Improves techlimits section (WIP #5, #8)

stats.bib

@@ -27,3 +27,17 @@ @report{FreedomInTheWorld2017
   institution = {FreedomHouse},
   URL = {https://freedomhouse.org/report/freedom-world/freedom-world-2017},
 }
+
+@report{BotsAndAutomationDuringUSElection,
+  author = {Bence Kollanyi
+    and Howard, Philip N.
+    and Woolley, Samuel C.},
+  title = {Bots and Automation over Twitter during the U.S.\ Election},
+  type = {Data Memo},
+  number = {2016.4},
+  year = {2016},
+  location = {Oxford, UK},
+  publisher = {Project on Computational Propaganda},
+  institution = {Univ.\ of Oxford},
+  URL = {https://www.politicalbots.org},
+}

techlimits.tex

@@ -3,18 +3,43 @@ \section{Technical limitations}
 We can only do so much with technology, and, of course, there are some 
 limitations to what technical solutions can achieve.
 In particular, we face two problems.
-The first one, the double agent problem, is caused by humans ability to deceive 
-each other, and consequently cannot easily be solved by technology.
-The second one, the Sybil attack, is a consequence of the unlimited ability of creating 
-arbitrary identities in online systems such as social networks.
-
+The first one is the double agent problem.
+It is caused by humans ability to deceive each other, and consequently cannot 
+easily be solved by technology.
 \label{DoubleAgentProblem}
-The double agent problem is the problem of one of the regime's agents 
-infiltrating the opposition by acting as if part of the opposition.
+In out context it is the problem of one of the regime's agents infiltrating the 
+opposition by acting as if part of the opposition.
 We cannot solve this problem, however, we might be able to reduce the damage.
 One design principle for privacy is data minimization, this strategy will help 
 Alice reduce the information that the double agent, and anyone else, can learn.
 
+The second problem is the Sybil attack.
+This problem is a consequence of the unlimited ability of creating arbitrary 
+identities in online systems such as social networks.
 \label{SybilAttacks}
-\input{sybil.tex}
+The Sybil attack is somewhat related to the double agent problem, but is only 
+a problem in electronic systems.
+The problem occurs when there is nothing that limits the creation of new 
+identities, thus the adversary can create multiple unlinkable identities.
+Usually the attack is aimed at reputation systems, where the adversary can use 
+its many identities to vouch for each other to falsely gain in reputation.
+In general, the problem can be summarized as that a rather small number of 
+people in the network control a large part of the identities in the network in 
+order to gain a disproportionate amount of influence~\cite{SybilAttack}.
+A concrete example is Trump's use of Twitter bots in the US 
+election~\cite{BotsAndAutomationDuringUSElection}.
 
+\Citet{SybilAttack} proved that this problem cannot be solved without 
+a \emph{logically} central control of the creation of identities.
+This means that we can not handle identity creation by letting the people who 
+are already in the network vouch for other identities, i.e.\ to build a network 
+of trust.
+The Sybil attack itself aims at compromising a considerable portion of the 
+identities, and thus, the more identities the attacker gains, the more new 
+identities it can create and vouch for.
+To prevent this kind of behaviour we rather need something like the national 
+identification systems present in most countries, where the state has ensured 
+a one-to-one correspondence between identities and physical persons.
+Fortunately, there are techniques that can mitigate the effects of the Sybil 
+attack without forcing us to use such a centralized identity system.
+We will return to these where relevant in the text.