[randomdata]

The private information is shared with Youtube/Google, so the assumption is that anyone who is an agent of Google is in on the secret. If it must only be in the hands of one or a small group of people at Google, you'd best go to those individuals directly, not through the overarching entity of Google as a proxy.

[throwaway2016a]

> so the assumption is that anyone who is an agent of Google is in on the secret

I think there is a difference here between "expectation" and "assumption".

Without the ability to do a third-party audit I agree the only reasonable assumption to make is that everyone is in on the secret and when dealing with sensitive information it should always be the assumption you go with.

However, as an expectation, I expect SaaS and social network providers (and by extension most of the HN crowd) to be better.

[randomdata]

There may be a difference, but it seems you have them flipped. It is a reasonable assumption to think that they have controls to limit who is able to see information[1], but one must go in with the expectation that every acting agent has access.

[1] Of course, since you don't know who the individuals are, you still have to place your trust in every single agent that works for the entity you chose to entrust. As such, nothing is gained by restricting access. It remains that if it is important that it be private with only one or a few, you must go to those individuals you trust directly. Granting them private information by proxy will always be subject to man-in-the-middle-ing.

[ImPostingOnHN]

I think you have it backwards: an expectation is a standard (the term is used loosely here) that someone should be meeting. We expect people to do the right thing, but sometimes must, as in this case, assume they are doing the wrong thing.

Applied here, the expected and right thing to do is follow the principles of least access. However, we must assume google is not doing this, because there is insufficient evidence that they are, and there is actual evidence that they don't have sufficient controls to limit who is able to see information.

[randomdata]

Right, expectation is the standard. The standard is that anyone who is an agent of the entity you have entrusted is also considered trustworthy. After all, giving full trust to an entity you only trust partially is nonsensical.

However, you make a fair point that it is reasonable to assume that entities you trust are willing to go above and beyond, for various reasons.

[ImPostingOnHN]

> Right, expectation is the standard. The standard is that anyone who is an agent of the entity you have entrusted is also considered trustworthy.

To clarify, I am the second person here telling you that that is not the expectation. The expectation, and/or the right thing to do, and/or "the standard we expect them to meet", is that Google follows the standard security principle of least privileged access, meaning each employee can only access data they need to see, with proper permission acquired beforehand, auditing during, and abuse-detection & alerting afterwards.

Unfortunately, they don't meet this expectation that we have of them. Your own expectations and/or standards might be lower, like you described.

[randomdata]

The expectation is placed on the entity the trust is given to. Unless you go to individuals directly, there can be so such expectation on individuals. There may be an assumption that the entity you have given trust to do will "do the right thing" with individual agents, but it there is no such expectation as you have already trusted the entity, meaning that you have already trusted its agents. What have you gained by keeping information away from people you have already entrusted with the information? If you cannot find the trust to give them, why are you giving it?