i was a pretty severe libertarian shitlord in my 'college years', and tbh if the left had been the hand-wringing fear-mongering mess it is today back in '03, i don't think i would ever have stopped
one important way to tell is that none of them have ever claimed that they loved America and wanted to do work to make it better just so that they could get in and enrich themselves
like sincerely, if you want to see some of the most dedicated, talented, and sincere technologists on planet Earth, 18f and the people they contract with would be a great place to start
remix friendly protocols trend this direction anyway: consider the old standard group blog, but executed as a shared git repo full of markdown and a static site generator
ugh i need to figure out who, if anyone, of who this account used to interact with, is still here/which other thing most of y'all fled to.
gotta be bsky right?
@rauchg I like this thought but why would it be any less profitable for AI companies to just participate in the review grift than it is for the existing review grifters?
a foreign power manipulates each of us every day. often more than one. you and i both believe state crafted misinformation and have since youth
trump is trash, don't get me wrong, but the ppl he manipulates _like_ the idea that he thinks like they do (i.e. is easily manipulated)
me: send this to the c api
rustc: no that's stupid and could create a hanging pointer
me: ok but the the api wants a potentially-hanging pointer
rustc: fuck you
me: `unsafe`
rustc: fuck you, *verbosely*
Telegram messages aren't e2e encrypted. It is also a "cloud messenger," meaning that all messages live on Telegram's servers rather than the user's device.
With one query, the Russian Telegram team can get every message the French president has ever sent or received to his contacts, every message those contacts have ever sent or received to their contacts, every message those contacts' contacts have ever sent or received, etc.
It's just plaintext, there are no limits to what they can do, like use an LLM to help go through all that material and pull out the dirt, map the relationships, figure out who's keeping secrets from who, etc.
For the French politicians and cabinet members, it is kind of too late to do anything. Even if they try to delete all their messages now, the Telegram team can just mark the messages as "deleted" so that they no longer display to the user -- but not actually delete the data they retain access to.
This could get really wild.
In short w3 is pointlessly overloaded as an API and while for backwards compatibility reasons, we can't get rid of the overloading, I don't think that should mean that new solutions that break out existing functionality into more purpose-built sandboxes aren't achievable
(we also *could* have a header that says "this document is not self-editing and you can unload the JavaScript runtime" but that seems like a harder sell in 2024 even though it also seems extremely useful to me)
While it's understandable that we cannot say that all documents are pure documents, we can definitely have an API that says " this is definitely not a document, it is an application, and you do not need to allow it access to tools to treat itself as a document"
we took a document viewer, put a real time renderer in it, gave the renderer a "screen/window" abstraction to draw into, but never finished the job of separating, even optionally, "this is an interactive render loop" and "this is a document"