Legal Eagle reminds us that winning the recent elections is not enough. Restoring Americans' faith in the process of governance and the rule of law is essential. We have do to that with (fair) hearings and trials - but to ignore the evil that has been done and hold no one accountable is to be complicit with it.
It's paywalled, but the TL;DR; is
"Pope Leo has voted in the US midterm elections – the first time a pontiff is known to have cast a ballot in any political election at any time in history."
HomePod mini 2. Sexy
I'm calling on all senior citizens using walkers to start blocking ICE vehicles by crossing slowly in front of them.
One of my civic duties is that I work the polls on elections. I'll be doing it 6am-9pm on 11/3.
We all know trump is going to challange results he doesn't like, so I want to take the chance now to praise the quality and integrity of the voting process. (at least in my county). As a technologist who knows more than a little about security, I'm impressed with the processes we use to prevent cheating. Also, in the years I've been doing this, I've never seen anything remotely like identity fraud.
I’m naming this big boy — Bobby
#slackisshit
If your company is slack first (not investment advice) then, essentially, slack == mail, so why do can't I turn off things like this.
The smug answer is "because it might be urgent and we should make their phone beep."
Sure, but that's a no-op because there is so much noise from slack that you turn off notification sounds anyway. I'll read your slack message along with mail, when I break to process all my inboxes. Unless it is from my manager. That beeps.
The kicker: the impact of an AI triggered humanity destruction we is infinite. There is no coming back from extinction. We can't just take our lessons and try better on the next iteration.
Because the impact is ~infinite, the Risk is huge,e ven for small values of the probability X. 10%, 30% or 50% are just the same for the risk calculus. Any non-trivially small X% is a bad deal we shouldn't take.
The only argument worth making by AI proponents is that the risk added by AI is ~0.
#ai
3/3
There is the other, more colloquial use of probability values, for non-repeatable events. It's much softer as the actual values are unverifiable, but still useful - and used all the time.
It's for risk calculations - the prob. of bad outcome P times the impact of that outcome, give you the Risk that you are taking. Eg, is it worth carrying an umbrella given 20% chance of rain? You take the annoyance of getting wet times 0.2, and compare that against the annoyance of lugging the ☔
2/X
#ai
"P(doom)=X%"
I'm surprised than Anti-AI people let that slide so easily and fall for discussion of what the value of X is.
What is a probability? It the amount of times when something would happen vs it wouldn't happen in the same starting conditions.
That's useful for coin flips or card games, but not quite applicable to non-repeatable reality. We get to experience each year, month or day a single time. We can't run through the next decade a hundred times and see how many we survive
1/x
#ai
I'm so old I remember when laptops had spinning rust drives in them and you tried to keep them still when working.
RE: https://fosstodon.org/@Blosc2/117359796684529053
The Blosc team is committed to help you survive the data deluge ☔️
I know you’re not supposed to feed wildlife, but I’ve been spraying sugar water on the window screen for the wasps.
This write-up is very intriguing: https://nathan.rs/posts/gzip-lm/
Actually, this isn't the first time that I see comparing LLM models with compression as a valid abstraction, and it isn't gonna be the last for sure.
IMO, there is indeed a blurry line among both aspects of theory of information. Creating good corpus/models, where methods/approaches of compression and LLMs complement each other, is going to be an interesting field to have fun in coming months/years.
9/18/2026I have had to close registrations on masto.social due to a huge influx of AI registrations. While manual approval is required for new registrations (thank goodness), seeing 1k+ accounts wanting to join from different IP addresses but containing similar format for email addresses in a 12 hour period was jaw dropping.
I need to come up with a solid rule to disallow AI registrations.
#spellingbee not taking MILF today. Come on dudes. That entered common usage 30 years ago.
RE: https://fosstodon.org/@Blosc2/117274732437947791
New Blosc2 is all about *interoperability*.
I saw Steve Winwood (78) and John Fogerty (81) in concert last night. I hope that when I am 80 I can still practice my craft with the skill they can.
RE: https://fosstodon.org/@Blosc2/117252592714449579
Our data capsules for remote data are premiering. Check them out!
Maybe this sign should not be next to the massage room.
