thyrsus

Budgets are based on predictions of the future, and "predictions are difficult, especially about the future".

Most file systems get cached in ram, which is unlikely to have happy consequences when scarce ram triggers use of the filesystem swap.

The article makes a convincing argument that Haskell compilation is too slow for the fast code generation of AI. But python?

I have yet to experience a RHEL major version change that did not blow up all my tiny simplistic python scripts. I see the following options for using python: * run inside the container it was developed in * build your own python interpreter and environment and libraries and never use python pieces from the OS (i.e., act like a container without using one) * keep different versions of the code for different OS versions and use AI to rewrite all the code for the new OS version

Start to consider third party dependencies, and none of those feels tractable without an AI assist.

I've dabbled in C device driver code and kernel version differences were my only problem, not C. My perl scripts never break. My bash scripts rarely break. My dabbling in erlang didn't suffer from language version differences. My little elisp hasn't broken. Only python has inflicted this level of version pain. I have a colleague who says java has the same version pain as python, and from what I've seen from Jenkins maintenance he may be right, but I don't have colleagues who want to read java code, so I haven't written my own.

That claim needs explanation. My understanding is that substantially less than 20% hold the controlling ownership of U.S. productive capacity.

As an ancient one (graduated college 1981), my use of AI is very conservative: look things up. Generate code I can read and understand in less than 30 minutes. This is working well for me, because when the AI botches the answer, I know quickly. It either works or fails fast: there's no importable function by that name, that keyword isn't in the language, that only works in a different version of the OS. I never ask it to do something I couldn't do myself in 10x the time (spent fixing typos or missing punctuation). If I ask it to do something I don't know how to do, I create tests - usually informal - to ensure that I understand what the code is doing. If the syntax is unfamiliar, I make it explain what it's doing, and then I informally test that explanation (usually toy examples at the command line). You must learn to do these things regardless of where the answers come from - the Internet, a journal, a book, a colleague. Otherwise >>when<< it fails, you will not be able to reason about the causes for the failure and how to find a correction.

I've seen copyright notices that explicitly forbid use for AI training. Would this "transformation" argument still hold in such cases?

For example:

No Generative AI Training Use

For avoidance of doubt, Author reserves the rights, and grants no rights to, reproduce and/or otherwise use the Work in any manner for purposes of training artificial intelligence or machine learning technologies to generate text, text to speech, voice, or audio including without limitation, technologies that are capable of generating works in the same style or genre as the Work, unless individual or entity obtains Author’s specific and express permission to do so. Nor does any individual or entity have the right to sublicense others to reproduce and/or otherwise use the Work in any manner for the purposes of training artificial intelligence or machine learning technologies to generate text, text to speech, voice, or audio without Author’s specific and express permission.

What is being licensed by the End User License Agreement (EULA) is the copyright on the code and its artefacts (executable bytes, etc.) - you can't have an EULA without having the copyright to license.

Are there currently circulating cryptocurrencies that use quantum resistant cryptography?

The first amendment protects from U.S. government censorship (with exceptions for national security and commission of otherwise illegal acts, such as child molestation, fraud, or yelling fire in a crowded theater). The suppression here is a civil matter in which the author signed a contract not to "disparage" and would suffer civil penalties (monetary, other negotiated constraints) for violating the contract.

I dealt with physical paper tape on only three or four occasions in the early 1980's, each time terrified of a jam or tear. It seems in this case it's a read-once operation, which is plausible. Read-many, not so much. Punch cards are orders of magnitude more reliable.

As a subscriber to The Guardian, I have a very different experience.

Explain to me how qualified immunity is better than any ill it is supposed to address? And how is it that if you sue the government and win, then the judgement doesn't automatically award reasonable legal fees?

The left hand scale is presumably dollars per ounce of silver. The right hand side is "margin requirement ($)". I can think of two interpretations of that, neither of which make sense to me. The first interpretation is that one recently needs a $25000 margin per $90 ounce of silver instrument - obviously absurd. The other interpretation is that once you have a $25000 margin you can buy unlimited silver instruments, which is only barely less absurd. How does reality work?