Re: On the 1752 date
And prior to 1750, the leap day was often the 24th Feb. (If I've got the date right.) With there being two days labelled 24th Feb in a leap year.
Don't you love calendars?
3789 publicly visible posts • joined 23 May 2011
Look at the costs and the tech it takes to train these models. Very few people can afford to do this or have access to enough of the right tech.
Maybe PRC. But maybe not. It depends on whether they really have trained their own LLMs or whether have just reverse engineered existing ones.
Sometimes, there are files that will defy that. I managed to create a directory called ` '.` You read that right: a directory whose name is a space followed by an apostrophe followed by a dot. I had to turn off path validation - so `rmdir "\\?\path\ '."`
Idle thoughts:
Currently talking to 172.16.8.3:2223 might mean you're talking to 172.16.8.3:192.168.6.76:1912 That's working. And the difference between ports and hosts has turned out to be unimportant.
So, if I was going to extend the address space, maybe the best way to do it would be to add a "subport" field; i.e. extend the address to the right. So the first 32 bits of the address are the global machine identifier and the rest is a "locally significant identifier". So 172.16.0.1.0:80 means talk to port "0,80" on 172.16.0.1.
Notationally, a fifth dot indicates the new protocol, and missing components to the right are assumed to be zero (out to whatever max size is deemed appropriate). So the gateway can scale to it's need. But conceptually, all we've done is extend the port size. So 256 gateways can be replaced with one.
Sure, it would bake in some of the problems IPv4 has and can't do half the fancy things IPv6 can. But it could solve some of the problems - like port churn. If a gateway wants, it can give you a permanent "locally significant identifier" for a certain type of connection. And things behind the gateway can carry on using IPv4. Most notably, a public facing host has the same address for both protocols.
Of course this is never going to happen. It was just an intellectual exercise.
"We find it difficult to believe that the US couldn't have developed its own one-way attack drone, so perhaps it was just more cost-effective to copy what was already known to work?"
You hit the nail on the head there. The usual suspects would have charged $billions and it would have ended up including a holographic projector to disguise itself by the time everyone had had their say. Much easier to point at an existing model and say "give us some of them". The Iranians gave them the specification.
I'm not aware of the details. But it's sound like there is a switch to do exactly that.
The authors of version x can never be sure what the authors of version x+n might decide to do and how features it doesn't understand might impact the meaning of what it does understand. Downgrading is not a common use case, either. So you can understand why this is "at your own risk".
Deferment leaves you with an empty slot that you're unlikely to be able to book at short notice if the system is working and everybody who wants a test has one booked already.
Allowing an instructor to sort this out by swapping between pupils helped the system run smoothly for a long time. I presume it all went to pot during lockdown. And then the scalpers smelt a money making opportunity.
That's why I had option (a).
But they are her lawyers, not the court; I don't see any problem with redacting information they shouldn't see. She could add a note explaining what she had done, pending the arrival of correctly redacted documents from the council.
Or, better yet, as they are her lawyers, she could seek legal advice from them on how to best handle it...
By your argument, a pack of cards can never be shuffled in a way that's "truly random", because they never produce a repeated symbol.
As I note above, banning repeated symbols is not an efficient way to encode your randomness---for the same entropy you need a much longer password---but it could still be perfectly random and achieve the same entropy as in an encoding which permits repetition.
Part of the trouble, here, is that good definitions of randomness are hard to come, and it's quite feasible to get very predictable strings to pass the statistical tests for randomness.
You're right that, if you're generating an n-character password from an alphabet of size s, your naive entropy has gone from s^n to s!-(s-n)!.
But clearly, you've blown a lot of entropy by doing that (I can't be bother to work out how much.) And do you really think the AI is shuffling the alphabet using a cryptographically secure source of randomness and then giving you the first n characters, rather than gently jiggling a pattern it's learnt?
"ONE RIDICULOUSLY GARGANTUAN FLAW in Wayland's basic design...missing the split between client and server that allows you to run the client on one piece of hardware, and the UI on another, over a regular network connection"
I'm old enough to have done that. And old enough to remember the CPU did all the graphics calculations and the memory in your video card was only slightly faster than main memory. Modern machines have a very different architecture. You're asking people to more-or-less switch off GPU acceleration (modulo bodged hacks to get some of that performance back) just for a niche use case.