Category: Writing

Categories
Links Writing

The Significance of a ‘Three Hop’ Analysis

Washington’s Blog has an excellent, if somewhat long, post that outlines the significance of the NSA’s ‘three hop’ analysis. It collects and provides some numbers behind basic communications network analyses, and comes to the conclusion that upwards to 2.5 million Americans could be “caught up in dragnet for each suspected terrorist, means that a mere 140 potential terrorists could lead to spying on all Americans. There are tens of thousands of Americans listed as suspected terrorists … including just about anyone who protests anything that the government or big banks do.”

Go read the full post. Some of the numbers are a bit speculative, but on the whole it does a good job showing why ‘three hop’ analyses are so problematic: such analyses disproportionately collect data on American citizens the basis of the most limited forms of suspicion. Such surveillance should be set aside because it constitutes an inappropriate infringement on individuals’ and communities’ reasonable expectations of privacy; it runs counter to how a well ordered and properly functioning democracy should operate in theory and in practice.

Categories
Links Writing

Facebook’s ‘Other’ Folder

David Pogue’s recent post on Facebook’s ‘Other’ folder notes how the company is effectively hiding a significant number of legitimate messages from its users in an attempt to prevent spam and ‘unimportant’ messages from disturbing subscribers. What follows are a few examples of legitimate messages that subscribers missed because they were placed in this folder:

  • “Notification of the death of a friend was hidden in my Other box. I had been very hurt at not being told, and actually missed her funeral.”
  • “I just checked my ‘Other’ folder and found out that I won a free high-end kitchen faucet for a contest I entered last year. Rats.”
  • “Just looked at my ‘Other’ messages and found one about a job opening — in 2011. Think it’s been filled?”
  • “Whoa! There’s tons of important messages in here. Former students of mine were trying to reach out to me. I can’t believe Facebook doesn’t notify you in any way about these.”
  • “Unbelievable! My husband’s wallet was lost and presumed stolen — someone had found it a year ago and sent us a Facebook message, which was hidden until now! Thanks so much.”
  • “Just checked and found a message from someone telling me that they found my lost wallet…a year ago. They really need to redo some thinking on that ‘other’ folder.”

The intent of Facebook’s filtering is noble, insofar as it’s meant to cut down on the cruft and spam that people inevitably get in their email inboxes on a daily basis. I’m sure that the logic is as follows: if we can get people to like using Facebook messages more than email, then we can convince people to rely on our corporate system and wean people off of their traditional email services. Unfortunately, it looks like Facebook’s filtering system suffers from flaws, just as their competitors’ systems do. Worse, and unlike most of their competitors, Facebook subscribers can’t access this folder from their tablets or smartphones without visiting Facebook via the web interface. So, for people that predominantly engage with Facebook using the company’s mobile applications, this folder is effectively invisible. Messages simply vanish into a black hole. This is a very bad thing.

While Facebook’s system makes sense, I suspect that a great many people are as ignorant of the ‘Other’ folder’s existence as the people who wrote to Pogue. This information asymmetry between the developers and users suggests a problem in the UX or UI, insofar as it shouldn’t be a shock that this folder exists. Good UI and UX will prevent subscribers from getting ‘shocked’ about the existence of hidden messages, and will help ensure that the service remains ‘sticky’ for its user base.

Network effects can stymie subscriber churn but they can’t stop it entirely. If Facebook undermines professional or personal networks because of how it handles suspected ‘unimportant’ messages, then the network effect that Facebook currently enjoys could be weakened and expose a part of Facebook’s flank to companies that are more attuned to people’s communicative interests and desires. It will be curious to see how/whether Facebook incorporates the information that arose from Pogue’s columns, and if they actually modify users’ interfaces such that the ‘Other’ folder is more prominently displayed. At the very least, something should change in the mobile applications so users can at least theoretically access all of those ‘unimportant’ messages.

Categories
Aside Links

AT&T’s Anti-Infringement Patent

AT&T’s recent patent to detect and act on network-based copyright infringement raises significant red flags for network neutrality advocates. However, we need to look beyond the most obvious (and nefarious!) red flags: when examining corporate surveillance prospects we need to reflect on the full range of reasons behind the practice. Only in taking this broader, and often more nuanced, view are we likely to come closer to the truth of what is actually going on, and why. And, if we don’t get closer to the specific truth of the situation, at least we can better understand the battleground and likely terms of the conflict.

Categories
Quotations

2013.7.19

Mark Zuckerberg runs a giant spy machine in Palo Alto, California. He wasn’t the first to build one, but his was the best, and every day hundreds of thousands of peopl eupload the most intimate details of their lives to the Internet. The real coup wasn’t hoodwinking the public into revealing their thoughts, closest associates, and exact geographic coordinates at any given time. Rather, it was getting the public to volunteer that information. Then he turned off the privacy settings.

…

If the state had organized such an informationd rive, protestors would have burned down the White House. But the state is the natural beneficary of this new “social norm.” Today, that information is regularly used in court proceedings and law enforcement. There is no need for warrants or subpoenas. Judges need not be consulted. Th Forth Amendment does not come into play. Intelligence agencies don’t have to worry about violating laws protecting the citizenry from wiretapping and information gathering. Sharing information “more openly” and with “more people” is a step backward in civil liberties. And spies, whether foreign and domestic, are “more people,” too.

Marc Ambinder and D.B. Grady. (2013). Deep State: Inside the Government Secrecy Industry. New Jersey: Wiley. Pp. 27.
Categories
Aside Links

Backdooring an ‘Encrypted’ Application

Persuant to my last post on cryptography and pixie dust, it’s helpful to read through Matt Green’s highly accessible article “How to ‘backdoor’ an encryption app.” You’ll find that companies have a host of ways of enabling third-party surveillance, ranging from overt deception to having access to communications metadata to compromising their product’s security if required by authorities. In effect, there are lots of ways that data custodians can undermine their promises to consumers, and it’s pretty rare that the public ever learns that the method(s) used to secure their communications have either been broken or are generally ineffective.

Categories
Writing

Pixie Dust and Data Encryption

CNet recently revealed that Google is encrypting some of their subscribers’ Google Drive data. Data has always been secured in transit, but Google is testing encrypting data at rest. This means that, without the private key, someone who got access to your data on Google’s Drive servers would just get reams of ciphertext. At issue, however, is that ‘encryption’ is only a significant barrier if the the third-party storing your data cannot decrypt the data when a government-backed actor comes knocking.

Encryption has become something like pixie dust, insofar as companies far and wide assure their end-users and subscribers that data is armoured in cryptographic shells. Don’t worry! You’re safe with us! Unfortunately, detailed audits of commercial encrypted products often reveal firms offering more snake oil than genuine protection. Just consider some of the following studies and reports that are, generally, damning[1]:

As noted in Bruce Schneier’s (still) excellent analysis of cryptographic snake oil, there are at least nine warning signs that the company you’re dealing with isn’t providing a working cryptographic solution:

  1. You come across a lot of “pseudo-mathematical goobledygook” that isn’t linked to referenced and reviewed third-party reviews of the cryptographic underpinnings.
  2. The company states that ‘new mathematics’ are used to secure your information.
  3. The cryptographic process is proprietary and neither you nor anyone else can examine how data is secured.
  4. Weird claims are made about the nature of the product, such that the claims or terms used could easily fit within the latest episode of a sci-fi show you’re watching.
  5. Excessive key lengths are trumpted as a demonstrated proof of cryptographic security.
  6. The company claims your data is secure because one-time pads are used.
  7. Claims are made that cannot be backed up in fact.
  8. Security proofs involve twists of linguistic logic, and lack demonstrations of mathematical logic.
  9. The product is somehow secure because it hasn’t been ‘cracked’. (Yet.)

Unfortunately, people have been conditioned by Hollywood and other media that as soon as something is ‘encrypted’ only super-duper hackers can subsequently ‘penetrate the codes and extract the meta-details to derive a data-intuition of the content’ (or some such similiar garbage). When you’re dealing with crappy ‘encryption’ – like showing private keys in plain text, or transmitting passphrases across the Internet in the clear – then the product is just providing consumers a false sense of security. You don’t need to be a hacker to ‘defeat’ particularly poor implementations of data encryption, you often just need to know how to read a file system.

Presently, however, there aren’t clear ways for consumers to know if a product is genuinely capable of securing their data in transit or at rest. There isn’t a clear solution to getting bad products off the market or generally improving product security, save for media shaming and/or the development of better cryptographic libraries that non-cryptographers (read: developers) can easily use when developing product. However, there are always going to be flaws and errors, and most consumers are never going to know that something has gone terribly awry until it’s far, far too late. So, despite there being a well-known problem, there isn’t a productive solution. And that has to change.

  1. The selection of studies were just chosen because they’re sitting on my computer now/I’ve referenced or written about them previously. If you spend a few minutes trawling Google Scholar using the search term ‘encryption broken’ you’re going to come across even more analyses of encryption ‘solutions’ that have been defeated.  ↩
Categories
Links

Constraints

Every product is constrained. Choose to optimise your compromises, whether you’re designing or just buying.

Matt has written one of the most succinct and clear pieces on product constraints. It’s well worth the time to read and subsequently mull over.

Categories
Writing

A Brief Comment on ‘Metadata’

We live in environments that are pervasively penetrated by digital systems. We carry personalized tracking devices with us everywhere (i.e. mobile phones) that have increasingly sophisticated sensors embedded in them. We rely on Internet-based systems for travel, work, and play. Even our ‘landline’ communications are pervasively turned into digital code when we call a friend or family member.

Every one of the previously mentioned transactions generates ‘non-content’ data: when and who we call, and for how long; which cellular towers we pass by; what (semi-)unique IP addresses are provided to websites we visit, and so forth. These identifiers can be used to trace our movements, practices, and who we communicate with: they are often far more revealing about ourselves than the pure content of our communications.

It’s with the reality of the surveillance potentials of metadata that we need to reorient how to talk about such ‘non-content’ data. It has become depressingly common to see elected officials and other authorities state that “it’s just metadata” as well as “we only use it for appropriate purposes.”

To the first statement, metadata can reveal incredibly sensitive infomation about individuals and about their community/communities. The collection and processing of such information therefore warrants a similar degree of care and concern as the processing of clearly personal information.

To the second statement, clarity around collection and use of metadata is needed. Moreover, data cannot be massively collected and ‘appropriate purposes’ just applied to how the data is subsequently parsed. The very collection of data itself needs to be targeted, justified, and enjoy significant oversight – arguably more oversight that ‘just’ the content of communications.

In a recent paper on metadata, Ontario Information and Privacy Commissioner Ann Cavoukian wrote:

we urge governments to adopt a proactive approach to securing the rights affected by intrusive surveillance programs. To protect privacy and liberty, any power to seize communications metadata must come with strong safeguards directly embedded into programs and technologies, that are clearly expressed in the governing legal framework. The purpose, scope, and duration of data collection must be strictly controlled. More robust judicial oversight, parliamentary or congressional controls, and systems capable of providing for effective public accountability should be brought to bear. The need for operational secrecy must not stand in the way of public accountability. Our essential need for privacy and the preservation of our freedoms are at stake.[1]

Commissioner Cavoukian is decidely correct that data collection, use, and intent must be carefully controlled. However, I would go a step further than the Commissioner has in her call for additional parliamentary oversight and control. In Canada, and unlike the United States and United Kingdom, there is not a committee of parliamentarians with security clearances to oversee how our intelligence and security authorities operate. Presently, the Canadian system predominantly enjoys only Cabinet-level political oversight: we need a broader set of eyes, and eyes that are not mindful of the ruling government’s optics, to evaluate the appropriateness of what our intelligence and security services are up to. So, in excess of Commissioner Cavoukian’s comments, we actually need to modify parliament such that oversight is even possible.

Reasonable people can disagree on the value and desire for national security and foreign intelligence services. Such disagreements should happen more prominently amongst parliamentarians and the public. However, there should be no disagreement that, in order to represent the public, at least some members of our legislative assemblies must know the extent of the government’s security and intelligence powers, capabilities, and practices.

Canada is a democracy and, as such, it is imperative that we establish a committee of parliamentarians to oversee how our security and spy agencies are collecting, using, and retaining the metadata and content associated with our communications. The actions that these agencies engage in are too significant to leave to Cabinet oversight alone.

  1. Ann Cavoukian. (2013). “A Primer on Metadata: Separating Fact from Fiction.” Office of the Information and Privacy Commissioner of Ontario. Available at: http://www.privacybydesign.ca/content/uploads/2013/07/Metadata.pdf. Pp. 10. Emphasis added.  ↩
Categories
Humour Links Writing

Definitions for the American Surveillance State

David Sirota of Salon has developed an excellent set of terms to speed along discussions about the contemporary American surveillance state. My own favorites include:

Least untruthful: A new legal doctrine that allows an executive branch official to issue a deliberate, calculated lie to Congress yet avoid prosecution for perjury, as long as the official is protecting the executive branch’s political interests. Usage example: Director of National Intelligence James Clapper avoided prosecution for perjury because he insisted that the blatant lie he told to Congress was merely the “least untruthful” statement he could have made.

And:

Modest encroachment: A massive, indiscriminate intrusion. Usage example: President Obama has deemed the NSA’s “collect it all” surveillance operation, which has captured 20 trillion information transactions and touches virtually all aspects of American life, a “modest encroachment” on citizens’ right to privacy.

The full listing of terms is depressingly cynical. However, the persistent – if often humorous – turn to cynicism may ultimately limit how politicians address and respond to Snowden’s surveillance revelations. What Snowden confirmed raises existential challenges to the potential to imagine, let alone actualize, a deliberative democratic state. The accompanying risk is that instead of addressing such challenges head on, citizens may retreat to cynicism rather than engaging in the hard work of recuperating their increasingly-authoritarian democratic institutions. We’re at a point where we need a more active, not more withdrawn and bemused, citizen response to government excesses.

Categories
Aside Links Quotations

How to Publish A  Story That Explains How to Use Social Media to Juice Your Story’s Popularity

emptyage:

I paid to have my latest Wired story promoted on social networks, like Twitter and Facebook, to try to show that a lot of the metrics* we use to measure a story’s success are bullshit. It worked. When the story went live today, the page appeared with more than 15,500 links on Twitter, and 6,500 likes on Facebook. The story is a part of Wired’s Cheats package for the latest issue of the magazine. It needed to go live online at the same time readers encountered it in print, and it needed to have all those social shares set up in advance. 

…

The entire package was going live at once. I could publish my story a little bit early, but the timing needed to be very close. I wanted all the public-facing stats (like the 15 thousand links and Twitter and 6,000 Facebook shares) to be live by the time the text appeared. Certainly, if someone found it in print or on the tablet, it needed those metrics to already be there. To make that happen, we cheated. 

…

This morning (or last night) at a little after 1 am, I added the story text, set it to the current time, and hit update. Now it showed up in RSS readers and I could openly tweet it form my main account. (I had originally used a secondary Twitter account I have for testing 3rd party stuff to link to it and score retweets.)

So now, the story goes “live” and as if by magic it has tens of thousands of social shares listed on it the instant real people start to encounter it. It worked. 

*As is site traffic, to a very large extent. My original idea was to use a botnet to throw traffic at it, but Wired’s lawyers said “no, no. Don’t do that.“ 

And, of course, people tend to associate lots of shares with an article’s significance or influence. Consequently, by ‘cheating’ ahead of time a content owner can add a false gravitas to the content in question. I’m curious to know how search companies that, in part, use social signals to surface content deal with this kind of ‘hacking the social.’