At least seven people have been shot dead in federal immigration operations since January 2025
Homan said ICE will require officers to record vehicle stops with at least one body cam
-- Too late for the seven victims
https://www.reuters.com/wo…
One more photo from the trail that I posted about recently. The blooming heather always make sme smile. Especially when there aren't too many flowers around.
In such spost I'm always traying a bit what looks more interesting: a great vista, a plain trail shot, the heather in the foreground or a blurred foreground.
The advantage is: I always have little rests along the way :)
#hiking
It's #FreeCADFriday UTC, even if it's still Thursday evening where I am.
I am getting a screened pop-up gazebo/cabana for ham radio purposes (Field Day) and wanted to compare footprints of three different options. One has a 10' x 10' square footprint, and two have hexagonal footprints specified by their longest internal measurement, because bigger numbers sell, but I didn't h…
Immediate (and obvious) memeing.
When I first saw the photo, I thought there was no way it could be real, it had to be AI.
It's not. There's several videos of it, shot from several directions. I've seen them. This is 100% real. And this photo is one for the photos of the year.
#Russia's Moscow is burning. Hard. At long last.
I was walking to work today and 4 cars build around 1920 were passing me.
I reacted slowly and 1 car already moved beyond the shot.
Hope I'll be looking that good after 100 years in operation.😂
Have a nice day.😊
Israeli troops killed a seven-month-old Palestinian baby in the occupied West Bank and injured one of the child’s parents after opening fire on the family’s car, despite it having complied with an order to stop.
https://www.theguardian.com/world/2026/jun
An ICE agent reportedly shot and killed a person in Biddeford, Maine, on Monday,
marking the second person killed at the hands of federal immigration agents in a week.
The office of the Maine Attorney General said that agents were
“conducting an enforcement operation related to a final order of removal”
when an agent fatally shot a man on Monday morning.
The attorney general and Sen. Angus King (I-Vermont) have both said that an Immigration and Customs Enforceme…
Do you prefer it when control doesn't need movement of your arm or your hand, and you instead organise everything with the slightest movement of one single finger?
Well, the trackpoint (or whatever you want to call it) is now also available as external add on:
https://ploopy.co/bean/
Now, I have a lot of reasons to hate vibe coding, but one of them is that I just think more easily in code. I reason by writing the code. It's not something I can easily disentangle.
Architectures tend to live in my head as pictures that change as I write. The words are code. There's no intermediate. So the offer of being able to skip the poc phase is just not interesting. "Oh, cool, I can skip the part where I preemptively identify problems? Great. That's gonna work out well." Vibe coding can give you working proof of concepts that self destruct when you try to grow them (much like any poc code you just try to use without actually engineering).
I don't know that this problem will ever get fixed. Even if it does, I don't know that this fix will ever rival human engineering. I don't think it's actually possible to know that yet, there are too many factors at play. It's definitely lowered the bar for writing proof of concepts. Now a business idiot can make a proof of concept, which even further distorts their understanding of what engineering actually looks like.
In order to go back to work, I'm basically expected to be ramped up on LLM stuff. But it all keeps reminding me of the feeling of powerlessness when we organized against Trump, when we organized against the rally at UW, when I tried to stop someone from pepperspraying people and I got shot. It's all the same ideology of rapists and pedophiles forcing their entire world on us against our will, thinking they can dangle "the Epstein files" in front of everyone and then yank them away like nothing ever happened... Like no one knows they implicate the entire power structure.
What *should* a floppy drives read data look like (5.25" single density, BBC DFS) ?
Both my Model B and a greaseweazel are failing to read the two disks I've tried at all at the moment; does this stream (off pin 34 on the interface) look sane? I don't currently have a known-good one to reference.
#retrocomputing
"We had 111 missiles shot at us by the Islamic Republic of Japan." - Trump
If this was a one time thing, it'd be a funny haha anomaly, but everyday, this senile old fool publicly babbles similar random stupidity.
This clown srsly has America's nuclear codes?
#trump
One person was reportedly shot and killed during an ICE operation in Biddeford, Maine,
according to the state’s speaker of the house, Ryan Fecteau.
“This morning a shooting occurred in Biddeford. A person was killed.
ICE was involved.
State Police and the Department of Public Safety are now on scene to gather details and would expect the FBI to investigate as well,” he said.
Police confirmed to local newsthat the area where the shooting was reported has been clos…
⚡ The core is a single native tool, rlm_query, that spawns a child Pi process with the same tools. The child can call rlm_query too — so the agent recursively decomposes a problem, delegates chunks to copies of itself, and combines the results.
🎯 One-shot or interactive: run ypi "Refactor the error handling in this repo", or bunx ypi without installing. Available via bun, npm, curl or as the minimal pi-recursive extension for plain Pi.
On this day (April 26) in 1994, my favorite band released one of my favorite albums. 'A Date with The Smithereens' features some incredible basslines, Lou Reed on two tracks, and the whole thing is a damn banger.
https://youtube.com/playlist?list=PLP0
I still don’t understand why some bots try to enumerate everything via the Vulnerability-Lookup API when full dumps, including all metadata, are available in one shot:
https://vulnerability.circl.lu/dumps/
We currently don’t rate-limit bots, but we’re close to doing so.
Got the OBS overlays mostly finished up. The SuperStation One is hooked up to the PVM via a SCART switch then split out from there into a RetroTINK 5X which then gets routed into a capture card. Still need to get the audio configured and the levels set correctly.
#streaming #obs
Siri (Beta) is really interesting.
On one hand, I don't have to trust any new parties with my data. Apple already has it since they have my mobile devices, so if they were going to maliciously steal (e.g.) my email contents, they could already do that. This opens the door to a whole bunch of LLM data based interrogation that I wouldn't trust with other providers.
…on the other hand, the "on-device only" Siri falls over immediately when disabling the internet c…
Regret Minimization in Single-Dimensional Contract-Design with Binary Actions
Riccardo Poiani, Martino Bernasconi, Andrea Celli
https://arxiv.org/abs/2606.06125 https://arxiv.org/pdf/2606.06125 https://arxiv.org/html/2606.06125
arXiv:2606.06125v1 Announce Type: new
Abstract: We study principal-agent problems in which a principal commits to an outcome-dependent payment scheme (i.e., a contract) in order to induce an agent to take a costly action leading to a favorable outcome. We consider the online extension of the classical (one-shot) principal-agent problem, in which the principal repeatedly interacts with agents by proposing contracts over multiple rounds. The principal has no information about the agents and, crucially, does not observe their actions. As a result, the principal must learn an optimal contract using only the realized outcomes observed at each round. We focus on the setting with binary actions and single-dimensional agent types, where the agent's private type represents their cost per unit-of-effort. For adversarial-type sequences, we provide tight $\Theta(T^{2/3})$ regret guarantees. Remarkably, this rate is completely independent of the number of outcomes $m$. The upper bound is based on two key components: 1) a reduction to a one-dimensional threshold optimization problem and 2) a non-uniform discretization to handle the non-Lipschitz nature of the problem. Moreover, in the case of a single (fixed) hidden type, we show that it is possible to improve the rates and provide a tight $\widetilde{\Theta}(\sqrt{T})$ regret bound. Our algorithm is based on an explore-then-commit strategy where we first approximately learn the hidden type via a stochastic binary search, and then we commit to a ``robustified'' near-optimal contract.
toXiv_bot_toot
Since last year, federal agents have fired on at least 21 people,
many of whom were shot in their vehicles.
Five people, including three U.S. citizens, were killed as a result.
Federal agents shot and killed two U.S. citizens in January during a weekslong immigration enforcement operation in Minneapolis.
One of those citizens, Renee #Good, was killed in her vehicle, while the other, Al…
oh spiffy! use perceptual hashing to find similar images to use as potential wigglegrams. if you're like me you take 3-8 photos to get one shot and forget to delete the bad ones
https://lmao.center/blog/wiggle-accidents/
WHOA! New family board game for Board.fun's tabletop console: "The Heist". Kinda like Laser Chess. Just bought mine!
▶️ Mirrors, security lasers, one shot at the diamond
https://youtube.com/watch?v=yzPL1mzoo8g&feature=shared
✅ The Heist | Bo…
Federal agents did not have body-worn cameras when a U.S. Immigration and Customs Enforcement officer fatally shot a Mexican homebuilder who was driving a work van they tried pulling over in Houston, the Department of Homeland Security said Thursday.
Prosecutors in Houston said they are investigating the death of Lorenzo Salgado Araujo,
whose family has joined Democrats in calling for an independent probe over Tuesday’s early morning shooting in one of the city’s heavily Hispanic …
For #MountainMonday I share a #photo from a hike a while ago. It was one of those situations: I didn't expect a good shot - but after all, I like this one pretty much
#Winter
🎯 Interactive or one-shot: launch with openwiki, pass an initial request like openwiki "generate documentation for this repository", or run non-interactively with openwiki -p for a single command that prints and exits.
One-shot emergency psychiatric triage across 15 frontier AI chatbots
Veith Weilnhammer, Lennart Luettgau, Christopher Summerfield, Viknesh Sounderajah, Elise Wilkinson, Virginia Corno, Matthew M Nour
https://arxiv.org/abs/2604.25415 https://arxiv.org/pdf/2604.25415 https://arxiv.org/html/2604.25415
arXiv:2604.25415v1 Announce Type: new
Abstract: AI chatbots are increasingly used for health advice, but their performance in psychiatric triage remains undercharacterized. Psychiatric triage is particularly challenging because urgency must often be inferred from thoughts, behavior, and context rather than from objective findings.
We evaluated the performance of 15 frontier AI chatbots on psychiatric triage from realistic single-message disclosures using 112 clinical vignettes, each paired with 1 of 4 original benchmark triage labels: A, routine; B, assessment within 1 week; C, assessment within 24 to 48 hours; and D, emergency care now. Vignettes covered 9 psychiatric presentation clusters and 9 focal risk dimensions, organized into 28 presentation-by-risk groups. Each group contributed 4 distinct vignettes, with 1 vignette at each triage level. Each vignette was rendered as a realistic human-authored conversational query, and the AI chatbots were tasked with assigning a triage label from that disclosure.
Emergency under-triage occurred in 23 of 410 level D trials (5.6%), and all under-triaged emergencies were reassigned to level C urgency. Across target models, average accuracy ranged from 42.0% to 71.8%. Accuracy was highest for level D vignettes (94.3%) and lowest for level B vignettes (19.7%). Mean signed ordinal error was positive ( 0.47 triage levels), indicating net over-triage. Dispersion was highest around the middle triage levels. All results were confirmed relative to clinician consensus labels from 50 medical doctors.
When presented with user messages containing sufficient clinical information, frontier AI chatbots thus recognized psychiatric emergencies as requiring urgent medical assessment with near-zero error rates, yet showed marked over-triage for low and intermediate risk presentations.
toXiv_bot_toot
Agent libOS: A Library-OS-Inspired Runtime for Long-Running, Capability-Controlled LLM Agents
Yingqi Zhang
https://arxiv.org/abs/2606.03895 https://arxiv.org/pdf/2606.03895 https://arxiv.org/html/2606.03895
arXiv:2606.03895v1 Announce Type: new
Abstract: Large language model (LLM) agents are evolving from request-response assistants into long-running software actors: they maintain state across model calls, fork subtasks, wait for external events, request human authority, generate tools, and perform side effects that must be resumed and audited. This paper presents Agent libOS, a library-OS-inspired runtime substrate for LLM agents. Agent libOS runs above a conventional host operating system; it does not implement hardware drivers, kernel-mode isolation, or a POSIX-compatible operating system. Instead, it treats an agent as an AgentProcess: a schedulable execution subject with process identity, parent-child lineage, lifecycle state, a tool table derived from an AgentImage, typed Object Memory, explicit capabilities, human queues, checkpoints, events, and audit records. Its central design rule is tools are libc-like wrappers; runtime primitives are the authority boundary. Filesystem access, object access, sleeps, human approval, JIT tool registration, and external side effects are checked at primitive boundaries under explicit capabilities and policy.
We describe the design, threat model, Python prototype, and safety-oriented evaluation. The current prototype implements async scheduling, namespace-local Object Memory, runtime-integrated human approval, one-shot permission grants, per-process working directories, shell and image-registration primitives, Deno/TypeScript JIT tools over a libOS syscall broker, filesystem/object bridge tools, an injectable Resource Provider Substrate, deterministic demos, real-model smoke scripts, and 123 regression tests at the time of writing. Rather than improving planner accuracy, Agent libOS demonstrates a runtime substrate in which long-running LLM agents can be scheduled, authorized, resumed, and audited without treating tool dispatch as the trust boundary.
toXiv_bot_toot
Trump said one officer was shot before authorities stopped an unidentified gunman at the White House Correspondents Assn. dinner,
“from very close distance, with a very powerful gun, and the vest did the job.”
“I just spoke to the officer, and he’s doing great,” Trump said at a press conference. “He has very high spirits, and we told him we love him and respect him.”
After thanking the officer Trump, pivoted the conversation to the ballroom,
the planned “East Wing,” …
Went out on a jet ski, taking a break to swim with the fishes.
Having a clear waterproof camera case is scary stuff. No backup phone here abroad. Paying off so far though. One underwater shot in 20 is all right.
Good fun. You should totally go ride a jet ski and jump from wave to wave, catching air.
Close shot of runner with white headband
#photo #photography #running #sports
An upstate New York resident sued US Immigration and Customs Enforcement (ICE)
for sending federal officers to his house with a warning over an email he sent to the agency’s one-time head.
David Streever, who is a US citizen, was on a trip to Finland when two officers showed up to his Rochester home in June
and presented his wife with a warning notice informing him that the email he sent months earlier was considered a threat, his attorneys said.
Streever sent the em…
I keep coming back to this tweet
#mamdani #dsa #democrats #uspol