ENISA Report - CTF Events
ENISA Report - CTF Events
MAY 2021
0
CTF EVENTS
May 2021
ABOUT ENISA
The European Union Agency for Cybersecurity (ENISA) has been working to make Europe
cyber secure since 2004. ENISA works with the EU, its member states, the private sector and
Europe’s citizens to develop advice and recommendations on good practice in information
security. It assists EU member states in implementing relevant EU legislation and works to
improve the resilience of Europe’s critical information infrastructure and networks. ENISA seeks
to enhance existing expertise in EU member states by supporting the development of cross-
border communities committed to improving network and information security throughout the
EU. Since 2019, it has been drawing up cybersecurity certification schemes. More information
about ENISA and its work can be found at www.enisa.europa.eu.
CONTACT
For contacting the authors please use [email protected]
For media enquiries about this paper, please use [email protected]
AUTHORS
Alastair Janse van Rensburg, University of Oxford
Richard Baker, University of Oxford
EDITORS
Ioannis Agrafiotis, Christina Skouloudi, Adrian Belmonte Martin, Filippos Papaioannou,
European Union Agency for Cybersecurity (ENISA)
ACKNOWLEDGEMENTS
We would like to thank all members of the International Cyber Security Challenge Steering
Committee for their valuable contributions.
LEGAL NOTICE
Notice must be taken that this publication represents the views and interpretations of ENISA,
unless stated otherwise. This publication should not be construed to be a legal action of ENISA
or the ENISA bodies unless adopted pursuant to the Regulation (EU) No 2019/881.
ENISA may update this publication from time to time.
Third-party sources are quoted as appropriate. ENISA is not responsible for the content of the
external sources including external websites referenced in this publication. This publication is
intended for information purposes only. It must be accessible free of charge. Neither ENISA nor
any person acting on its behalf is responsible for the use that might be made of the information
contained in this publication.
COPYRIGHT NOTICE
© European Union Agency for Cybersecurity (ENISA), 2021
Reproduction is authorised provided the source is acknowledged.
For any use or reproduction of photos or other material that is not under the ENISA copyright,
permission must be sought directly from the copyright holders.
1
CTF EVENTS
May 2021
EXECUTIVE SUMMARY
This report addresses the contemporary use of capture-the-flag (CTF) competitions around the
world. It first provides background on such competitions, their structure and variations. Analyses
of recent competitions is then conducted, comprising an in-depth qualitative analysis of notable
events (22 in total) and a high-level statistical analysis of a large dataset of public events of all
levels (879 in total).
team size
challenge categories
scoring methodology
hosting of event online vs. in-person
use of qualifier rounds
inclusion of peripheral activities
communication channels for media strategy
The report discusses the findings and proposes topics for consideration during event design. In
particular:
Team sizes: Hard limits may not be necessary and unbounded team sizes are seen in
notable events.
Parallel Competitions: Running parallel events with a different focus (different audience
or different challenge type) can broaden appeal easily.
Range of Media: Public engagement strategies benefit from a range of media. Inclusion of
CTF specific venues (such as that used in the statistical analysis) is recommended to best
reach the CTF community.
2
CTF EVENTS
May 2021
TABLE OF CONTENTS
1. INTRODUCTION 5
1.1 SUMMARY 5
1.2 STRUCTURE 5
2. BACKGROUND 6
2.1 JEOPARDY 6
2.1.1 Format 6
2.1.2 Scoring 8
2.1.3 Discussion 9
2.1.4 Variants 9
2.2 ATTACK-DEFENCE 10
2.2.1 Format 10
2.2.2 Scoring 10
2.2.3 Discussion 10
2.2.4 Variants 11
3. REPORT METHODOLOGY 12
4. ANALYSIS RESULTS 19
3
CTF EVENTS
May 2021
5. DISCUSSION 38
5.1.1 Popularity 38
5.1.2 Online and in-person 38
5.1.3 Mentors and coaches 38
5.2.1 Format 39
5.2.2 Challenge categories 39
5.2.3 Challenge sources 39
5.2.4 Jeopardy variants 40
5.4.1 Qualification 41
5.4.2 Team allocation 41
5.4.3 Resultant publications 41
5.4.4 Data Sharing 41
5.4.5 Writeups 41
6. CONCLUSIONS 43
6.8 POST-EVENT 44
7. BIBLIOGRAPHY 46
4
CTF EVENTS
May 2021
1. INTRODUCTION
1.1 SUMMARY
This report describes capture-the-flag (CTF) competitions and their contemporary use in cyber
security education and recreation worldwide. It provides background on CTFs and describes the
various competition formats and practices.
Based on the results of the analysis, different events and their practices are compared and
contrasted; to find commonalities and identify design decisions and their trade-offs. The report
concludes by making recommendations for consideration during event design.
1.2 STRUCTURE
This report is structured into the following sections:
Background: An outline of CTF competitions and formats to give background for later
sections. In particular, the challenge types, scoring systems and formats of both
"Jeopardy" and "Attack-Defence" events are detailed.
Methodology: A full description of the methodology used in this report, including data
themes, an explanation and justification of data sources, and methods used in the
statistical analysis.
Results: A per-theme analysis of the manually gathered data on each individual event,
together with summaries of the results of the statistical data gathered.
Discussion: A discussion of the results, including trends and consideration of the
differences between the two datasets produced by this report.
Conclusions: Conclusions drawn for running future CTF events based on the data
and analysis gathered in this report.
5
CTF EVENTS
May 2021
2. BACKGROUND
CTFs take many forms, but the most common forms are Jeopardy and Attack-Defence [3],
which are described in detail below. The information in this section is informed by the
experience of the authors in participating in and organising events, by the results of the
analyses of the report, and by the related work cited.
2.1 JEOPARDY
Jeopardy CTFs [2] [4] [5], the most common format for current competitions, contain a set of
custom-made, artificial challenges created specifically for the event. Participants tackle each
challenge as a stand-alone problem, and a solving a challenge results in a flag, which can be
submitted to a scoring system for a number of points.
2.1.1 Format
Jeopardy challenges are highly variable and the only key property is that they should grant
access to one (or more) flags when solved. Challenges are typically completely independent of
each other (although they may have multiple parts), and participants are able to solve them,
broadly, in any order. Each challenge consists of some downloadable files, a remote server, or
both. When remote servers are present, participants are not able to impact the performance of
the server for other teams, so that each team participants in the competition independently.
Challenges that consist only of files will usually contain the flag embedded into the
downloadable file and participants are expected to use any means to find the flag within the
files. When a remote server is present, participants are typically expected to perform a remote
exploit on the server to recover the flag, and there may be downloadable files (such as
challenge source code, or compiled binaries), that aid them in the attempt.
Challenges vary across a large and diverse number of categories, and the flexible format of the
contest allows for a wide degree of variation in challenges. Some common categories are
expected in any CTF, while less-common categories may appear depending on the theme or
format of the event. In particular, more difficult events tend to be more focused on categories
that replicate real exploits (such as pwn, web and re), while beginner-focused events have
more scope for categories that explore more gamified areas (often under the grouping of misc
challenges). Some of the most common categories include [6]:
2.1.1.1 pwn
Deriving their name from "owning" services, these challenges replicate attacks on real
vulnerable services. In some cases, these challenges may involve performing the compromise
of a piece of known-vulnerable software, but more often they are custom binaries that are
developed to showcase a particularly interesting exploit. These challenges are solved by
participants by interacting with a remote server, usually over a command line. In easier
challenges, participants may be given access to the compiled binary running on the target, or
even the source code directly. In harder challenges, participants will be given no information
6
CTF EVENTS
May 2021
other than access to the remote service. An example challenge might give participants access
to a remote server which runs a binary vulnerable to a buffer overflow. Overflowing the buffer
would cause the binary to crash and consequently output the flag.
2.1.1.3 web
Web challenges follow the same pattern as pwn challenges, but involve a target that runs a
web server, usually serving over HTTP, and often backed by other common web technologies
such as a database. Participants exploit the web server and cause it to provide the flag. In
contrast to pwn challenges, these challenges rarely involve low-level exploits of the web server
itself; participants will rarely gain shell access to the server. Instead, these usually involve
attacks on the web pages themselves, with SQL injection and PHP vulnerabilities common
themes in this category.
2.1.1.4 crypto
Cryptography challenges are similar to re challenges but involve reverse engineering of a
particular cryptographic protocol or implementation. These challenges can take the form of
encrypted messages that must be broken, or access to a flawed implementation of a
cryptographic protocol on a service which can be exploited to cause it to reveal secret keys.
2.1.1.5 forensics
Forensics challenges involve participants investigating an unknown piece of data, usually
determining the format of the data and then finding or building a tool capable of reading the
information. In these challenges, the flag is usually contained but obfuscated inside the
downloadable files and participants must understand the format in order to read the flag. There
are many varieties of challenges in this category, including reverse engineering corrupted file
formats, mounting obscure drive formats and finding hidden files, and investigating patterns in
recorded web traffic.
2.1.1.6 misc
Miscellaneous challenges that do not fall into other categories. These challenges are popular
for newer participants and beginner CTFs, where they can help participants get used to the
format of CTF competitions and teach introductory skills. Miscellaneous challenges are typically
more gamified than other categories and often offer a chance for organisers to include more
interesting, though less realistic, challenges. Examples vary hugely but might include reading
the source code for a simple maze generator and solving the maze, or interacting with a service
in a highly unusual way, such as via images.
2.1.1.7 programming
Some challenges are designed particularly to test the participants ability to write code. These
are less common and challenges of this nature are less favoured, in preference for challenges
in other categories that may require a programmed solution. This may be a response to the
popularity of programming-specific hackathon competitions, distinct from security-focused CTF
events.
7
CTF EVENTS
May 2021
2.1.1.8 physical
In-person events may include physical challenges that involve the participants interacting with
something in the real world. This may be security-focused, or intended to encourage
participants to interact with each other or explore the area. Examples of security-focused
challenges include performing wire-taps of Ethernet cables, intercepting WiFi traffic, and the
perennial-favourite of picking locks. Interaction-focused challenges might include giving parts of
a flag to each team so they must trade parts with each other, or hiding flags on QR codes
around the event location. Interaction-focused challenges are less preferred in higher skill
competitions in favour of more competitive challenges.
Flags are usually a string of text in a standardised format. This enables participants to know
when the challenge is successfully solved, and means that the entire process is automated
without the need for a judging process. Flag formats are standardised so that participants do not
need to guess what the flag is once the challenge is solved. While some competitions may use
answers to questions as flags (e.g., "which IP tried to attack the server?") this is less popular
and can be a frequent source of confusion and ambiguity. Most competitions embed flags into
challenges in such a way that a successful solution to the challenge results in a clear flag that
can be submitted. To this end, flags often begin and end with a published string, for instance
often taking the format `[CTF Name]{[Flag Text]}`. In this case, the `Flag Text` contains a
challenge-specific string, typically taking one of two formats:
A random string of text: This has the benefit of being un-guessable and requiring
participants to solve to find the entire flag rather than a partial string. In some
challenges, participants may be able to recover half the flag easily, but not the whole
flag.
A word or phrase: In some competitions, the flag text is a short phrase, often a joke or
a play on words involving the challenge and its solution. While enjoyable for
participants, these can be susceptible to guessing, particularly where participants are
able to get part of the flag and infer the rest.
2.1.2 Scoring
Each successful challenge completion results in the participant gaining access to a flag [6]. This
flag can be submitted to a scoring system which then rewards the participant with some number
of points. The number of points rewarded varies according to the rules of the competition, and
may be affected by:
A base score for the challenge, which may be constant across challenges or selected
by the organiser to reflect the perceived difficulty of the competition.
The time taken to solve the challenge, with challenges decreasing in value as time
goes on; in some cases, special challenges may be released that must be solved
within a given period.
The number of teams that have already solved the challenge. This is mostly commonly
a reward for the first team to solve that particular challenge (sometimes referred to as a
"first blood" award), but may be a decreasing amount of points awarded as more
teams solve the challenge.
The number of teams that ever solve the challenge. Many CTFs award the final points
based on the number of teams that have solved the challenge by the end of the
competition. This means that the value of a challenge varies over time (including for
teams that have already solved it). By doing this, the number of successful solves of a
challenge acts as a proxy for the difficulty of the challenge, and teams are rewarded
more for solving challenges that fewer teams solve.
8
CTF EVENTS
May 2021
In addition to awarding points, some competitions award prizes to the first solutions of particular
challenges; this is often done in conjunction with sponsor-provided challenges, who will provide
a prize for the team that is first to solve the challenge.
Each team's score is the sum of their awarded scores for each challenge, and the winning team
is decided by the highest score at the end of the competition.
2.1.3 Discussion
Jeopardy CTFs offer an excellent platform for engaging participants of all levels. Because
participant teams interact with the challenges independently of each other, participants are not
blocked from solving problems by the success of other teams. As a result, even the lowest-
scoring teams are able to engage with, and benefit from, the experience. The wide variety of
categories and difficulties makes it easy for organisers to ensure that all participants are catered
to, irrespective of background or skill level. The flexible nature of challenges also make these
competitions suitable for a variety of time-frames, particularly when participants are spread
across time-zones. Participants can start and stop their participation during the contest as
required, with no pressure to participate at te same time as other teams. Teams can distribute
tasks amongst team members, either as individuals or groups, and approach multiple problems
at once.
In contrast to other formats, however, Jeopardy competitions are more gamified and less
representative of realistic security skillsets, although this can largely be mitigated by organisers
choosing suitable challenges. When teams participate in Jeopardy contests it is possible (and
not uncommon) for each member to tackle different categories of challenges. Consequently, the
co-operation between team members may be minimal and there may be little interaction
between team members during a competition.
2.1.4 Variants
There are a number of Jeopardy variants currently in use, with varying degrees of popularity
9
CTF EVENTS
May 2021
2.1.4.3 Quiz-style
Some competitions adapt the Jeopardy format slightly by asking questions about the challenges
to participants which must be answered to score points. These may be objective, and judged
automatically, or subjective and be given to a jury to award points.
2.2 ATTACK-DEFENCE
In an Attack-Defence CTF [2] [5], teams are given access to a set of target hosts [7]. The
objective of each participant team is to take and maintain control over as many of the target
hosts as possible. To enable this, challenge organisers will deploy or create a range of
vulnerable services, ensuring that each target contains one or more vulnerabilities.
Teams must balance the need to attack other hosts and accrue more points, with the need to
patch vulnerable services on hosts they already control -- preventing other teams from
compromising those hosts instead. This was the earliest CTF format, having grown from
'wargame' activities in military and hacker communities. Because of the complexity required in
setting up and running events of this format, together with the comparatively-high security risks
involved, Attack-Defence CTFs are more common for invitational or private events, and
infrequent in public events.
2.2.1 Format
In this format, participants are tasked with successfully compromising (and subsequently
securing) target servers. Each server contains one or more vulnerable services, which may be
based either on real-world vulnerabilities or novel vulnerabilities created by the organisers.
Participants in an Attack-Defence CTF are tasked with performing compromises on systems
designed to look and act like real servers running real services.
Attack-Defence CTFs are also unusual in that participants are expected to hold control of the
target, and so to perform defensive actions such as patching or mitigating vulnerabilities. Teams
may be expected to deploy specific patches to vulnerable software, which might range from
updating off-the-shelf vulnerable software, through to writing and applying patches directly to
custom services. They may also be expected to perform general network-hardening measures,
such as updating firewall rules, resetting or strengthening passwords, and disabling unwanted
or untrusted services or users.
Teams may begin the competition already in control of some or all of the target hosts. In some
styles, every host is always under the control of its original owner, and teams are rewarded for
repeatedly performing exploits against other hosts over time, encouraging teams to fix their
vulnerabilities to prevent attacks from gaining further points from that attack.
2.2.2 Scoring
Points are typically awarded on a regular interval (for example, every minute), with each team
receiving a certain number of points for each host they control at that moment. This encourages
participants to compromise, and subsequently protect, the servers in the contest. Competition
rules typically require services to remain active and available in order for points to be awarded,
to prevent teams from simply disabling vulnerable services.
2.2.3 Discussion
Attack-Defence CTFs offer a very practical model of real security scenarios, with participants
gaining experience of both red-teaming and blue-teaming. Competitions in this form are often
less artificial or gamified, particularly in contrast to the more-popular Jeopardy contests. Attack-
Defence competitions are more suitable for spectators and live events, as observers can
witness the changing control of servers throughout the event. Participants, tasked with a more
rounded set of objectives, must also manage their time, splitting their attention between seeking
10
CTF EVENTS
May 2021
targets, compromising them, and defending their own servers. This makes them especially
suitable to team events, particularly when building or testing team cohesion is desired.
Events in this format can be more daunting to new-comers and have a higher barrier to entry.
Because of the requirement to fit challenges into the format of a practical vulnerability in a
service, challenge designers have less freedom and so are less able to make challenges that
cater to beginners. Compounding this, the directly-adversarial nature of Attack-Defence
challenges, where participants seek to directly compromise the servers of other teams, have
significant consequences when participant skill levels are imbalanced; overly-capable teams
have the ability to quickly overtake servers and make them all-but-impenetrable to the other
teams, resulting in contests that are decided very quickly. As participants can only make
progress by capturing the servers of other teams, it is possible that some teams are then unable
to perform any successful exploits during the entire contest.
Establishing and running successful Attack-Defence CTFs presents more difficulties than that of
formats such as Jeopardy. Participants are granted access to the target hosts, usually via a
Virtual Private Network (VPN) architecture, with the aim of isolating malicious traffic and
preventing accidental attacks on non-targets. Despite these precautions, participants must still
take care to target their exploits onto strictly in-scope targets.
2.2.4 Variants
2.2.4.2 Wargames
Wargames are closely related to Attack-Defence CTFs, but are less gamified and have a
stronger focus on capability-building, particularly in the context of training security teams with
realistic experience. Wargames are frequently defender-focused, in contrast to the greater focus
on attackers in Attack-Defence competitions. When red-teams are present, they are usually
considered part of the organisers and not participants in their own right. While these are
significant and important events, they have a different role to most CTF competitions and so are
not typically categorised as CTF events.
11
CTF EVENTS
May 2021
3. REPORT METHODOLOGY
To perform the analysis presented in this report, three sources were used. Firstly, the survey of
members from the International Cyber Security Challenge (ICSC)1 provided overall guidance
towards the events and methodology used throughout the rest of the data gathering. Secondly,
a manual selection, analysis and coding of important events was performed. Details on the
selection criteria and coding used is explained below. Thirdly, to build context into the report
statistical analysis of existing events was performed on a large dataset. For this purpose, the
selection criteria were broader in order to sample as many events as possible.
The core data source for this report is the manual analysis of a set of CTF events. Events were
chosen via selection criteria and then researched via public information available about the
event. Before events were found, a set of key themes and questions within each theme were
determined based around key event information, the specified requirements of the report, the
requirements of the ICSC Steering Committee, and the data likely to be accessible and
practical. Data was initially stored as unstructured text per-question. Once data had been
gathered for each event and question pairing, the results for each question were aggregated
and grouped across events into a set of non-exclusive tags. At this point, the data was
standardised into a database containing the set of appropriate tags for each question and event.
This structured data formed the basis for subsequent analysis and presentation in this report.
1
The International Cyber Security Challenge is a CTF-type of event to be hosted by ENISA in Athens in late 2021. Several
agencies, universities and governmental institutions from all regions participate in the Steering Committee which is
responsible for the organisation of the event.
12
CTF EVENTS
May 2021
were used to invite participants to in-person events); they had been run in-person in previous
years; or they had been intended to run in-person but had been forced to move online to due
disruption from COVID-19. Because of the significant logistical differences between online and
in-person events, data relating to online-only events was considered less applicable to the
purposes of this report. The comparative abundance of online events, however, make them
beneficial when considering challenge formats, which are typically more consistent between
online and in-person events. In particular, the context-building statistical analysis considered
more online events, and the statistical differences between online and in-person events are
detailed later.
3.1.1.3 Significance
Events were selected only if they were significant according to one of the following tests: they
had a considerable number of participants, either as part of their final event or qualifiers; they
were used as a qualification event into a significant event, such as the European Cyber Security
Challenge; they were run, officially or unofficially, as a national-level contest, particularly when
used as a selection method for a national team.
3.1.1.6 Location
Location was not used as a selection criterion and events from anywhere in the world were
included. This was a particular specification of this report, and effort was made to ensure that
events represented locations across the world.
13
CTF EVENTS
May 2021
On top of these documents, information was gathered from other sources published by event
organisers:
Almost every event found contained a public webpage detailing the event, either as a
post-event summary or as a record of pre-event marketing and information. These
websites were a source for general information about events but often lacked specific
details, except when contained within the documents listed above.
In some cases, particularly open-format events, access into the platforms used during
the event was still available. This enabled direct data gathering, particularly when
concerned with challenge formats and information. Where platforms contained a record
of organiser communication during the events, a considerable about of information
across data themes could be gathered directly or by inference from records of public
communications.
In some events, little event information was published by the organisers. This was more
common for events that were not open to participation applications, such as in directly
invitational events. While the majority of these events would fail to pass the stated availability of
information selection criteria, some events had sufficient data published about them from third-
party sources. In particular, news articles reporting the outcome of the event frequently contain
some information about the structure and format of the event.
In some cases, direct experience with the events being researched was available and in these
cases this was used to inform the data gathering.
3.1.3 Themes
Once events had been selected and filtered according to the selection criteria, they were
researched individually through the sources described above. During the research of each
event, data in a specific set of themes was gathered. Furthermore, each theme consisted of a
set of questions. Each question was answered in free-text by researchers during the
compilation process, before any coding was performed. Each theme was established in order to
address the general concerns of this report. Questions were determined by considering the
information that would be most beneficial towards the aim of the analysis and report, with
concern for the practicality of gathering the data via the identified sources.
Questions were gathered into themes in order to aggregate limited information into workable
collections, and to provide structure to the data gathering and analysis process. Consequently,
this report is laid out with these themes in mind, and analysis is broken down by theme to aid
comprehensibility and make clear the connection between gathered data and conclusions
drawn.
Age: Did the event have specific age requirements for participants, either as an upper-
or lower- bound on the ages of participants? Were teams all required to be the same
age? Were teams of different ages put into different competition categories?
14
CTF EVENTS
May 2021
Qualifications: Did the event require participants to hold any particular qualifications?
Were they required to pass any tests, either formal or informal, in order to attend? This
does not include successfully competing at other events.
Gender: Did the event have competition rules designed to encourage female
participation? Did the event have competition structure designed to encourage female
participation, such as grouping participants by gender?
Socio-economic: Did the event have competition rules designed to equalise socio-
economic disadvantages? Did the event make considerations for participants with
socio-economic disadvantages?
Format: Which CTF formats did the challenges come from? Did the event contain only
one format, or contain components of multiple formats?
Challenge Categories: Which categories were challenges drawn from? In the case of
Jeopardy-style events, which categories were challenges from? In the case of Attack
Defence-style events, which aspects of Attack-Defence competitions were part of the
competition?
Scoring: What scoring methods were used for scoring participants? Were there
multiple ways to score points through challenges? Were there any non-challenge ways
to earn points towards winning the competition?
Platform: What platform was used by the event? Was the platform custom-made?
Was the platform an existing off-the-shelf solution?
Prizes: Was there a prize for the winning team or teams? Were there prizes for
particular challenges? Were there prizes for other parts of the competition, such as
providing challenges or writing write-ups?
Length: How long was the competition period? Was it broken up into multiple period,
or continuous? If the event was online, were participants able to begin their
competition period on-demand?
15
CTF EVENTS
May 2021
Team Size: Was there a maximum team size? Was there a minimum team size? Were
teams formed by participants in advance of the competition or formed by the
organisers after selection of participants?
Mentors and Coaches: Were teams allowed a mentor or coach? Were teams required
to have a mentor or coach? Were teams assigned a mentor or coach by the
organisers?
Qualifiers: Did the event have a separate qualifier round? Did participants have to
achieve the top scores in the qualifiers to compete, or was the qualifier only part of the
selection process? Was the qualifier a specific event run by the same organisers or
was it based on other events run by different organisers?
Parallel Contests: Were there other competitions, other than the primary competition,
running at the event? Were they targeted at a different audience? Did they have a
different theme or challenge categories? Were they an extension of the main
competition with further challenges of the same type?
Online or In-Person: Was the event run entirely online? Was the event entirely in-
person? Was the event a mixture of both, such as having an online portion leading into
an in-person event? Was the event simultaneously in-person and online, for instance
to cater to different groups or to host different formats? Did this differ from previous
events, in particular because of COVID-19 restrictions or concerns?
Challenge Providers: Did the organisation create all challenges? Did the event use
existing challenges, such as from a platform or challenge provider? Were participants
expected to provide challenges?
Other Activities: What other activities were organised by the event, in addition to the
competition? Were there any social activities? Were there sponsor-led activities? Were
there activities designed to benefit participants, such as career events?
Catering: Was the event catered by the organisers? Was catering only during the
competition period or during the entire event period, such as dinners after competition
days?
16
CTF EVENTS
May 2021
Transport and Accommodation: Did the organisers provide funding for transport to
the event? Did the organisers provide funding for accommodation expenses for
participants?
Challenge Distribution: Did organisers release the challenges publicly once the
competition was over? Did the organisers explicitly prevent participants from sharing
challenges themselves?
Solution Distribution: Did organisers release solutions publicly once the competition
was over? Did the organisers explicitly prevent participants from sharing challenge
solutions themselves? Did the organisers actively promote participants who created
write-ups, for instance by having prizes or featuring solutions?
Data Release: Did organisers release data gathered during the competition? Did
organisers publish participant statistics, such as participant numbers or breakdowns?
As part of the tag creation process, data for each question underwent aggregation into suitable,
granular categories. This was performed to facilitate later analysis and to avoid over-specificity
in cases where considerable amounts of data were present. This was particularly important
where events had very little available data.
In some cases, multiple question groups were aggregated together to provide more content
across all events. The rest of this report is structured around discussion of each theme, and
data is presented in these aggregated groupings so that each grouping provides sufficient detail
for meaningful analysis and discussion.
17
CTF EVENTS
May 2021
performance across multiple competitions. It is also public and free-to-access, both for readers
and CTF organisers, such that announcing a CTF event is accessible even with minimal
financial resources. As such it is a de facto publication platform for CTF events and therefore a
data source that is likely to be accurate and complete. The website also offers an open API to
allow analysis of the data.
The dataset was compiled from data retrieved via the open API and cover the period
01/01/2015 -- 06/12/2020; including data on events and contained challenges.
From manual inspection, event records were overwhelmingly complete and consistent, with only
a small number of empty or invalid entries. Challenge records were less complete, with only
approximately 73% of events having challenge records associated to them. Challenge records
would only be expected for events in a Jeopardy format, but records were incomplete for even
this subset. Nevertheless, the data were considered to be easily sufficient.
Event
18
CTF EVENTS
May 2021
4. ANALYSIS RESULTS
The analysed CTF competitions were variously organised by governments, universities, for-
profit companies and community groups (see Figure 1). Government events were organised
either by national-level governments or supra-national bodies (such as the EU bloc), with no
local-government events in the analysis. University-run CTFs often had government support,
either through partnerships or via research-body funding.
Some university events sought to perform research in cybersecurity education, using the event
as a study, and publish findings as research papers. In other cases, the delivery of that
education and the promotion of cybersecurity careers was the main goal. Of the four
commercial events, three were operated by large, well-known technology companies, while one
was run by a cybersecurity recruitment agency. Community events were often, but not always,
attached to security conferences.
19
CTF EVENTS
May 2021
The majority of the analysed events were intended either for the general public or for students in
tertiary education (see Figure 2). However, a small number specifically targeted school-aged
children to promote early cybersecurity education. The structure of these varied; with one
restricting entry solely to children, while others were open to wider age groups in another stream
of the competition. Most public CTFs were open to wide participation among hobbyists and
professionals. Four competitions were noted to target only skilled professionals in the area,
either with specific entry restrictions or de facto due to low entrant limits and high challenge
difficulties.
20
CTF EVENTS
May 2021
21
CTF EVENTS
May 2021
22
CTF EVENTS
May 2021
23
CTF EVENTS
May 2021
Popular challenge categories were represented across the examined events, with Crypto,
Exploitation, Forensics, Web and Reverse Engineering commonly appearing (see Figure 11).
As noted above, defensive categories were also well-represented, both as defence against
targeted attacks and defence against malware. Where reported, major CTFs typically operated
custom platforms (Figure 12), although external hosted services were occasionally seen
(indeed, the same HackingLab hosted platform was used in two cases).
It is suspected that popular open-source platforms were utilised for these custom
arrangements, rather than software developed from scratch, although hard data was not
available to support that. Scoring was primarily seen to be fixed, with specific point values given
for Jeopardy challenges or successful Attack-Defence captures and holds (Figure 13). Most
variants upon scoring included either modified point values, an element of manual grading or
additional points for special cases (such as 'King-of-the-Hill' in Attack-Defence). In one notable
case, Qihoo 360 WCTF operated primarily on a Jeopardy model, but supplemented the main
solution points with an additional round in which solutions were evaluated by a technical jury
and awarded bonus points.
24
CTF EVENTS
May 2021
25
CTF EVENTS
May 2021
It was found that winners' prizes were commonly awarded (see Figure 14). These were either
items of consumer technology, cash prizes, or invitations to other prestigious events. There was
wide variation in this rule, however. The Pwn2Own competition operates as a live bug bounty
event, in which successful attacks have direct commercial applications, and as such carries
cash prizes up to $80,000 (USD).
By contrast, the very well-respected, DEF CON CTF offers prizes with low monetary value -- but
enormous prestige in the community. Ancillary prizes were occasionally awarded for contributed
challenges and stand-out actions ("the je ne sais quoi award" in Cambridge2Cambridge), but
these were not common (Figure 15).
26
CTF EVENTS
May 2021
Most events were short, with 2--3 day CTFs marginally more popular than single-day CTFs
events (see Figure 16Figure 16). As many of the analysed events were large, in-person
competitions; operating short, focused events is understandable -- both in terms of cost and
available participant time. Online competitions typically ran for longer periods; consistent with
being background or hobby activities rather than full-time pursuits.
27
CTF EVENTS
May 2021
28
CTF EVENTS
May 2021
From those analysed, more events incorporated qualifying rounds than did not, although this is
skewed by the selection of renowned competitions and 'finals' events with limited memberships
-- open competitions typically have far more entrants (see Figure 18). Competitions operating
in-person or online were equally matched (Figure 19). However, in 2020, a handful of events
were moved online due to COVID-19, despite otherwise being held in-person. Events were split
between allowing a mentor with a team and prohibiting this (Figure 20). Where allowed,
mentors were variously an accompanying adult (for children), an employer or a prior competitor
at the event.
CTF organisers typically provide at least technical support for entrants, with hints being provided
in some cases. Few CTFs made a concrete statement of the communication policy and the
availability of hints was very rare (Figure 22). A handful of events operated parallel competitive
contests alongside the main challenges (Figure 21) and these were either targeting another
audience (i.e., a student tier of a professional competition) or adding additional challenges of a
different type (e.g., hardware attacks, physical security challenges or a social engineering
exercise during social events).
29
CTF EVENTS
May 2021
30
CTF EVENTS
May 2021
Where information was available, CTF challenges were mostly provided by event organisers,
with a small number incorporating challenges from sponsors (see Figure 26). A handful of
events asked participating teams to produce challenges, or used those provided by a hosted
platform.
31
CTF EVENTS
May 2021
Events were advertised over a range of media (see Figure 27). Events usually (although not
universally) hosted a public website. Twitter was used nearly as frequently to disseminate
information. Live updates, via Twitter, blogs, Discord channels or a video feed were a common
feature. Live streaming of competitions is rare, but seen both for in-person events (National
Collegiate Cyber Defense Competition, Qihoo 360 WCTF) and one online event (GoogleCTF).
In the latter case, it was a competition requirement for finalist teams to stream their work from a
designated computer and these streams are available on YouTube.
32
CTF EVENTS
May 2021
4.2.6 Post-event
In the examined events, the amount and type of information released by CTF organisers was
varied. Challenges and solutions were officially released for 28% and 23% of events,
respectively (see Figure 28, Figure 29). However, these numbers should be considered with
some context. Challenges accessed after a competition were sometimes broken (e.g., where a
webservice must be attacked, this was often found to be unavailable), while unofficial solutions
(in the form of participant write-ups) were often plentiful even if no official solutions were
released.
Events with an education focus, rather than an entertainment focus, sometimes released more
detailed data (see Figure 30), or even published academic papers about the design and
operation of the event (see Figure 31). Both were rare overall, however.
33
CTF EVENTS
May 2021
The number of CTF competitions has grown every year, even reporting the highest numbers in
2020, despite the COVID-19 pandemic (see Figure 32). Competitions are primarily conducted
online (>73%), but in-person events take place all over the world (Figure 33). Figure 34 shows
the locations of in-person events.
34
CTF EVENTS
May 2021
The majority of events took place over either a single day, or a few days (see Figure 35). A
notable benefit of short multi-day events is to minimise the effect of time-zones to promote
worldwide participation. Nevertheless, even the predominantly-online CTF Time events were
still mostly bounded to only a small multi-day duration.
35
CTF EVENTS
May 2021
As with the events studied in the manual analysis, the vast majority of competitions from the
CTF Time dataset use a Jeopardy format (>87%), over Attack-Defence or Hack Quest formats
(see Figure 36). Again however, Attack-Defence is more popular for in-person events;
accounting for nearly 25% of the total events, while online competitions only use an Attack-
Defence format approximately 5% of the time.
The competitions listed on CTF Time overwhelmingly specify no entry criteria; with all teams
welcome to apply (see Figure 37). A small number are restricted to (high-) schoolchildren or
university teams, but this is less common than in the manual analysis. A handful of
competitions, usually larger and more renowned events, apply a prequalification criterion. This
is typically an initial public competition, from which the best-placing teams are selected for
invitation to the main competition. Alternatively, teams may be scouted among other notable
competitions and invited without a direct qualification round. While the use of qualification
rounds was seen often in the manual analysis, it is far rarer in the full CTF Time data, likely due
to the logistical effort required and the focus on accessibility, over exclusivity and performance,
that is common of community events.
In the CTF Time difficulty weighting, events were significantly clustered around a low-to-medium
difficulty level (approx. 25), although with instances of events having nearly every difficulty
weighting. This is visualised as a histogram of weightings in Figure 38. This is consistent with
36
CTF EVENTS
May 2021
the focus on accessible challenges that can be enjoyed by a range of participants, along with
the likelihood that many community organisers cannot devote sufficient resources to create very
challenging tasks as part of their events. Many events included in the manual analysis do not
appear on CTF Time (particularly government events), however those that do have an average
weighting of 53.05 and include DEF CON CTF, HITCON CTF and PlaidCTF, which all have
weightings over 90 in recent years.
37
CTF EVENTS
May 2021
5. DISCUSSION
This section contains a discussion of the results found in the course of this report and their
relevant for organisers of future events.
5.1.1 Popularity
The data gathered in this report shows that CTFs are a hugely, and increasingly, popular event
format. CTF events are carried out all over the world, with participants from a range of age
groups and skill levels. The number of events has consistently grown year-on-year since the
earliest year in the data gathered. Recommendations of the educational benefits of security
competitions2 have resonance with the success of events aimed at school- and university-age
groups. The data gathered in this report suggests that there are more events targeted at these
groups than at professionals; which suggests the value of CTFs as introductory and skill-
building opportunities. CTFs have also been used to encourage the traditionally-
underrepresented group of female participants. Some events were seen to offer benefits to
teams with female members, although this was not a popular approach. Indeed, in the authors'
experience of one such event, their female colleagues felt this practice commented negatively
on their contribution. In other cases, gender diversity measures focused on providing access to
role models and tailored mentorship as peripheral activities, rather than employing measures
that altered the competition.
2
Chothia and Novakovic, ‘An Offline Capture the Flag-Style Virtual Machine and an Assessment of Its Value for
Cybersecurity Education’.
38
CTF EVENTS
May 2021
introduction to the CTF format is highly productive and can enable participants to gain
experience from avenue that they may be otherwise have been dissuaded from using.
5.2.1 Format
While Jeopardy is the dominant format both online and in-person, Attack-Defence events have
had a persistent presence since the earliest events. Mixed-format events, containing both
Jeopardy challenges and Attack-Defence targets provides a middle-ground that combines the
benefits of both styles. In the manually-gathered dataset, only one of the events surveyed was
purely Attack-Defence without any Jeopardy components. The considerably higher costs
associated with setting up and managing Attack-Defence infrastructure may be a considerable
factor in the decisions of event organisers. Further, the comparative difficulty of scaling-up
Attack-Defence contests may further explain their lack of use in online formats, which typically
have larger participant numbers. Attack-Defence may also be associated with higher skill
requirements for participants, as their less-gamified structure may increase the knowledge and
tools needed to successfully perform compromises. This is somewhat supported by the
participant-perceived difficulty of the events, with Attack-Defence events having the highest
average difficulty, followed by Jeopardy. Hack Quests, were on average considered to be
significantly easier than standard Jeopardy, suggesting their typical use as highly-structured
introductory events for new players.
Challenges are predominantly technical, and while some events mentioned non-technical
categories, the survey was not able to find any specific instances of non-technical challenges.
Some examples from prior experience include social engineering challenges, such as
interacting with an automated email service to send it phishing emails, lockpicking and
interacting with other teams to gather all the parts of a flag. Physical security challenges such as
lockpicking are sometimes incorporated into the main, scored competition, while others are
conducted as an unscored peripheral activity (such as a guided exercise, from the authors'
experience, in which participants performed a physical wiretap of a network cable).
39
CTF EVENTS
May 2021
40
CTF EVENTS
May 2021
5.4.1 Qualification
Few CTF events had explicit requirements on who could compete, aside from educational
status; one event required participants to pass a short online test that examined technical skills
and English-language proficiency. One event, which was on the borderline between a wargame
exercise and a CTF, was invitational and only included participants who worked professionally
as cyber incident responders. Roughly half the events used a qualification round to filter
participants down for final events. In most cases, qualification rounds were run by the same
organisers and were often online or run in a decentralised fashion, with individual schools or
universities hosting qualifiers. In some cases, events that are significant in themselves are used
as qualifiers to a future event, and successful winners of the main event go on to participate in a
further competition. This was particularly prevalent in the dataset due to selection biases, and
many surveyed events act as national qualifiers for the ECSC. In the case of DEF CON,
participant teams were able to join from successful participation in the DEF CON qualification
round or by winning other selected other events (including the previous year).
5.4.5 Writeups
CTF culture places significant importance of the creation, study and dissemination of post-event
writeups. Each writeup explains the details of a challenge and its solution from the perspective
of a participant or team. These are widely shared and repositories exist that collect them
together (for instance, CTFTime allows users to submit writeups for challenges once a contest
is ended). This is reinforced by some event organisers, who encourage (and sometimes reward)
participants who submit high-quality writeups to them. In some instances, teams must supply
writeups to challenges they solve in order to be presented with their prizes. This may act as a
deterrent to cheats, or may ensure winning teams have a thorough understanding of the
challenges they have solved. Organisers may also release solutions to their challenges
41
CTF EVENTS
May 2021
themselves so that participants can understand the technicalities behind the challenges after
the event. On the other end of the spectrum, some competitions explicitly forbid the release of
writeups or the sharing of challenge files after the event. This may be intended to enable them
to re-use challenges in later years, although it is unlikely that determined participants would be
unable to find copies of the solutions to previous years. In the manual dataset, we found no
direct evidence that any of the events surveyed forbade sharing of either challenges or
writeups.
42
CTF EVENTS
May 2021
6. CONCLUSIONS
In this section, we provide the following recommendations for organisers of future competitions.
These conclusions are informed by the analysis and discussions presented in this report. These
recommendations are made in the context of the survey results. Our conclusions are as follows:
Teams may expect to have a mentor or coach present, particular when younger participants are
involved. In cases where a mentor is involved, it may be important to provide clear rules and
roles for them. In particular, their communication with their team during the event may give an
unfair advantage to competitors based on the expertise and willingness of their mentor. As a
result, we suggest that mentor roles are clearly defined, if included at all.
43
CTF EVENTS
May 2021
scale well into larger competitions with higher skill levels and suggest that objective scoring
systems have higher integrity.
Equally, while some events include participant-submitted challenges, these require a distinct
skillset to that of typical CTF participation. Including participant-submitted challenges also
creates an uneven playing field, where each team has a different set of challenges to solve and
consequently this may somewhat dilute the competitive aspect. Sponsor-provided challenges
provide a potential opportunity for more competitive challenge variety. In these instances, care
should be taken that challenges fit with the format and style of the rest of the competition; in our
experience, unrestricted sponsor challenges can fit poorly with the rest of the competition,
possibly due to sponsor inexperience with the format.
6.8 POST-EVENT
To further encourage and engage the community beyond the event itself, organisers may wish
to support participants (and non-participants) who wish to go over the challenges in their own
time. This has two primary avenues; firstly, making challenges available after the event, and
secondly, ensuring that writeups or solutions for challenges are available. Organisers may not
44
CTF EVENTS
May 2021
need to do a great deal to facilitate this, as the community has a strong tradition of publishing
writeups for most events. However, organisers who go further to facilitate this, such as by
amplifying or sharing participant writeups, or by providing awards to the best writeups, may
compound the benefits. Releasing challenges after the event is also beneficial as they can be
used as a training and teaching tool, particularly for participants seeking to prepare for future
iterations of the event. Some thought may need to be given to this in cases where challenges
involve servers, and this may not be possible in cases where hardware is involved.
In addition to supporting the spread of challenges and writeups, organisers should consider the
beneficial impact they can have on the academic community. This could be performed indirectly,
by sharing data about the competition, such as demographic data on participants, or about the
challenges themselves. Such data is of particular interest in Attack-Defence formats, where
attack traffic data may closely mirror real-world attack traffic. Further, organisers may wish to
follow the example of other events and directly publish their conclusions that follow from their
own experience running an event.
45
CTF EVENTS
May 2021
7. BIBLIOGRAPHY
46
CTF EVENTS
May 2021
A ANNEX:
CTF EVENT PROFILES
This annex provides short profiles on each capture-the-flag event included in the manual
analysis:
US Cyber Challenge
URL: https://www.uscyberchallenge.org
Organiser Type: Government
National US programme designed "to identify, attract, train and recruit the next generation
of cybersecurity professionals". Run as a public-private partnership.
47
CTF EVENTS
May 2021
ACSC-ASEAN
URL: https://www.cyber.gov.au/acsc/view-all-content/news/acsc-asean-strengthening-regional-
cyber-security
Organiser Type: Government
Australian-organised event, inviting competitors from the Association of South East Asian
Nations (ASEAN), with a specific focus on cybersecurity professionals and a stated intention
to promote collaboration.
PicoCTF
URL: https://picoctf.com
Organiser Type: University
US online competition with focus on school-age children, intended to promote cybersecurity
education. Also used as a research platform to develop effective skills training.
Pwn2Own
URL: https://cansecwest.com
Organiser Type: Community
Exploit-finding competition for popular consumer software and products; essentially a live
bug-bounty programme. Attached to the CanSecWest conference.
Country 2 Country
URL: https://www.c2c-ctf.org/
Organiser Type: University
Joint event organised by InterNational Cyber Security Center of Excellence (INCS-CoE)
members, aiming to host five CTFs over five years (in the UK, Israel, USA, Japan and
Australia). Developed from the Cambridge2Cambridge events. Intended to promote
international collaboration and develop cybersecurity skills in university students.
48
CTF EVENTS
May 2021
Global Cyberlympics
URL: https://www.cyberlympics.org/
Organiser Type: Commercial
International competition run by the EC-Council Foundation, with stated goals of: "Capacity
Building, Raising Awareness, Global Peace & Child Online Protection".
PlaidCTF
URL: https://play.plaidctf.com
Organiser Type: Community
Public CTF run by the PPP CTF team attached to Carnegie Mellon University
HITCON
URL: https://ctf2020.hitcon.org/
Organiser Type: Community
Public CTF organised by Hacking-in-Taiwan. Attached to the HITCON conference.
Google CTF
URL: https://capturetheflag.withgoogle.com/
Organiser Type: Commercial
Public CTF run by internet company Google Inc.
49
CTF EVENTS
May 2021
B ANNEX:
MANUAL ANALYSIS DATA
TABLES
Government
Commercial
Community
University
Event Name
50
CTF EVENTS
May 2021
Professional
University
Unknown
Children
Public
Event Name
B.3 AGE
Young Adult
Young Teen
Elder Teen
Unknown
Adult
Event Name
51
CTF EVENTS
May 2021
Professionals
No restriction
Unknown
Invitation
Students
Children
Cyber
Event Name
Language Skills
Technical Skills
No restriction
Event Name
52
CTF EVENTS
May 2021
Region Within
No restriction
Countries
Country
Country
Multiple
Event Name
B.7 GENDER
Categorisation
No restriction
Participation
Gender
Female
Benefit
Event Name
53
CTF EVENTS
May 2021
B.8 SOCIO-ECONOMIC
Considerations
No restriction
made
Event Name
54
CTF EVENTS
May 2021
B.9 ETHNICITY
restriction
Unknown
No
Event Name
Jury Evaluation
Attack/Defence
Exploit Finding
Vulnerable VM
Questions
Unknown
Jeopardy
Patching
Defence
Event Name
55
CTF EVENTS
May 2021
Recon/Opsec/Intel
Physical Security
User-Submitted
Human Factors
Administration
Hash-breaking
Programming
Virtualization
Pen Testing
Automation
Forensics
Hardware
Networks
Patching
Malware
Defence
Privacy
Exploit
Crypto
Mobile
Attack
Misc
Web
IoT
OS
Re
Event Name
56
CTF EVENTS
May 2021
B.12 PLATFORM
Hosted Service
Unknown
Custom
Event Name
57
CTF EVENTS
May 2021
B.13 SCORING
Manual Grading
Based on Solve
Miscellaneous
Attack Points
KoTH Points
First Blood
Unknown
Count
Event Name
Sponsor-
related
Yes
No
Event Name
58
CTF EVENTS
May 2021
Best Submitted
Challenge
Unknown
Other
No
Event Name
Weeks or
Longer
Event Name
59
CTF EVENTS
May 2021
Minimum
Unknown
1.0
2.0
3.0
4.0
5.0
6.0
Event Name
Maximum
Unknown
10.0
1.0
3.0
4.0
5.0
6.0
8.0
Event Name
60
CTF EVENTS
May 2021
B.18 QUALIFIERS
Other Contests
Invitational
Yes
No
Event Name
In-Person
Online
Event Name
61
CTF EVENTS
May 2021
B.20 MENTOR/COACH
Adult/Teacher
Competitors
Employer
Unknown
Previous
No
Event Name
Different
Theme
Extra
No
Event Name
62
CTF EVENTS
May 2021
Unknown
Platform
Support
Hints
Help
Event Name
Recruitment
Unknown
Briefings
Social
Meals
Event Name
63
CTF EVENTS
May 2021
B.24 CATERING
Unknown
Yes
No
Event Name
B.25 TRANSPORT
Unknown
Yes
No
Event Name
64
CTF EVENTS
May 2021
Participant-
Organisers
Unknown
Platform-
Sponsor-
Provided
Provided
Provided
Event Name
Telegram
CTFTime
Website
Discord
Twitter
Flickr
Blog
IRC
Event Name
65
CTF EVENTS
May 2021
Unknown
Yes
No
Event Name
No
Event Name
66
CTF EVENTS
May 2021
Unknown
Yes
No
Event Name
No
Event Name
67
TP-02-21-396-EN-N
ABOUT ENIS A
The European Union Agency for Cybersecurity (ENISA) has been working to make
Europe cyber secure since 2004. ENISA works with the EU, its member states, the
private sector and Europe’s citizens to develop advice and recommendations on
good practice in information security. It assists EU member states in implementing
relevant EU legislation and works to improve the resilience of Europe’s critical
information infrastructure and networks. ENISA seeks to enhance existing expertise
in EU member states by supporting the development of cross-border communities
committed to improving network and information security throughout the EU. Since
2019, it has been drawing up cybersecurity certification schemes. More information
about ENISA and its work can be found at www.enisa.europa.eu.
ISBN: 978-92-9204-501-2
DOI: 10.2824/313553