It's also one of the few obvious tells we have left.
View quoted note →
ChipTuner
ChipTuner@gitcitadel.com
npub1qdjn...fqm7
Building software they don't like. Free, as in freedom.
Low-level and server engineer: libnoscrypt, NVault, vnlib.
Staff @GitCitadel
https://geyser.fund/project/gitcitadel
DM's don't work. Just tag me.
I keep hearing that gptoss-120 is actually usable for agentic work. True? It may cause me to break the sat's piggy bank for some more vram this year.
If you have experience, what context window size has been working well for you? 128k would be nice, but not sure i can afford that XD
Still holds true
View quoted note →
I need to stop looking at my dashboard, but it's wild how this bot crawling is 1:1 request/ip address. They're choosing to spare me. I feel like control over 20,000 unique IP addresses could do a lot of damage. This is new btw. Completely generic agents, completely "random" paths, a new IP address per request.
I'm seeing 2-3k unique IP addresses per hour, over 10k unique IP's in the past 12 hours.
People ask me why it takes me so long to get anywhere, or why I don't get on the freeway. MF it's because I'm driving the back-roads without Flock government surveillance cameras.


DeFlock
Find license plate readers (LPRs) near you.
You look away from your logs for 25 seconds and your mail server gets blasted.
Oh, they just turned most popular smart TVs into a massive AI botnet... That's cool. I wonder if that has anything to do with the 2500 unique IP addresses I see per day.
https://www.techspot.com/news/111492-smart-tv-apps-quietly-scraping-web-data-ai.html

The Verge
Your smart TV may be crawling the web for AI
Some TV apps let you watch programming with fewer ads, as long as you allow your TV to participate in a global proxy network.
Smart TV Apps Quietly Use Your Internet to Train AI Models
Bright Data embedded SDK in Smart TV apps to harvest web data through users
I wonder if I will break 3000 commits across my oss projects before the summer :/