Kermit, so fancy! I remember the excitement of discovering ZMODEM and how much better it was than the old standard of XMODEM.

This sort of thing is one of the reasons I chose a RainMachine irrigation controller over other options, because they specifically marketed their cloud-independent firmware design. It was vindicated a couple years ago when they started going defunct and grasped for recurring revenue by billing for proxied remote access, but even then they emphasized that everything else would continue to function without their servers.

The onus is on the consumer to reward cloud-independent designs like this. While it has been sad to see RainMachine’s collapse, my device indeed just keeps working. Hopefully it isn’t ultimately killed by firmware or app security vulnerabilities since it’s now thoroughly unmaintained.

deleted by creator

Software engineering is so often dominated by a move fast and break things mentality, driven by a rush to deploy and scale and profit, with the ability to fix problems with later updates. It’s a very immature process compared to every other engineering domain, because fix-it-later is much more difficult, expensive, and dangerous when it’s a bridge, building, airplane, or anything else tangible (although Boeing did a great job of destroying engineering process and accountability after the MBAs took control away from the engineers).

The work detailed in this Signal blog post is clearly slow and methodical, with continual checks for correctness and curiosity for optimal solutions driving careful experimentation. Building on existing proven PQ standards and keeping their refinements open for public academic feedback is wonderfully responsible. Building formal correctness proofs into CI and blocking trunk merges is spectacular.

They’re doing everything right, even years after Moxie Marlinspike’s departure. Bravo! Working this way is very expensive and requires absolute support from upper management. I’m definitely a fanboy for Meredith Whittaker and the direction she’s running the organization. Hell yeah!

Neither the article nor the RSL website makes clear how pricing or payment works, which seems like a huge miss. It’s not obvious if a publisher can price-differentiate among content, or even choose their own prices at all.

RSL makes an analogy:

Collective licensing organizations like ASCAP and BMI have long helped musicians get paid fairly by working together and pooling rights into a single, indispensable offering.

I’d like to get excited about this because AI companies suck, but if the best example they have is that ASCAP helps “musicians get paid fairly” I’m afraid this isn’t a solution that most content creators will celebrate.

That’s why it’s spelled Froot.

The new offering includes custom national security tools, AI-powered science and health applications, and cleared engineering support for classified environments.

What happens in the SCIF stays in the…oh fuck it, never mind - send it to the cloud for processing. What could go wrong?

Very interesting paper, and grade A irony to begin the title with “delving” while finding that “delve” is one of the top excess words/markers of LLM writing.

Moreover, the authors highlight a few excerpts that “illustrate the LLM-style flowery language” including

By meticulously delving into the intricate web connecting […] and […], this comprehensive chapter takes a deep dive into their involvement as significant risk factors for […].

…and then they clearly intentionally conclude the discussion section thus

We hope that future work will meticulously delve into tracking LLM usage more accurately and assess which policy changes are crucial to tackle the intricate challenges posed by the rise of LLMs in scientific publishing.

Great work.

Almost everything in that list of new features sounds negative to me. A few are neutral, and one might be positive depending on how it’s implemented (having the phone monitor a phone call while sitting on hold). Pretty disappointing, Tim Apple.

From the article:

While drug companies profit from the sales of unproven drugs, everyone else — patients, insurers, and the government — pays a heavy price. In just four years, from 2018 through 2021, the taxpayer-funded health insurance programs Medicare and Medicaid shelled out $18 billion for drugs approved on the condition that their manufacturers produce confirmatory trials that had yet to be delivered.

I’m guessing their citation only includes Medicare and Medicaid because those have publicly-available data for the study to review, but I have to assume that private insurers pay a ton as well. I can see your point that insurance denials result in angry sick people, but there’s not really a lot of nuance in “that medication has never been shown to be safe and effective for your (or any) condition.”

I dunno. Everyone sucks here.

American health insurance companies are famously miserly, and this seems like a great area to use penny pinching for good. Where the hell are the insurance CFOs who should be demanding efficacy proof instead of being swindled along with the masses?

Ubiquiti is a pretty good answer, but the airMax line is better for this sort of thing than UniFi. A pair of Nanobeams would be perfect, and standalone without need of an additional controller.

They have updated it so that you don’t need to use your phone number as the identifier you share with other people so that they can message you. You can now give out a username and your new contact will not be able to learn your phone number.

As for Signal itself knowing what your phone number is, I don’t see that as much of a problem, because they intentionally don’t know anything useful about you. They publish redacted subpoenas and their responses so you can see just how little data they can provide. They don’t know who your contacts are so there’s no social graph to be drawn.