> Digital distribution and lock-in are a disaster for historical preservation.
Agreed. This reality slapped me in the face quite hard after I was given an ipad 2 (circa 2011). I figured I'd use it as a couch-side device for spotify. Soon realized nothing worked on it, even with old version apps loaded. Server side shut it down.
That's partially because of web security, though, if I'm not mistaken. I have an old Windows PC and an old Mac mini that I use as servers and, outside of local stuff that's bridged to the outside network (and heavily crippled), I can't do much on them because of the changes to the TLS protocol for SSL certs. That means that the browsers won't connect to most modern servers, any apps that hardcoded instances of TSL can't connect to the internet, and, if I managed to get something to connect, it would be a huge security risk for anything on those machines.
So, if the connection is being shut down on the server side, it's likely because of the TLS version mismatch. I thought it would be a simple solution just via updated software but was told (by peeps on Reddit, so it's likely to be total nonsense) that, although it can be updated, it won't run reliably on older hardware unless that hardware can also support newer OSs. Win7 and below (?) don't support it and MS won't provide updates with it and MacOS versions before Lion don't have it. I've heard you can also update OpenSSL by breaking the symlinks to the OS's install and get it working again but that's more hassle than I cared to put in.
Also, most of those old iPads are able to be jailbroken. You can update OpenSSL on them and probably get Spotify working again.
Yes. Although I would disagree with the statement "it would be a huge security risk for anything on those machines"
Older TLS protocols have vulnerabilities yes, but most of them require a very motivated attacker with the ability to do a MITM.
I wrote a blog post about why I have decided to support older protocols, https://blog.nyman.re/2021/02/07/usability-security.html , the tl.dr. is that only allowing new TLS just means more forced obsoletion. For Banks and other sensitive things, yes it makes sense, for your personal blog, not so much.
I mean... Google still allows TLS 1.0 , if it was a "huge security risk", do you think they would?
The updating OpenSSL/breaking symlinks sounds interesting, do you have more info?
Personally I have a old iPhone 5S running iOS6 which I use for listening to podcasts. I ran in the the TLS issue there and my solution was to use a proxy https://bitbucket.org/ValdikSS/oldssl-proxy
Works well, and until maybe a year ago or I could even browse Apple's Podcast Store, but at some point that stopped working so now I'm stuck with the ones I have there.
> That’s also why it takes 2 days to send money to people, in an era of instant communication.
I always assumed anti-fraud had something to do with it. Give the victims time to notice and the institutions the ability to reverse transactions without loss
> I always assumed anti-fraud had something to do with it.
The banks must have had a good PR campaign going because I was under this impression for a long time too. Something didn’t click when I learned that mistyping a single digit in an ACH transfer means your money is gone forever…
You can get professionally printed proxies [1]. I've been told there's often a "don't ask, don't tell" policy around many unofficial mtg tournaments regarding proxies. It obviously upsets some people, but others see it as requirement for new players to enter legacy format (and keep it alive).
I've been told that many players buy counterfeits specifically for old school tournaments so they don't put their very expensive pieces at risk to theft or damage. I can't verify this, but it makes a lot of sense to me ¯\_(ツ)_/¯
I don't know. I heard this narrative a lot more during the trump era, and I rarely hear it now. I have a pet theory that it comes up in popular Canadian culture when a republican is in office and dies down when a democrat is there.
Maybe it’s changed, but the “Canadian identity” has long been this amorphous thing many Canadians quietly fret over because nobody is quite sure what it is but they're sure it exists.
Since US and Canada are so similar, a lot of the discussion of identity are around the differences, which to me is a weird way to talk about a national identity.
It also comes up a lot when someone mentions something positive about the US that Canada struggles with, say economic growth - then it becomes a common refrain of “yes, but at least X is better in Canada than the US”.
It’s also used commonly in discussions in healthcare - any change is described as “Amerification” of the healthcare system even those the change would just make Canada more like European systems.
Maybe it’s unavoidable being right next door to the US, I don't know.
Can you unpack this a little please? Is it possible to ELI5 the mechanisms involved that can "push" a rule set out? I would have assumed the rules apply globally/uniformly across the entire prompt
Thanks! So is patching this as simple as not allowing the entire space of X for user prompt? i.e. guaranteeing some amount of X for model owner's instructions
No. The input and the output are the same thing with transformers. Internally, you're providing them with some sequence of tokens and asking them to continue the sequence. If the sequence they generate exceeds their capacity, they can "forget" what they were doing.
The "obvious" fix for this is to ensure that the their instructions are always within their horizon. But that has lots of failure modes as well.
To really fix this, you need to find a way to fully isolate instructions, input data, and output.
>So is patching this as simple as not allowing the entire space of X for user prompt?
>No
Isn't the answer yes?
>The "obvious" fix for this is to ensure that the their instructions are always within their horizon.
That's what I take GP to be suggesting. Any possible failure mode that could result from doing this is less serious than allowing top-level instructions to be pushed out, surely?
Agreed except ChatGPT (3.5 at least, haven't tried 4) is unable to provide primary sources for its results. At least when I tried, it just provided hallucinated urls
I wanted to hear this again. Leaving it here for the next person: https://www.youtube.com/watch?v=Fm5FE0x9eY0