SECCON 2016 finals

SECCON 2016 Finals in Tokyo: Format and Results

SECCON 2016 Finals gathered 24 teams in Tokyo for a twelve-hour contest that mixed jeopardy with king-of-the-hill. This guide covers the organizer, the format, the scoreboard top and what the event is like.

Hexfield editorial

OrganizerSECCON
Dates28-29 Jan 2017
FormatOn-site, Tokyo
Teams24 finalists
Event
Byte map of x86 shellcode with scattered high-value opcode bytes, repeated zero bytes and a short run of NOP padding
Shellcode shows a compact, uneven byte map with a NOP run, the kind of pattern exploitation tasks produce.

Organizer and series

SECCON is a Japanese security contest series run by the SECCON organizing committee, which also runs conferences and training under the same name. The competition works in two layers. An online qualifier is open to teams from all over the world, and an on-site final is played in Tokyo. The 2016 contest year ended with the international final on 28 and 29 January 2017 at Tokyo Denki University, listed on CTFtime as event 401.

The numbers explain the standing of the contest. The online qualifier counted 1,834 teams from 99 countries, and only a small group of them reached the final. The on-site field had 24 teams, so the competition was a selection of about one in seventy-six of the qualifier entrants.

Format of the final

The final is a twelve-hour contest with two components. The first is a jeopardy board, which the organizers call the attack part: tasks in categories such as reverse engineering, exploitation, crypto and web carry point values, and a solved task returns a flag. The second is a king-of-the-hill part, called the defense part, in which teams compete to take and hold control of a shared target. The king-of-the-hill tasks are scored over time, so the team that holds a position longest collects the most, and the leader can change several times during the day.

The mix matters for the final table. A jeopardy-only board rewards raw solving speed on tasks that have fixed answers. A hill component adds interaction: teams watch each other’s moves, react to patches and try to dislodge an opponent. Teams that are strong in exploit writing and fast in automation tend to benefit, because a working exploit can be reused against changing targets.

Playing on site

Twenty-four teams share one venue, which was a university campus in Tokyo. As at any on-site final, teams bring their own laptops, the organizers provide the network and scoring system, and the schedule is fixed. Twelve continuous hours is long enough for fatigue to affect the second half, and the king-of-the-hill component asks for attention at the end, when the table is closest. A team that spends its best people on the jeopardy part in the morning needs a plan for who handles the hill in the evening.

The tasks tend to be heavy on binary work. Expect compiled targets for x86 and x86-64, shellcode constraints such as forbidden bytes or short buffers, and custom protocols. For readers who want to practice, a byte map of shellcode and a digram of a small ELF reveal the structure that disassemblers then confirm. The Veles tool is built for this kind of look at raw data.

The published top of the board

CTFtime shows CyKOR in first place with 3,433 points. PwnPineappleApplePwn finished second with 2,234, eee third with 1,899, 217 fourth with 1,798, and binja fifth with 1,469. The sixth-placed team scored 1,464, only 5 points behind fifth. Plaid Parliament of Pwning followed on 1,259 and Dragon Sector on 1,257.

Three observations follow from that table. The winner’s advantage is large, 1,199 points ahead of the runner-up, a margin that probably owes something to the king-of-the-hill part, where scores accumulate over time. Places five and six are separated by a single small margin, and places seven and eight by 2 points, so the middle of the table was decided by very small differences. And teams with a long record in the global ranking, such as Plaid Parliament of Pwning and Dragon Sector, finished below the middle of the final, which shows how much the combined format changes the usual order.

What it is like for a new team

A first-time finalist must prepare for a different contest from the qualifier. The qualifier is a jeopardy that any team can play from home over a weekend. The final needs people who can write a reliable exploit in minutes, who can patch or defend a target and who can keep the team’s tooling stable for twelve hours. A practical plan is to define roles in advance: two people on the jeopardy board, one on the hill, one coordinating and keeping notes. Practice with a king-of-the-hill game before the event, because the dynamics are unfamiliar to teams that have only played jeopardy.

Travel arrangements and visa planning should start early. Foreign teams usually spend the day before the event setting up the network connection and checking the venue.

More on how these events fit into a season is on the team page, and the blog lists related material.

Where the event went later

SECCON continued with a new qualifier and final each year, run on the same two-layer structure. Every edition has a CTFtime page with the date, the number of teams and the scoreboard. The contest has also kept a place in the Japanese security community as an event for students and professionals, with the international final inviting teams from across Asia, Europe and North America.

Questions readers ask

How did teams reach the SECCON finals?

Through an online qualifier that drew 1,834 teams from 99 countries. The best of them were invited to the final round at Tokyo Denki University on 28 and 29 January 2017, where 24 teams played on site. Japanese domestic teams run their own parallel track.

What is the format of the final?

The final lasts twelve hours and combines a classic jeopardy board with a king-of-the-hill component. In the jeopardy part teams solve tasks for points. In the king-of-the-hill part teams compete to hold a target, which adds interaction and changes the order of the table.

How strong was the winning margin?

CyKOR won with 3,433 points, with PwnPineappleApplePwn second on 2,234. That is a gap of nearly 1,200 points, about half of the runner-up's total. The positions from third to eighth are far more tightly packed.

New to capture the flag? Formats, teams and a first-month plan in one place.

Read the CTF guide