bargain and plead guilty and serve X years in prison, and maybe liberated after Y years for good conduct. But, oh wait, being convicted of terrorism actually sentences you to 30 years[1] in the U.S. so you'll most likely rot in a cell and be forgotten by everyone except your captors.

[1]: https://code.dccouncil.gov/us/dc/council/code/sections/22-31...

Jokes aside, there is only so much complexity that can be held. Try writing a financial regulatory report from the data present in your enterprise ERP (whose schema you'll eventually have to feed prior or in addition to your prompt).

Granted you can also do it in small steps using only subsets of the whole schema, but that still requires non-trivial oversight by a human expert. Plus, user needs are not set in stone and are not exempt from contradictions. The human might challenge those contradictions, the machine will not.

Infinite-length context could address prompt size related issues, but hallucinations on imaginary columns and flat-out wrong calculations, which cannot be avoided by humans either, will happen past a certain point.

Anyway I am not sure any sensible CFO or Financial department manager would give results produced by an unauditable automated system that takes in English and spits out numbers on a spreadsheet: if errors are missed or bugs arise, who's responsible? (the AI _is_ the expert, remember)

I wonder if the system is still in place. Last time I checked the Jenkins folder was occupying 270 GB (!) of the 10 TB shared FS, most likely because FS block size was 1 MB.