Summer Games Done Quick 2024 Recap: Takeaways and Interviews with the Zelda Runners

This post was co-authored by Kristen G. Rosario.

This year’s Summer Games Done Quick (SGDQ) featured three Zelda titles — A Link to the Past, Twilight Princess, and Ocarina of Time. We here at Zelda Dungeon loved watching the stream, especially since it showcased some of the series’ most beloved titles. But beyond the action on the stream, our team also had the privilege of getting exclusive interviews with the runners from each game.

So, kick back and relax (antithetical for speedrunners, we know), because there is a lot of coverage!

---

A Link to the Past, All Dungeons (Swordless) by Glan

 

Summary and Reactions

Glan is a prominent name in the A Link to the Past speedrunning community. With amazing resilience and persistence from Glan, their run ended up being one of best at SGDQ this year. This category is no joke. It is chock-full of frame-perfect glitches, many of which result in notable time losses. Regardless, Glan put on an amazing show. So, what does the category entail?

In “All Dungeons (Swordless),” runners simply have to do as the category name states: beat all dungeons (nabbing all Crystals and Pendants along the way), fight all dungeon bosses, and get to the end of the game without obtaining the sword. Fighting Ganon is not required. Instead, one just has to rescue Zelda and hit the credits. Glan’s attempt was played on the Japanese 1.0 version of the game because that is where the most glitch potential lies. We recommend reading our interview below and watching the playback to get visuals on the most intricate glitches!

“Swordless” is a mind-boggling category that uses numerous glitches like Wrong Warps, but the most shocking to watch was Glan’s use of the “Exploration Glitch.” What casual players don’t know is that there is a hidden, explorable layer under the visible map in A Link to the Past. This allows players to play dungeons out of order and collect tons of items at weird times. Glan’s run included several unlucky deaths and moments of intense recovery after a couple of glitches wouldn’t cooperate. If SGDQ proved anything, it was that the knowledge these runners have about their titles is beyond comprehension. It was a joy to watch Glan strategize on the fly to find a way to complete the speedrun.

The run was full of laughs and support from the Zelda speedrunning community. Glan finished the run in 1h 13m 06s, with commentary help from twinmo23 and chexhuman.

Interview With Glan

Are you curious about the nitty-gritty details of this insane run? Well, long-time ZD marathoner Moogle got quite a few answers out of Glan after the marathon, earning insight into Glan’s thoughts on the run, his preference toward glitch-based categories, and even what it takes to practice an A Link to the Past run like the one he did at SGDQ:

We just want to get your impressions about how everything went — obviously not exactly according to plan, but such is the reality of live events. Regardless, you put together some nice backups and showed off some crazy visuals. Did you have any sort of contingency plans prepared in advance, and what was your thought process on some of the on-the-fly backups?

Glan: So, how I felt about my run is a bit of a longer answer. Obviously at the moment, I was frustrated. As one usually does, I had gone into the run hoping that it would go really well and [hoping that] I could show off a really clean execution of one of the game’s hardest categories. The first part of the run mostly went very smoothly and I was feeling good, but then it went off the rails quite a bit. After the first rough segment with Thieves’ Town, I knew it would go pretty far over estimate, so I wasn’t feeling good. It was especially tiring for me that a lot of what went wrong happened as kind of a chain reaction from one mistake which I didn’t even know was a problem, which was after Palace of Darkness. We perform an upward teleport on the overworld screen outside Eastern Palace to reach the potion shop. 

You can do a teleport on two Y coordinates; normally we do it on the higher one because it has a cleaner visual cue. But I did it on the lower one because I was already lined up there. I wasn’t aware that doing it from the lower pixel meant you couldn’t reach the screen transition, which made me have to save & quit, and then a lot of the other stuff came as a consequence of that. So for a lot of the run, I couldn’t help but think about how much time I had lost all because of one thing that I didn’t even know about, and honestly I felt embarrassed. I did eventually accept it and kept going. The goal was to show off the route and the glitches involved, and the audience cheering helped a lot. I knew people were not thinking negatively and were making light of the situation.

However, my perspective shifted more as I talked to my friends and other people. Everyone said they enjoyed my run a lot. And one particular thing that people have said to me really stood out: they were impressed that I was able to keep my composure and back up everything that went wrong. Since the run, I’ve been thinking about it a lot. Again, everyone would like to go up on the GDQ stage and perform a really good run with little to no mistakes. But in some ways, that desire is more for them than for the audience. If I had truly sat down and performed a perfect Swordless run, most of the audience would never have gained a sense about how difficult the category is. 

So I’ve come to appreciate the experience in another light: by having so much go wrong, I was really able to convey how difficult of a category it truly is. Take for example fighting Agahnim. One, I said that I couldn’t get hit in the fight or I’d end up in EG again, which then happened. It’s one thing to simply say, “Oh, something bad will happen,” but it’s another to actually have people see what happens. The other thing is exactly what people said to me: as a viewer, seeing the improv and backup strats is impressive. In many ways, I think it was actually more impressive to viewers to see things go wrong and to see that I know how to recover.

I know from my limited experience with A Link to the Past glitches how tight some of the tech can be and how easy it is to crash or softlock the game. I’m curious how you think it compares in difficulty with more standard speedruns like No Major Glitches (not exactly a 1:1 comparison, I know) and why you seem to prefer the more broken categories.

Glan: Most of the backups I had to do were basically on the fly. I did have some things planned, or some backups that I knew from practicing. The main contingency I had prepared was a backup save for the early game. Basically you can’t save in the earliest progression state of the game. This state normally ends when you reach Uncle, but since we never go to him in Swordless, it doesn’t end until you bring Zelda to the sanctuary. Not being able to save means any death or softlock situation would force you to reset the entire run and start from the beginning. 

Eastern is mostly safe, the main risk there being getting Zelda text during the Pendant cutscene, which is easily avoidable by waiting for it before grabbing the Pendant. Escape poses more risk, with the possibility of softlocks from clipping into the moat or swim clipping to the castle wall. The clip inside the castle is very difficult as well, and running out of bombs would basically force doing Escape normally and would screw up some routing after Sanctuary. Because of this, I made a backup save at Uncle, since luckily it is possible to talk to him without actually getting the sword. We just don’t do it during the actual run because it’s slow and costs bombs and arrows, but it’s very nice to have been able to make a backup save there.

How does this run compare in skill and difficulty to other A Link to the Past runs? And why A Link to the Past?

Glan: In terms of run difficulty, it’s hard to fully compare Swordless and other glitched categories to runs like [No Major Glitches (NMG)]. I think both are hard in different ways. NMG at a very high level, when trying to get super-optimized times, is extremely challenging; you need to learn how to handle all sorts of enemy RNGs at basically muscle memory level to get through everything seamlessly. For glitched runs, that level of technical skill isn’t as necessary. However, the glitches themselves can be quite difficult, especially performing them quickly, which gives these categories a much higher barrier-to-entry. They have a much different skillset too. Instead of needing to perfect movement and combat, you need to be really good at things like buffering, timed inputs, or unusual input sequences. Being really good at smoothly changing controller grips for various tricks is super helpful for glitched runs. 

Glitched runs also have a very different mental game. In NMG categories, you’ll lose time mainly by just having some sloppy movement or combat here or there. Probably the worst thing that can happen is that you die and have to redo some rooms. In glitched runs, these can still happen, but then there’s even more things that can happen: you can fail tricks and have to redo them, and the worst case is running out of important resources like magic or bombs. Pretty much every glitched category has tight bomb routing for at least part of it, and sometimes magic too, meaning consistency in performing tricks is incredibly important, on top of being fast at performing tricks.

What types of categories appeal to you the most? We’re also curious about your involvement in route planning and theory testing things like misslotting, and about, in a general sense, how much technical knowledge is required in discovering and practicing what appears to be totally arbitrary item spamming.

Glan: I’m not sure I can give an exact answer for why I prefer the more glitch-heavy categories. A Link to the Past was my first speed game, and even then, my introduction to playing the game fast was through the randomizer. I guess from that I got accustomed to not playing the game “normally.” Even before actually doing speedruns, though, I ended up getting very interested in all the glitches the game had to offer. Overall I’m a rather technically-inclined person; I was interested in programming from a young age, studied computer science, and now work as a programmer. I’ve always had an interest in learning how things worked. I was that kid who took apart his pens and put them back together just to understand the internal mechanism of the button. 

So I guess that kind of interest carried into playing this game, as well. I just love seeing all the things that can be done in this game when you peel it back a few layers. In terms of the actual speedruns themselves, I actually don’t care for the general movement tech in the game very much. There are other games where I really enjoy running glitchless because I find optimizing the movement to be very fun and challenging, but that has never really been the case for me with A Link to the Past. On top of that, this isn’t actually a super nostalgic game for me either, which means I don’t feel a lot of sentimental value in playing through “normally” compared to some other games.

Overall, I really just have a lot of fun using the glitches and thinking up ways to route things. One other thing that I really enjoy about glitched categories is that they open up the possibility of doing certain parts of the game with unintended item sets, whether that be more than what you should have, or less. Both can create really interesting scenarios and strats, and build your skillset in a much wider way than anything NMG can offer.

I used to do a ton of routing myself, but I haven’t done as much recently. All the full-game glitched categories right now have routes that were developed by Hotarubi. I was around when misslotting was found and did contribute to some of the earliest applications in runs, including how to do it in Swordless. Much of the very technical stuff like finding new interactions was a bit beyond me for a while though. I actually kind of had a hard time understanding it at first. A lot of that initial development, like discovering hookpushing and weirdshots, was done by Pinkus, Tojso, Yuzu, and Hotarubi. I’ve mostly worked to find more setups and applications, especially consistent setups.

Misslotting is an incredibly complicated glitch with so, so many different applications, and I always wish I could explain it in more detail than I can during marathon runs. But doing so would take the entire length of the run. To try to give a general sense of how deep the knowledge goes without having to explain too much: in general the weird effects that come from misslotting are due to the game only allocating five bytes for certain ancilla property arrays, when ten ancilla slots exist in total. This is done because it’s intended that only certain, very basic ancilla can spawn in the higher slots. So, if we force something into a higher slot that is only expected to live in a lower slot, it will write its properties outside those five-byte arrays into the memory addresses following them, which might be other property arrays or just other assorted data. 

This can lead to what we call “five-slot relationships” where two ancilla that are five slots apart in memory will both try to use the same byte for different properties, allowing some things like weirdshots. Fully understanding all of this and knowing how to make use of it requires knowledge of what properties each ancilla has, what values get used for those properties, and how it uses them. As well as just their general behavior and how they interact, and how the game decides what slots to use to spawn them in. Understanding all of this is also pretty important for doing glitched runs too, to be able to identify what went wrong when you fail a setup or how to adapt if something does go wrong.

 

Twilight Princess Co-op Randomizer performed by gymnast86 and spikevegeta

The Twilight Princess segment was truly a spectacle to behold, as gymnast86 teamed with resident GDQ staff member and fellow speedrunner spikevegeta as they took on a unique challenge known as a “Co-Op Randomizer.” The requirements of this randomizer run are that each runner is given the same seed, along with four random dungeons they must complete. Gymnast and Spike needed to accomplish their given tasks and then beat the game in order to call time on the run.

The four dungeons included Forest Temple, Lakebed Temple, Snowpeak Ruins, and City in the Sky. During the run, more dungeons were on the line to be added via two separate donation incentives, with the first revealed to be Goron Mines, and the second being Arbiter’s Grounds. Both of these goals were met, which added $57,500 USD to the marathon’s grand total. The two additional dungeons were each done separately, with spikevegeta taking on Goron Mines, while gymnast86 took on Arbiter’s Grounds.

What made this co-op run so interesting is how both gymnast86 and spikevegeta took on the run. Though they were using the same randomizer seed, they each tackled the default four dungeons in a different order. For instance, gymnast86 went for the Lakebed Temple first, while spikevegeta started with the Forest Temple instead. The biggest benefit of running it this way is that they can better map out how each other’s runs should go. The same can be said for assisting each other with all the randomized items scattered throughout the chests in Twilight Princess.

The challenges didn’t stop there, as additional donations given during this run added unique tasks for the runners. During the beginning portion of the run, for example, gymnast86 was given the challenge to beat the Lakebed Temple boss, Morpheel, without wearing the Zora Armor. For spikevegeta, it was the opposite, as Link had to wear Zora Armor while going through Snowpeak Ruins.

The last stretch of their run was nearly in sync as both of them managed to go through each of the final bosses within a few seconds of each other. Spikevegeta accomplished his part of the run at 3h 08m 40s 50ms, while it would take gymnast86 just ten seconds more to complete his, with the overall completion time being 3h 08m 50s 60ms.

Overall, this was a really fun and unique run, backed by not only the commentary of lunarsoap and phantom5800, but by the enthusiastic nature that spikevegeta is best known for. If you want to get started on your own Twilight Princess Randomizer, you can head on over to TP Randomizer where they have everything you need to get started.

Interview With gymnast86

Moogle also just so happened to get a few answers out of gymnast86 after the marathon about his thoughts on the run and about how he prepared for it:

Could you share some of the key challenges you faced during the run and how you overcame them?

gymnast86: The biggest challenge with a randomizer run is usually figuring out what the best strategy to do on the fly is as we’re collecting items. We also don’t end up using all the techniques that we practice since we only go through a portion of the game for each randomizer seed that we play.

How does your preparation differ when you’re getting ready for a randomizer compared to a standard speedrun? Did you approach this co-op randomizer differently in terms of strategy or practice sessions?

gymnast86: Preparation isn’t a whole lot different from standard speedruns. For myself, it’s just doing a lot of runs until we get to the point where I’m comfortable enough to do the run on a stage. This co-op run wasn’t different in that aspect.

Can you describe how you and Spike coordinated during the run? Were there any specific strategies or moments that particularly stood out to you?

gymnast86: The biggest part of our coordination was probably just figuring out what each of our opening plays were, but after the first 20 minutes or so, we just play things by ear and talk with each other about who should go where in the moment. We could also of course see each other’s gameplay, so there wasn’t usually any confusion about where certain items were that either of us found.

Prior to the event, was the randomizer seed tested to ensure it was suitable and balanced for the GDQ segment? From your experience, how important is seed testing in ensuring a smooth and entertaining run for both the runners and the audience?

gymnast86: The seed was vetted by trusted community members before the event. We as the runners didn’t know anything about the seed ahead of time though. Vetting a seed beforehand is usually pretty important, but there may be some games that can get away without any vetting. In this specific case for Co-op, the vetting tried to ensure that it was possible for me and Spike to split up what we were checking and that the playthrough of the seed was not just a linear path where one of us would be following the other.

What were the most memorable highlights of your co-op rando segment?

gymnast86: The most memorable moment was definitely when I was talking about how annoying it was that I didn’t have a sword when traversing through Gerudo Desert and then promptly found a sword while complaining about it.

 

Ocarina of Time, Defeat Ganon No SRM by dannyb

If you know anything about glitch-hunting in Ocarina of Time, you’ve heard of dannyb. He is one of the greatest minds in the community, toppling this category at the beginning of 2024 by pure accident. So, let’s explain what this category is all about. The “Defeat Ganon” category (previously Any%) only cares about beating the game’s end boss. It doesn’t matter how you get there. All glitches are on the table, except for arbitrary code execution (ACE) and stale reference manipulation (SRM). As the name implies, this exploit is not used.

To make this run happen, donors needed to meet an astounding $100,000 USD incentive. For a game of this caliber, this was easily met.

Danny commentated the run, alongside goodgortho and adef. Throughout, they did a phenomenal job explaining the history and complexities of this legendary category. Danny discussed the old method used before 2024, dubbed “Ganondoor,” which involved wrong-warping. Though he intended to not use it, the run took an interesting turn.

Everything looked like a standard Ocarina speedrun — despite only collecting a Deku Stick to fight Ganon and doing frame-perfect inputs to get Bombchus from Bottom of the Well within the first nine minutes of the game — until he showed off the category’s new strat called “Ganonfloor.” We recommend watching the playback, as the commentators even prepared a script to explain this extremely complex out-of-bounds, loading-zone-manipulation glitch, called a Dynawarp.

Danny perfectly executed the hardest glitch in the run. He skipped straight to the final boss fight, and, much to his surprise, unexpectedly died. This ironically meant he actually had to use the old Ganondoor route involving the collapse. Even with the time loss, the commentators and attitude of the crowd, donors, and Danny himself were phenomenal representations of the speedrunning community. Danny completed his run in 23m 20s.

Interview With dannyb

Another member of our team, Connor Dennin, reached out to Danny for some thoughts about the run.

With the massive development with the Defeat Ganon No SRM category, how does this change your perspective on other categories for Ocarina of Time?

dannyb: There’s a lot to consider! The new Dynawarp glitch can warp you anywhere, do it repeatedly, and be stored within a Farore’s Wind warp point so that you can pick the glitch back up later in a run once you end it the first time. 

So, the potential for faster routes is enormous, especially in longer speedruns that go many different places in the game, such as 100%. Routing a long run is already an exercise in trying to solve the Traveling Salesman problem, and with Dynawarp that seems to be true tenfold.

We have a lot of work to do! We also need to consider whether we want this glitch to be allowed in all categories which could benefit from it. Some categories already have some restrictions on what glitches can be performed, so we need to discuss within the community whether Dynawarp fits the vibe of those categories or not.

Defeat Ganon was solved so quickly since it was the simplest possible use case: warp to exactly one place exactly one time as fast as possible.

Editor’s Note: Moogle offered a fascinating point about this perspective:

“[This] first response brings up a wider interesting topic about how ‘rulesets’ form and change over time, having to decide if a new discovery fits the ‘vibe’ of a current category or ruleset, and how change (or lack of change) can impact players willingness to run a category. Lots of sub topics there: community/committee votes (deciding who is qualified to vote), button remapping, RNG removal, cutscene removal, prepared files, legacy and obsolete categories, or even obsolete routes within current categories. As we get bigger we can crowdsource more of this stuff by polling people for examples of a time when a rules vote had the community split.” — Moogle

What is the risk factor in completing a run like this and how costly can some of these glitches be?

dannyb: I suppose you mean mostly in terms of potential time loss for the run? Doing any run at a live event is always much scarier than doing regular attempts on my own stream. Most of the time, if I make a costly mistake, I just end the run and try again from the beginning. But the expectation of finishing no matter what changes the approach entirely.

I spent a long time coming up with backup strategies and contingency plans for every possible thing that could go wrong, so that I’d be prepared if I made a mistake. The lucky thing about this particular run is that it offers great opportunities to make save points, so that if I really mess up, I can just reset the console and start over from that save.

The main risk in this run was performing Dynawarp itself. It’s a very tricky glitch to pull off, but I saved right at the start of the Dodongo’s Cavern dungeon, where the glitch takes place. So, worst case scenario, I was losing a couple of minutes to failing Dynawarp, and just resetting from the beginning of the dungeon.

My death to Ganon was pretty funny considering it was the thing I was least worried about actually occurring, but ended up costing me quite a bit of time in the end. But I’m glad I played it off well, and got some extra time to talk to the audience about what went wrong, and some other fun facts that I wouldn’t have had time for otherwise.

With the donation incentive being $100,000 USD for the showing of your run, what was your experience finally knowing that you were finally able to showcase it at SGDQ?

dannyb: I was so pumped that the 100k incentive was met. Truth be told, I wasn’t even really worried about it. Ocarina of Time is always a moneymaker at charity events because it’s developed a reputation as a game that always has some crazy new magic trick to show. 

I’m excited to have been the one to showcase our community’s hard work on the biggest stage in speedrunning. It was extra cool for me since I haven’t run at a GDQ event in eight years (I ran Glitchless Ocarina of Time at SGDQ 2016), and I was proud to see growth in my ability to stay comfortable in front of a crowd and explain tough concepts while doing a run.

---

Thanks for tuning into our SGDQ wrap-up! What did you think of these amazing runs this year? What was your favorite, and what were you most impressed by? Let us know your thoughts in the comments below!

The interviews above have been edited for length, clarity, and style guide compliance.