rpm91

My guess would be that the amounts of compute required for these evaluations constrain the sorts of facilities where they can be run. It's certainly much easier to use existing datacenter infrastructure than build dedicated facilities for evaluation. It also sounds like they wanted it to be able to use software package registries, given that the initial compromise was a "proxy and cache for package registries".

That said, these are both solvable problems. I'd hope that the frontier labs are thinking long and hard about the possibility of using airgapped facilities, given that relying on a lack of vulnerabilities in the software used for containment is unlikely to be a viable strategy, especially if models' exploitation capabilities continue to improve.

I tried the detector on several of my own Stack Overflow answers, and the fourth or fifth one I tried was flagged as "85 % machine / likely machine-written" with the default settings, well above the default sensitivity threshold of 42%. When I turn off the option to strip LaTeX commands (the source is Markdown), that jumps to 96% machine.

Admittedly, a Stack Overflow answer is somewhat outside the realm of scientific writing, so it's still possible that the detector may be accurate within that domain. That said, it's a cautionary tale on the hazards of applying classifiers like this outside of the domain that they were trained on.

I think calling the "vast majority" of any country "pure evil" is painting with a pretty over-broad brush.

Specifically here, social media sentiment tends to be poor evidence of anything broader, due to being dominated by the vocal minority of people who post the most, as well as a non-trivial number of bots.

Perhaps the link should be changed to this, as this has more details, formal confirmation from a reputable source, and contradicts some aspects of the original source, such as that the website will go down at midnight, which wouldn't really make sense—presumably they'd want to sell the remaining merchandise.

Edit: or the link in the reply below, which is an even better article (and explains a potential reason that they may not have to sell any remaining merchandise).

All you've done is deploy the existing documentation on your own service and change the links on the GitHub repository to point to your service rather than the GitHub documentation (with no indication of how one would keep this in sync with updates).

Sure, it makes navigating this version of the docs slightly easier, but this doesn't seem to be a good-faith attempt at actually improving the documentation so much as promoting your service.

Apple claimed that Huseby was allegedly importing “counterfeit” iPhone screens. Huseby denied this, saying that he simply used refurbished iPhone screens that he never advertised to the public as “genuine” parts from Apple.

I'm no fan of Apple's crackdown on repairers in general, but this sure sounds like he had third-party screens that he called "refurbished iPhone screens," which sounds pretty misleading to me. If someone told me something was a "refurbished iPhone screen," my assumption would be that it was originally a genuine Apple part, and it doesn't sound like that's what he was using.

Just leaving out the word "genuine" doesn't make it not-misleading.

This is referring to the sales tax in the buyer's state. I'm not aware of any state that requires sales tax in the seller's state for orders that were not made or received in that state.

Today. I have a friend who insists on using either that or Snapchat, and Snapchat doesn't have a web client, so Riot wins for that reason alone.

True to form, I experienced yet another bug with the encryption moments ago: on web, search appears to be completely broken in e2ee rooms. It just says "No results", even for words that were said in literally the last message (and messages not far back in history from the previous day, so unlikely to be a caching issue).

Do you know why the decision was made to rewrite rather than do an incremental migration, given the increased pain to users? I have some guesses from looking at the codebase, but I'd love to know the actual answer.

For the client experience, it's not even close: Slack's product experience is far superior. I love the idea of Riot as a decentralized e2ee group chat client, but the implementation is so buggy and far from being production-ready that any attempt to use it seriously is an exercise in frustration. In comparison, Slack is mildly annoying at worst, and usually just works.

If it were one or two little things, I'd file some bugs or even try to fix the issues myself in the Android app, but my experience has been that I have at least one experience-breaking bug or UX issue at least every other time I interact with the client. Especially if you have encryption enabled.

I'll give the author the complaints about modern scroll bars, which drive me up a wall, but the complaints about Firefox and Chrome really feel like grasping at straws to find something to dislike.

"This new take on the [Firefox] URL bar pops up when you least expect it, is very hard to get rid of and, as a bonus, covers the tried, tested and highly usable bookmarks toolbar below it."

What? It pops up when you click on it, or when you hit Ctrl+ L. That seems perfectly expected, and is how URL bars have worked as long as I can remember. And it only covers the bookmarks toolbar (should you choose to enable it) when you're using the address bar and the dropdown is visible. This is like complaining that the Save File dialog covers the tried-and-true document editor in Microsoft Word, but even more ridiculous, because you can, in fact, access your bookmarks through typing into the megabar. It's an excellent UI that provides keyboard access to a wide variety of browser features, all in one location - search, URLs, bookmarks, and even a fuzzy search of open tabs.

And the Chrome tooltips...behave exactly like normal tooltips, but with slightly different styling to make them more useful for showing the title and domain name. How is this a "crime[] against basic concepts of desktop UI design"?

(disclosure: I work for the company that makes one of those applications)

I believe that line in the Q&A is referring to the "other COVID-19 tests, where the test result is available in a few minutes". It's ambiguous and confusingly phrased, but it seems like Bosch directly detecting the pathogen, in contrast to other rapid tests (which don't, presumably, use PCR, as that takes longer than "a few minutes").

My perl is super rusty, but you've got me curious...what happened?

A backup method that just copies it _somewhere else on the same device_ is not really a very useful backup. It involves manually moving files around if you want a real backup, which feels very antiquated in 2020. Why isn't there some way to automatically store my (encrypted!) backup via a cloud backup solution, whether that's Google Drive, Dropbox, or just allowing it to be backed up as part of an Android backup?