The Other Side of the Coin:

User Experiences with Bitcoin Security and Privacy

Katharina Krombholz, Aljosha Judmayer, Matthias Gusenbauer, and Edgar Weippl

SBA Research, Vienna, Austria

(firstletterfirstname)(lastname)@sba-research.org

Abstract. We present the first large-scale survey to investigate how users ex-
perience the Bitcoin ecosystem in terms of security, privacy and anonymity. We
surveyed 990 Bitcoin users to determine Bitcoin management strategies and iden-
tified how users deploy security measures to protect their keys and bitcoins. We
found that about 46% of our participants use web-hosted solutions to manage at
least some of their bitcoins, and about half of them use exclusively such solu-
tions. We also found that many users do not use all security capabilities of their
selected Bitcoin management tool and have significant misconceptions on how to
remain anonymous and protect their privacy in the Bitcoin network. Also, 22% of
our participants have already lost money due to security breaches or self-induced
errors. To get a deeper understanding, we conducted qualitative interviews to ex-
plain some of the observed phenomena.

1 Introduction
With a current market capitalization of more than 3.5 billion USD, Bitcoin is the most
successful cryptographic currency at this time. Bitcoin is utilized for roughly 130.000
transactions per day [6] and has gained significant news coverage. With the success of
Bitcoin, several other cryptographic currencies were developed either based on Bitcoin
or from scratch.
Although the popularity of cryptographic currencies is increasing, they are not yet a
mass phenomenon. One of the reasons is that Bitcoin forces its users to deal with pub-
lic key cryptography. Furthermore, Bitcoin shifts the responsibilities for most security
measures to the end user compared to centralized monetary systems. Even though there
is a great variety of software available for managing bitcoins, user-experience is still
not obviating the need to deal with the technical fundamentals and to perform backups
to recover their virtual monetary assets in case of a loss. Hence, these systems are not
resilient to human errors. Reports from online forums and mailing-lists show that many
Bitcoin users already lost money due to poor usability of key management and secu-
rity breaches such as malicious exchanges and wallets. This motivates our research on
human interactions with the Bitcoin ecosystem.
Bitcoin users have a huge variety of tools available to manage their virtual assets.
These tools are commonly referred to as wallets. A wallet was originally defined as a
collection of private keys [8]. Hence, a piece of paper with a private key on it or even a
mental representation can be considered a wallet. However, most of these tools provide
functionality beyond storing keys, such as performing transactions. In contrary to other
public key crypto-systems, e.g. PGP/GPG, Bitcoin is not fully communication channel
agnostic. In case of Bitcoin the interaction with the Bitcoin network is an integral part to
operate in the distributed system. In contrast to other signing systems, Bitcoin tools need
to keep state information on performed transactions and account balances respectively.
As a first step to accommodate these misconceptions on Bitcoin wallets, we intro-
duce the term Coin Management Tool (CMT) as an extension to the current narrow
definition of a wallet. We define a CMT as a tool or a collection of tools which allows
users to manage one or more core tasks of cryptocurrencies. Throughout this paper we
are therefore referring to Bitcoin management, as it better describes user activities when
interacting with the Bitcoin ecosystem. Bitcoin security and privacy aspects have aleady
been studied in the research literature [7, 10, 14–16]. A first look on the usability of Bit-
coin key management has been presented in [8]. However, we are the first to conduct a
comprehensive user study to collect evidence on user experiences with Bitcoin security
and privacy.
In this paper, we present a comprehensive user study (n = 990) to cover human-
computer interaction aspects of the Bitcoin ecosystem. The goal was to understand how
users interact with Bitcoin and how they manage their virtual assets. We furthermore
studied experiences and perceptions related to security, privacy and anonymity in the
Bitcoin network. To collect user-reported data, we conducted a comprehensive online
survey with 990 participants and qualitative interviews with a subset of 10 participants.
Additionally, we extended the evaluation criteria from [8] and provide a method to
categorize CMTs depending on the level of control and verifiability a user can exercise
with the respective client.
We gathered interesting insights on how users interact with the Bitcoin network and
what privacy and security measures they deploy to protect their keys and coins. We
found that the first- and third-most used CMTs (Coinbase, Xapo) are web-hosted tools
where users shift security responsibilities to a third party. We also found that about a
third of their users are not aware whether their CMT data is encrypted or backed-up.
Among the participants who use a web-hosted solution, 50% indicated to use it exclu-
sively while the other half used additional local clients to manage their coins. Regard-
ing risk scenarios and their likelihood to occur, the second-highest risk was attributed
to vulnerabilities in web-hosted CMTs (after value fluctuation and followed by theft via
malware).
We also found that many users have misconceptions about how to remain anony-
mous. About 25% of our participants reported to use Bitcoin over Tor which has already
shown to be disadvantageous in certain cases [1, 3]. 22.5% of the participants reported
to have lost their bitcoins due to security breaches. About half of them consider this
loss as their own fault and the majority of them was not able to recover their bitcoins
and lost money permanently. Our work contributes research on user-centric concerns
of Bitcoin management, as according to Bonneau [7] Bitcoin is one of the cases where
practice is ahead of theory.
The main contributions of this paper are (1) a user study consisting of an online
survey and qualitative interviews, and (2) a method for categorizing Bitcoin CMTs.
2 Bitcoin Background
The Bitcoin currency is based on a distributed P2P system which synchronizes a public
ledger of all transactions among all Bitcoin clients. As a consequence, every full client
in the Bitcoin network is able to see the entire history containing all prior transactions.
Thereby it is possible to determine the current balance of every account. The account
information in Bitcoin basically consists of a hash over a public key which can be
compared to an account number, the so-called Bitcoin address. The protocol does not
require a link between account information and personal data. An individual can have
more than one account, hence Bitcoin provides a certain degree of pseudonymity [1, 9].
To transfer n bitcoins from account A, which is under control of Alice, to an-
other account B, which is under control of Bob, a new transaction is created by Alice.
Thereby, Alice creates a transaction message with the amount of bitcoins she wants to
send to Bob and includes the hash of the public key of Bobs account B as a destination
before signing it with her secret key skA . Alice publishes this transaction in the Bitcoin
network so that every participant knows that Alice now has n bitcoins less on her ac-
count A and Bob has received the difference on his account B. When this transaction
is successfully propagated in the network, Bob can create new transactions from his
account B to another account and spend the previously received bitcoins. This chaining
mechanism works fine for passing over arbitrary amounts of bitcoins from one account
to another, except in the special case of the first transaction in a chain, because this is
where new bitcoins come into existence [20].
Bitcoins are created during the so-called mining process. In this procedure every
miner collects transactions which have recently been propagated in the P2P network.
Then they try to successfully create a new block out of all unconfirmed transactions that
have not yet been included in a block of the block chain. A block essentially consists of a
collection of valid transactions1 , a nonce value, and a proof of work. The proof of work
is a partial pre-image attack on SHA-256 over the whole block as input. For the attack
to succeed, the hash has to be a value smaller than the current difficulty in the Bitcoin
network. In other words, the SHA-256 hash has to start with a certain number of zero
bits. The number of zero bits is referred to as difficulty. Since SHA-256 is categorized as
a cryptographic hash function [21], it is easy to verify a previously calculated SHA-256
sum of a block, but it is considered infeasible to generate a specific block that produces
a given hash value. To achieve this, the nonce field is constantly incremented to search
for a hash value that fulfils the described property. This brute-force process of searching
is called mining. If one client in the Bitcoin network finds such a combination of valid
transactions and nonce that yields a desired result, he/she publishes this new block in
the Bitcoin network and gets rewarded with newly created bitcoins.
The reward comes in form of a new transaction of (currently) 25 bitcoins that has no
predecessor and is included as a special so-called coinbase transaction by the creator
of the respective block. This coinbase transaction also includes the public key/bitcoin
address of the creator and marks the first transaction of a new chain of Bitcoin transac-
tions [4, 5, 17, 20].

1
More precisely a Merkle-Tree Hash over those transactions, for details see the specifications [4,
5, 17]
3 Related Work
We build upon already existing work by contributing the first user study with Bit-
coin users. Eskandari et al. [8] presented a first look at the key management of Bit-
coin by providing a set of evaluation criteria for Bitcoin wallets and a cognitive walk-
through [23] of selected wallets. The work by Eskandari et al. [8] can be considered a
first look at the usability of Bitcoin.
Moore et al. [19] conducted an empirical analysis of Bitcoin exchange risks. They
examined the track record of 40 Bitcoin exchanges and found that 18 had been closed,
with customer account balances often wiped out. They also found that popularity is a
strong indicator to predict the lifetime of an exchange, i.e. popular exchanges have a
longer lifespan.
Baur et al. [2] conducted exploratory interviews with individuals of distinct groups
and found that most stakeholders perceived the ease of use still as rather low. They also
found that the experienced usefulness varies according to the user group.
However, no empirical study has been performed to examine user perceptions of
Bitcoin security, privacy and anonymity. For a cryptographic currency like Bitcoin, pub-
lic key cryptography is required. Regarding the usability of key management and en-
cryption in the context of e-mail various studies have shown that there are numerous us-
ability issues regarding the successful usage of public key cryptography [11,12,22,24].
At this time, for neither domain a fully usable concept has been successful. Human as-
pects of key management have already been studied in other domains [11–13, 22, 24].
For the Bitcoin ecosystem however, secure key management alone is not sufficient, as
communication is not channel-independent but an integral part of the security concept.
4 User Study Methodology
The goal of this study is to empirically investigate end user perceptions and behavior in
the Bitcoin ecosystem with an emphasis on security practices as well as coin and key
management with the involved security risks. We designed an online questionnaire and
additionally conducted qualitative interviews. We derived specific research questions
from already existing literature on Bitcoin (as discussed in Section 3) as well as from
a qualitative content analysis of threads from online forums and mailing lists. Further-
more, we revised the available Bitcoin wallets2 and their capabilities and used them
as inspiration for our questions and the design of the security and privacy risk scenar-
ios. We focus on Bitcoin as it was by far the most popular cryptographic currency at
the time we conducted this study (July 2015). While the online survey was intended
to broadly measure self-reported Bitcoin management behavior and risk perception, the
interviews were conducted to get a deeper understanding on key usability issues, causes
of common security incidents and if and how they managed to recover their keys.
4.1 Research Questions
We sought answers to the following questions regarding users’ perceptions of Bitcoin
management and Bitcoin-associated security risks:

– Q1: What are the main usage scenarios of Bitcoin?

2
bitcoin.org
 – Q2: How do participants manage their Bitcoins? What are participants’ current
practices and how do they deal with security, privacy and anonymity?
– Q3: How do participants perceive Bitcoin-associated security risks?
– Q4: What security breaches have affected users and how did they recover their
Bitcoin keys and bitcoins?
– Q5: What are the main usability challenges that users have to deal with when using
Bitcoin?

We were also interested in categorizing CMTs in a way that users can quickly make an
informed decision based on the level of security, privacy and control they prefer? Our
categorization can be found in Appendix A.
5 Online Survey
We conducted our online survey over July 8-15, 2015. Our survey consisted of both
closed- and open-ended questions and covered the following topics: (1) Bitcoin us-
age and management, (2) CMT choice and usage, (3) security, privacy, anonymity and
backup behavior, (4) risk perception, and (5) demographics. The full set of questions is
presented in Appendix B. The open-ended questions were coded independently by two
researchers independently. After agreeing on a final set of codes, we coded all answer
segments for the final analysis. Coding refers to categorizing qualitative data to facili-
tate analysis [18] and is a common practice in human-computer interaction research.
5.1 Recruitment
We hosted our survey at soscisurvey.de3 . To restrict our participants to Bitcoin users
only, we deliberately designed our study to exclude all non-Bitcoin users. As it is diffi-
cult to construct such a restricted sample on platforms like Amazon Mechanical Turk,
we decided to use Bitcoin mailing lists and forums for recruiting. Furthermore, we
compensated participants in Bitcoin. The reward for a completed questionnaire was 4.2
mB(= 0.0042 B ≈ 1.22 USD at that time) After completing the survey, the participants
were instructed to enter a valid Bitcoin address to receive the payment. This ensured
that everyone who wanted to receive bitcoins as a reward is a Bitcoin user and hence
exactly our target audience. Even participants who had not used Bitcoin before had to
create a Bitcoin address to receive the compensation.
To motivate participants to spread the word and thus recruit further participants, we
displayed a link for re-distribution at the end of the survey. All participants that recruited
others received an additional 1 mB(≈ 0.29 USD). Table 1 shows that this additional
incentive scheme was successful since we received a high number of participants this
way. As Table 1 shows, the top 5 re-distributors of the link recruited about one quarter
of the overall sample. Initially we distributed the link to our survey over the following
channels: bitcointalk.org forum4 , bitcoin-list mailing list5 , twitter.com6 and an Austrian
3
https://www.soscisurvey.de/
4
https://bitcointalk.org/index.php?topic=1114149.0
5
http://sourceforge.net/p/bitcoin/mailman/bitcoin-list/
?viewmonth=201507
6
https://twitter.com/bit_use
bitcoin mailing list7 . We aimed for maximum transparency to avoid that our call for
participation would be misinterpreted as scam. Therefore, we proved on the initial page
of our survey that we indeed hold a respectable amount of bitcoins8 , by providing our
Bitcoin address9 together with a signature with the according private key (see appendix
C for the signature).
We recruited 1,265 participants over July 8-15, 2015 via these channels. The total
sample size after filtering out 275 participants due to incomplete or duplicated sub-
mission, or invalid entries, was 990. Of these, 85.2% claimed to be male (m), 10.5%
claimed to be female (f). 4.3% of our participants preferred not to provide their gender.
Ages ranged from 15 to 72 (median = 28.56). About half of our participants reported
to have an IT-related background. According to the collected IP addresses, most of our
participants filled out the survey in the US, followed by the UK and Germany. 7.6%
accessed the survey site over Tor (Figure 1). These numbers can of course be biased by
VPN usage.

United States
United Kingdom
Germany
Indonesia
Canada
Russian Federation
Tor
0

50

100

150

200
Fig. 1. Countries from which our participants accessed the survey site.

Table 1. Most refereed links.

Reference Occurrences Reward in BTC / EUR / USD

455975 91 0.0952 / 24.78 / 27.18
1295 58 0.0622 / 16.19 / 17.76
699324 51 0.0552 / 14.37 / 15.76
932181 28 0.0322 / 8.38 / 9.19
637623 21 0.0252 / 6.56 / 7.19

5.2 Validity of our Dataset

Since the survey was designed to be anonymous and we only required a valid Bitcoin
address, we had to take special care to avoid abuse. We semi-automatically verified
the authenticity of our dataset and were able to exclude 116 submissions we suspected
to be fraudulent, and 160 incomplete submissions. Nevertheless, there is still a chance
that we missed some manual double submissions. However, due to our deployed coun-
termeasures and the high quality of submitted data (e.g., the open-text questions) we
7
http://bitcoin-austria.at/
8
We purchased our 6.3965 BTC at https://coinfinity.co/
9
https://blockchain.info/address/12yeU5ymM67SL5UWVSwErAgwVwwaTd1Nma
suspect that the overall number is negligible. Among other, we deployed the following
countermeasure to make automation harder: reCAPTCHA: The last page of our survey
contained a text box to enter a Bitcoin address for receiving the compensation and a
Google reCAPTCHA. This together with the relatively low overall amount of compen-
sations helped to mitigate fully automated submissions. Since reCAPTCHA adapts the
difficulty depending on the source IP address, some Tor users complained about hard-
to-solve CAPTCHAs. Meta data: The meta data like source IP address and information
on the user’s browser was used to pinpoint simple double submission attempts. Time:
We considered submissions below a certain threshold fraudulent since it is impossible
to provide reasonable answers under a certain lower bound. Open-text questions: In
suspicious and borderline cases we manually checked the open-text questions to see
if the user had meaningful contributions to the survey. Reference links: The reference
links also provided a good insight when users attempted to submit multiple surveys and
always referenced their initial survey. Bitcoin address: The uniqueness of a Bitcoin
address was also an indicator for double submissions.
In case we detected double submissions, we accepted only the first submission for
our dataset as well as for our compensation scheme. All subsequent submissions were
excluded. In conclusion, we did not encounter fully automated submissions and that
most fraudulent attempts can be attributed to simple manual double submissions. More-
over, the Bitcoin community has proven to be very forthcoming. There have been cases
in which participants deliberately did not include Bitcoin addresses, and commenting
that they would like to help by saving the reward in order to recruit more participants.
The demographics of our sample correspond with data on the general Bitcoin pop-
ulation10 .
6 Qualitative Interviews
To get a deeper understanding of the findings from our online survey, we conducted an
additional field session with qualitative interviews.
6.1 Design and Recruitment
We recruited participants via a local Bitcoin mailing list and conducted a two-hour field
session at a local bar that accepts bitcoins. All interviewees are regularly using Bitcoin
and had previously completed our online questionnaire. Two researchers were present
during the field session, one conducted the interview and the other one took notes. As
all participants were very particular about preserving their privacy, we chose not to
audio-record the interviews.
For the evaluation of our qualitative data, we focused on the exploration of ideas
and insights of the participants. Some of the numbers gathered from the interviews will
be used as rough indicators to discuss and complement the results from our quantitative
survey. We interviewed 10 participants in total. All participants were male and frequent
users of Bitcoin and other crypto-currencies. All of them reported to have an IT-related
background. The purpose of the qualitative interviews was mainly to complement our
quantitative results and to explain phenomena and trends from our online survey. After
10
http://www.coindesk.com/new-coindesk-report-reveals-who-
really-uses-bitcoin/
10 participants, we reached saturation and little to no further insights were gained, so
we concluded the study.
6.2 Coding
After the interviews, we went through the collected data and produced an initial set of
codes. We traversed the data segments collected from each participant for each ques-
tion and also included statements that did not directly evolve from a question. Two
researchers performed the initial coding independently of each other to minimize the
susceptibility of biased interpretation. After the initial coding process, we revised the
retrieved codes and discussed recurring themes, patterns and interconnections. After
agreeing on a final set of codes, we coded the entire interview data. We coded all data
segments, regardless if they emerged directly from a question or a continuative discus-
sion.
7 Results
In this section, we present an analysis of the participants’ responses addressing our
research questions defined in Section 4.1. At the beginning of each section we analyse
the results from our online survey, whereas at the end we compare these results with
our qualitative interviews and try to correlate and explain our findings.
7.1 General Bitcoin Usage (Q1)
Most participants reported to use Bitcoins for tips and donations (38.0%), followed
by virtual goods, such as web hosting, online newspapers (33.3%), online shopping
(27.5%), altcoins (26.5%), gambling (26.5%) and Bitcoin gift cards (19.9%). About 5%
self-reported to buy or have bought drugs with bitcoins. 30.2% of our sample reported
to use Bitcoin at least once a week, 25% stated that they use Bitcoin at least once a
month and 19% at least once a day. The remainder of the participants indicated to use
Bitcoin at least once a year or even less. These results suggest that the majority within
our survey frequently uses Bitcoin.
We also asked our participants about the amount of bitcoins they are currently hold-
ing. About half of the participants did not want to specify. According to their reports,
our sample holds approximately 8000 B in total. The majority of users (70%) started
to use Bitcoin between 2013 and 2015. 17% started between 2011 and 2012. 58.0% re-
ported to use other crypto currencies in addition to Bitcoin, most frequently Dogecoin
and Litecoin. The most popular Bitcoin exchanges in our sample are BTCE (20.9%),
Bittrex (14.0%) and Bitstamp (13.0%). 11.4% of our participants are currently mining
bitcoins. Most of them started mining after 2014. Many of those who started earlier
have stopped mining as they currently consider it infeasible. 195 (19.7%) participants
claimed to be running a full Bitcoin server that is reachable from the Internet. The
top-mentioned reason for running a Bitcoin server was to support the Bitcoin network
(60.5%), followed by fast transaction propagation (46.6%), network analysis (30.3%)
and double-spending detection (26.1%).
All participants from our qualitative interviews are frequent Bitcoin users, and some
of them are active in the local Bitcoin association. Most interviewees mentioned that the
decentralized nature of Bitcoin was among the main reasons to start using Bitcoin. The
second-most mentioned reason was simply curiosity. One participant who used to live
in Crimea at the time the Ukrainian-Russian conflict started mentioned socio-political
reasons. He used to work for a US company at that time and needed a safe and cheap
option to receive his salary in Crimea. He furthermore wanted to make sure to not lose
any money due to the annexation to the Russian Federation. In his opinion, Bitcoin
was the best option and according to him, many people started using Bitcoin at that
time in Crimea. Some participants also mined Bitcoins some years ago when it was still
profitable to mine at small scale.
7.2 Practices of Bitcoin Management (Q2)
Bitcoin Wallets and Backup Behavior. Table 3 shows the most widely used Bitcoin
wallets. The participants could mention multiple wallets as it is a common scenario that
users use more than one wallet. The table also shows the number of participants from
our sample who use a certain wallet as well as the percentage. Furthermore, Table 4
shows whether the users protect their wallets with a password and if these wallets are
encrypted. Our findings show that the majority of users protect their wallets with a
password. In case of web clients, we observed a lack of background knowledge. For
example, 47.7% of Coinbase users in our sample say that their wallet is encrypted and
34% claim that they do not know if it is encrypted. We observed a similar trend for Xapo
which is the third-most used wallet in our sample. Just like Coinbase, it is also a web-
hosted tool and, similarly to Coinbase, only about half of the users say it is encrypted
and about a third does not know if it is encrypted. Regarding backups, only a third of
Coinbase users and 43% of Xapo users backup their wallets. 33.9% of Coinbase and
28.5% of Xapo users do not know whether their wallet is backed up. We also found
that Bitcoin users with more than 0.42B(100 USD) do not backup their CMT more
often than users with less bitcoins. This effect is statistically significant in our sample
(χ2 (1) = 5.1, p = 0.02).
We also asked our participants whether they create additional backups in case their
primary backup gets lost or stolen. In our sample, Bitcoin Core users have the highest
rate of additional backups. 64% of them indicated to make a secondary backup of their
wallet. Table 2 shows self-reported properties of wallet backups. According to our data,
none of our participants stores a backup on an air-gapped computer. The most reported
backup properties were encryption and password protection. According to our sample,
197 backups are stored in a cloud.
59.7% of our participants only use one wallet to manage their bitcoins. 22.7% use
two, and 10.6% use three wallets. The remaining 7% use four or more wallets. The
maximum number of wallets a participant reported to use was 14. This participant justi-
fied this high number by reporting that he wanted to try out the wallets before choosing
those that met his requirements best. About half of our participants who used a web
client did this exclusively to manage their bitcoins. The other half used a web client
in addition to a local client. To our surprise our results show that most coins of our
participants are stored in Armory11 . The Armory users in our sample have about 3818
B in their Armorys, where the top five users reported to have 2,000 B, 885 B, 300 B,
230 B and 150 B. The highest reported number of bitcoins stored in a participant’s web
client was 100 B. The reported sum of all coins stored in Coinbase is 238 B, in Xapo
11
https://bitcoinarmory.com/
it is 157 B. Figure 2 illustrates the accumulated bitcoins per wallet as reported by our
participants.

Fig. 2. Self-reported wallet usage and accumulated hosted bitcoins per wallet.

Table 2. Backup properties in absolute mentions in descending order; a user can have multiple
wallets and multiple backups.

Backup properties Mentions

My backup is encrypted. 662
My backup is password protected. 629
My backup is stored on external storage (e.g. USB drive). 430
My backup is stored on paper. 334
My backup is stored in the cloud (e.g. Dropbox). 197
My backup is stored on an air-gapped device. 0

Table 3. Properties of the most frequently used wallets mentioned by our participants.

CMT Number Percent B

Coinbase 314 31.7 238
Bitcoin Core 236 23.8 752
Xapo 179 18.1 157
Electrum 125 12.6 226
MyCelium 97 9.8 62
Table 4. Properties of the most mentioned CMTs. The three blocked columns contain information
on whether the CMT is encrypted, if it is backed up, whether there exists an additional backup
and the mentions in percent (Yes, No and I don’t know (IDK)). The rightmost column contains
the sum of bitcoins stored in a respective CMT by our participants.

Encrypted? Backup? Additional Backup?

CMT Yes No IDK Yes No IDK Yes No IDK
Coinbase 47.5 18.5 34.0 35.5 30.6 33.9 30.3 66.9 2.8
Bitcoin Core 72.8 16.1 11.1 76.3 14.0 9.7 64.0 32.2 3.8
Xapo 51.4 19.0 29.9 43.0 28.5 28.5 41.3 57.5 1.2
Electrum 72.8 15.2 22.0 77.6 16.0 6.4 55.2 44.0 0.8
MyCelium 61.9 21.6 16.5 83.5 12.4 4.1 52.6 47.2 0.2

Anonymity. We found that 32.3% of our participants think that Bitcoin is per-se anony-
mous while it is in fact only pseudonymous. 47% thinks that Bitcoin is not per-se anony-
mous but can be used anonymously. However, about 80% think that it is possible to
follow their transactions. 25% reported to have used Bitcoin over Tor to preserve their
anonymity.

We also asked participants if they take any additional steps to stay anonymous. 18%
reported to frequently apply methods to stay anonymous on the Bitcoin network. Most
of them reported to use Bitcoin over Tor followed by multiple addresses, mixing ser-
vices, multiple wallets and VPN services. As shown by Biryukov et al. [1, 3] using Bit-
coin over Tor creates an attack vector for deterministic and stealthy man-in-the-middle
attacks and fingerprinting.
7.3 Risk Perception (Q3)
We were also interested in user perceptions of risks associated with Bitcoin. We pro-
vided the participants with 11 risk scenarios. We selected the risk scenarios based on
findings from scientific literature and evidence from online resources. For each risk
scenario, we provided an easy-to-understand description and asked the participants
whether they think the risk is likely or unlikely to occur. Figure 3 shows the partici-
pants’ risk estimation. Our results show that the participants consider value fluctuation
as the highest risk, followed by vulnerabilities in hosted wallets and Bitcoin theft via
malware. Our participants estimated the risk for cryptographic flaws as the lowest, fol-
lowed by double-spending attacks and DoS attacks on the Bitcoin network.
7.4 Security Breaches (Q4)
About 22.5% indicated to have lost bitcoins or Bitcoin keys at least once. Of those,
43.2% mentioned that it was their own fault (e.g., formatted hard drive or lost a physi-
cal device with Bitcoin keys). 26.5% reported that their loss stemmed from a hardware
failure (e.g., a broken hard drive), followed by software failure (24.4%; e.g. keyfile cor-
ruption) and security breaches (18% e.g., malware, hacker).
 Fig. 3. User perceptions of risk scenarios in percentage of participants (N = 990).

The majority (77.6%) among those who lost bitcoins did not want to indicate whether
they were able to recover their keys. Of those who provided an answer, 65% were not
able to recover their keys. Overall, our participants reported to have lost about 660.6873
bitcoins. However, it must be taken into account that we did not ask when the coins were
lost. Hence, interpreting this result we must take into consideration that the Bitcoin ex-
change rate is highly volatile and it is therefore hard to provide an overall estimation in
USD. About 40% of our participants reported to have lost money due to a self-classified
major security breach. 13.1% of our overall sample reported to have lost bitcoins in
HYIPS (high-yield investment programs) and pyramid schemes. 7.9% lost money at
Mt. Gox.

We also gave our participants the opportunity to describe how they dealt with the
incident. Most participants stated that they did not do anything to recover their keys and
simply accepted the loss. Some argued that the financial loss was not worth the effort to
take further steps or that they felt helpless as they didn’t know what to do. Those who
actually took action most frequently mentioned that they filed claims and contacted the
exchange or online wallet provider. Those who lost money to a malicious online wal-
let reported to have moved to other types of wallets instead of hosted/online wallets.
The participants who lost money in HYIPS mostly stated that they started to use less
risky investments and learned from their previous mistakes. Irrespective of the security
breach, many participants reported to have spread the word over forums on the Internet
and shared their experiences with other affected users.
Participant Statements

– “I follow the ‘do not invest more than you’re ready to lose’ rule.” (P3848)
– “I just had to accept that my money was stolen ... and that I learned my lesson to
never use exchanges as wallets. Keep everything in your own hand.” (P3763)
– “Just learned from it. It was exceedingly stupid on my part.” (P853)

Eight participants from our qualitative interviews reported that they have already
experienced an intentional or accidential key and/or Bitcoin loss. Three participants
were affected from the Mt. Gox security breach and two of them reported to have filed
a claim on Kraken12 . One participant reported to have lost a physical Casascius13 Bit-
coin but then stopped searching for it as it was only worth about 9 USD at that time.
Others also mentioned to have lost their keys due to device failure, corrupted HDDs, or
software failure.
7.5 Perceptions of Usability (Q5)
Even though most participants of our qualitative interviews were very much concerned
about security and privacy aspects of Bitcoin management, eight of them said that
they would recommend web wallets and deterministic wallets to non-tech-savvy Bit-
coin users. Convenience and easiness of use were highlighted as the main benefits. One
participant said that he would definitely recommend a wallet where the private key is
stored on a central server to make key recovery easier and to obviate the need for com-
prehensive backups as well as that mnemonics would help. Six participants also said
that they would recommend MyCelium14 as the most usable wallet. Those who had
already used MyCelium consider the paper backup procedure as the most usable and
secure way. To create a paper backup with MyCelium, the user has to print out a tem-
plate that contains some parts of the key and then lets the user fill out the empty spots
manually. Some participants expressed initial discomfort when they used paper wallets.
Most interviewees also highlighted the need for fundamental education in early
years of childhood. P2 said that Bitcoin is inherently complex, that the fundamental
idea of public key cryptography should be taught in school and monetary systems are a
matter of culture.
Two participants also highlighted that user interfaces should be simplified and min-
imalized. Many participants stated that for a fast proliferation of Bitcoin, simple and
intuitive UIs are more important than security. They argued that computers proliferated
even though most people do not know how computers work and that security is not
necessarily an argument for large-scale adoption. They provided examples such as cars
in the 1940s, computers, credit cards and WhatsApp. They also said that the amount
of money that is circulating in the Bitcoin network is low enough to take the risk of
loosing it and compared this scenario to the risk of loosing cash. Some participants also
proposed a dedicated device with an intuitive UI for key management and think that
such an artifact would be the most secure and usable option.

12
https://www.kraken.com/
13
https://www.casascius.com/
14
https://mycelium.com/
Participant Statements

– “It somehow didn’t feel right for me to go out of the digital realm.” (P6 on paper
wallets)
– “Children learn about our monetary system in their very early days in primary
school. This is why society knows how to use cash and credit cards. I’m sure it
could be the same thing with a decentralized crypto-currency.” (P7)

8 Discussion
The goal of this paper was to answer the research questions provided in Section 4.1
in order to understand how users interact with the Bitcoin ecosystem. As this is the
first-ever user study focused on user experiences with Bitcoin security and privacy, we
gathered useful insights. In the following we discuss our results in the context of already
existing works in the field.
Regarding Bitcoin management tools and practices (Q2), we found that two of the
most widely used CMTs were web-hosted solutions that obviate the need for users to
deal with key management and backups. Our results show that our participants had clear
preferences regarding their choice of CMT. In contrary, this is not the case for Bitcoin
exchanges. Our data shows that the Bitcoin exchanges chosen by our participants were
almost evenly distributed. Even though our data reveals a clear tendency towards web-
hosted solutions, these CMTs do not host the majority of our participants’ bitcoins.
According to our participants’ self-reported data, the highest amount of accumulated
bitcoins is hosted in Armory. At the time of writing, if used correctly, Armory is one of
the most secure solutions.
For the two most widely used web-hosted CMTs, about a third of our participants
are unaware of whether their wallet is encrypted or backed up. In such a scenario, users
shift responsibilities to a third party. Even though this seems to be a convenient and
usable solution for non-expert users, it implies that the user trusts these third parties to
take care of their security. About 50% of web client users indicated to use an additional
local client to store their virtual assets. According to our results, users that have a higher
number of bitcoins do not necessarily back up their wallets more often. Also, MyCelium
users back up their wallets more often than others. Hence we conclude that backup
motivation and respectively fatigue depend highly on usability and not on the number
of coins.
As the answer to Q4 indicates, participants have already lost money to malicious
hosted-wallet providers. Also, our participants perceived vulnerabilities in hosted wal-
lets as the second highest among our risk scenarios (Q5). Some participants from our
qualitative interviews said that they would recommend inexperienced users to start with
a hosted wallet due to the usability benefits as for most other solutions users are re-
quired to have at least a basic understanding of the underlying basics of Bitcoin and the
blockchain.
Bitcoin is a pseudonymous system, whereas a wide-spread myth says that it is per-se
anonymous. More than a third of our participants still believe in this myth and reported
that they think that Bitcoin is fully anonymous. About half of our participants are aware
that Bitcoin is not per-se anonymous, but that it can be used anonymously. Regarding
anonymity measures, many users reported to use Bitcoin over Tor, which in fact creates
an attack vector for deterministic and stealthy MITM attacks, as shown in [3].
Our results also suggest that our participants trust the cryptography behind Bitcoin
and are aware of risks according to value fluctuation and software vulnerabilities. Poor
usability and the lack of knowledge are major contributors to security failures. Almost
a fourth of our participants indicated that they had already lost bitcoins or Bitcoin keys
at least once (Q5). To our surprise, almost half of those who lost bitcoins due to a self-
induced error which indicates that state of the art CMTs are sometimes still difficult
to use or require users to manually take care of security tasks, such as backups and
encryption. Our results also indicate that the Bitcoin ecosystem is mostly utilized for
tipping and donations as well as acquiring digital goods, but to some extend also for
criminal activity and adventurous gambling.
9 Conclusion
In this work we presented the first user study to examine how users interact with the
Bitcoin ecosystem in terms of security and privacy. We conducted an online survey with
990 Bitcoin users and qualitative interviews with a subset of 10 participants. Further-
more, we introduced the term Coin Management Tools (CMTs) to describe tools that let
users manage their virtual assets (keys) and interact with the Bitcoin network. Addition-
ally, we proposed a method for categorizing CMTs according to the degree of control
and verifiability a user can exercise with this client.
We found that managing bitcoins is still a major challenge for many users, as many
of them do not apply sufficient security measures such as encryption and backups. We
found that many participants were not even aware of security features provided by their
used CMT. Two of the most widely used CMTs among our participants were web-
hosted solutions. About half of their users reported to use such solutions exclusively,
while the other half also used local clients. Even though web clients ought to be a usable
and convenient solution, they require a certain level of trust and shift the responsibilities
of encryption and managing backups to a third party. We also found that 22.5% of our
participants have already experienced security breaches and lost bitcoins. About half of
them mentioned a self-induced error as the reason, which highlights that users find it
still difficult to manage their bitcoins in a secure way.
We believe that our insights and suggestions are an important first step towards
improving the usability of Bitcoin security. In order to guarantee secure interactions
with the Bitcoin ecosystem to both expert and non-expert users, we must re-think the
concept of Bitcoin management, since it is more than just the secure handling of secret
keys. Bitcoin is a decentralized system where the interactions between peers and the
propagation and verification of messages and data is important. If this aspect is ignored,
Bitcoin would just consist of signed numbers without value.
Acknowledgements
This research was funded by COMET K1, FFG - Austrian Research Promotion Agency
and by FFG Bridge Early Stage 846573 A2Bit. We would also like to thank Martin
Mulazzani, Artemios G. Voyiatzis and Matthew Smith for their useful comments and
feedback. Furthermore, we would like to thank Elizabeth Stobert for her valuable feed-
back and for her help in recruiting participants.
References
1. Alex Biryukov and Dmitry Khovratovich and Ivan Pustogarov. Deanonymisation of clients
in Bitcoin P2P network. CoRR, abs/1405.7418, 2014.
2. A. W. Baur, J. Bühler, M. Bick, and C. S. Bonorden. Cryptocurrencies as a disruption?
empirical findings on user adoption and future potential of bitcoin and co. In Open and Big
Data Management and Innovation, pages 63–80. Springer, 2015.
3. A. Biryukov and I. Pustogarov. Bitcoin over tor isn’t a good idea. arXiv preprint
arXiv:1410.6079, 2014.
4. Bitcoin community. Bitcoin developer guide, Oct. 2014. Accessed: 2014-10-14.
5. Bitcoin community. Bitcoin protocol specification, Oct. 2014. Accessed: 2014-10-14.
6. Blockchain.info. Bitcoin currency statistics, Apr. 2014. Accessed: 2014-04-05.
7. J. Bonneau, A. Miller, J. Clark, A. Narayanan, J. A. Kroll, and E. W. Felten. Sok: Research
perspectives and challenges for bitcoin and cryptocurrencies. 2015.
8. S. Eskandari, D. Barrera, E. Stobert, and J. Clark. A first look at the usability of bitcoin key
management. In Workshop on Usable Security (USEC), 2015.
9. M. H. F. Reid. An analysis of anonymity in the bitcoin system. In 2011 IEEE International
Conference on Privacy, Security, Risk, and Trust, and IEEE International Conference on
Social Computing, 2011.
10. J. Garay, A. Kiayias, and N. Leonardos. The bitcoin backbone protocol: Analysis and appli-
cations. In Advances in Cryptology-EUROCRYPT 2015, pages 281–310. Springer, 2015.
11. S. L. Garfinkel, D. Margrave, J. I. Schiller, E. Nordlander, and R. C. Miller. How to make
secure email easier to use. In Proceedings of the SIGCHI conference on human factors in
computing systems, pages 701–710. ACM, 2005.
12. S. L. Garfinkel and R. C. Miller. Johnny 2: a user test of key continuity management with
s/mime and outlook express. In Proceedings of the 2005 symposium on Usable privacy and
security, pages 13–24. ACM, 2005.
13. S. Gaw, E. W. Felten, and P. Fernandez-Kelly. Secrecy, flagging, and paranoia: adoption
criteria in encrypted email. In Proceedings of the SIGCHI conference on human factors in
computing systems, pages 591–600. ACM, 2006.
14. A. Gervais, H. Ritzdorf, G. O. Karame, and S. Capkun. Tampering with the delivery of blocks
and transactions in bitcoin. Technical report, Cryptology ePrint Archive, Report 2015/578,
2015. http://eprint. iacr. org.
15. S. Goldfeder, R. Gennaro, H. Kalodner, J. Bonneau, J. Kroll, E. W. Felten, and A. Narayanan.
Securing Bitcoin wallets via a new DSA/ECDSA threshold signature scheme. Accessed
2015-06-09.
16. E. Heilman, A. Kendler, A. Zohar, and S. Goldberg. Eclipse attacks on bitcoin’s peer-to-
peer network. In 24th USENIX Security Symposium (USENIX Security 15), pages 129–144,
Washington, D.C., Aug. 2015. USENIX Association.
17. Krzysztof Okupski. Bitcoin protocol specification, Oct. 2014. Accessed: 2014-10-14.
18. J. Lazar, J. H. Feng, and H. Hochheiser. Research methods in human-computer interaction.
John Wiley & Sons, 2010.
19. T. Moore and N. Christin. Beware the middleman: Empirical analysis of bitcoin-exchange
risk. In Financial Cryptography and Data Security, pages 25–33. Springer, 2013.
20. S. Nakamoto. Bitcoin: A peer-to-peer electronic cash system. online, Dec 2008.
21. NIST. FIPS 180-4: Secure Hash Standard (SHS), Mar. 2012.
22. S. Sheng, L. Broderick, C. A. Koranda, and J. J. Hyland. Why johnny still can’t encrypt:
evaluating the usability of email encryption software. In Symposium On Usable Privacy and
Security, 2006.
23. C. Wharton, J. Rieman, C. Lewis, and P. Polson. The cognitive walkthrough method: A
practitioner’s guide. In Usability inspection methods, pages 105–140. John Wiley & Sons,
Inc., 1994.
24. A. Whitten and J. D. Tygar. Why johnny can’t encrypt: A usability evaluation of pgp 5.0. In
Usenix Security, volume 1999, 1999.

A CMT Categorization
In this section we discuss the term Coin Management Tool and provide a methodol-
ogy to categorize CMTs according to the degree of control and verifiability a user can
exercise with his respective client. The proposed scheme is tailored to Bitcoin-like cryp-
tocurrencies, but explicitly designed in an utmost generic way so that it can be applied
to other derived cryptocurrencies as long as they are not fundamentally different in
their design. Our approach used the evaluation framework from [8] as a starting point.
A categorization according to our scheme allows users to quickly distinguish clients ac-
cording to their underlying security model and hence allows users to make an informed
decision on the level of confidence and trust they can put into an individual client.
A.1 Definitions
When Bitcoin was in its infancy bitcoind was the only available Bitcoin client which
performed all required tasks: mining management, P2P network communication and
blockchain management, key management and virtual asset management. With the in-
creased popularity of Bitcoin and cryptocurrencies in general, more and more software
was developed which focused on a subset of individual tasks of the original implemen-
tation. Moreover, the design of Bitcoin allows users to use it even if they do not run
mining software or a full P2P client (full node). As a result there exists software with
varying feature sets where the handling of public-private key pairs is the most sensi-
tive and hence the most common core feature. A Bitcoin wallet was originally defined
as a collection of private keys15 . Since this definition is very narrow, we introduced
the broader definition of a Coin Management Tool (CMT) to account for the other ar-
eas without whom most cryptocurrencies would not work. Especially the network and
blockchain layer of Bitcoin and other cryptocurrencies is not only important for the in-
tegrity of the system as a whole, but has a significant impact on the security and privacy
of each and every end user.
A.2 Categorization
To categorize CMTs, we first identified critical CMT tasks which are directly related to
security and privacy issues. This covers aspects regarding key management like gener-
ating keys/addresses and signing transactions, as well as P2P network communication
and blockchain management like handling connections as well as verifying and stor-
ing the blockchain. These core tasks can be used to divide CMTs into five different
categories. A client can be in more than one category depending on its configuration.

– cat. 0: A client which runs on a user-controlled device and is able to perform

key management operations, but cannot perform any P2P network communication.
Therefore, it is not a stand-alone solution. This includes some dedicated hardware
clients/wallets and cold-storage clients which require a second online device for
transaction processing.
15
https://en.bitcoin.it/wiki/Wallet
– cat. I: A client which runs on a user-controlled device and performs all P2P network
communication and blockchain verification related tasks, keeps a copy of the full
blockchain, is able to perform all key management related operations and executes
the mining algorithm. In other words, this is a client which can perform all required
tasks to operate a cryptographic currency (e.g., the Bitcoin core implementation
bitcoind when the option setgenerate true is set).
– cat. II: A client which runs on a user-controlled device and performs all P2P tasks
related to network communication and blockchain verification, keeps a copy of the
full blockchain and is able to perform all key management related operations. This
type of client is sometimes referred to as thick-client or full-node.
– cat. III: A client which runs on a user-controlled device and performs certain
P2P tasks related to network communication and blockchain verification but does
not keep a copy of the full blockchain, although, it is able to perform all key-
management-related operations. This type of client is sometimes referred to as
thin-client or mobile-client/wallet and includes so-called SPV-clients/wallets (Sim-
plified Payment Verification) e.g., Electrum.
– cat. IV: A client which does not run on a user-controlled device and where all tasks
are performed by a trusted third party on behalf of the user. This type of client is
sometimes referred to as hosted- or web-client/wallet. Thereby it is not relevant
if the key management is handled in the browser (e.g., via JavaScript) since this
would require the user to download (and verify) the script code from the website of
the third party every time he/she wants to use it.
B Interview Questions
Questions with answer options as ”( )” are multiple choice checkboxes whereas answer
possibilities marked alphabetical e.g. ”a)” are single selections.
B.1 BTC Demographics
Q1 Please input which year you started using Bitcoin:
a) 2009 b) 2010 c) 2011 d) 2012 e) 2013 f) 2014 g) 2015
Q2 Select which main features are responsible for you using Bitcoin (multiple selec-
tions possible):
( ) The opportunity of financial gain
( ) Curiosity
( ) Anonymous nature
( ) Decentralized nature
( ) A friend/colleague suggested to me to start using Bitcoin
( ) The possibility to internationally transfer money with relatively low fee
( ) The possibility to accept bitcoins for my services or for my products
( ) Other:

Q3 What is the estimated sum of bitcoins you are holding?

a) I hold approximately b) I do not want to specify
Q4 Please provide what services or products you pay for with bitcoins (multiple selec-
tions possible):
( ) Bars, restaurants
( ) Bitcoin gift cards
( ) Donations, tipping
( ) Drugs
( ) Gambling sites
( ) Hotels, travel
( ) Online marketplaces and auctions
( ) Online shopping (Newegg, ...)
( ) Altcoin (e.g. Litecoin, ...)
( ) Physical stores that accept bitcoins
( ) Underground marketplaces
( ) Virtual goods (webhosting, online newspapers, ...)
( ) Medium for currency exchange
( ) Other:

Q5 What do you think are the most likely risks associated with Bitcoin?
Q6 Please select the crypto currencies you are holding or using besides Bitcoin (mul-
tiple selections possible):
( ) I do not use other crypto currencies
( ) BanxShares
( ) BitShares
( ) BlackCoin
( ) Bytecoin
( ) Counterparty
 ( ) Dash
( ) Dogecoin
( ) Litecoin
( ) MaidSafeCoin
( ) MonaCoin
( ) Monero
( ) Namecoin
( ) Nxt
( ) Peercoin
( ) Primecoin
( ) Ripple
( ) Startcoin
( ) Stellar
( ) SuperNET
( ) Vertcoin
( ) YbCoin
( ) Other

Q7 Select the Bitcoin exchanges you have used in the past or you are using on regu-
larly (multiple selections possible):
( ) None
( ) BanxIO
( ) Bitcoin Exchange Thailand
( ) Bittrex
( ) Bitcoin Indonesia
( ) bitcoin.de
( ) Bitfinex
( ) Bitstamp
( ) BitX South Africa
( ) BTC-e
( ) BTC38
( ) BTCChina
( ) CCEDK
( ) Cryptsy
( ) Gatecoin
( ) hibtc
( ) Kraken
( ) Mt. Gox
( ) OKCoin
( ) Poloniex
( ) QuadrigaCX
( ) The Rock Trading
( ) VirWox
( ) Other:

Q8 What do you think are the greatest benefits of Bitcoin?

Q9 How often do you perform Bitcoin transactions?
a) At least once a day b) At least once a week c) At least once a month d) At least
once every six months e) At least once a year f) Less than once a year

B.2 BTC Wallets

Q10 Please tick which wallets you are ¡b¿currently¡/b¿ using (multiple selections pos-
sible):
( ) Airbitz
( ) Armory
( ) Bitcoin Core
( ) Bitcoin Wallet (Schildbach Wallet)
( ) BitGo
( ) Bither
( ) breadwallet
( ) Circle
( ) Coinapult
( ) Coinbase
( ) Coinkite
( ) Coinomi
( ) Electrum
( ) Green Address
( ) Hive
( ) Ledger Nano
( ) mSIGNA
( ) MultiBit
( ) Mycelium
( ) Ninki
( ) TREZOR
( ) Xapo
( ) Not in list

Q11 Why did you choose to use multiple wallets to manage your bitcoins?

B.3 Wallet Usage

For every selected wallet in Q10 we asked the following questions.

Q12 Why did you choose hwallet-namei to manage your Bitcoins?

Q13 How many bitcoins do you have approximately in this wallet?
a) I hold approximately htextfieldi bitcoins. b) I do not want to specify
Q14 Is this wallet password protected?
a) Yes b) No c) I do not care d) I do not know
Q15 Is this wallet encrypted?
a) Yes b) No c) I do not care d) I do not know
Q16 Is this wallet backed up?
a) Yes b) No c) I do not know
B.4 BTC Mining

Q17 Are you currently mining bitcoins?

a) Yes, since b) No, but I have mined from-to c) No, I have never mined bitcoins
Q18 How many bitcoins have you mined in total?
a) I mined approximately b) I do not want to specify
Q19 Do you or have you participated in mining pools?
a) Yes b) No
Q20 Please tick the names of the mining pools you have or are participating in (multi-
ple mentions possible):
( ) 21 Inc.
( ) AntPool
( ) Bitcoin Affiliate Network
( ) BitFury
( ) BitMinter
( ) Bitsolo
( ) BTCChina Pool
( ) BTC Guild
( ) BTC Nuggets
( ) BW.COM
( ) EclipseMC
( ) Eligius
( ) F2Pool
( ) GHash.IO
( ) Kano CKPool
( ) KnCMiner
( ) MegaBigPower
( ) P2Pool
( ) Slush
( ) Telco 214
( ) Other:

B.5 BTC Server

Q21 Do you run a full Bitcoin server that is reachable for others from the Internet?
a) Yes b) No
Q22 Please provide some reasons on why you operate a full Bitcoin server (multiple
selections possible):
( ) Fast transaction propagation
( ) Double-spending detection
( ) Network analysis
( ) Support the Bitcoin network
( ) Other
B.6 BTC Security Risks
Q23 How would you estimate the risk of monetary loss for Bitcoin compared to credit
cards?
(7 Point Likert-Scale from ”High” to ”Low”)
Q24 How high do you think is the risk of becoming a victim of a successful double
spending attack?
(7 Point Likert-Scale from ”High” to ”Low”)
Q25 How high or low would you estimate the risk for malware that steals your Bit-
coins?
(7 Point Likert-Scale from ”High” to ”Low”)
Q26 How would you estimate the risk of monetary theft in case the device with your
wallet gets lost or stolen?
(7 Point Likert-Scale from ”High” to ”Low”)
Q27 How would you estimate the risk of de-anonymization?
(7 Point Likert-Scale from ”High” to ”Low”)
Q28 How high do you think the risk of cryptographic flaws is?
(7 Point Likert-Scale from ”High” to ”Low”)
Q29 How high do you think is the risk of security vulnerabilities in hosted/web wallets
or Exchange services?
(7 Point Likert-Scale from ”High” to ”Low”)
Q30 How high do you think is the risk of key loss due to a device failure?
(7 Point Likert-Scale from ”High” to ”Low”)
Q31 How high do you think is the risk that the Bitcoin network is temporarily not
available?
(7 Point Likert-Scale from ”High” to ”Low”)
Q32 How high do you think is the risk of a centralization of mining?
(7 Point Likert-Scale from ”High” to ”Low”)
Q33 How high do you think is the risk of a strong fluctuation in the Bitcoin exchange
rate (e.g. BTC to USD and vice versa)?
(7 Point Likert-Scale from ”High” to ”Low”)

B.7 BTC Anonymity

Q34 Do you think that Bitcoin usage is anonymous?
a) Yes, Bitcoin is fully anonymous b) No, Bitcoin is not anonymous c) Not per se,
but it can be used in an annonymouse manner
Q35 Do you think it is possible to follow your transactions?
a) Yes b) No
Q36 Have you ever used Bitcoin over Tor¡b title=”Tor is free software and an open
network that helps you defend against traffic analysis, a form of network surveil-
lance that threatens personal freedom and privacy, confidential business activities
and relationships, and state security. More info at www.torproject.org”¿[?]¡/b¿?
a) Yes b) No
Q37 Do you take additional steps to ensure your privacy using Bitcoin?
a) Yes b) No
B.8 BTC Security Breaches
Q38 Have you ever lost your bitcoins or Bitcoin keys?
a) Yes b) No
Q39 Please select the reason for your key/Bitcoin loss (multiple selections possible):
( ) Hardware failure (e.g. hard drive broke, etc)
( ) Software failure (e.g. keyfile corruption, etc)
( ) Self induced event (e.g. hard drive formatted, physical device lost, etc)
( ) Malicious event (e.g. malware, hacker, etc)
( ) Other

Q40 Have you been able to recover your keys?

a) Yes, b) No,
Q41 How many bitcoins did you loose due to this incident?
a) bitcoins b) I do not want to specify
Q42 Please select the security incidents you have been affected by (multiple selections
possible):
( ) None
( ) Mt. Gox incident
( ) Silk Road bust
( ) inputs.io hack
( ) Pony botnet malware
( ) Pyramid schemes / HYIPS (High yield investment programs)
( ) Mining hardware scams (Labcoin, Active Mining Corporation, Ice Drill, AsicMiningE-
quipment.com Dragon-Miner.com, ...)
( ) Mining pool scams
( ) Scam wallets
( ) Bitcoin exchange scam
( ) Other:

Q43 How did you deal with the incident?

Q44 What was the approximate value of your lost bitcoins in USD?
a) USD b) I do not want to specify c) I do not know

B.9 Demographics
Q45 Please provide your age:
Q45 Please provide your gender:
a) Female b) Male c) Do not want to specify
Q46 Please select your highest completed level of education:
a) Did Not Complete High School b) High School/GED c) Some College d) Bach-
elor’s Degree e) Master’s Degree f) Advanced Graduate work or Ph.D. g) Not Sure
Q47 Do you work or study in a computer science related field?
a) Yes b) No
Q48 How would you describe yourself in terms of privacy behaviour?
A continouos slider between ”I am not concerned about my privacy” and ”I would
describe myself as a privacy fundamentalist”
B.10 End
Q49 You can enter your Bitcoin address in the textfield below. Please make sure that
your address is correct in order to receive your incentive.
Q49 This is the place where you can provide suggestions, complaints or any other
information we may have forgotten to ask in the questionnaire.

C Address Signature
./bitcoin-cli signmessage 12yeU5ymM67SL5UWVSwErAgwVwwaTd1Nma \
"https://www.soscisurvey.de/BTC_study/"
HzzNxFmeRhbhAwVZ4DsraBkXkW7JYjO0tAlIPAnHB2z5P12eddFilWXJmwGm\
PkgS/v8W0DNr0Z1qLwroPbWWMoE=

D Reference link issue

We had a problem in our implementation of this last page of the survey which also
showed the link to the survey containing a random reference which should identify this
particular participant in our rewarding scheme. If the CAPTCHA was not solved suc-
cessfully the side reloads itself and would also calculate and show a different reference
link. The references link will only be stored and linked to this particular participant if
the CAPTCHA is entered correctly. Therefore, all users which just copied the first link
and then entered a wrong CAPTCHA distributed a link we where not able to attribute
correctly at the end of the survey.