token_progressive
maybe not the only progressive here
No bio...
User ID: 1737
token_progressive
maybe not the only progressive here
5d ago
Inference costs can be covered with a reasonably priced subscription.
[Citation needed]
This is currently not true. Maybe if you freeze the models and wait for silicon to improve for a few years it will be. But the LLM companies are constantly increasing their inference costs as FLOPs go down in price.
At least according to Ed Zitron's analysis. Maybe you just don't believe his numbers.
token_progressive
maybe not the only progressive here
19d ago
No, I meant what I said. There's no difficulty finding the stale public keys. The stale private keys should be published. DKIM can't provide strongly deniable authentication, but publishing the old private keys would give a weak version of it.
token_progressive
maybe not the only progressive here
19d ago
This issue came up with the Hillary Clinton and Hunter Biden email leaks. While basically no one signs their emails, modern anti-spam technology means nearly all modern emails are signed by the sending server using DKIM. Theoretically, a privacy-conscious email host would regularly rotate their DKIM keys and publicly post their stale ones (i.e. make sure it's trivial to forge old emails), but in practice GMail does not do this and a quick web search finds no one recommending this.
In short, given the full headers, it's possible to cryptographically verify emails really were sent by the user they appear to have been sent by. (Obviously, someone could have gained access to the account who is not the owner of the account or there might be other reasons why the server's signature might not correspond to the human who wrote the email, but those are a lot less likely than the ease of forging an unsigned email.)
https://www.wheresyoured.at/ is his blog. He also has a podcast https://www.betteroffline.com/ but the AI industry analysis there is essentially him reading/summarizing his blog posts.
More options
Context Copy link