\documentclass{article}

\usepackage{../style}

\usepackage{../graphics}

\begin{document}

\fnote{\copyright{} Christian Urban, King's College London, 2014, 2015}

%https://www.theguardian.com/technology/2016/oct/04/yahoo-secret-email-program-nsa-fbi

%https://nakedsecurity.sophos.com/2015/11/12/california-collects-owns-and-sells-infants-dna-samples/

%http://randomwalker.info/teaching/fall-2012-privacy-technologies/?

%https://josephhall.org/papers/NYU-MCC-1303-S2012_privacy_syllabus.pdf

%http://www.jetlaw.org/wp-content/uploads/2014/06/Bambauer_Final.pdf

%http://www.cs.cmu.edu/~yuxiangw/docs/Differential%20Privacy.pdf

%https://www.youtube.com/watch?v=Gx13lgEudtU

%https://fpf.org/wp-content/uploads/Differential-Privacy-as-a-Response-to-the-Reidentification-Threat-Klinefelter-and-Chin.pdf

%http://research.neustar.biz/2014/09/08/differential-privacy-the-basics/

%=====

%Tim Greene, Network World, 17 Dec 2015   (via ACM TechNews, 18 Dec 2015)

%

%Massachusetts Institute of Technology (MIT) researchers' experimental

%Vuvuzela messaging system offers more privacy than The Onion Router (Tor) by

%rendering text messages sent through it untraceable.  MIT Ph.D. student

%David Lazar says Vuvuzela resists traffic analysis attacks, while Tor

%cannot.  The researchers say the system functions no matter how many parties

%are using it to communicate, and it employs encryption and a set of servers

%to conceal whether or not parties are participating in text-based dialogues.

%"Vuvuzela prevents an adversary from learning which pairs of users are

%communicating, as long as just one out of [the] servers is not compromised,

%even for users who continue to use Vuvuzela for years," they note.  Vuvuzela

%can support millions of users hosted on commodity servers deployed by a

%single group of users.  Instead of anonymizing users, Vuvuzela prevents

%outside observers from differentiating between people sending messages,

%receiving messages, or neither, according to Lazar.  The system imposes

%noise on the client-server traffic which cannot be distinguished from actual

%messages, and all communications are triple-wrapped in encryption by three

%servers.  "Vuvuzela guarantees privacy as long as one of the servers is

%uncompromised, so using more servers increases security at the cost of

%increased message latency," Lazar notes.

%http://orange.hosting.lsoft.com/trk/click?ref=znwrbbrs9_5-e70bx2d991x066779&

%%%%

%% canvas tracking

%%https://freedom-to-tinker.com/blog/englehardt/the-princeton-web-census-a-1-million-site-measurement-and-analysis-of-web-privacy/

%%%

%% cupit re-identification attack

%% https://nakedsecurity.sophos.com/2016/05/20/published-personal-data-on-70000-okcupid-users-taken-down-after-dmca-order/?utm_source=feedburner&utm_medium=feed&utm_campaign=Feed%3A+nakedsecurity+%28Naked+Security+-+Sophos%29

%Differential privacy

%=====================

%https://www.wired.com/2016/06/apples-differential-privacy-collecting-data/

%Differential privacy, translated from Apple-speak, is the

%statistical science of trying to learn as much as possible

%about a group while learning as little as possible about any

%individual in it.

%As Roth notes when he refers to a “mathematical proof,”

%differential privacy doesn’t merely try to obfuscate or

%“anonymize” users’ data. That anonymization approach, he

%argues, tends to fail. In 2007, for instance, Netflix released

%a large collection of its viewers’ film ratings as part of a

%competition to optimize its recommendations, removing people’s

%names and other identifying details and publishing only their

%Netflix ratings. But researchers soon cross-referenced the

%Netflix data with public review data on IMDB to match up

%similar patterns of recommendations between the sites and add

%names back into Netflix’s supposedly anonymous database.

%As an example of that last method, Microsoft’s Dwork points to

%the technique in which a survey asks if the respondent has

%ever, say, broken a law. But first, the survey asks them to

%flip a coin. If the result is tails, they should answer

%honestly. If the result is heads, they’re instructed to flip

%the coin again and then answer “yes” for heads or “no” for

%tails. The resulting random noise can be subtracted from the

%results with a bit of algebra, and every respondent is

%protected from punishment if they admitted to lawbreaking.

%https://www.cis.upenn.edu/~aaroth/Papers/privacybook.pdf

% Windows 10 data send back to Microsoft (Cortana)

%Here’s a non-exhaustive list of data sent back: location data, text

%input, voice input, touch input, webpages you visit, and telemetry

%data regarding your general usage of your computer, including which

%programs you run and for how long.

% Businesses are already using customised pricing online based on

% information they can glean about you. It is hard to know how

% widespread the practice is; companies keep their pricing strategies

% closely guarded and are wary of the bad PR price discrimination

% could pose. However, it is clear that a number of large retailers

% are experimenting with it. Staples, for example, has offered

% discounted prices based on whether rival stores are within 20 miles

% of its customers’ location. Office Depot has admitted to using its

% customers’ browsing history and location to vary its range of offers

% and products. A 2014 study from Northeastern University found

% evidence of “steering” or differential pricing at four out of 10

% general merchandise websites and five out of five travel

% websites. (Steering is when a company doesn’t give you a customised

% price, but points you towards more expensive options if it thinks

% you will pay more.) The online travel company Orbitz raised

% headlines in 2012 when it emerged that the firm was pointing Mac

% users towards higher-priced hotel rooms than PC users.

%%% government will overwrite your wishes if it is annoymous

%% https://www.lightbluetouchpaper.org/2016/12/05/government-u-turn-on-health-privacy/

%% corporate surveilance / privacy - report and CC3C talk

%%      http://crackedlabs.org/en/networksofcontrol

%%      https://media.ccc.de/v/33c3-8414-corporate_surveillance_digital_tracking_big_data_privacy#video&t=2933

\section*{Handout 6 (Privacy)}

The first motor car was invented around 1886. For ten years,

until 1896, the law in the UK (and elsewhere) required a

person to walk in front of any moving car waving a red flag.

Cars were such a novelty that most people did not know what to

make of them. The person with the red flag was intended to

warn the public, for example horse owners, about the impending

novelty---a car. In my humble opinion, we are at the same

stage of development with privacy. Nobody really knows what it

is about or what it is good for. All seems very hazy. There

are a few laws (e.g.~cookie law, right-to-be-forgotten law)

which address problems with privacy, but even if they are well

intentioned, they either back-fire or are already obsolete

because of newer technologies. The result is that the world of

``privacy'' looks a little bit like the old Wild

West---lawless and mythical.

We would have hoped that after Snowden, Western governments

would be a bit more sensitive and enlightened about the topic

of privacy, but this is far from the truth. Ross Anderson

wrote the following in his blog\footnote{\url{https://www.lightbluetouchpaper.org/2016/02/11/report-on-the-ip-bill/}} about the approach taken in

the US to lessons learned from the Snowden leaks and contrasts

this with the new snooping bill that is considered in the UK

parliament: 

\begin{quote}\it 

``The comparison with the USA is stark. There, all three

branches of government realised they'd gone too far after

Snowden. President Obama set up the NSA review group, and

implemented most of its recommendations by executive order;

the judiciary made changes to the procedures of the FISA

Court; and Congress failed to renew the data retention

provisions in the Patriot Act (aided by the judiciary). Yet

here in Britain the response is just to take Henry VIII powers

to legalise all the illegal things that GCHQ had been up to,

and hope that the European courts won't strike the law down

yet again.''

\end{quote}

\noindent Unfortunately, also big organisations besides

governments seem to take an unenlightened approach to privacy.

For example, UCAS, a charity set up to help students with

applying to universities in the UK, has a commercial unit that

happily sells your email addresses to anybody who forks out

enough money for bombarding you with spam. Yes, you can opt

out very often from such ``schemes'', but in case of UCAS any

opt-out will limit also legit emails you might actually be

interested in.\footnote{The main objectionable point, in my

opinion, is that the \emph{charity} everybody has to use for

HE applications has actually very honourable goals

(e.g.~assist applicants in gaining access to universities),

but the small print (or better the link ``About us'') reveals

they set up their organisation so that they can also

shamelessly sell the email addresses they ``harvest''.

Everything is of course very legal\ldots{}ethical?\ldots{}well

that is in the eye of the beholder. See:

\url{http://www.ucas.com/about-us/inside-ucas/advertising-opportunities} 

or

\url{http://www.theguardian.com/uk-news/2014/mar/12/ucas-sells-marketing-access-student-data-advertisers}}

Another example: Verizon, an ISP who is supposed to provide

you just with connectivity, has found a ``nice'' side-business

too: When you have enabled all privacy guards in your browser

(the few you have at your disposal), Verizon happily adds a

kind of cookie to your

HTTP-requests.\footnote{\url{http://webpolicy.org/2014/10/24/how-verizons-advertising-header-works/}}

As shown in the picture below, this cookie will be sent to

every web-site you visit. The web-sites then can forward the

cookie to advertisers who in turn pay Verizon to tell them

everything they want to know about the person who just made

this request, that is you.

\begin{center}

\includegraphics[scale=0.16]{../pics/verizon.png}

\end{center}

\noindent How disgusting! Even worse, Verizon is not known for

being the cheapest ISP on the planet (completely the

contrary), and also not known for providing the fastest

possible speeds, but rather for being among the few ISPs in

the US with a quasi-monopolistic ``market distribution''.

Well, we could go on and on\ldots{}and that has not even

started us yet with all the naughty things NSA \& Friends are

up to. Why does privacy actually matter? Nobody, I think, has

a conclusive answer to this question yet. Maybe the following

four notions help with clarifying the overall picture

somewhat: 

\begin{itemize}

\item \textbf{Secrecy} is the mechanism used to limit the

      number of principals with access to information (e.g.,

      cryptography or access controls). For example I better

      keep my password secret, otherwise people from the wrong

      side of the law might impersonate me.

\item \textbf{Confidentiality} is the obligation to protect

      the secrets of other people or organisations (secrecy

      for the benefit of an organisation). For example as a

      staff member at King's I have access to data, even

      private data, I am allowed to use in my work but not

      allowed to disclose to anyone else.

\item \textbf{Anonymity} is the ability to leave no evidence of

      an activity (e.g., sharing a secret). This is not equal

        with privacy---anonymity is required in many 

        circumstances, for example for whistle-blowers, 

        voting, exam marking and so on.

\item \textbf{Privacy} is the ability or right to protect your

      personal secrets (secrecy for the benefit of an

      individual). For example, in a job interview, I might

      not like to disclose that I am pregnant, if I were a

      woman, or that I am a father. Lest they might not hire

      me. Similarly, I might not like to disclose my location

      data, because thieves might break into my house if they

      know I am away at work. Privacy is essentially

      everything which ``shouldn't be anybody's business''.

\end{itemize}

\noindent While this might provide us with some rough

definitions, the problem with privacy is that it is an

extremely fine line what should stay private and what should

not. For example, since I am working in academia, I am every

so often very happy to be a digital exhibitionist: I am very

happy to disclose all `trivia' related to my work on my

personal web-page. This is a kind of bragging that is normal

in academia (at least in the field of CS), even expected if

you look for a job. I am even happy that Google maintains a

profile about all my academic papers and their citations. 

On the other hand I would be very irritated if anybody I do

not know had a too close look on my private live---it

shouldn't be anybody's business. The reason is that knowledge

about my private life can often be used against me. As mentioned

above, public location data might mean I get robbed. If

supermarkets build a profile of my shopping habits, they will

use it to \emph{their} advantage---surely not to \emph{my}

advantage. Also whatever might be collected about my life will

always be an incomplete, or even misleading, picture. For

example I am pretty sure my creditworthiness score was

temporarily(?) destroyed by not having a regular income in

this country (before coming to King's I worked in Munich for

five years). To correct such incomplete or flawed credit

history data there is, since recently, a law that allows you

to check what information is held about you for determining

your creditworthiness. But this concerns only a very small

part of the data that is held about me/you. Also

what about cases where data is wrong or outdated (but do we

need a right-to be forgotten).

To see how private matter can lead really to the wrong

conclusions, take the example of Stephen Hawking: When he was

diagnosed with his disease, he was given a life expectancy of

two years. If employers would know about such problems, would

they have employed Hawking? Now, he is enjoying his 70+

birthday. Clearly personal medical data needs to stay private.

To cut a long story short, I let you ponder about the two

statements which are often voiced in discussions about privacy:

\begin{itemize}

\item \textit{``You have zero privacy anyway. Get over 

it.''}\\

\mbox{}\hfill{}{\small{}(by Scott Mcnealy, former CEO of Sun)}

\item \textit{``If you have nothing to hide, you have nothing 

to fear.''}

\end{itemize}

\noindent If you like to watch a movie which has this topic as

its main focus I recommend \emph{Gattaca} from

1997.\footnote{\url{http://www.imdb.com/title/tt0119177/}} If

you want to read up on this topic, I can recommend the

following article that appeared in 2011 in the Chronicle of

Higher Education:

\begin{center} 

\url{http://chronicle.com/article/Why-Privacy-Matters-Even-if/127461/} 

\end{center} 

\noindent Funnily, or maybe not so funnily, the author of this

article carefully tries to construct an argument that does not

only attack the nothing-to-hide statement in cases where

governments \& co collect people's deepest secrets, or

pictures of people's naked bodies, but an argument that

applies also in cases where governments ``only'' collect data

relevant to, say, preventing terrorism. The fun is of course

that in 2011 we could just not imagine that respected

governments would do such infantile things as intercepting

people's nude photos. Well, since Snowden we know some people

at the NSA did exactly that and then shared such photos among

colleagues as ``fringe benefit''.  

\subsubsection*{Re-Identification Attacks} 

Apart from philosophical musings, there are fortunately also

some real technical problems with privacy. The problem I want

to focus on in this handout is how to safely disclose datasets

containing potentially very private data, say health records.

What can go wrong with such disclosures can be illustrated

with four well-known examples:

\begin{itemize}

\item In 2006, a then young company called Netflix offered a 1

      Mio \$ prize to anybody who could improve their movie

      rating algorithm. For this they disclosed a dataset

      containing 10\% of all Netflix users at the time

      (appr.~500K). They removed names, but included numerical

      ratings of movies as well as times when ratings were

      uploaded. Though some information was perturbed (i.e.,

      slightly modified).

      Two researchers had a closer look at this anonymised

      data and compared it with public data available from the

      International Movie Database (IMDb). They found that

      98\% of the entries could be re-identified in the

      Netflix dataset: either by their ratings or by the dates

      the ratings were uploaded. The result was a class-action

      suit against Netflix, which was only recently resolved

      involving a lot of money.

\item In the 1990ies, medical datasets were often made public

      for research purposes. This was done in anonymised form

      with names removed, but birth dates, gender and ZIP-code

      were retained. In one case where such data about

      hospital visits of state employees in Massachusetts was

      made public, the then governor assured the public that

      the released dataset protected patient privacy by

      deleting identifiers. 

      A graduate student could not resist cross-referencing

      public voter data with the released data that still

      included birth dates, gender and ZIP-code. The result

      was that she could send the governor his own hospital

      record. It turns out that birth dates, gender and

      ZIP-code uniquely identify 87\% of people in the US.

      This work resulted in a number of laws prescribing which

      private data cannot be released in such datasets.

\item In 2006, AOL published 20 million Web search queries

      collected from 650,000 users (names had been deleted).

      This was again done for research purposes. However,

      within days an old lady, Thelma Arnold, from Lilburn,

      Georgia, (11,596 inhabitants) was identified as user

      No.~4417749 in this dataset. It turned out that search

      engine queries are deep windows into people's private

      lives. 

\item Genome-Wide Association Studies (GWAS) was a public

      database of gene-frequency studies linked to diseases.

      It would essentially record that people who have a

      disease, say diabetes, have also certain genes. In order

      to maintain privacy, the dataset would only include

      aggregate information. In case of DNA data this

      aggregation was achieved by mixing the DNA of many

      individuals (having a disease) into a single solution.

      Then this mixture was sequenced and included in the

      dataset. The idea was that the aggregate information

      would still be helpful to researchers, but would protect

      the DNA data of individuals. 

      In 2007 a forensic computer scientist showed that

      individuals can still be identified. For this he used

      the DNA data from a comparison group (people from the

      general public) and ``subtracted'' this data from the

      published data. He was left with data that included all

      ``special'' DNA-markers of the individuals present in

      the original mixture. He essentially deleted the

      ``background noise'' in the published data. The problem

      with DNA data is that it is of such a high resolution

      that even if the mixture contained maybe 100

      individuals, you can with current technology detect

      whether an individual was included in the mixture or

      not.

      This result changed completely how DNA data is nowadays

      published for research purposes. After the success of 

      the human-genome project with a very open culture of

      exchanging data, it became much more difficult to 

      anonymise data so that patient's privacy is preserved.

      The public GWAS database was taken offline in 2008.

\end{itemize}

\noindent There are many lessons that can be learned from

these examples. One is that when making datasets public in

anonymised form, you want to achieve \emph{forward privacy}.

This means, no matter what other data that is also available

or will be released later, the data in the original dataset

does not compromise an individual's privacy. This principle

was violated by the availability of ``outside data'' in the

Netflix and governor of Massachusetts cases. The additional

data permitted a re-identification of individuals in the

dataset. In case of GWAS a new technique of re-identification

compromised the privacy of people in the dataset. The case of

the AOL dataset shows clearly how incomplete such data can be:

Although the queries uniquely identified the older lady, she

also looked up diseases that her friends had, which had

nothing to do with her. Any rational analysis of her query

data must therefore have concluded, the lady is on her

death bed, while she was actually very much alive and kicking.

In 2016, Yahoo released the so far largest machine learning

dataset to the research community. It includes approximately

13.5 TByte of data representing around 100 Billion events from

anonymized user-news items, collected by recording

interactions of about 20M users from February 2015 to May

2015. Yahoo's gracious goal is to promote independent research

in the fields of large-scale machine learning and recommender

systems. It remains to be seen whether this data will really

only be used for that purpose.

\subsubsection*{Differential Privacy}

Differential privacy is one of the few methods that tries to

achieve forward privacy. The basic idea is to add appropriate

noise, or errors, to any query of the dataset. The intention

is to make the result of a query insensitive to individual

entries in the database. That means the results are

approximately the same no matter if a particular individual is

in the dataset or not. The hope is that the added error does

not eliminate the ``signal'' one is looking for in the

dataset.

%\begin{center}

%User\;\;\;\;    

%\begin{tabular}{c}

%tell me $f(x)$ $\Rightarrow$\\

%$\Leftarrow$ $f(x) + \text{noise}$

%\end{tabular}

%\;\;\;\;\begin{tabular}{@{}c}

%Database\\

%$x_1, \ldots, x_n$

%\end{tabular}

%\end{center}

%

%\begin{center}

%\begin{tabular}{l|l}

%Staff & Salary\\\hline

%$PM$ & \pounds{107}\\

%$PF$ & \pounds{102}\\

%$LM_1$ & \pounds{101}\\

%$LF_2$ & \pounds{97}\\

%$LM_3$ & \pounds{100}\\

%$LM_4$ & \pounds{99}\\

%$LF_5$ & \pounds{98}

%\end{tabular}

%\end{center}

%

%

%\begin{center}

%\begin{tikzpicture} 

%\begin{axis}[symbolic y coords={salary},

%             ytick=data,

%             height=3cm]
%\addplot+[jump mark mid] coordinates
%{(0,salary)   (0.1,salary) 
% (0.4,salary) (0.5,salary)  
% (0.8,salary) (0.9,salary)};
%\end{axis}
%\end{tikzpicture}

%\end{center}

%

%\begin{tikzpicture}[outline/.style={draw=#1,fill=#1!20}]
%  \node [outline=red]            {red box};
%  \node [outline=blue] at (0,-1) {blue box};
%\end{tikzpicture}

\ldots

\subsubsection*{Further Reading}

Two cool articles about how somebody obtained via the Freedom

of Information Law the taxicab dataset of New York and someone

else showed how easy it is to mine for private information: 

\begin{center}\small

\begin{tabular}{p{0.78\textwidth}}

\url{http://chriswhong.com/open-data/foil_nyc_taxi/}\smallskip\\

\url{http://research.neustar.biz/2014/09/15/riding-with-the-stars-passenger-privacy-in-the-nyc-taxicab-dataset}