This is the third and last post about Privacy in a Networked World. See the first post (Snowden and Schneier) here and the second post (John DeLong and John Wilbanks) here.

“The Mete and Measure of Privacy,” Cynthia Dwork, Microsoft Research

This was probably the presentation I was least able to follow well, so I’ll do my best to recap in English-major language; please feel free to suggest corrections in the comments. Dwork talked about the importance of being able to draw conclusions about a whole population from a representative data set while maintaining the confidentiality of the individuals in the data set. “Differential privacy” means the outcome of data analysis is equally likely independent of whether any individual does or doesn’t join the data set; this “equally likely” can be measured/represented by epsilon, with a smaller value being better (i.e. less privacy loss). An epsilon registry could then be created to help us better understand cumulative privacy loss.

Dwork also talked about targeted advertising. Companies who say “Your privacy is very important to us” have “no idea what they’re talking about” – they don’t know how to (and may have little interest in) keeping your data private. And when you hear “Don’t worry, we just want to advertise to you,” remember – your advertiser is not your friend. Advertisers want to create demand where none exists for their own benefit, not for yours. If an advertiser can pinpoint your mood, they may want to manipulate it (keeping you sad, or cheering you up when you are sad for a good reason). During this presentation, someone posted a link on Twitter to this article from The Atlantic, “The Internet’s Original Sin,” which is well worth a read.

Dwork quoted Latanya Sweeney, who asked, “Computer science got us into this mess. Can computer science get us out of it?” Dwork’s differential privacy is one attempt to simplify and solve the problem of privacy loss. Slides from a different but similar presentation are available through SlideServe.

“Protecting Privacy in an Uncertain World,” Betsy Masiello, Senior Manager, Global Public Policy, Google

Masiello’s talk focused on what Google is doing to protect users’ privacy. “It’s hard to imagine that Larry and Sergey had any idea what they were building,” she began. Today, “Everything is mobile…everything is signed in.” Services like Google Now send you a flow of relevant information, from calendar reminders to traffic to weather. In addition to Google, “the average user has 100 accounts online.” It’s impossible to remember that many passwords, especially if they’re good passwords; and even if they’re good passwords, your accounts still aren’t really safe (see Mat Honan’s  2012 article for Wired, “Kill the Password: Why a String of Characters Can’t Protect Us Anymore“).

To increase security, Google offers two-factor authentication. (You can find out what other sites offer 2FA by checking out twofactorauth.org. Dropbox, Skype, many – but not all – banks, Facebook, LinkedIn, Tumblr, and Twitter all support 2FA.) Masiello said that after news of hacks, they see more people sign up for 2FA. “It’s an awareness problem,” she said. In addition to 2FA, Google is encrypting its services, including Gmail (note that the URLs start with https). “E-mail is still the most common way people send private information,” she said, and as such deserves protection.

“Encryption is the 21st century way of protecting our personal information,” said Masiello. Governments have protested companies who have started using encryption, but “governments have all the tools they need to obtain information legally.” As Cory Doctorow has pointed out many times before, it’s impossible to build a back door that only the “good guys” can walk through. Masiello said, “Governments getting information to protect us doesn’t require mass surveillance or undermining security designed to keep us safe.” The PRISM revelations “sparked a very important debate about privacy and security online.” Masiello believes that we can protect civil liberties and national security, without back doors or mass surveillance.

“Getting security right takes expertise and commitment,” Masiello said. She mentioned the paper “Breaking the Web” by Anupam Chander and Uyen P. Le, and said that we already have a good set of guidelines: the OECD Privacy Principles, which include collection limitation, data quality, purpose specification, use limitation, security safeguards, openness, individual participation, and accountability. As for Google, Masiello said, “We don’t sell user data; we don’t share with third parties.” All of the advertising revenue is based on user searches, and it’s possible to opt out of interest-based ads. (Those creepy right-sidebar ads that used to show up in Gmail, having mined your e-mail to produce relevant ads, appear to be gone. And good riddance.)

Finally, Masiello talking about developing/setting industry standards for privacy and security that would facilitate innovation and competition. But privacy isn’t even the main concern in the future: it’s identity – what it means, and how we construct it.

“Sur-veillance, Sous-veillance and Co-veillance,” Lee Rainie, Director of Internet, Science and Technology Research, Pew Research Center

Lee Rainie definitely walked away with the “Most Neutral, Fact-Based Presentation” Award. Rainie has spoken at library conferences in New England before, but I – perhaps unwisely – chose to go to other sessions, so this was the first time I saw him speak, and he was great. Furthermore, all of his slides are available on SlideShare. He started off with a few findings:

1. Privacy is not binary / context matters

2. Personal control / agency matters

3. Trade-offs are part of the bargain

4. The young are more focused on network privacy than their elders (this is only surprising if you haven’t read danah boyd’s excellent It’s Complicated: The Social Lives of Networked Teens, and in fact Rainie quoted boyd a few slides later: “The new reality is that people are ‘public by default and private by effort.'”)

5. Many know that they don’t know what’s going on

6. People are growing hopeless and their trust is failing

The Pew Research Center has found that consumers have lost control over companies’ use and control of their data; they have adopted a transactional frame of mind (e.g. giving up control of personal data in exchange for the use of a platform or service). In general, trust in most institutions has gone down, with the exceptions of the military, firefighters, and librarians(!). But there is a pervasive sense of vulnerability, and users want anonymity – mostly from advertisers, hackers, and social connections, rather than the government (see slide below).

This slide supports the argument for privacy, especially against the “nothing to hide” argument: people desire – and deserve – privacy for many reasons, the least of which is to avoid the government or law enforcement. (Mostly, someone on Twitter pointed out, we want to avoid “that guy.”)

As for the future of privacy, people are feeling “hopeless.” Rainie remembered saying, in the early 2000s, “There’s going to be an Exxon-Valdez of data spills…” and there have been many since then, but little has been done to protect consumer privacy. “How do we convince people to have hope?” he asked.

Panel: “What Privacy Does Society Demand Now and How Much is New?” Danny Weitzner (moderator), Kobbi Nissim, Nick Sinai, Latanya Sweeney

Fortunately, the moderator and panelists have different initials. The questions and responses below are paraphrased from the notes I took during the panel session.

DW: What sort of privacy does society demand now? Is privacy different now?

NS: Access to your own data has always been a art of privacy; also the right to correct, erase, and transfer. Your data should be useable and portable.

KN: The ability to collect a lot of data all the time is new. There is a different balance of power (companies have too much).

LS: Privacy and security are just the beginning. Every American value is being changed by technology. Computer scientists aren’t trained to think of social science effects and the power of technology design.

DW: Cryptography and math are a foundation we can trust if implemented properly, as Snowden said this morning.

LS: I dislike choosing between two things. We need a cross-disciplinary approach, a blended approach.

NS: Any great company should constantly be trying to improve user experience. How does privacy/security get integrated into design?

KN: Aim for mathematical solutions/foundations. We need to re-architect economic incentives, regulations, how all the components work together.

DW: Where will the leadership and initiative come from? Government?

KN: Academia, research. We need to find ways to incentivize.

LS: Economic [incentives] or regulations are necessary for privacy by design. They’re all collapsing…every single one of them [Facebook, the IRS] is heading for a major disaster.

DW: People care about control of their data, yet the information environment is increasingly complicated.

LS: Society benefits from technology with certain protections.

KN: Regulations we have today were designed in a completely different era. We may be in compliance, and still we have damaged privacy severely.

LS mentioned HIPPA, NS mentioned the Consumer Bill of Rights, DW mentioned “Privacy on the Books and on the Ground.”

DW: Privacy practices and discussion are/is evolving in the U.S.

LS: A huge dose of transparency would go a long way. This is the new 1776. It’s a whole new world. Technology is redefining society. The Federal Trade Commission could become the Federal Technology Commission.

DW: Are you optimistic? Are we heading toward a positive sense of privacy?

NS: Yes, by nature I’m optimistic, but complexity and user experience (user accounts, passwords) frustrates me. Entrepreneurs do help change the world.

KN: The genie is out of the bottle. This forces us to rethink privacy. Nineteen-fifties privacy has changed and isn’t the privacy we have today, but that doesn’t mean that privacy is dead. Privacy is a sword and a shield.

DW: We’re at the beginning of a long cycle. It’s only been a year [and a half] since Snowden. What do we expect from our government and our companies? How powerful should government and private organizations be? Marketing/advertising issues are trivial compared to bigger issues.

LS: The cost of collecting data is almost zero, so organizations (public and private) collect it and then figure out how to use it later. They should be more selective about collection. If we can expose the harm, it will lead to change.

Question/comment from audience: A lot of people are not aware they’re giving away their privacy (when browsing the internet, etc.).

LS: We need transparency.

NS: We need regulation and consumer protection.

 

This is the second post about Privacy in a Networked World. The first post, about the conversation between Bruce Schneier and Edward Snowden, is here.

“Privacy in a Networked World,” John DeLong, Director of the Commercial Solutions Center, NSA

Other than the length and volume of applause, it’s difficult to measure an audience’s attitude toward a speaker. I’ll venture, though, that the audience of Privacy in a Networked World was generally pro-Snowden; the attitude toward John DeLong can perhaps be characterized as guarded open-mindedness laced with a healthy dose of skepticism.

DeLong’s talk was both forceful and defensive; he wanted to set the record straight about certain things, but he also knew that public opinion (in that room, at least) probably wasn’t in his favor. (He said repeatedly that he did not want to have an “Oxford-style debate,” though his talk wasn’t set up as a debate in the first place.) “Let’s not confuse the recipe with the cooking,” he said, in a somewhat belabored analogy where the NSA’s work was the cooking and the law was the recipe. (I cook a lot at home, and I’ll just say, I can usually tell when I’m looking at a bad recipe, and opt to adapt it or not make it at all.)

DeLong quoted at length from Geoffrey R. Stone’s “What I Told the NSA.” (Stone was a member of the President’s Review Group in fall 2013, after the Snowden revelations.) Stone’s conclusions were not altogether positive; he found that while the NSA “did its job,” many of its programs were “highly problematic and much in need of reform.” But it’s the Executive Branch, Congress, and FISA who authorized those programs and are responsible for reforming them. Stone added, “Of course, ‘I was only following orders’ is not always an excuse….To be clear, I am not saying that citizens should trust the NSA. They should not. Distrust is essential to effective democratic governance.”

DeLong said, “The idea that the NSA’s activities were unauthorized is wrong, wrong in a magnificent way.” He emphasized that the NSA is not a law enforcement agency, it’s an intelligence agency. He spoke in favor of people with different backgrounds and expertise – lawyers, engineers, mathematicians, privacy experts, etc. – coming together to work out solutions to problems, with respect for each others’ abilities. “Technology,” he said, “always comes back to how we as humans use it.” At present, “We do not have technologies that identify privacy risks….Privacy engineering could be one of the most important engineering feats of our time.”

DeLong talked about rebuilding the nation’s confidence in the NSA. “Confidence is the residue of promises kept,” he said. “More information does not necessarily equal more confidence.” (Someone on Twitter pointed out that much depends on the content of the information.) The talk was a good reminder not to villainize the entire NSA; part of DeLong’s forcefulness was undoubtedly on behalf of his co-workers and staff whom he felt were unfairly maligned. And technology that could identify privacy risks, built by people who have different perspectives and backgrounds, would be excellent. But do we need technology that identifies privacy risks, or do we need stronger oversight and better common sense? Mass surveillance erodes trust in government and hasn’t been terribly effective; what more do we need to know to put a stop to it?

“Privacy and Irony in Digital Health Data,” John Wilbanks, Chief Commons Officer, Sage Bionetworks

John Wilbanks gave a fast-paced, interesting talk about health data. The “irony” in the title of his talk soon became clear when he gave the example of Facebook’s mood manipulation experiment compared to a study of Parkinson’s disease. The sample size for Facebook was many times larger, with a constant flow of information from “participants,” as opposed to a much smaller sample population who filled out a survey and answered questions by phone. “What does our society value?” Wilbanks asked. This question can be answered by another question: “What do we surveil?”

Wilbanks showed a graph representing cardiovascular disease and terrorism: there is 1 death every 33 seconds from cardiovascular disease – “That’s like 9/11 every day” – and yet there’s not nearly the same kind of “surveillance” for health that there is for terrorism. Participating in a research study, Wilbanks said, is like “volunteering for surveillance,” and usually the mechanisms for tracking aren’t as comprehensive as, say, Facebook’s. Of course, privacy laws affect health research, and informed consent protects people by siloing their data; once the study is concluded, other researchers can’t use that data, and there’s no “network effect.”

Informed consent, while a good idea in theory, often leads to incomprehensible documents (much like Terms of Service). These documents are written by doctors, reviewed by lawyers, and edited by committee. Furthermore, said Wilbanks, people in health care don’t usually understand issues of power and data. So, he asked, how do we run studies at internet scale and make them recombinant? How do we scale privacy alongside the ability to do research? Wilbanks demonstrated some ideas to improve on traditional informed consent, which could also allow research subjects to get a copy of their own data and see which researchers are using data from the research in which they participated.

Obviously there are risks to individuals who share their personal health data, but there can be advantages too: more scientists having access to more data and doing more research can lead to more breakthroughs and improvements in the field of medicine.

Last year, Wilbanks talked about privacy and health data on NPR; you can listen to the segment here.

Still to come: Microsoft, Google, Pew, and a panel on “What Privacy Does Society Demand Now and How Much is New?” 

This is the first post about Privacy in a Networked World, the Fourth Annual Symposium on the Future of Computation in Science and Engineering, at Harvard on Friday, January 23.

A Conversation between Bruce Schneier and Edward Snowden (video chat)

Bruce Schneier is a fellow at the Berkman Center for Internet & Society, and the author of Data and Goliath. Edward Snowden was a sysadmin at the NSA who revealed the extent of the government’s mass surveillance. The conversation was recorded (no joke) and is available on YouTube.

I have to say it was an incredibly cool feeling when Snowden popped up on the giant screen and was there in the room with us. There was sustained applause when he first appeared and also at the end of the conversation, when he was waving goodbye. Schneier started by asking Snowden about cryptography: What can and can’t be done? Snowden replied, “Encryption…is one of the few things that we can rely on.” When implemented properly, “encryption does work.” Poor cryptography, either through bad implementation or a weak algorithm, means weak security. End points are also weak spots, even if the data in transit is protected; it’s easier for an attacker to get around crypto than to break it.

Snowden pointed out a shift in the NSA’s focus over the last ten years from defense to offense. He encouraged us to ask Why? Is this proper? Appropriate? Does it benefit or serve the public?

The explosion in “passive” mass surveillance (collecting everything in case it’s needed later) is partly because it’s easy, cheap, and simple. If more data is encrypted, it becomes harder to sweep up, and hackers (including the NSA) who use more “active” techniques run a higher risk of exposure. This “hunger for risk has greatly increased” during the War on Terror era. Their targets are “crazy, unjustified….If they were truly risk averse they wouldn’t be doing this…it’s unlawful.”

Snowden said that the NSA “is completely free from any meaningful judicial oversight…in this environment, a culture of impunity develops.” Schneier said there were two kinds of oversight: tactical oversight within the organization (“did we follow the rules?”) and oversight from outside of the organization (“are these the right rules?”). He asked, “What is moral in our society?”

Snowden asked if the potential intelligence that we gain was worth the potential cost. He stated that reducing trust in the American infrastructure is a costly move; the information sector is crucial to our economy. The decrease in trust, he said, has already cost us more than the NSA’s budget. “They are not representing our interests.”

Schneier, using his NSA voice, said, “Corporations are spying on the whole internet, let’s get ourselves a copy!” (This was much re-tweeted.) “Personal information,” he said, “is the currency by which we buy our internet.” (Remember, if you can’t tell what the product is, you’re the product.) It’s “always amusing,” he said, when Google complains about the government spying on their users, because “it’s our job to spy on our users!” However, Schneier thinks that the attitudes of tech companies and standards bodies are changing.

These silos of information were too rich and interesting for governments to ignore, said Snowden, and there was no cost to scooping up the data because until 2013, “people didn’t realize how badly they were being sold up the river.” Schneier said that research into privacy-preserving technologies might increase now that there is more interest. Can we build a more privacy-preserving network, with less metadata?

“We’ve seen that the arguments for mass surveillance” haven’t really held up; there is little evidence that it has stopped many terrorist attacks. Schneier cited an article from the January 26, 2015 edition of The New Yorker, “The Whole Haystack,” in which author Mattathias Schwartz lists several recent terrorist attacks, and concludes, “In each of these cases, the authorities were not wanting for data. What they failed to do was appreciate the significance of the data they already had.”

Unlike during the Cold War, now “we all use the same stuff”: we can’t attack their networks and defend our networks, because it’s all the same thing. Schneier said, “Every time we hoard a zero-day opportunity [knowing about a security flaw], we’re leaving ourselves vulnerable to attack.”

Snowden was a tough act to follow, especially for John DeLong, Director of the Commercial Solutions Center for the NSA, but that’s exactly who spoke next. Stay tuned.