The core idea
Speedrunning is the practice of completing a video game, or a defined part of it, as quickly as possible. The goal is a faster time on a clock, judged against an agreed set of rules. Some people do it for personal improvement, some to compete for a place on a leaderboard, and some to perform live for an audience.
What makes it more than “playing quickly” is that the rules define what counts as finishing, which shortcuts are allowed, and how time is measured. Without that, comparing two runs would not mean anything.
Why runs need categories
Different goals produce very different games, so communities split runs into categories. Each category has its own rule set and its own leaderboard. Common speedrun categories include:
- Any%. Reach the end of the game by any permitted means, with no requirement to collect everything. This is usually the fastest category and often involves the most skipping of content.
- 100%. Finish the game after completing everything the community defines as “complete”, such as collecting every item or finishing every side task. The exact definition varies between games.
- Low%. Finish with as little collected or upgraded as possible, which makes the game harder.
- Glitchless. Finish without using certain glitches or bugs, so the route follows what the designers intended more closely.
- Single segment or full-game. Whether the run has to be done in one continuous attempt, or can be stitched together from separate parts.
- Individual levels. Timing one stage or section rather than the whole game.
Many games also have categories built on a particular restriction, such as using only certain characters or never taking damage. Because communities decide the categories, the same game can have different ones on different leaderboards, and a category’s name does not always mean exactly the same thing from game to game.
Rules, timing and proof
Each game’s community normally keeps a written rule page. It specifies:
- Where the clock starts and stops. For example, from the first input after a file is created until the final hit on the last boss, or from selecting “new game” until the credits begin.
- Which version and platform are allowed. Different releases of a game can behave differently, so rules often name the version.
- What counts as cheating. Altering the game’s code or using unapproved tools is generally banned in standard categories.
- What evidence is required. Runners usually submit a recording, sometimes with extra proof such as a view of the hands on the controller or a live stream, and volunteer moderators check submissions.
A related point is the difference between “game time” and “real time”. Some communities remove loading screens from the clock, since load speed depends on hardware and would otherwise unfairly reward one setup over another. Others use real time, because it is simpler to verify. Both approaches exist, and a community’s choice is usually part of its category rules.
How runs get faster
Speedrunners generally combine several kinds of knowledge:
- Route planning. Choosing the order of events that saves the most time. A route can be refined over years as players find improvements.
- Execution. Performing difficult inputs precisely and repeatedly.
- Movement techniques. Using the game’s physics in ways designers did not necessarily plan. Understanding how game physics works helps explain why jumping, sliding or collision rules produce exploitable behaviour.
- Glitches and skips. Taking advantage of bugs to pass through walls or skip sections. These arise because games are complex programs, and some bugs are well understood and repeatable.
- Luck management. In games that use random numbers, runners may learn to influence or avoid unfavourable outcomes. The details depend on how the game’s systems generate chance.
- Practice tools. Saving at chosen points to rehearse a hard section many times; this relates to how save checkpoints work.
The fastest times often come from refinement more than from one trick. A world-record run is typically the result of many players building on each other’s discoveries.
Tool-assisted speedruns
A separate branch is the tool-assisted speedrun, or TAS. Here, players use software to play frame by frame, slow the game down, and rewind, creating a sequence of inputs that is then played back at normal speed. The result can show things humans may not be able to perform live, so TAS runs are usually kept on separate lists from unassisted ones. They are often treated as a demonstration of what a game’s rules allow rather than as a human competition.
Events and community
Speedrunning has a visible community presence. Runners often perform on live platforms, and charity marathon events such as the Games Done Quick series have featured many games being finished in rapid succession. The scene overlaps with game streaming and, to some extent, with competitive play, though it is mostly about beating a clock rather than another player directly.
Communities also tend to document games thoroughly, including maps, glitches and routes. That documentation is valuable to historians and to anyone interested in how games actually work.
Why speedrunning appears in a retro and preservation context
Older games are popular for speedrunning because they are well understood and often have deterministic behaviour. Runs also depend on the exact version of the software and hardware, so the community cares about accurate copies and, in some cases, original consoles and CRT displays. Questions arise about whether game emulation is accurate enough to be accepted for records, and many communities have specific rules about it. In that sense, speedrunning is both a hobby and a kind of living record of how games behave.
A note on difficulty
Speedrunning can look effortless in a polished performance, but runs often fail far more times than they succeed, and learning a category can take a long time. Starting with an unfamiliar game, a simple category and the community’s rules page is a typical way into it.