My understanding is that tokens are basically words, and that when you ask a question it charges for all the tokens it consumes, produces, or processes. There’s a lot of internal processing for each request, where the input text is summarized in different ways and combined with previous parts of the conversation, so it’s not as straightforward as “word count of what you say plus what it says”.
Here’s a link to the Krebs on Security article that Gizmodo used as a source: https://krebsonsecurity.com/2026/05/cisa-admin-leaked-aws-govcloud-keys-on-github/
Requires subscription
Yeah, the definition of “conscious” really is a puzzle. My guess is that the “nothing is conscious” model has a great deal of crossover with the “free will doesn’t exist” one; for both of those, I don’t consider them useful models even if they end up being true: if I’m not actually conscious and just think I am, I might as well behave as though I am.
Regardless, we really do need to define what exactly we mean by “conscious” before we can have a meaningful discussion about it. Where’s Socrates when we need him?
Huh, interesting; that’s a good question. I’m not actually sure about that; it’d be a good thing for me to dig into more. Thanks for the thought!
I mean, they specifically point to post-quantum cryptography and advise people to move towards it in the article:
Google said: “We’ve adjusted our threat model to prioritise post-quantum cryptography migration for authentication services – an important component of online security and digital signature migrations. We recommend that other engineering teams follow suit.”
The issue here is not that there aren’t solutions; it’s that organizations are not interested in taking the time and effort to move towards them. I’ve been beating this particular drum at my org for about a year, and I’ve gotten zero traction. This is a concern because moving to New encryption means taking all the data you’ve got, decrypting it, and re-encrypting it. That’s not fast when you’re talking hundreds of terabytes.
With keycloak you can have a single local password to all your selfhosted apps: you sign in to keycloak, then you sso into everything else from there. I’m building that out on my homelab right now, and it’s working fine.
One place I’m familiar with actually just deidentifies data when they say they delete it. They also have ways to re-identify if needed.
“plotting” != “conducting”
They explicitly don’t:
The law does not require photo ID uploads or facial recognition, with users instead simply self-reporting their age, setting AB 1043 apart from similar laws passed in Texas and Utah that require “commercially reasonable” verification methods, such as government-issued ID checks. Assemblymember Buffy Wicks, who authored the bill, said this “avoids constitutional concerns by focusing strictly on age assurance, not content moderation,” in a press release. The bill passed both chambers unanimously, 76-0 in the Assembly and 38-0 in the Senate.">
I’m still hopeful about https://veilid.com/ but I suspect that’s a long way out; the protocol is still in development stages, let alone implementations of it.
[citation needed]
Don’t do it; they’re a security nightmare: https://krebsonsecurity.com/2025/11/is-your-android-tv-streaming-box-part-of-a-botnet/
There are safer ways to acquire content to view.
Syncthing is great for syncing things like keepass dbs
It’s absolutely feasible; it’s just not painless. The network effect is real. This is something most of us here on the fediverse should be well familiar with: I’d wager we’ve all left a big proprietary centralized platform for a smaller one with less content that doesn’t exactly fit our needs. Yes, it can suck; yes, you’ll miss out on content and discussions that you would have seen before; yes, you’ll be fine in a year anyway.
That said, it sounds like Discord is doing some backpedaling at the moment, and that the vast majority of people won’t have to do this. If that’s so, there’s some time before it enshittifies further. I’m going to use that to check out alternatives. My first step will be spinning up Matrix and a Discord bridge to see if that meets my needs. Then some of the other platforms, probably Stoat and Spacebar. I’m going to treat them as beta tests and give lots of feedback to the dev teams, which I hope will help incentivize development. I might even throw a few bucks at promising projects.
The ultimate end of centralized platforms is decay. It’s time to get serious about Discord replacements, at least for me.
Sounds like it’s experimental at the moment:
Voice/Video when?
Currently there is experimental voice/video WebRTC support in Spacebar. UDP connections are not currently supported.
This is a very difficult feature to get working, especially given that we must implement it the exact same way as Discord.com for client compatibility, so if you find any bugs please open an issue in Spacebar server.
We would also be incredibly thankful for any assistance.
https://docs.spacebar.chat/faq/
So might not be ready for primetime. Voice is pretty critical to how I often use Discord. Video less so, but still important.
There’s also https://spacebar.chat/, which I’ve been hearing some buzz about, but not really here; not sure if that’s because there are known problems with it or what.
Edit: it looks like voice and video are just experimental right now, so might not be a good replacement for many: https://docs.spacebar.chat/faq/
Last time I tried to play around with Stoat, I never even got a verification email, so taking an hour sounds like progress I guess? I’ll have to try again.
Polymorphic malware is probably one of the easier things to do with LLMs, so static scanners seem of limited use.