

Hopefully this episode gives the Danish government reason enough to get away from American owned tech.


Hopefully this episode gives the Danish government reason enough to get away from American owned tech.


The irony of this being about the apple and Google app stores is palpable.


It’s kind of wild to criticize people for not providing the conversation prompt logs they had with LLMs and then publish code on GitHub generated by an LLM without publishing their own LLM conversation log.


“Infini-attention” isn’t perfect memory, it’s highly compressed representations of the entire history. (https://arxiv.org/html/2404.07143v1).
Much like quantization. The longer the context gets the worse the compression makes recall.


For a lot of people for a long time your insurance card (that didn’t have a photo) was the only “identification” you had. Otherwise you had to bring your school ID, work ID etc.
Most people don’t have drivers licenses cause they take the train. When you sign up for banks etc you usually have to get a bunch of official documentation from the local ward office with your information.
Proof of identity in Japan has always been a bit of a hazy problem. You sign most documents still with a family stamp, so the idea of what legally is defined as identifying is kind of vague.
Most local offices aren’t networked up, so when you move you have to register with your local ward office and the japanese beauricratic army goes and gets the previous ward office to fax over your info.
“My Number” is the japanese governments attempt to get all that stuff wired together in one database.


Yes the “my number card” (national ID) was mostly a volunteer thing but now that it’s needed for health insurance it’s required of everyone


Japan just hooked up the national health insurance to the national ID.


For those interested:


deleted by creator
You can’t, if the code is open source it can be cloned to not fit in the license no matter what kind of license or fancy shenanigans you do.
The argument most MIT/BSL proponents have is that companies will be more likely to directly contribute if the project doesn’t have GPL “poisoning”.
I usually split the difference and license LGPL for everything.
the GPL v2 doesn’t have any less restrictions around strong copyleft (requiring that a company publish changes for components).
Maybe you’re thinking of the fact that GPL requirements don’t cross the kernel module syscall boundary?
This “poisoning”, effect is the reason the LGPL and AGPL licenses exist.


Only 20k nbd
It’s a shame Moore’s law doesn’t seem to hold anymore.


Research from a few years ago was able to measure gait (so a person’s height and build etc) from the wifi shadow of a single router.
I assume 3 is to get the super accurate placement.


They said they can’t do that in the article:
but at the same time, we cannot ‘take over’ the application identifiers for the open-source apps we distribute, as that would effectively seize exclusive distribution rights to those applications


They’re both “immutable” in the sense that they’re setting up either read-only Filesystem Hierarchies (as in bazzite, which uses ostree) or Symlinking their entire filesystem hierarchy to a read-only “store” (as in nixos).
Bazzite uses something called ostree to “diff” the filesystem hierarchy much like git does, while Nix basically makes giant read-only store of files and hashes them, then weaves them all together into a “view” of a filesystem that gets symlinked into the context of a running program.


It’s not just that, it’s also the fact they scored the responses based on user feedback, and users tend to give better feedback for more confident, even if wrong, responses.


I know of matrix, what are some other alternatives?
Also a protocol that got falsely maligned during the crypto days was secure scuttlebutt, and people should be talking about it more.


And then when those AI also have issues do we use the AI to check the AI for the AI?
Ten bucks says you downloaded a compromised VST.