The third-party list on page 12 is not small. The real-time api architecture creates a live, per-query link between a specific user event and every broker in the chain. Batch transfers or delta shares would break that linkage. Zero-knowledge proofs (also mentioned in the study) can prove age without handing anyone a name, document, or photo.
There's no reason Aristotle or Veratad should see who the underlying requestor is. Yoti should receive the verification request, strip the context, make the request - that's it. The fact that it isn't structured that way and they are tagging on additional metadata suggests per-query economics, which creates a direct incentive to route more verifications through more parties, exactly backwards from data minimization. I'm not going to call it a rev share, but the architecture is consistent with one.
While I agree with your claim that it is likely the number of third parties info is being routed to is likely related to per-query economics, I want to note that ZKP are not magic. They tend to either be worthless at preventing fraud or require so much additional context as to question how much privacy is really being preserved.
While ZKP is more useful in limiting how much info is provided and, depending on implementation letting you make sure of the full scope of information acquired....if it literally only validates age then there's nothing other than logistics preventing a single adult from authenticating the entire world.
Totally agree. Was just trying to emphasize that there are better ways to do this if privacy and security are something that Yoti actually cares about. ZKP is not a magic bullet.
Has there been a lot of progress with ML in Rust? I don't really keep up with it because it seems like every crate ends up getting abandoned and I just gave up caring.
Yeah Claude is really, really good there. You tell it the distro and the problem and it will solve it. Saved me a lot of pain when it came to swapping out an encrypted boot drive and was good about emphasizing the order of operations required for what I would consider a higher risk/complexity situation.
JetBrains has several niches it excels in. DataGrip is by far the most important tool in my toolbox, as it allows me to work with every database type imaginable in one place (Databricks, Postgres, MSSQL, Oracle, etc.).
And people wonder why there is so much push back against AI. The last thing leadership should do when laying off people is use the term AI. It's the most tone deaf thing you can do.
We don't live in the same world as they do. Saying AI out loud makes line go up, not down. Investors are still eating this shit up, for now at least...
Effort is still being put into it. Just this weekend YouTube put the 4K Coachella streams behind SABR. I could still get 1080p easily but 4K required some fanangling.
I failed to run in LM Studio on M5 with 32gb at even half max context. Literally locked up computer and had to reboot.
Ran gemma-4-26B-A4B-it-GGUF:Q4_K_M just fine with llama.cpp though. First time in a long time that I have been impressed by a local model. Both speed (~38t/s) and quality are very nice.
There's no reason Aristotle or Veratad should see who the underlying requestor is. Yoti should receive the verification request, strip the context, make the request - that's it. The fact that it isn't structured that way and they are tagging on additional metadata suggests per-query economics, which creates a direct incentive to route more verifications through more parties, exactly backwards from data minimization. I'm not going to call it a rev share, but the architecture is consistent with one.
reply