Speedrunning is when someone plays a video game or part of a video game to finish it as quickly as possible. This activity often includes using planned paths, which may involve doing parts of the game out of order or using mistakes in the game to skip sections or complete them faster. Tool-assisted speedrunning, or TAS, is a type of speedrunning that uses special software or tools to control the game precisely or slow it down to make movements perfect.

Many online groups focus on speedrunning specific games. Leaderboards that track how fast people finish games are the main way speedrunners compete. Races between two or more speedrunners are also common competitions. Videos and live streams of speedruns are shared online on websites like YouTube and Twitch. Speedruns are sometimes shown at marathon events, which are gaming events where many people perform speedruns in different games.

History

Speedrunning has been part of video games since the beginning, similar to trying to get high scores. It became more popular in 1993. Before this, games like Metroid (1986) and Prince of Persia (1989) encouraged players to track their times when completing certain tasks. The game Wolfenstein 3D (1992) introduced a "par time" feature, which was based on the best times set by John Romero for each level. His best times were also included in the game's hint book, which players could order through the same system used to buy the game. Romero wanted players to try to beat his times.

The strong speedrunning community started with the 1993 game Doom. This game kept the "par time" feature from Wolfenstein 3D and added a way for players to record and share their gameplay using files called demos. These files were small and easy to share on early internet systems. In 1994, a student named Christina Norman created a special server for sharing demos, called the LMP Hall of Fame. This inspired Frank Stajano to create a list of titles for Doom players who completed specific challenges. In 1994, Simon Widlake started COMPET-N, a website that tracked the fastest times for Doom levels.

In 1996, Quake was released as a follow-up to Doom. Like Doom, it allowed players to record demos and attracted speedrunners. In 1997, Nolan Pflug created a website called Nightmare Speed Demos (NSD) to track Quake speedruns. He also shared a full-game speedrun called Quake Done Quick, which helped more people learn about speedrunning. Quake speedruns used movement techniques like "bunny hopping," which later appeared in games like Counter-Strike and Team Fortress. In 1998, NSD merged with another site to create Speed Demos Archive (SDA).

For five years, SDA only hosted Quake speedruns. In 2003, it shared a 100% speedrun of Metroid Prime by Pflug. Later that year, SDA began accepting speedruns from all games. Unlike earlier sites, SDA only showed the fastest speedrun for each game, not leaderboards. Before SDA expanded, speedrun videos were mostly sent to Twin Galaxies, but these videos were rarely shared publicly, making it hard to verify records. This led to the creation of websites where players could share runs, discuss strategies, and work together to improve times.

In 2003, a video of a tool-assisted speedrun (TAS) of Super Mario Bros. 3 became popular online. The video was made by a Japanese user named Morimoto, but it did not explain that it used special tools. Many people thought it was done by a human, and some felt tricked when Morimoto later explained the process. This led to the creation of TASVideos in 2003 by a user named Bisqwit, a site for sharing TAS runs.

The rise of video-sharing and streaming websites in the 2000s made speedrunning more accessible. In 2005, YouTube allowed speedrunners to share videos and discuss strategies on SDA forums. In 2011, Twitch, a livestreaming site for gaming, was launched. Livestreaming helped verify and preserve speedruns, and some speedrunners believe it encouraged more teamwork. In 2014, Speedrun.com was created, with easier rules than SDA, to centralize leaderboards for many games. As speedrunners used Speedrun.com and social media like Skype and Discord, SDA became less important in the 2010s. Tools like speedrunning timers, which compare current and best times, were also developed to help players improve.

Methodology

Routing is an important process in speedrunning. Routing means creating the best possible order of actions and steps in a video game. A route might include skipping parts of the game that are usually needed to move forward. Skipping a required part of a game is called sequence breaking, a term first used in the 2002 game Metroid Prime. Glitches in video games can be used to skip parts of a game, or to skip cutscenes and improve a player's speed or damage. Some people, called glitch-hunters, look for glitches that help speedrunners. In some games, players might use a technique called arbitrary code execution, which lets them add their own code to the game's memory. A type of glitch called a "credits warp" causes the game's credits to play, and this may need arbitrary code execution. Early leaderboards on Twin Galaxies did not allow glitches or sequence breaks, but as speedrunning moved to independent leaderboards, these methods became more common.

A tool-assisted speedrun (TAS) is a speedrun that uses software and tools to create a "perfect" playthrough. According to TASVideos, tools include playing the game frame by frame, using saved game states to retry parts of the run, and editing game data. These tools help remove limits from human reflexes and allow for the best gameplay. The run is recorded as a series of controller inputs meant to be played back in order. While TASes are usually recorded on an emulator, they can also be played on real console hardware through a process called console verification. Some inputs used in TASes, like pressing left and right at the same time on a Nintendo Entertainment System, cannot be done with standard controllers. Unassisted speedruns are sometimes called real-time attack (RTA) speedruns. Because TASes are not played in real time by a human, they are not considered competitors of RTA speedruns.

Speedruns are grouped into categories that add rules for runners. Categories often require completing certain parts of the game. Each game may have its own categories, but some are popular across many games. Common categories include:

• Any%: finishing the game as quickly as possible without any requirements.
• 100%: completing the entire game, which may involve collecting all items or meeting other goals.
• Low%: finishing the game while completing the least amount of content possible.
• Glitchless: avoiding all glitches during the run.
• No Major Glitches: finishing the game as quickly as possible without using glitches that break the game.

Speedrunners compete by setting times on online leaderboards. According to Wired, the main website for speedrun leaderboards is Speedrun.com. As of July 2021, the site lists leaderboards for over 20,000 video games. Runners often record their speedruns for timing and verification, and may include a timer in their videos. Timers track splits, which are the time between the start of the run and completing a section or goal. Verification is usually done by moderators who check submissions and decide if a speedrun is valid.

Community

According to many speedrunners, community is a key part of the hobby. Matt Merkle, director of operations at Games Done Quick, says that speedrunners "value the cooperation the community encourages," and many speedrunners have said that their mental health has improved because of their involvement in the community. Erica Lenti, writing for Wired, said a sense of community is vital to speedrunning because it motivates players and helps develop routes and tricks used in speedruns. Milan Jacevic highlighted "years of research" and collective community efforts that contribute to world records.

Speedrunners use media-sharing sites like YouTube and Twitch to share videos and livestreams of speedruns. The speedrunning community is divided into many sub-communities focused on speedrunning specific games. These sub-communities can form their own independent leaderboards and communicate about their games using Discord. Many communities have used the centralized leaderboard hosting site Speedrun.com since its founding in 2014.

Speedrunning marathons, a form of gaming convention, feature a series of speedruns by multiple speedrunners. While many marathons are held worldwide, the largest event is Games Done Quick, a semiannual marathon held in the United States. As of January 2022, it has raised over $37 million for charity organizations since its inception in 2010. The largest marathon in Europe is the European Speedrunner Assembly, held in Sweden. Both events broadcast the speedruns on Twitch and raise money for various charity organizations. Speedruns at marathons are done in one attempt and often have accompanying commentary. Many people consider marathons to be important to runners and spectators in the speedrunning community. Peter Marsh, writing for the Australian Broadcasting Corporation, says that the Games Done Quick events provide an inclusive space for women and the LGBTQ community in contrast to the related cultures of gaming and Twitch streaming. Alex Miller of Wired says the events have played an important role in connecting people and supporting international humanitarian organization Médecins Sans Frontières during the COVID-19 pandemic.

Races between two or more speedrunners are a common competition format. They require players to be skilled at recovering from setbacks during a speedrun because they cannot start over. Occasionally, races are featured at marathons; a 4-person Super Metroid race is a popular recurring event at Games Done Quick marathons. The Global Speedrun Association (GSA) has organized head-to-head tournaments for multiple games, including Celeste, Super Mario 64, and Super Mario Odyssey. In 2019, GSA organized an in-person speedrun race event called PACE. Their efforts have drawn criticism from some speedrunners who believe that they "undermine the community spirit," citing cash prizes as incentives to avoid collaboration with other speedrunners and ignore games without prize money. Video game randomizers—ROM hacks that randomly shuffle item locations and other in-game content—are popular for speedrun races as well. Tournaments and other events have been organized for randomizer races, and they have been featured at speedrun marathons.

Cheating

Splicing is the most common way people cheat in speedrunning. In speedrunning, a run is usually recorded continuously, but when splicing is used, the run is made from different video clips recorded at different times. Sometimes, these clips include gameplay from other players or from tool-assisted speedruns (TAS).

At SGDQ 2019, a speedrunner named "ConnorAce" used a spliced run to unfairly claim the world record for the "NoAbility%" category in the game Clustertruck. This action prevented the true record holder from being invited to the event. The run was questioned because it was not officially submitted to Speedrun.com, and the video was not publicly shared on YouTube before ConnorAce was accepted into SGDQ. In October 2019, the run was exposed by a YouTube documentarian named Apollo Legend.

In typical splicing cases, difficult parts of a game can be repeated and edited together to create a run that appears continuous. This can greatly reduce the time needed to complete a run. However, splicing is not considered cheating if the runner clearly states the run is made of multiple segments. For example, a community-made multi-segment compilation for Super Mario Bros. is not considered cheating.

When using "TASbotting," a player records their controller inputs as a tool-assisted run on an external device. This device then plays the inputs on a real console. Like splicing, individual segments can be combined. For tool-assisted runs, inputs can be made frame by frame. If these inputs are real and seem like they were done by a human, they are harder to detect than splicing. However, if a TAS is played on an emulator instead of original hardware, it may be noticed because some emulators do not perfectly copy the original hardware, causing problems when replayed on a console.

Changing game timers, especially in computer games, is another common cheating method. This is often easy to spot because top speedruns are closely examined by other players. These checks include a process called "retiming," which compares the recorded time to the time of other runs. This can reveal differences between a run's time and those on the leaderboard.

A variation of splicing includes speeding up cutscenes or shortening black screen transitions. These methods may be detected by moderators, though some games might allow for hardware-related variations.

Another cheating method involves playing the game in slow motion or using frame-by-frame advancement, similar to tool-assisted speedrunning but without the ability to redo inputs. This is effective for games requiring precise movements.

Traditional cheats, like GameShark devices, can be used to improve a character's speed or health. However, these are usually easy for moderators to detect. Modifying internal game files to change random events (RNG) is harder to spot.

One famous example of file modification involved a speedrunner named Dream in 2020. His Minecraft speedruns had outcomes so unlikely that they were suspected of cheating. Experts estimated the chance of these outcomes happening naturally was about 1 in 20 sextillion. Dream admitted to cheating about five months after his runs were rejected, claiming he did not know he was using a modified game. In late 2022, another player, MinecrAvenger, was also found to have used similar methods.

Lying about a time is another simple form of cheating. A well-known example is Todd Rogers, whose records were questioned for being impossible or lacking proof. In 2002, his record for the game Barnstorming was removed after it was found to be unverifiable. His 1982 record for Dragster, which was 5.51 seconds, was later proven to be impossible based on game analysis.

To prevent cheating, some games require a video showing the player's hands on a controller or keyboard ("handcam") alongside the screen recording. This helps moderators confirm that the inputs were made by a human. Other methods include forensic audio analysis to detect signs of video splicing. Runs without high-quality audio are often rejected.

Other detection methods include using mathematics, as seen in the Dream case, or human moderation of inputs in games like Doom and TrackMania. Cheat detection software for TrackMania analyzed over 400,000 replays and found hundreds of cheaters who used slowdown tools. This included a player named Burim "riolu" Fejza, who was later removed from an eSports team after the scandal.