SI-generated
Some weeks in speedrunning are about a single record falling by a few frames. This was not one of those weeks. The big story across the community was discovery: a new Ocarina of Time remake arrived looking familiar, and runners immediately started asking the only question that matters—what happens if we push on every suspicious wall, platform, bean, and animation?
At the same time, Minecraft cracked a barrier that had been sitting in plain sight, several beloved classics kept shaving away at apparently impossible limits, and a blindfolded Super Mario 64 run reminded everyone that “knowing the game” can mean something far more literal than memorizing a route. Even the community’s lighter conversations circled back to the same idea: speedrunning is part competition, part research lab, and part group project conducted in public.
So this week’s mood is equal parts new-game chaos and veteran precision. We have movement tech that may already be living on borrowed time, a first sub-30 Minecraft run, world records in games from Mega Man 2 to Banjo-Kazooie, and a convincing argument for giving speedrunning’s documentary history its own live showcase.
Here are the biggest stories from the week.
Ocarina of Time Remake Has Entered Its “Please Stop Patching This” Era
The Ocarina of Time remake dominated the week’s discussion not because runners found one spectacular shortcut, but because they found an entire ecosystem of them. The top post showed that roll canceling—at first glance the obvious way to make Link’s movement more efficient—is still slower than simply rolling. That is a wonderfully speedrunning kind of result: an elegant technique loses to the old, ugly option because the game engine has already decided that momentum is king.
The testing was careful, too. The comparison used macros to make both inputs as consistent as possible, with turbo-assisted rolling and a precisely timed roll-cancel sequence. In other words, this was not a casual “it felt faster in my run” claim. It was a tiny laboratory experiment, and the conclusion was devastatingly practical: if you are routing the remake, the new trick is not your default transportation plan. The comments immediately started sorting out the consequences, with runners pointing toward Z-targeted forward jumps followed by a roll as the movement to watch instead.
There is a delicious irony in the remake apparently trying to discourage Link from rolling everywhere while still leaving rolling as the faster answer. One of the highest-rated reactions asked whether running has any speed advantage at all; another joked that the roll cancel’s main benefit is letting Link keep his sword dangling. That combination—serious frame analysis delivered alongside complete disbelief at the design—is exactly the community at its best. The game may be new, but the speedrunning instinct is ancient: measure everything, trust nothing, and keep the funny animation if it saves time.
Hookshot Jump Is Back, and the Remake Is Already Looking Guilty
The more explosive Ocarina discovery was the return of Hookshot Jump. Once Adult Link has the hookshot, a setup involving Z-targeting, positioning against a wall, and a carefully timed release can launch him high enough to reach the Shadow Temple route early. Anyone familiar with the original game or its 3D remake will recognize the shape of the trick immediately: the remake has inherited the series’ long tradition of treating collision detection as a suggestion.
What makes this version especially interesting is that the jump appears to have a much higher launch than its older counterparts. That is not just a nostalgic callback; it could create remake-exclusive Hookshot Jumps in places where the original games offered no useful line. The thread’s discussion quickly broadened from “is the old trick back?” to “what counts as an actionable object?” Beans, chests, and other objects that cannot be picked up may all become little springboards, and a separate discovery showed a windmill platform can be used for actor jumping as well.
There is a familiar tension underneath the excitement. Runners are delighted that the new game is broken in the same ways its predecessors were broken, while also noticing that these particular tricks look easy enough to reproduce that a patch could arrive at any moment. That makes the current period unusually valuable. Every clip is both a route idea and a time capsule. If Nintendo closes one gap next week, the early routing videos will still show how quickly the community mapped the game’s physics before the walls went back up.
Ganon’s Castle Has a Front Door, Apparently
The remake’s discoveries kept stacking up with an early route into Ganon’s Castle. The clip is the kind of thing that makes a casual viewer assume the game has been completely solved: Link reaches a late-game landmark far ahead of schedule, apparently skipping the intended parade of checks and keys. For a speedrunner, of course, reaching the place is only the first question. The important follow-up is what the game lets you do once you arrive.
That distinction showed up in the comments. The likely immediate prize is the Gold Gauntlets, but without Light Arrows and the Longshot, the castle’s trials and the final fight still present very real problems. In other words, the discovery may be a route-changing breakthrough, or it may be an extremely stylish detour to a locked door. Speedrunning history is full of both. A skip can be technically real and strategically useless until someone finds the one missing piece that turns a sightseeing trip into a world-record route.
Even the smaller discoveries matter here. A Zelda skip toward Death Mountain seems, for now, to be more useful for bypassing a guard conversation than for unlocking the mountain itself, while a moving magic platform appears to line up uncannily with Link’s speed and may have implications for repeated jumpslashes. None of these clips is guaranteed to survive contact with a full route, but together they sketch a game whose early category is still being invented in real time. The remake is not waiting politely for runners to finish their research.
Minecraft Finally Breaks the 30-Second Wall
Amid all the Ocarina physics, Minecraft supplied the week’s cleanest headline: nsla’s set-seed Any% run reached 28.492 seconds, the first time the category has gone sub-30. The time is so small that it sounds like a typo until you remember what set-seed running has become. The modern route is not simply “spawn, find a structure, and hope.” It is an exercise in seed selection, path compression, precise combat, and exploiting a game that can generate an entire adventure in the time it takes another runner to clear a tutorial.
The post’s context was almost as useful as the record itself. Set-seed glitchless had crossed below a minute years ago, but a lot had changed in the meantime: new seeds, new tools, new understanding, and a community willing to keep improving a category whose best runs are over almost before the audience has settled into the chair. The discussion was not large, but it captured the oddness of this corner of speedrunning. One commenter called the run an awful viewing experience, which is either a criticism or the highest possible compliment to a strategy that leaves no room for theatrics.
A sub-30 time changes the emotional scale of the category. A mistake that costs a second is not a small error; it is a substantial percentage of the run. Every route decision has to survive a world where the clock is already sprinting before the viewer can identify the biome. The record may fall again quickly, but that does not make the milestone disposable. Breaking a round-number barrier is a map of how much work has gone into making a supposedly random game behave like a rehearsed stage play.
The Classics Refuse to Become Historical Footnotes
While new discoveries stole the attention, the old guard kept doing what it does best: making “optimized” look temporary. In Super Mario Bros., stem_kuki improved a personal best to 4:55.297, climbing from 25th place to 10th. That is not a world record, but it is exactly the kind of result that gives a leaderboard life. A four-tenths improvement in a mature game can represent a huge amount of work, and moving fifteen places at once is a reminder that the competitive story is not reserved for the person currently sitting at the top.
The reactions had a bit of the usual Mario vertigo. One commenter admitted that seeing a Super Mario Bros. post now creates the expectation that it will be the post—the one carrying another astonishing Niftski time. That expectation says something flattering about the level of the category, but it can also obscure performances like this one. The game is not a museum piece waiting for one celebrity to touch it. It is a living leaderboard with runners still finding cleaner inputs and better ways to convert practice into a finished run.
Banjo-Kazooie offered the other classic-game jolt, with a new 40:47 Any% world record arriving only days after the previous record. The run is now close enough to 40 minutes that the next barrier is impossible to ignore, and the comments immediately asked whether sub-40 is possible. That is how a category’s atmosphere changes: yesterday’s record is celebrated, today’s record is analyzed, and tomorrow’s round number starts looming over every split. The runner’s own note—“Super close to the first 30 minute time!” in the original post’s phrasing—may have been hopeful, but the community heard a challenge.
Blindfolded Mario 64 Turns Practice Into a Sixth Sense
Bubzia’s 53:03 world record in blindfolded Super Mario 64 was the week’s most immediately astonishing performance. The category already asks for a kind of memorization that makes ordinary speedrunning look casual, but the run carries an extra twist: it was intended as a derust session before an upcoming marathon appearance. After more than a year of practicing 120 Star, the runner found that the larger route had sharpened the 50 Star game enough to produce a world record on the second derust attempt.
That is a beautiful example of skill transfer. Practice is rarely as tidy as “repeat the exact category until the timer obeys.” Learning a longer route can improve spatial memory, movement confidence, and recovery instincts in a shorter one. The result is not magic, although it may look like it from the outside. It is accumulated knowledge surfacing in a format where the audience cannot even watch the runner confirm what the screen is doing.
The run apparently left time on the table, including two missed Bowser throws. The sole comment on the post was already looking ahead, urging Bubzia to keep running and push the time down. That response feels right: blindfolded speedruns are impressive as spectacles, but they are also ruthlessly competitive. Once a runner proves a line is possible, the record stops being a stunt and becomes a target. The blindfold is part of the premise; the optimization remains completely real.
Mega Man 2 and Trials Rising Find Different Ways to Break Physics
Kuto1k’s 26:29.5 world record in Mega Man 2 Any% on Difficult crossed a smaller but no less meaningful threshold: the first-ever 26:2x. In a game this established, a new second boundary is a statement about refinement. The route is familiar enough to be legible and demanding enough that every boss, weapon cycle, and movement decision still has teeth. The headline is short; the history behind it is not.
Trials Rising offered a more unusual kind of breakthrough. A first-ever TAS beat the human world record by 3.500 seconds, made possible because the game’s formerly nondeterministic physics were finally understood well enough to make tool-assisted work viable. That is not merely a faster run; it is a change in what kind of question the game can answer. If the physics can be reproduced, measured, and manipulated consistently, routes that once lived in the realm of “maybe the game will do the thing” can become testable strategies.
The community’s questions were telling. People wanted to know what changed and where the game’s runs are tracked, because a technical breakthrough is only useful if it can connect to a living scene. A TAS does not erase human competition; it gives runners a sharper picture of the ceiling. In a physics-heavy game, that picture may be the difference between chasing a mysterious ideal and practicing a sequence that has finally been proven to exist.
The Community Wants a Summoning Salt Block at GDQ
Not every important speedrunning conversation is about a time. One of the week’s most charming proposals imagined a Games Done Quick “Salt Block”: five runs based on games featured in Summoning Salt videos, assembled into a single showcase of speedrunning history. The pitch works because those documentaries already give viewers a narrative language for records—the old strategies, the breakthroughs, the rivalries, and the moment a seemingly unreachable barrier finally gives way.
The strongest response added one condition: Summoning Salt should be there to commentate the block. That suggestion gets at the appeal. A live program could show not just the current best run, but the progression that made it meaningful. The thread imagined a short game with multiple eras represented, so the audience could see old routes and new optimizations side by side. It would be part competition, part museum exhibit, and part extremely fast history lesson.
Other comments broadened the idea beyond one creator, pointing to marathon favorites with especially strong couch commentary. That matters because speedrunning is easier to understand when the people on the couch make the invisible visible: why a reset happened, why a trick is hard, and why losing half a second in one room can change the entire run. The Salt Block idea is really a vote for that kind of context. The community does not just want to watch records; it wants to remember why they happened.
What ties this whole week together is the gap between how quickly a discovery appears and how long it takes to understand its value. A roll cancel can look like a breakthrough and turn out to be slower than rolling. A Hookshot Jump can be spectacular but need a route to give it purpose. A castle skip can open a door without solving the rooms beyond it. A sub-30 Minecraft run can last less than half a minute while representing years of accumulated knowledge.
That is the rhythm of speedrunning in 2026: new games arrive with old ghosts in their collision boxes, mature games keep producing fresh personal victories, and the audience moves fluidly between frame data and jokes about Link’s sword. The leaderboard is only one part of the story. The other part is the public process of figuring out what the game is actually willing to let people do.
Whether next week’s big news is a patched exploit, another round-number record, or a couch commentary dream that somehow makes it onto a GDQ schedule, the community will be ready with a clip, a macro, and a suspiciously detailed spreadsheet. See you next week.
Source: reddit.com