closetheloopdev

I remember when Opus 5.5 came out, we had a discussion about whether it was nerfed a week later [1]. It's not possible to do so with Gemini 4 models since none has been made available the day it is announced.

Without us being able to access Carbon when it is announced, there's no way to know the Carbon we get a few weeks later is the same thing as what has been benchmarked here. I think benchmarks should verify what is benchmarked is exactly the same as what's been released.

  [1] https://news.ycombinator.com/item?id=49901736

The problem is that these names are random and never-ending. They are not categorized like Claude and GPT, both of which have four classes of models each. That's eight classes that I can remember.

There is no way I will be able to (nor do I want to) remember more than four names just for Google or any other single provider, and there's no value in doing so for me. There's nothing unique or distinguished that Gemini does that neither Claude nor GPT can.

I used to land on SO questions and answers through Google. Nowadays I ask ChatGPT or Claude instead of Google, and then go to the links they give me to verify. The links are either to the official docs or some blog posts. Never have I seen LLMs recommend any SO links.

Despite that move, the gravitational pull was quite strong on their stock today after the earnings call.

True. And if Carbon (6) is better than Argon (18), wouldn't that make Hydrogen (1) the ultimate Gemini model, or will the next model jump to another random element?

  O, B, Sc, U, Re?
  Ra, Nd?

I agree. Announcing a new model without immediate availability or a definite release date seems confusing than just doing it later. When Carbon is made GA, OpenAI or Anthropic might just release their new model, and we'd be like: Carbon what?

Having said that, I don't see people excited about using Argon so far, so that might be why Google is announcing this to hype Gemini up a bit.

I created a proxy using the 6 compromised eMudhra CA keys (2 of the 8 are code-signing and not for TLS). I initially tried just one (the one with d=4) like in the commit, but it did not seem to work for anything. Once I added all 6 keys, I was able to intercept some.

I was mainly interested in testing which IoT devices are affected and protecting them, not trying to sniff their traffic since it's been a long day. My proxy thus just intercepted and forwarded the traffic upstream using a fresh TLS connection that does not use the compromised keys.

Thank you! Working on the proxy now ...

After my shock and anger subside, I can think more clearly now. I see many have mentioned using this to perform MITM on the IoT devices.

On the other hand, can we not utilize that to also protect the IoT devices? Can we not create a MITM proxy on our own network that proxies all of our IoT devices's traffic and then forwards them to the trusted HTTPS, MQTTS, ... endpoints upstream? Since the proxy already revokes these bad certificates, the upstream traffic should be protected. We should be able to revoke any other bad certs in the the future as well.

After my shock and anger subside, I can think more clearly now. I see many have mentioned using this to perform MITM on the IoT devices.

On the other hand, can we not utilize that to also protect the IoT devices? Can we not create a MITM proxy on our own network that proxies all of our IoT devices' traffic and then forwards them to the trusted HTTPS, MQTTS, ... endpoints upstream? Since the proxy already revokes these bad certificates, the upstream traffic should be protected. We should be able to revoke any other bad certs in the the future as well.

Do we call this human's slop? Should we be so hard on AI when humans do something like this? I'm pretty sure LLMs did not and would not allow such things to happen.

On the bright side, hopefully with AI we won't see silly slops like this.

I'd need to see a real video of such a thing first :-)

Do you have a link to it? I could not see the landing well in any of the videos since there was a lot of white cloud when it landed.