I am all for using proper typographic symbols, but it is unclear what place the precomposed ellipsis U+2026—what I assume you mean by “true ellipsis”—has in that canon, especially with the compressed form it takes in most fonts.
I integrated Turnstile with a fail-open strategy that proved itself today. Basically, if the Turnstile JS fails to load in the browser (or in a few specific frontend error conditions), we allow the user to submit the web form with a dummy challenge token. On the backend, we process the dummy token like normal, and if there is an error or timeout checking Turnstile's siteverify endpoint, we fail open.
Of course, some users were still blocked, because the Turnstile JS failed to load in their browser but the subsequent siteverify check succeeded on the backend. But overall the fail-open implementation lessened impact to our customers nonetheless.
Fail-open with Turnstile works for us because we have other bot mitigations that are sufficient to fall back on in the event of a Cloudflare outage.
Only if they are able to block the siteverify check performed by our backend server. That's not the kind of attack we are trying to mitigate with Turnstile.
The lettering on the proposal is very interesting. I assume it was hand drawn. I can't find any typeface quite like it. Google Lens suggested it is similar to Copperplate Gothic, which it is and at once is not even close.
Neutraface has some similarities, including the low crossbar height. Fittingly, it appears to be used in materials published by the Frank Lloyd Wright Foundation.
Shazam-like fingerprinting for text. The complete LLM outputs wouldn't need to be stored, just the fingerprints along with some mechanism for trusted timestamping (could be Blockchain).
This has been done for a very long time. Blockchains are definitely not required (this isn't just the usual hate from HN of Blockchain, it just actually doesn't make sense here).
Fingerprinting by shingling (windows of text) with some normalization steps is pretty typical in plagiarism or similarity detection. A big database of docid-shingleid pairs along with weights for their frequency is often a very simple and fast way to do this analysis.
The big part is getting OpenAI/anthropic/etc to do it on their data and provide a service for that, but there's obviously a lot of unwanted consequences - specifically storing of all user data (even if the shingled and docids are hashes, it's still info).
> I let it expire for a few days, now they are charging me $100/year!
It sounds like your domain name entered the redemption period. The high fee to restore a domain in the redemption period is mandated by the registry, not the registrar. So Gandi is not price gouging; they are required to collect that fee for restoring the domain.
