Replying to
@AkaSci@fosstodon.org making something permanent. Ideally use UTC everywhere.
Remote
Brian Koopman
@brian@social.briankoopman.com
mastodon 4.6.8Assoc. Research Scientist at Yale working on the Simons Observatory and ACTPol. Focusing on data acquisition software and Cybersecurity. PhD from Cornell.
Bari sax player in The Simulators and Viva Mayhem.
Profile picture is a selfie taken in a highly reflective aluminized mylar polarization wiregrid inside of the old Atacama Cosmology Telescope.
Posts auto-delete after 6 months.
#music #ska #physics #selfhosted #cats #cybersecurity #infosec
0 Followers
0 Following
8 Posts
Joined November 15, 2022
Website:
Just upgraded my old Unifi USG with a new UCG-Max. Pretty simple switch over. Just had to get the UCG-Max online with the WAN connection and restore all my settings from backup. Everything else "just worked". Step one in my plan to upgrade my home network to 2.5G!
I'm still thinking about the "AGI on top of LLMs" talk I saw on Monday, especially after the Claude leak.
The speaker was describing LLMs as "trying to predict the next word on the Internet", and claimed as fact that as LLMs got better, "in order for them predict that next word, they have to understand the universe".
He was talking to a room full of physicists and was clearly trying to mirror the famous Carl Sagan quote, but it just came off to me as completely unhinged
Replying to
@mttaggart@infosec.exchange I just saw the second variant. It seems to also be mitigated by blocking the esp4, esp6, and rxrpc modules suggested in this dirtyfrag repo. Tested on Ubuntu 24.04.
Replying to
@mttaggart@infosec.exchange yeah, someone dropped a quick summary in https://github.com/V4bel/dirtyfrag/issues/1#issuecomment-4400768488 -- seems to be for IPsec. Still reading up. None of those modules were loaded on systems I'm running though
Been in a bit of a funk lately, but as a result have been trying to get out of it with lots of exercise. Just went fairly hard on the erg with some interval training, feeling good 💪🏻🚣🏻♂️
Replying to
@hugovk @brettcannon I had the same thought when reading this (not being familiar with pylock.toml) -- "why doesn't pip do this?"
To contrast yesterday's physics colloquium, I attended a talk and conversation with Adam Becker about his book, "More Everything Forever" tonight.
One take away for me, that I also saw discussed here in recent days, is the importance of calling 'bullshit' on AI, and more broadly on the ridiculous claims made by folks like Sam Altman.
I'm not much of a poster here, but think I should start working on that