Speedrunning or Playing the Playing the Game

I spend a lot of time watching other people play video games. I like playing video games myself but watching other people play has a cathartic, sublime quality to it. It's how I imagine people feel about golf or snooker. I like seeing how people think their way through problems and make choices and I like, that with increasingly better CGI, watching a video game may as well mean watching an ok movie. Some really skilled players can be clever with the camera and build cinematic narrative into the way they play. Machinima has its origins here and has since expanded to include, for instance, VaatiVidya's lengthy but addictive expositions on theories in the Dark Souls games all set to stunning visuals. In these streams, the 'players' use the game's architecture, its physics and mechanics to create alternative narratives unintended by the games' developers.

These videos can be deeply impressive but an even richer vein of games-like media is found in  the art form of speedrunning. Just as with machinima makers, runners develop a nuanced understanding of the architecture and mechanics of a game but for different purposes. Speedrunning is essentially the act - or art - of completing a game or part of a game in as short a time as possible. In its simplest iterations, this can simply mean being very good at playing the game, accomplished in its principles and behaviours, able to lazily and mechanically respond to the actions of enemies and the environment that have been rehearsed a thousand times so as to optimise the 'run' - the journey from the beginning of the game to the end.

At its best, Speedrunning is responsible for some of the most unfalteringly stunning acts of mastery and showmanship that I've seen in any discipline. And I mean mastery in a very real way. Unlike, for instance, tennis, cricket, football, cycling or any other form of sport or game where mastery is measured by and conflated with just being very good at performing the sport within the confines of the rules, speedrunning rewards 'play' in the truest sense of flexing the edges of the technical construction of the game. Speedrunning is the art of exploitation of simulated environments.

Speedrunning comes in a wide variety of forms, records are held all over the place with Speedrun.com being an easily searchable database of games and records. Games are generally accompanied by a video that shows the run with lists and descriptions of techniques and constraints on the run. A culture has even grown up os intense analyses and breakdowns of specific runs. There are events and festivals, variants on 'GDQ' - Games Done Quick - being the most popular.

Speedrunning challenges the notions of how we 'play' and 'beat' games. Generally, in normal 'play,' these terms are used to mean 'the game was followed and finished within the confines of the way the developers intended.' Speedrunning, on the other hand, challenges the world the developers have built; it abuses and exploits glitches, cuts, tricks, shortcuts and hacks to defy the world that was created in the first place. This is true play, the pushing and testing of the boundaries that structure the world (of the game.) The most well-run games are dissected and broken down to their very code, every element of their construction is poured over by a community of rabid runners looking for any microsecond edge over their competitors. Wikis spring up to log techniques, routes and strategies as well as to debate the finer points of what exactly constitutes a run. Meanwhile, the streams will generate hundreds of thousands, if not millions of views.

The leaderboard on speedrun.com of Legend of Zelda; Ocharina of Time. This twenty year old game is still continually played by speedrunners to shave microseconds off play time. The tabs at the top are different conditions for runs; 'any%' means any level of completion is acceptable as long as the game is finished. This is a common category across most games. 'no IM/WW' means no item manipulation or 'wrong warping' which are techniques specific to the mechanics of Zelda; OoT. 

It may read as piratical and anarchic, but the world of speedrunning is incredibly rule-bound and has fervent and effective testing and validation procedures to insure the communities against 'cheating.'

Speedrun fans will look for evidence of cheating in video and audio tracks that are uploaded. In this case a Donkey Kong player was caught 'splicing' by analysis of the audio track. 

Cheating isn't cheating in the sense that we might understand it by the rules of normal gaming where, for instance, clipping through the map, using terrain to move around boss fights or overpowering your character would be at best considered poor sportsmanship and at worst outright cheating. Cheating in speedrunning is essentially the disingenuous reporting of a run. 'Splicing,' one of the most heinous of cheats, is the editing of the run together from smaller pre-recorded parts. Fans will spend hours pouring over videos to watch for signs of splicing such as skipped frames or mis-matched inventories. In fact, the debates over techniques for detecting splicing in a certain game - which can also mean looking for in-game cues - are as varied as the techniques used for speedrunning. In the table below, the frames of a loading sprite animation in Super Meat Boy which has a regular 40 frame cycle of 'up' and 'down' are analysed from screen to screen to detect any irregularities that might be due to splicing.

60 FPS
60 FPS
60 FPS
60 FPS
60 FPS
Forest Exit
16 Down
17 Down
1 Down
13 Up
16 Up
Hospital Enter
4 Down
3 Down
19 Down
7 Up
4 Up
Hospital Exit
14 Up
16 Up
8 Down
18 Up
15 Down
Factory Enter
6 Up
4 Up
12 Down
2 Up
5 Down
Factory Exit
10 Down
15 Down
10 Down
12 Up
12 Up
Hell Enter
10 Down
6 Down
10 Down
8 Up
8 Up
Hell Exit
4 Up
11 Up
9 Down
6 Down
7 Down
Rapture Enter
16 Up
9 Up
11 Down
14 Down
13 Down
Rapture Exit
20 Up
9 Down
7 Down
15 Up
7 Up
End Enter
21 Down
11 Down
13 Down
5 Up
13 Up

Photoshopping footage is obviously a common cheat with some cheaters changing numbers that appear on screen at the end of levels. This has led to the emergent behaviour of moving the cursor of the console or PC around and over the times shown on screen to show that it isn't photoshopped.

In this video, Karl beats a 15-year-old record for the Dam level in Goldeneye at 52 seconds. There's almost no play in it in the way the game is conventionally understood but a fascinating and dense understanding of its construction to the point where he is following ground textures for a route. Goldeneye was a common source of Photoshopped finishing times for years because of the ease of just copying the numbers on screen. Pixel pattern analysis (as in the below image) is performed at the by fans to detect when numbers have been copy-pasted. 

Another cheat is using an emulator to simulate one piece of hardware on another without reporting it. This can give the significant advantage of faster frame rates on a modern PC or the ability to slow the frame rate down for faster responses which can make all the difference when every microsecond counts.

Then there are nuances on where cheating begins. 'Menuing' for example is when a runner has to perform actions quickly through the game menu. This could be something rudimentary like moving inventory items around or something more sophisticated like rapidly saving and loading to reset an enemy or using a 'quitout' to respawn in a different location. Item manipulation to exploit glitches is allowed as any others are which are permitted by the architecture of the game. However, most western audiences will not allow the use of a turbo controller (a controller that will automate the rapid or continuous pressing of buttons) while Japanese speedrunners are respected for their mastery turbo controller menuers.

Even 'glitchless' runs where none of these tricks are used are up for debate and still causing controversy on 20 year old games to this day. The GIF above shows a technique which is a hot point of contention in Zelda players. It's a 'frame perfect' technique - a technique where a sequence of actions has to be performed to the exact frame in the right order - where there can be as many as sixty frames a second (!!!) - that was previously only thought possible with 'tools' - editing the programming of the game.

Arguments over what contributes 'cheating' in an activity that is about bending rules is responsible for one of the oldest schisms over questions of appropriate methodology in speedrunning: One of the capabilities of early games like Doom and Quake was to record 'demos' - these were essentially recordings of controller directions that could be re-played by other players. These demos made for much smaller files than streaming video at a time when bandwidth couldn't handle the transfer of vast amounts of video footage. Speed Demos Archive began as a forum for Quake players to share demos of their fastest runs through the game. This quickly spun out in to other games including Metroid 2002 where a common trick was to use 'secret worlds' or 'out of bounds' play' to speed up runs. These are techniques that involve the player accessing parts of the game architecture that were never originally intended to be accessed by the game developer including going outside the level or into unfinished or cut parts of the map. However, debate ensued for years on the SDA forums about what exactly constituted being 'out of bounds' with the design of different games leading to vastly different interpretations.

This out of bounds Portal run simply involves using some acrobatics to take a path through the game that players wouldn't normally take were they 'following' the path set by the game...

...while this Nier Automata technique involves clipping through the map to move outside the physics area of the game. 

The leader of the SDA community - Radix - was keen to lay down a common rule set for all speed runs across all games and so outlawed going out of bounds while Twin Galaxies, a more general game records site, allowed individual rules for each game based on its architecture. This became much more popular and is now the standard way speedrunning is done, with the community around each game discussing and suggesting rules for runs and popularly arbitrating on what is considered 'cheating.' Looking through Speedrun shows how each game allows for its own variables. Some games still won't allow certain techniques like glitching because it means that the run becomes basically void. For instance, as in the run below, it's possible to use glitches to complete early Pokemon games in a time of 0.00 which voids any notion of competition and makes a glitched run a pointless exercise. However, the community around it has set variables that still allow for challenges.

The sheer level of accomplishment in the community of speedrunners is staggering, the dedication to iterative play over, and over, and over again to find ever more perfect runs beggars belief. I can gladly extoll the virtues of certain techniques, runners or games for hours but there's something richly and politically nascent about speedrunning that isn't identified by the community. (FYI, I can't speedrun, or anything like it.)

The sophisticated mastery of a system of the type a leading runner accomplishes allows them to explore beyond the bounds of the architecture and mechanics of a game, to see it as an artefact contained by its construction and thus manipulable in all the permitted ways. They change their relationship with the developer and the game and their position in it. Any game is a system of interacting parts that fit together to perform certain functions. Speedrunners understand these mechanics so keenly that they are able to turn the system to new uses, to use the game in toto as their own playground, not one defined by the developers. It reminds me somewhat of a section of the preamble in Georges Perec's Life; A User's Manual:
...puzzling is not a solitary game: every move the puzzler makes, the puzzle-maker has made before; every piece the puzzler picks up, and picks up again, and studies and strokes, every combination he tries, and tries a second time, every blunder and every insight, each hope and each discouragement have all been designed, calculated, and decided by the other.
In speedrunning is a model for interacting with systems, playing them properly, not just operating within their confines as I've previously identified being the failure of contemporary political action. Speedrunners play the playing of the game. As we build a rule increasingly structured by software and the confines of simulations and predictions, a mentality that shifts to re-constructing the rules and playing in a true sense has enormous potential.

February, March, April 2018

I opened up this draft which I hadn't looked at since February and there was a whole load of paragraphs here about the turbulence of my relationship with social media. Essentially I never felt the need to quite or anything dramatic like that because I don't really use it properly. Anyway, all of that became irrelevant in recent weeks when the whole world decided it was going to sit up and have the same debate.

I've been keeping my head down at the moment, I have a course to run and a PhD to do. There's lots of other stuff but those two things are the least glamorous and most important. They essentially require looking at spreadsheets and blinking cursors day after day. It's not pretty but it matters and it's a case of just settling in for a long ride of three or so years to really build something that lasts and is important.

I was in Amsterdam the other week (surprise) visiting the Sandberg for the first time. I was there to give a talk and do some tutorials on a short course but I also got the opportunity to do some spying. I've worked with lots of people before from there but never had the chance to visit. I always like going to other universities and speaking to the staff and students about their problems, they tend to throw my own battles into a new light. The Dutch higher ed sector, particularly in arts and design has been going through some pretty interesting times with DAE students laying the line down on their management and concerns about levels of corporate interest. It's sometimes energising and sometimes disheartening to hear about the types of concerns they express in an atmosphere that is a lot closer to the ideal art school model than the target-driven, marketised and constricted confines of the mega-bureaucracy that I find myself playing. The issue of student representation is one that is brought up again and again in the Netherlands with no proper unionisation or quality assurance rocess involving student perspectives. Here, we're on the opposite ends with a system that has excellent representation of students in all levels of decision making but as a result is sometimes paralysed into inaction out of the fear of upsetting students with change.

Haunted Machines +
I've done a couple of talks recently starting to build on the Haunted Machines work and into my PhD research. There was Arrational Machines at Thingscon and then Pure Machines at Future Sessions in Manchester. That whole things was conducted in the shadow of the Facebook revelations which made a lot of folks decidedly uncomfortable and resulted in some quite heated and enlightening exchanges. I don't know if it was recorded but I think the Q&As throughout the day were very rich. On that note, I'm not sure if I remembered to tell you about a podcast that Natalie and I did last year after Haunted Machines. Natalie now has a newsletter. I thought newsletters were a 2016 thing but then I found out yesterday that wearing a green hat in China mean someone is cheating on you. There's all sorts of things I don't know.

Synthesising Obama 
I did a quick project for Dirty Furniture for an exhibition they're touring. I was asked to chose an artefact that I felt represented the contemporary age and I chose a piece of software, the Synthesising Obama machine learning package. The notion is that these objects will go in time capsules so I had an interesting dilemma of how to make it legible for years to come. Acrylic takes around 40,000 years to degrade so that was an easy choice. The exhibition is travelling to London for Clerkenwell design week so once it's there and I can get some proper photos I'll do a proper writeup.

I've been working on a little research/curatorial project with Z33 since January. They've set up a series of research schools to do projects under certain themes and I've been involved in the School of Time. As part of this they're exhibiting research at Milan this year, I've selected some diagrams and fragments of text that go into a plotter to be drawn into people's notebooks. If you're in Milan go check it out. Also go and check out Wesley's work with Superflux - he's done a new critical data vis project and will be doing a talk. There's also some IDC action at Atelier Clerici this year with Virginie Tan, a recent grad, presenting her work.

Beyond crunking out a good chunk of PhD I've been doing various other bits of writing. I'm in Paris next week for Cumulus to report on the Interact project I've been involved in for the last few year, Strange Telemetry wrote a chapter for Economic Science Fictions from Goldsmiths Press which is now available at all good bookstores and I've got a couple of essays in catalogues and books coming up that I'll let you know about when they're out.

Coming Up 
Natalie and I are running another series of event with Impakt. First on the 20th April in Osnabruck at the European Media Arts Festival and then on the 21st April in Utrecht. The events, both called 'Deep Fakes or Rendering The Truth' are opportunities to talk about the effects of advancing simulation and rendering on notions of truthfulness and objectivity in images. We've got a great lineup of speakers - Anna Riddler, Lucy Hardcastle, Luba Elliot, Igor Schwarzmann and others. Hopefully see you there. After that I'm in Geneva for Mapping Festival where I'll be sharing the stage with Julian Oliver for the first time in four years in a conversation with Daphne Dragona.

That's it. That's all I've got time for. There are probably other things.