What is your opinion on Extra Credits’ video on speedrunning? I know you’ve mentioned it before that it was god awful, but could you go into more detail about it?
Goddamnit, I hate having to do this, but fine, I will.
First mistake is EVER mentioning the term “glitchrun”. There’s NEVER rules for whether you’re allowed to glitch past large sections, there’s only ever rules forbidding you from doing so. Skipping things does not require permission, it is allowed by default. Using the term “glitchrun” implies that glitches are the aberration, not the standard.
Doing it all in one sitting should have said single segment, When they said using tools, they should have said frame advance and rerecording instead of emulators and save states, because those are allowed in realtime runs for some (especially old) games that people don’t own consoles for, or any PC FPS game practically, like dishonored and half life in the RTA categories.
Targetting speedrunners probably isn’t as profitable from the marketing side as they make out. The audiences for speedruns, barring the GDQ events, aren’t terribly big.
I’ve actually been trying to sum up what it means to make a good speedrun game much like EC attempted to here, and I’m not impressed with their criteria or suggestions.
I can certainly agree up front that narrowing focus for the speedrun community seems to be a bad idea, both because it’s targetting a small market demographic, and because whenever I see a developer attempt to intentionally make a speedgame, it always sucks, especially as a speedgame, because they focus superficially on speed elements instead of depth, and end up with a flat game for it (like dust force, which has level design way too limiting to really be an interesting speedgame if you ask me, versus Ori and the Blind Forest, which continues to surprise me).
Then he remarks that most speedrun game are, “reflex based precision games” Which I guess is true, but super vague. I mean, doesn’t that include nearly all realtime games?
Also no, you can’t really speedrun something like hearthstone, because that involves other people, same reason people generally don’t speedrun MMOs, or MMO runs can’t really have any sort of standardized ranking chart. Warframe got some runners, but that game is still ridiculously inconsistent.
He’s correct that glitches and shortcuts aren’t what makes speedrunning great, but then he says the key is “tight controls.” This is vague as fuck. This doesn’t mean anything. People speedrun games that don’t have “tight controls,” that’s the awful games done quick segment. That’s surgeon simulator. What makes a speedgame great is a deep system with a wide variety of mechanical interactions that can be exploited in such a way as to go slower or faster based on skilled play and understanding of the system.
The point isn’t to leave in glitches or add shortcuts, making a good speedgame requires a total shift in perspective about the way glitches are viewed. It means evaluating glitches as they come up in much the same way as you would evaluate a mechanic. It means appealing to a higher principle of game design. It needs to be evaluated whether a glitch overshadows or invalidates any existing mechanics, whether it allows one to bypass challenges without being challenging in of itself. Glitches should be revised in much the same way as mechanics, or even made easier if they’re too hard.
The idea with shortcuts isn’t to just stick some shortcuts in, because that’s solved easily and becomes boring quickly. A better idea is to make your out of bounds areas navigable, which most speedgames already do to some extent, and to design your levels in horseshoe shapes. Horseshoes have the end destination next to the starting point. This means that along the path of the horseshoe there are a lot of potential points where players can skip across to the other side of the horseshoe. Beyond this, design your levels so that there are many possible paths through them, making it difficult to find the optimal shortcut path. Add tradeoffs between paths that aren’t strictly time based, like ammo drops, items, abilities, enemies, that make it tricky to determine which path is the fastest forward. In more metroidvania style structures, have more interconnection between areas (you can see this in quake maps too), so people need to route around backtracking. Almost everything Dark Souls gets right as a speedgame isn’t due to glitches, which is why current patch dark souls 2 is still a successful speedgame. It works because there are many paths through the levels and the world, and every path has differently time consuming obstructions and valuable pickups, making optimization of the game difficult.
Telling people to focus development on tight controls isn’t giving people ANY type of useful information. Of the games shown onscreen here it’s a lot easier to point out that all of them have multilayered level design with branching/interconnected routes through levels, and a combination of advanced movement mechanics (Doom: straferun, Quake: bunnyhop, SMW: cape flying, DKC: a lot of things, Megaman X: dashing/walljumping/powerups, Super Metroid: running/walljump/mockball/shinespark) Tight controls is undefined here, an example isn’t even given of non-tight controls, or what type of game is even bad for speedrunning (ys games, most beat em ups).
Versatility and consistency is starting to get at what I’m talking about, but they’re still poor terms to describe what I’m going for. Yes the system should be as consistent as possible across the board, so that players can actively model how the system works to predict the outcomes of their actions, keep the game fair. Versatility is the wrong term though. It’s about giving the player a variety of options, actions the character can take, that each have differing, but slightly overlapping, functions, so that they will be good in different circumstances. In Mirror’s Edge you have walking, running, sideboosting, wallboosting, and kick glitching, which scale up in speed at about that order. If it’s possible, you’d like to be kick glitching all the time because it’s fastest, but you don’t always have the right setup to do it, so you need to fall back on whatever the fastest form of movement for the terrain is. Each form of movement has its own unique niche in the system, so working out exactly when to use which is tricky. There are some comparison videos on youtube of different methods of getting across areas in mirror’s edge, I can’t link them right now, when this becomes a blog post I’ll do that.
Consistency and versatility aren’t hard to get at the same time. Consistency is just a matter of avoiding randomness or subpixel type business with variables too small for the player to accurately track. Complex AI systems work in games like dark souls and FEAR (okay, they’re not that complex). Bigger issues again are things like the drop chance on the black knight halberd, or Link between worlds runs getting screwed by good/bad rupee drops. Then you get games like Animal Crossing where practically the whole game is decided by the random layout of the town, and people will still play it, because it’s cool to have one chance to memorize a town map and greet every person in town in the fastest order possible. The bigger issue is if the game is boring, and doesn’t have a lot of tradeoffs in approaches to problems, like assassin’s creed, which people still don’t speedrun to my knowledge.
The problem isn’t complexity or moving parts, it’s moving parts that the player has no feedback about, that they cannot observe in action. Hidden information is the problem for consistency.
The remark on randomness is especially infuriating, him saying, “make sure the randomness basically evens out over play” THE FUCK DOES THIS MEAN? How can randomness even out over play? How the fuck does this apply to a game other than Tetris or other puzzle games? It’s totally nonsensical for most speedgames. Roguelikes don’t have this. Only like Hotline Miami has anything close to this, and the better remark here would be to make randomness something the player can react to and counteract.
The Tetris example here is further infuriating because the RNG in Tetris is literally rigged to make it even out every 7 blocks. It has been for at least 20 years. You’re guaranteed to get one of each block every 7 blocks, preventing runs of bad blocks. He makes no mention of this, probably because he didn’t know because EC doesn’t do any fucking research.
Patching is a big deal, but this goes back to leaving bugs in your game intentionally. You need to be a good enough designer to realize whether bugs should be left in or not. To determine if they’re a positive addition. If the game is on PC, then players will hang onto their preferred version and continue to play it even if you patch the game. If the game is on console, then players can only play current patch or what’s on the disc, so patching a game can mean erasing a delicate balance for speedrunners.
Patches can also help revise runs and tease out more interesting play, like how the elimination of the kiln skip created more interesting dark souls speedrun routes, though that category likely would have been created in the end anyway. The addition of new warps helped out too.
Dark Souls 2 running probably wouldn’t have lasted as long as it did if the parrywalk glitch wasn’t patched out, and thankfully the old build is still viable for it. I wish binoboosting was still in the game, but it’s a multiplayer game and you can’t totally leave everything like that in.
Dishonored requires an older version of the game to speedrun, same for portal and half life 2, these patches are fucking pointless and did nothing to help the game, only hurting speedrunners.
Ori and the Blind Forest has patched the game a lot and left everything speedrun related totally untouched. Good on them.
I hate having to watch EC videos. They don’t try. They don’t want to convince me that they try. They don’t want to try. I wish I could just not watch a video by them ever again. (Editor’s note: Yeah, I later watch another video by them.)