Not news to me, either, Drew.

Aaw, I should have told you. Thereâ(TM)s some equation that relates energy, mass, and the speed of light. Probably most physics postdocs have heard of it.

Thanks, Will-O. I quite remember this cia guyâ(TM)s rubber stamp collection with dozens of great stamps: NOFORN, EYES-ONLY, and different fonts of Secret and top secret. Damn, but the guard at the place seized my paper with all those stamps.

On the other hand, I still have a VIP parking pass for the CIA headquarters â" Lets ypu park right at the front steps. Valid, if you have a time machine going back to April 5, 1988.

My warm thoughts fly over to you, Kill (owatt-) Hour â" last weekâ(TM)s eclipse fund me beneath stratus clouds just east of Buffalo, having a wonderful (though cloudy) day with 4 generations of my family. Itâ(TM)s a joy to see how my home town has evolved â" memories of climbing the Michigan Avenue Lift Bridge at midnight and watching them tap the redhot coke ovens at Bethlehem/Lackawanna Steel mills.

Following up, I am honored by the attention and kindness of fellow nerds and online friends. When I first started on that chase in 1986, I had no idea wrhere it would lead me.

A curious accounting error led me through Unix internals, tcp/ip protocols, early Arpanet connections, and backwards to a group of computer hackers working for then Soviet & Stassi agencies. Along the way, I met people from the FBI, NSA, CIA, AFOSI, and plenty of very smart computer jocks.

It was a time of analog phones and dial up modems; when you would carry coins in your pocket to make calls on the street.

Since then, thanks to the support of online friends and math folk, I have explored and shared interests in topology and math. Along the way, Iâ(TM)ve made plenty of mistakes and bloopers; pretty much the same as student times. Goofups in grad school are easier to sweep aside!

To all my friends: May you burdens be light and your purpose high. Stay curious!

- Cliff

Iâ(TM)ve been away from slashdot for a while, and Iâ(TM)m now on a post-eclipse trip on the east coast.

With good fortitune (and Amtrak), Iâ(TM)ll be home in 10 days; Iâ(TM)ll then fill the tsunami of Klein bottle orders that havve arrived in the past few hours. Over a dozenâ" Iâ(TM)ll be catching up for a few days!

Smiles all around,
-Cliff on a rainy Saturday in Potsdam, NU

The example videos actually show Sora generating multiple shots for a single video. I think the larger problem is similar to why it's difficult to get GPT-4 to write a novel or DALL-E to draw a comic book: it's hard to get current models to maintain consistency across a long time. You need some way to get characters and places to look recognizably the same in different shots, and that's not something our current technology handles well.

On movies with long takes, Hitchcock's Rope is a classic example. Birdman is a more recent film which plays with the idea (the film appears to be a single continuous take, albeit with intentionally obvious use of CGI to achieve that effect).

This is broken thinking. Processes never go wrong due to a single mistake. If you think there was only one mistake, then the second mistake was not designing the process to have a check for that category of mistake. If this computation was so important, surely there should have at least been a second set of eyes on it. I guess we're reading this article because there was a second pass on the computation, it just happened months too late.

Aren't QR codes just text in a format that's easy to read for a machine? How are the risks any different from going to a random website, which should be limited to browser bugs or whatever you do on the website (the internet has a lot of warnings about phishing via QR codes... which doesn't seem relevant to viewing restaurant menus unless they also do ordering through the smartphone)? i.e. you'd have the same risks if the menu link were given as a URL shortener or otherwise obfuscated link.

I believe you that a bug in the QR code processing code existed, but I can't find any reference to its existence in a quick web search. Other than that, my phone shows the text of the QR code and asks me what to do with it before taking an action, even if I use the camera app it came with to read the QR code. Is that not what every phone does?

This makes no sense. If you and the staff think they should be paid more for serving more customers, then pay them a commission and increase the price of the food enough to cover the commission. That would come out approximately the same as tipping except for it involves the prices on the menu being the real prices and no weird perverse incentives created by random customers deciding how much the staff should be paid.

This whole post is SPOILERS.

I did feel like this wasn't stated explicitly in the TV episodes, so you had to infer it from Holston dying after he took his helmet off, but the video of it looking lovely out there is a fake video shown on the helmet of the suit. The toxic wasteland is the truth. (Notice there's never a camera shot of the outside not filtered through the helmet camera or the cafeteria screen, so it's never made explicit.) Also, it's strongly implied that prior to the start of the series, cleanings have been decades apart, so people wouldn't have a good memory of a prior cleaning... but I agree it didn't make sense that Holston still cleaned knowing what he knew. I do recall when I saw the screens in the TV series, they were a lot bigger and clearer than I had imagined from the description in the books,

That said, a lot of the worldbuilding, including the people deciding to clean doesn't really quite make sense. Like how all the stuff they have has survived hundreds of years (or, equivalently, where they're fitting the manufacturing capacity to remake it).

Discord logins aren't by username, so they don't have that problem. Logins are keyed to email and/or phone number. The database master key is a string of digits (visible to the bot API, never intentionally shown to the user, although the mobile app shows it when it fails to load a username).

The username + 4 digits thing is used in Discord to simultaneously handle having a default "display name" of "@username" while also have a user-facing global identifier of "@username#1234" (you can also set a per-community ("server") display name to override your default display name, but people in that community can always see your global profile). The proposed change is to split the two into an arbitrary display name (similar rules to the current "username") and a user-facing global identifier that they're now calling a username so you'll have to be "@username1234" (or come up with something more creative if you want). But that user-facing global identifier will be pushed even further into obscurity since it will be decoupled from the default display name.

Making public IDs incrementing numbers makes them guessable and makes IDs leak information about the order accounts were created (Slashdot considers this a feature, not a bug), which can be a problem. In this case, that could let you easily spam friend requests to a lot of accounts by just enumerating them. Sure, you could (and should) have additional layers of security preventing an attacker from making use of successfully guessing an ID, but making the IDs not guessable in the first place is generally considered a good design pattern.

(There might also be some distributed system / race condition details of not requiring your servers to agree on what the next ID is?)