5 ms·
> The difference between “access without authorization” and “exceed[ing] authorized access.” I never really understood why this distinction is considered so co
by AnthonyMouse 11y ago
> The difference between “access without authorization” and “exceed[ing] authorized access.”
I never really understood why this distinction is considered so contentious. The law itself defines "exceeds authorized access" as "to access a computer with authorization and to use such access to obtain or alter information in the computer that the accesser is not entitled so to obtain or alter."
Which isn't that helpful because it's approximately what you would expect it to mean if it wasn't defined. But I'm not sure where the ambiguity between "access without authorization" and "exceeds authorized access" is supposed to come in. The plain meaning is clearly that in one case you have no legitimate access (i.e. the system doesn't allow anonymous access and you have no account) and in the other case you have some legitimate access but not to do what you did. Kerr makes the argument that it is possible for "access without authorization" to imply "exceeds authorized access", but that is one place where the statutory definition is useful: The definition of "exceeds authorized access" first requires you "to access a computer with authorization..."
The real trouble with the CFAA is that it doesn't make the scope of authorization clear in either event. The canonical way people know whether they're authorized to do something to a computer system is that it allows them to do it. If you aren't authorized then it comes back with "access is denied" and you can't do it.
So the only way to break the law is to get the computer to do something it isn't supposed to let you. But where is the definition of that? How are you supposed to know what the computer is supposed to do, if the normal way of knowing that is to look at what it actually does, and the only cases that matter are the ones where that doesn't apply? There may be some obvious cases (e.g. logging in with someone else's account), but by what rule or principle are these cases supposed to be distinguished from others?
- adanto6840 11y agoThis really hits the nail on the head. Does sending an HTTP request and receiving an HTTP 200 response code indicate that you're authorized? IMO, yes. What if you send 5 million requests over the course of X timeframe, each of which also receives an HTTP 200 response. Have you "exceeded authorized access" at that point? IMO you haven't -- but right now the answer is basically, "maybe, TBD on a case-by-case basis" which is a decidedly poor answer. And, as you eloquently summarized, that's precisely the problem with this part of the CFAA. Unfortunately I don't think that the outcome of this specific case will do much to clarify this aspect of CFAA. That said, the final opinion could possibly shine a bit of light in the right direction if we're lucky, but I'm not holding my breath for a black-and-white interpretation any time soon...
- cookiecaper 11y ago>Does sending an HTTP request and receiving an HTTP 200 response code indicate that you're authorized? IMO, yes. It'd be a boon for anyone who depends on data scraping (which includes basically everyone in some form or another) if the Court decided that issue in that way, but I agree that it seems really unlikely that that will be broached in this case. Meanwhile, big companies will be able to scrape to their hearts' content while small companies are forcibly shut down for the same when one such big company retains a big law firm to intimidate an upstart they're afraid of with the threat of a lawsuit that costs millions just to litigate, excluding any potential damages. If the small company were to attempt to use the same law(s) to defend themselves, they'd be guaranteeing themselves said millions in legal fees, as the big company will aggressively defend itself, so it's not really an option. License to bully for the rich, bupkis for everyone else. That's how legal gray areas work.
- snowwrestler 11y agoThat would certainly make life easier on technologists--just read the HTTP code and if it's 200 you're good. But the legal concept of authorization doesn't rest on technology, and there's actually a good reason for that: if the technology was that reliable, then you wouldn't need the law at all. That is, if servers always returned a 401, 403, 511, etc. status code for unauthorized access attempts, then it would be impossible for unauthorized access to occur. And we don't tend to bother with laws restricting impossibilities. But we all know the reality: people are imperfect and therefore so is their technology. A server could return a 200 status on a page that you are not authorized to see.
- AnthonyMouse 11y ago> But we all know the reality: people are imperfect and therefore so is their technology. A server could return a 200 status on a page that you are not authorized to see. That's the whole problem. 99+% of the time the status code is correct and the server will give you the data if it's supposed to but not if it isn't. It's generally quite reliable and people are expected to and do rely upon it -- that's what it's for. But 100% of the cases that end up in court will be in the <1% where the server made a mistake. If we had a better way to deterministically evaluate authorization than what the server is using then we could use it to improve the server and therefore ought to. So we have a choice: Either it is impossible to deterministically evaluate whether access to a particular server is authorized, or we defer to the server's determination (and improve the server where possible). In the latter case this law is obviously then redundant and silly, but that doesn't make it the wrong choice.
- kej 11y ago> The canonical way people know whether they're authorized to do something to a computer system is that it allows them to do it. If you aren't authorized then it comes back with "access is denied" and you can't do it. I think you're going astray by looking at this from a computer perspective instead of from a human perspective. The canonical way to know if you're authorized to access x is if the owner of x has given you permission to access it. The fact that the "Employees Only" door isn't locked doesn't imply that everyone is authorized to go back there, and the "Authorized Vehicles Only" driveway doesn't need a roadblock to be meaningful. There are certainly ambiguities and room for this statute to be overused, but there are also situations that are clearly on one side or the other.
- AnthonyMouse 11y ago> The canonical way to know if you're authorized to access x is if the owner of x has given you permission to access it. But that doesn't work here. You will never meet the "owner" of google.com. You send it packets and it sends some back. There is no other relationship or interaction from which to infer authorization. > The fact that the "Employees Only" door isn't locked doesn't imply that everyone is authorized to go back there, and the "Authorized Vehicles Only" driveway doesn't need a roadblock to be meaningful. It is not the common case that a private party can put a sign on a wall saying whatever they want and have anyone charged with a crime for not following it. Moreover, there is no sign and there is no door or driveway. The Internet is not a physical place. The analogy would be if you sent mail to someone's assistant requesting some information and the assistant sent back a letter saying you aren't allowed to have this but here it is anyway. At best you're getting mixed signals and the fault is on the party sending them rather than the party receiving them. > There are certainly ambiguities and room for this statute to be overused, but there are also situations that are clearly on one side or the other. If the best you can say about a law is that unambiguous cases exist, that is not saying very much, and it does nothing for all the cases where clarity is unfindable.
- kej 11y ago>But that doesn't work here. You will never meet the "owner" of google.com. You send it packets and it sends some back. There is no other relationship or interaction from which to infer authorization. And yet even without a phone call from Larry Page, we all know that we're authorized to point our web browsers at google.com, and not authorized to ssh into Google's internal networks. Just like we'd all agree that we're allowed to drive on the public roads up to the entrance to Google's campus but not allowed to force our way into the server room. >If the best you can say about a law is that unambiguous cases exist, that is not saying very much, and it does nothing for all the cases where clarity is unfindable. When there are grey areas, the legal system will look at the whole situation and consider what a reasonable person would conclude. Maybe you were sending billions of packets because you misconfigured your network testing code and it was an honest mistake with no harm done, or maybe you sent billions of packets because you wanted to trigger a vulnerability in the server to discover some secret information. The legal system can take that into account. Overly specific laws that don't leave room for that kind of judgement lead to trivial ways to skirt the law.
- rayiner 11y ago> The canonical way people know whether they're authorized to do something to a computer system is that it allows them to do it. I am not convinced you have to throw common sense out the window just because you're dealing with a computer. The computer doesn't need to reject your attempts to, e.g., see other peoples' account information for you to know that's outside the scope of authorized use. The presumption when you're dealing with any other sort of property is that you have license to do what you can reasonably believe the property owner is giving you permission to do. You can browse around a shop, but the property owner doesn't need to lock the back room for you to know you don't have authorization to go there. What you're saying is that for the specific case of computers, that presumption should be flipped: you have authorization to do whatever you're not actively prevented from doing.
- AnthonyMouse 11y ago> I am not convinced you have to throw common sense out the window just because you're dealing with a computer. The computer doesn't need to reject your attempts to, e.g., see other peoples' account information for you to know that's outside the scope of authorized use. It seems you're advocating for the "I know it when I see it" test. If this common sense is so common then we can write down what it is and make sure everybody is on the same page, right? > What you're saying is that for the specific case of computers, that presumption should be flipped: you have authorization to do whatever you're not actively prevented from doing. The practice is flipped. In practice you are actively prevented from doing anything you aren't authorized to do, absent some flaw in the server. Why shouldn't the rule follow the practice?
- rayiner 11y ago> The practice is flipped. In practice you are actively prevented from doing anything you aren't authorized to do, absent some flaw in the server. Why shouldn't the rule follow the practice? Because it puts the burden on the property owner to secure their systems, instead of on the user to not behave in anti-social ways. In an ideal world, you wouldn't need computer security--people should just do what they're supposed to do.
- tptacek 11y agoThe canonical way people know whether they're authorized to do something to a computer system is that it allows them to do it. This sounds reasonable at first but falls apart quickly. The nature of virtually every software security exploit is code that convinces a computer to allow you to do something that you shouldn't be able to do. You are a URL bar away from many thousands of SQL injection flaws, right this very instant. By your logic, any exploitation of those bugs should be lawful.
- AnthonyMouse 11y ago> You are a URL bar away from many thousands of SQL injection flaws, right this very instant. By your logic, any exploitation of those bugs should be lawful. It is a common mistake to think that some terrible law is required because otherwise some terrible act would be legal. Pretty much everything seriously bad (and also most neutral or good things) is illegal six ways from Sunday. And if you can think of some actually terrible specific thing which is not against any other law than the CFAA, certainly let's pass a law against that thing in particular.
- tedunangst 11y agoWhen in doubt, the answer is no. At least until you ask. Is it really that hard to send an email, "Hey, do you mind if I pound your site with my vuln scanner?", before doing so?