> The goodness of an action's consequence and whether that action is a violation of privacy are orthogonal. Yep, I agree. What I meant is that the expectation

> The goodness of an action's consequence and whether that action is a violation of privacy are orthogonal. Yep, I agree. What I meant is that the expectation of privacy tends to be determined by expected necessity and consequences. People tend to expect privacy where they expect sharing/collection/distribution of information to have (potentially) bad consequences for them and there is no reasonable need to do so that would justify the risk. And my point is that the expectation of privacy is not a function of just the data, but also of the (intended, potential) use. You expect your letter to be delivered, therefore you expect that the address information on it will be used for that purpose (after all, having the letter delivered is your goal, and you are willing to give up as much of your privacy as is reasonably necessary to reach that goal), but you do not therefore expect that same information to be published in the newspaper, hence it's neither "public" nor "private", it's "for those involved in delivering this letter for the purpose of delivering this letter", just as medical information is "for those involved in making you healthy for the purposes of making you healthy". > You seem to consider any inappropriate action which in some way involves personal information to be a violation of privacy. Nah, I don't think that is quite it. It's more about the collection and processing of information for purposes that are not in the interest of the person the information is about in a way that reveals information about them that they could not reasonably have expected to be revealing when they revealed whatever information is being collected and/or analyzed. Privacy ultimately is about an individual's ability to predict and have control over what other people and institutions know about them while still being able to function in society, though not primarily "control" in the sense of "forcing to know or forget", but in the sense of "being able to estimate the consequences of sharing some information", as in "being in control of your life". The ability to function in society might require forcing at least computers to "forget", though, as open complete surveillance, while somewhat predictable, prevents society from functioning well. > Privacy is about the ability to control which bits of your personal information area are available to which individuals, not what people do with your personal information after they have it (unless of course what they do includes sharing it with more people). I think there is uselessness lurking somewhere in that definition ;-) Does a government database count as an individual? Does government employees looking at the database constitute sharing? Does that control, at least in some cases, encompass new knowledge derived from previous knowledge in ways the affected person could not possibly have envisioned? If not, how do you resolve the conflict with the person's supposed control, if yes, how do you resolve the conflict with other people's supposed freedom to do with the information they have whatever they want (apart from sharing it)? Would blackmailing you with information about you be a violation of your privacy, if the person blackmailing you does not actually intend to share the information with anyone? How about just collecting information with the goal of possibly blackmailing you later (though again without any actual intention to disclose any of it)? Somewhat out of order: > I do not, and I believe my definition more closely matches general usage. I'm not sure what the general usage is, but in any case I think the general public is far too confused about what information is and what automated information processing can do and can be expected to do in the near future for many to have any coherent, let alone reality-conforming concept of privacy (just look at the repeated high-profile cases of failed anonymization of datasets! their goal obviously is "privacy", and yet they fail miserably ...). I might agree that your definition is probably close to a reasonable working model of privacy in the pre-computer era. But if you apply that too literally to today's world, I would expect you to end up with a practically useless concept that ignores many real-world problems that result from the collection and analysis of personally identifiable information. If you want to insist on such an outdated definition, well, there obviously is nothing fundamentally wrong with that, I just don't think there is much value to such a concept, and I'd rather define some privacy2 instead that has meaningful applicability today. > That's a different example, and I suppose it would depend on what diagnostic equipment we're talking about. For example, I do believe there is expectation of privacy when it comes to X-raying on a public walkway. Well, how about a camera that can detect heart problems from the pattern of blood flow in your face as it can be detected using high-speed high-resolution video in bright sunlight? (No clue whether such a thing exists, but I think we can agree something similar in some way or another should be technically within reach at least?) Also, what if the camera is actually just a high-speed high-resolution camera that doesn't do any diagnosis itself, but the video it records can later be analyzed on some computing cluster?