/biz/ - Business & Finance

That fud at the end is legit. There's not enough tee decentralization since there's only so many big name chip companies. Hopefully it doesn't fuck up the entire defense in depth thing they have.

>>11577362

It doesn't at all - Sergey could have BTFO these guys but for some reason he's either too nice or he doesn't understand their questions because they're so unknowledgeable.

For the guy who keeps talking about TEE and how it's bad, you can just choose not to fucking use TEE nodes then. Also you could pick multiple different TEE's, or some TEE and no TEE, or any fucking combination of this you desire.

>>11577806
This but I will say that if you pass private keys to any TEE that is compromised, that could be a disaster scenario.

Asking if Intel can be trusted is like asking if your money in the bank is safe.

>>11577858
Intel is not immune to attacks against SGX.

>>11577847

Agree, but again what Sergey even mentioned is that it's unlikely the main use of a zeroday or other exploit in a TEE would be to steal a private key, you could attack in many other ways not related to crypto that would be more profitable or damaging.

>>11577362
It was indeed a good question but sergey had answered that already in another question: you can have 5 high rep nodes, pay big bucks (change that for 5 tee environment), perhaps doing several requests, and then a shit load of others, doing basic confirmation for that results. the other anon just said it better. this is a combinatory issue. you can combine to achieve more security, and tee's are part of it. all nodes executing through tee's? yeah that might be problematic, but that's for the contract builder to decide. to simply say its unsafe to have that possibility is wrong.

Also he said more will announce they are working with them ehehehehehehheheheheehehe

>>11577131
Good stuff, even if that pajeet tried fudding at the end