How Apple Could Lose By Winning: The DOJ's Next Move Could Be Worse
Since the conflict over smartphone security, long simmering between Apple and the FBI, burst into the headlines last month, many of us who advocate for strong encryption have watched the competing legal arguments advanced by the parties with a certain queasiness. Many of the arguments on Apple’s side?whether offered by the company itself or the myriad groups who have weighed in with friend-of-the-court briefs?have turned critically on the government’s unprecedented invocation of the hoary All Writs Act to compel the company to write and authenticate a novel piece of software effectively dragooning Apple engineers into government service.
But there has always been an obvious alternative?a way to achieve the FBI’s aim of circumventing iPhone security features without requiring any Apple employees to write a line of new code: the Lavabit Option.
That is, instead of asking Apple to create a hacking tool that would permit the FBI to attempt to brute-force a phone’s passcode without triggering escalating delays between guesses or deletion of encrypted data, they could simply demand that Apple turn over the source code and documentation the FBI would need to develop its own custom version of the iOS boot ROM, sans security features. Then, they require Apple to either cryptographically sign that code or provide the government with access to its developer credentials, so that the FBiOS can run on an iPhone.
That hypothetical possibility is raised explicitly by the Justice Department in a footnote to its most recent motion in its ongoing litigation with Apple, which explains that the FBI had not gone that route because it “believed such a request would be less palatable to Apple.” Having tried it the easy way, the FBI suggests it’s happy to do things the hard way: “If Apple would prefer that course, however, that may provide an alternative that requires less labor by Apple programmers.”
Apple’s latest reply brief clearly registered the company’s dismayed response to this legal shot across the bow:
The catastrophic security implications of that threat only highlight the government’s misunderstanding or reckless disregard of the technology at issue and the security risks implicated by its suggestion.
Such a move would signal a race to the bottom of the slippery slope that has haunted privacy advocates: A world where companies can be forced to sign code developed by the government to facilitate surveillance. In this case, that means software to brute force a passcode, but could as easily apply to remote exploits targeting any networked device that relies on developer credentials to authenticate trusted updates. Which is to say, nearly any modern networked device. It entails, quite literally, handing the government the keys to the kingdom.
What’s particularly worrying is that, while this approach is massively more troubling from a security perspective than funneling such requests through the company itself on a case-by-case basis, it would likely stand on a less shaky legal foundation.
Apple’s arguments throughout this case have stressed the unprecedented nature of the FBI’s attempt to conscript the firm’s engineers, noting that the All-Writs Act invoked by the government was meant to enable only the particular types of orders familiar from common law, not grant an all-purpose power to “order private parties to do virtually anything the Justice Department and FBI can dream up.” The trouble is, an order to turn over information in the “possession custody or control” of a private party is just such a traditional order. Such demands are routinely made, for instance, via a subpoena duces tecum requiring a person or company to produce documents.
It’s likely that Apple’s developer keys are stored in a Hardware Security Module that would make it difficult or impossible to produce a copy of their firmware signing key directly to the government. But that might not be much legal help. In a separate iPhone unlocking case in New York, magistrate judge James Ornstein recently rejected the government’s argument that a previous All-Writs Act case, New York Telephone Co., required Apple’s compliance. In that case, Ornstein noted, the government’s
agents would normally have been able to install the authorized pen register without the company’s assistance but for the fact that the subject telephone’s wires were so placed as to prevent the agents from gaining surreptitious access. The agents thus needed the telephone company not to provide technical expertise they lacked, but only to step out of the way and let them perform their authorized surveillance on company property.
But that sounds much closer to what would be involved in a case where Apple is required to authenticate government-written code: Just “step out of the way” and let the FBI access the HSM containing the keys used to sign updates.
Similarly, many of the First Amendment arguments raised by Apple and the Electronic Frontier Foundation?to the effect that “code is speech” and the requirement that Apple create new software amounts to “compelled speech”?would also fall by the wayside. They might still advance such arguments with respect to the “endorsement” implicit in using company credentials to sign software, but a court may not find that as intuitive as the idea that “compelled speech” is involved in requiring engineers to devise wholly novel and potentially complicated software.
Many of Apple’s other arguments, of course, would remain untouched: There’s the idea that Congress has established a comprehensive statutory framework specifying the means of law enforcement access to digital content via laws like the Communications Assistance for Law Enforcement Act and the Electronic Communications Privacy Act, making the All-Writs Act an inappropriate mechanism to seek authority withheld by Congress. Nor would a “sign our code” approach affect any of Apple’s claims about the broader security harms inherent in the creation of developer-authenticated tools to break security. But the long list of legal barriers to the FBI getting its way would surely be significantly reduced.
That means it’s not just important that Apple win in this case?it matters how it wins. If the company emerges victorious on grounds fundamentally tied to the mandate to create software rather than the demand to authenticate it, it could prove a pyrrhic victory indeed, opening the door for the government to insist on doing things the “hard way,” and inaugurating an era of government scripted malware signed to look like genuine updates.
I'll echo those who are disturbed by the tenor of some of the comments here. It's often extremely difficult to prove either rape or sexual assault beyond a reasonable doubt, as would be required in a criminal trial. In the absence of third-party witnesses or unambiguous physical evidence, many genuine victims will find it simply impossible to meet that standard of proof. (Remember, someone can be acquitted if even if a "preponderance of the evidence" suggests they are guilty.) Isn't that bad enough without demanding that victims in those many cases be silent about their experiences unless they can meet the evidentiary standard appropriate to a criminal prosecution?
In this case in particular, applying the criminal standard seems outrageous, since Alkon can reasonably understood to be complaining that the government has decided to permit and immunize conduct that should (and in other circumstances would) be classified as rape or sexual assault. To permit a defamation suit under those circumstances is to say, in effect, that the government may bar people from naming an abuse of power, simply by immunizing itself from prosecution for that abuse.
The anonymous commenter at #2 makes a reasonable point?that there are increasingly many ways to consume paid video content other than "buying DVDs"?but I suspect that this is a point that cuts against the industry's attempt to inflate the loss estimates, which so often do assume downloads are substituting for disc sales (as opposed to, say, a Netflix streaming subscription that runs $8/month however many movies you watch).
But of course, many of those downloaders are actually already paying for Netflix or cable, but looking for the pirated version of a movie that isn't yet available as part of their paid service. I suspect the net loss to the film industry from *these* downloaders is in many cases pretty close to zero, because their alternative to downloading would not be making an additional purchase, but rather watching one of the thousands of other movies that are already part of their paid plan. (Or borrowing a DVD from a friend. Or watching something free-with-ads on Hulu or Snagfilms or YouTube.) A realistic cost estimate require a realistic assessment of the likely substitutes for the pirated product.
The author....
...is Will Wilkinson. And I do think it's the absence of bylines that creates the impression of "inconsistency." Anywhere else, you'd just see that one writer expressed one view in the magazine, and another writer for their blog expressed a different view, which would seem quite normal.
Interpretive Charity
Much as I tend to agree this is a bad ruling, this framing is seems unfair to the judge. He may not be anyone's idea of a techie, but it's not like he somehow just fails to comprehend that WiFi technology uses radio signals. He's questioning whether it falls within the scope of what *Congress* meant by the term in 1986, when the distinction was supposed to ensure that nobody would be liable for ordinary operation of consumer gear that might pick up a neighbor's cordless phone. Googling "Rebecca Black greatest living singer" or "Barack Obama muslim kenya birth certifictate" could surely be described as an "unreasonable search," but that's not what the phrase means in the context of the Fourth Amendment.
A good heuristic...
...for evaluating claims like Pascazi's,even if you don't have any expertise in foreign trademark laws, is to ask: "If there were any validity to his little theory, why aren't actual trademark lawyers *already* doing this whenever some promising startup fails to immediately register a trademark in every first-to-file country where it might eventually find a market?" One possibility is that Pascazi is vastly more clever than all the hundred of actual specialists in IP law. The other is that his avarice outran his comprehension. Take your pick as to which seems more plausible.
The fact checking system at the New Yorker is pretty much the gold standard in journalism; I'd be astonished if this went anywhere.
easy vs possible
Eh, I don't know if this is a "privacy issue" in the sense of some serious failure on their part. But a lot of privacy has more to do with ease of information sharing than whether it's possible *at all*, and more to do with the clarity of norms than explicit prohibitions. Someone COULD copy the contents of a private e-mail (or, by hand, the contents of a private letter) and forward it to hundreds of friends. But that would be both effortful and rude. If I share a photo with my "Friends" circle, I realize they COULD save and reupload it if there's not sharing functionality built in... but they'd have to be big jerks (and ergo probably not "Friends") to make the effort to do so if I've signaled via my settings that I don't expect these pictures to be more widely circulated. It's not a question of the service "violating my privacy," which is the unhelpful frame of most of these stories, but facilitating social signalling about the information norms we expect friends, peers, and colleagues to respect.
Hmm
Gotta say, after looking at the album art, I'm inclined to side with Spacey on this one... It does actually look apt to confuse buyers. ("Oh look, Kevin Spacey made an album! I think it's called Cassettes Won't Listen?") http://www.cassetteswontlisten.com/preorder/pics/EVINSPACEY.jpg
Weird
I've been a professional writer & blogger for about a decade now, have been pretty routinely quoted or linked by other folks throughout that decade, and have never -- not a single time in ten years -- had any of the hundreds of people who quoted me ask permission to reproduce anything short of the entirety of a long article. Gannon's notion of what is "common" is utterly bizarre; I have never heard any of the many writers I know describe anything remotely similar. Like Nina Paley, it would be an absolute nuisance if I had to read (let alone respond to) an email every time somebody excerpted something I'd written. For someone like Andrew Sullivan or Cory Doctorow, I have to assume it would be totally crippling.
Is that even true?
Presumably the overseas distributors are in breach of their contracts with the publisher if they're shipping to people in the U.S., but as I understand the First Sale doctrine, a domestic BUYER who purchases a UK copy, whether to read or resell it, ought to be in the clear. I'm missing the part in the story where anybody actually, you know, makes a copy.
Re: This is paranoia
I'm not sure which part is supposed to be paranoid. We know full well that law enforcement officers seek location data from cell phones during investigations. The inference from the redactions has to do with the fairly narrow question of what level of precision they can achieve using a 2703(d) order (plus, perhaps, something like Triggerfish) as opposed to a rule 41 warrant. Actually, we know the DoJ's position is that they believe they're entitled to full GPS data with a 2703(d), but that they prefer to ask for a warrant lest a court disagree. And as I realized after the initial post, we already know that cell site/sector data is used to get a rough fix for more precise triggerfish tracking. The trivial inference that the same data might be used to get a better fix *during* the call for the same purpose strikes me as quite conservative.
Ryan hits on one important reason this seems unhelpful: Often you want a site like Yelp to help you choose between two or three nearby places, not to suggest something from an unconstrained universe of options. But even in that case, you might just ask *a friend*, whose likes and dislikes you probably know to be reasonably similar to yours, for a recommendation. But if you don't know much about how well the preferences of the endorsers match up with your own, then the ratio of positive to negative reactions among patrons is probably a hell of a lot more instructive than a raw number of recommendations.
Ding ding ding
CeeVee nails it. I'm willing to bit that the demographics of any instant messaging service's user base don't look a whole lot like the general population in a host of ways. Ditto social networking services like FB. But there's an even more obvious form of sample bias at work here: Most people sign up for IM services because they know that lots of their friends and acquaintances use IM. So of course IM users will be fairly tightly connected--that's how they got there. The real takeaway here should be that if even in this pool that's basically self-selected for high social connectivity, the average chain length is 6.6, we can infer that the average for the general population is likely to be significantly higher.
The "expansion" alluded to is not the telecom immunity provision. All the various FISA reform bills permit the surveillance of communications between U.S. parties and persons overseas without court order, at the discretion of the Attorney General and the Director of National Intelligence. That is unambiguously, whether one approves of it or not, an expansion of the foreign intel surveillance powers granted by FISA.
Was it really that confusing? Ok, how about: "requiring that intelligence agencies obtain a FISA warrant in order to intercept purely foreign-to-foreign communications, when such interception is done at facilities within the United States"
(1) Radio is such a stunningly inapt analogy for cable deployment that I'm a little stunned more than one person brought it up. Maybe consolidation in radio is bad, and an interesting conversation could be had about reasons and remedies, but the economics and the issues involved are so wildly different that I'm pretty dubious about how naturally any conclusions there cross apply.
(2) People raise a bunch of issues that go back to the monopoly power of cable providers, which just seems to ignore the central point of the post. Monopoly power in cable is local. A national cap does exactly nothing to reduce local monopoly. At best, it changes the identity of the local monopolist. Possibly it even exacerbates the problem by eliminating incentives for providers to compete through expansion.
Just to clarify, what I'd meant to suggest is that the "hedge" function is already, at least to some extent, one labels serve--and perhaps one that will become less relevant in the future.