I wish that I had known of Tetris Attack as a kid. My teen years were consumed by Pokemon Puzzle League on the N64. It wasn't until much later in life that I received any exposure to Tetris Attack. After starting my speedrunning career with Pokemon Puzzle League in 2013, my efforts redirected towards grinding this game at the start of 2014. I had done races of the game in 2013, but I had focused more on grinding Pokemon Puzzle League and GBC's Pokemon Puzzle Challenge, among other things, before moving forward with this project with the goal of improving on Craig 'cyghfer' Gordon's 2010 Very Hard run of 4:19.

In July of 2014, I had worked my time down to 4:22 and found myself in a bit of a rough patch where I started to believe that the grind would never end. I don't normally practice before my livestreams because I usually don't have the time to. However, on one fateful Monday night, I came home from work and found that the lovely Edobean had started to stream runs of Tetris Attack. Her Tetris Attack streams entertain me, even if she's not the very best at the game, so I try to catch her every time she runs. For reasons I have long since forgotten, I decided to pick up my SNES controller, start recording and play along with her. After a handful of attempts, I ended up finishing a run at about 8:00 on my recording timer. Knowing full well that it takes about 15 seconds from power-on or a soft-reset to start a run and cyghfer's run timed out in real time to about 8:06, I proceeded to freak out a little bit before timing my run. Minutes later, I streamed the video of my newest accomplishment: 3:57, a 22-second improvement over a time that had stood for over four years. Most people would be done for good after a time like that, but I am not most people. I did have a 32 second Lunge Fish stage that I was NOT pleased with. I wanted more. My real time was 7:44; I made it my final goal to cut the 7:30.

After more grinding woes, attention to other games and a shift to a somewhat-organized streaming schedule to maintain my sanity, my February 2016 began with an out-of-the-blue 3:53. Nearly two months later, on a night after a small hiatus where I became confident enough to essentially call my own shot (in a game/series where doing so has historically led to so many backfires), I cranked out this 3:37 in less than an hour of attempts. cyghfer stated in his SDA comments that he would have be done after sub-3:45. Welp, I guess I'm done now. What a journey that was.

Before going in-depth with my approach to the game, let's take a look back at the history of Tetris Attack speedrunning that came before me, as provided to me via second-hand information and the Wayback Machine.

The game may have been released in 1996, but talk about speedrunning the VS COM portion of the game did not start until 2004. Back then, TetrisAttack.net was the place to go to talk all sorts of strategy for every mode of the game. (The site has since went down and been relaunched as a husk of its former glorious form. I wish I had known about in its prime.) The first run of VS COM's Very Hard mode did not hit SDA until 2007 with Bas 'Iron Knuckle' van de Ven's 6:19, a run that relied on making as many large chains as fast as possible as soon as possible to ensure that the opponent's screen would stay filled with garbage no matter how many panels were cleared. Darkwing Duck followed in 2009 with a 5:12 run that also relied on large chains to fill the opponent's screen, but kept them to a maximum of x9 and also tried to incorporate a few combos here and there. In 2010, cyghfer put a stricter limit on his chains and added more combos to his strategy on his way to recording his 4:19. There were no other improvements to the category until my 3:57 time in July 2014, though I should backtrack and emphasize the word "recording" in that last sentence. During my initial broadcast of the 3:57 run, cyghfer got word of it during his Super Mario RPG stream and stated that he did get a 3:54 that he was not able to record. (This is the sub-4:00 time cyghfer referenced in his comments.) I have no reason not to believe him. The only improvement prior to this 3:37 was my 3:53 referenced earlier. Improvements in these types of games are really, really hard to come by at the highest level.

As evidenced by the progression of this category over the years, combos have taken on a much greater role in the overall speed of the run. The goal of this speedrun is to stack a certain height of garbage onto each of your opponents as fast as you can. Creating two rows of garbage at once will always be faster than creating one row at a time--combos are one of the easiest ways to do this. (This can also be achieved to some extent with time-lag chains. I will go into more detail on those later.) Darkwing and cyghfer picked up on this in their runs, but there is an artificial intelligence (AI) quirk that reveals itself only when the implications of this are fully explored. The AI's general rule is to search for matches starting at the TOP of the screen. Combine that with the fact that matching panels while they touch a garbage block turn the bottom row of that garbage block, every garbage block of the same color touching it, every garbage block of the same color touching those blocks, and etc. into panels and you can create a situation where the AI clears garbage at the bottom of the screen and, after the garbage blocks are finished transforming, subsequently moves towards the top of the screen to make another match. (Garbage from combos made during a chain fall on top of the chain garbage.) Most of the AIs are too slow to get to the top of the screen in time and just lose. In this run, Hookbill would have been able to demonstrate this behavior had he been given enough time to lower his stack. It is because of this behavior that I aim for at least three rows of combo garbage in my chain to allow for a potential vertical match in that situation. It can be hard to stack combos effectively, though. A 4 combo is the easiest combo to get, but it creates a garbage block that is three columns wide. As the field is six columns wide, making another 4 combo will result in another three-wide block falling right next to the first one. For this reason, I look to start my chains with at least a 5 combo to ensure better stacking. I particularly love the 8 combo as it is the first combo to create two rows of garbage.

Another behavior of the garbage blocks that can hinder your ability to stack garbage quickly was previously noted in cyghfer's comments. When you finish one chain and start another before the garbage from the first chain drops, the combo garbage made by the first chain will not drop until the second chain finishes. This can be particularly infuriating when you are pressured to send garbage as fast as you can and proved to be one of my biggest stumbling blocks in learning this run as I transitioned from my years of experience with Pokemon Puzzle League.

My general strategy in each stage is to execute an x5 chain (four rows) with three rows of combos during that chain. I also raise my stack until I make the combo to maximize my chances of getting the chain I need. I have found that seven rows of garbage will generally be enough to keep an opponentfs stack filled if they donft combo or chain too much. One of the biggest pitfalls I see in most other runners of this game is that they lose track of their opponent's stack and will continue their chains far longer than necessary, wasting precious seconds. Even so, whether your opponents don't combo or chain too much plays a big part in how long the stages can go. Combos and chains grant a bonus "stop time" that prevents the rise of the stack for a time. You also get stop time while panels are exploding, while garbage is transforming, and while the screen shakes. (Note: bigger chains = bigger shakes.) If you have stop time, then you can't actually lose. As much as you would like to minimize the amount of stop time your opponents get, the frame-based random generation of the panels puts that largely out of your control.

I will add more links to my chain if I am short on combos or see that the opponent has cleared more panels than usual, but other than that, there are no good backup strats available to you in order to keep the stages down to a reasonable time. Once you give even the smallest amount of space to your opponent, the time you need to make more combos & chains will allow the opponent to clear the garbage and/or make combos & chains of their own, clearing more panels and, in turn, forcing you to make more combos & chains to send more garbage. It's a vicious spiral that consumes a large number of runs.

Once I get to Raphael, I have to be more cognizant of the fact that opponents will start to be more aggressive with raising their stack, sometimes reducing the amount of garbage you need to send to win. On one occasion, I defeated Kamek in 7 seconds with only a 7 combo and an x2 chain. It's not something that can be relied on in the late-game, but I did manage to channel some of that magic in this run.

Without further delay, allow me to (attempt to) give a play-by-play of what was going through my mind during the course of this run. The decisions you have to make to go this fast have to be processed so quickly that I find myself playing most effectively by instinct, meaning I am probably skipping a few steps in the normal process. This breakdown will most likely turn into a post-game analysis of all of the options I had available and all of the mistakes that were made, whether they actually cost me time or not. It gets hard to tell at points whether time losses are under your control or not.

Stage 1 - Lakitu - 0:14

This was often a place of heavy resets. I would frequently restart my run if this stage ended up a 0;25 or longer. The attempt prior to this run reset a Lakitu at 0:23. Considering that I would have needed to average less than 0:20 per stage to improve my time and that the late-game AI becomes so fast to the point of being luck-based to get anything under 0:20, my standards for the first few stages have to be ridiculously high in anticipation of detrimental late-game luck.

As far as this run goes, if there was anything to take away from the previous comments about myself, it:s that I love me some 8 combos. After lamenting briefly about the empty column that sometimes shows up in the random arrangement of the starting panels that limits your options, my eyes went right to those blue and green panels in prime arrangement for an 8 combo. Once I matched those, I saw the yellow panels lined up below the combo and looked for another yellow panel to fall into the match for the x2. Below, my eyes turned to the two purple panels in the bottom right as I knew I had another purple ready to match with them for the x3. Upon further review, I had more red panels available to me that I could have lined up instead, but I'm not sure I would have recognized the follow-ups needed to continue the chain for long enough. I would have had to use the two purple panels in the second row from the bottom for the x4 and then bring either a green or blue panel from the left side to line up in the third column for the x5.

It's at this point that I should elaborate on some of the movement tech available to you. You can tap the d-pad button for more precision with the cursor or, do what I did twice in this stage, hold it for a quicker scroll. This is a technique I often use to move quickly from the top to the bottom of the field, using the bottom as a buffer to stop and move up to the next row, if needed, as I did with the red panels for the x4.

From there, I don't remember how I saw the purple panels line up like they did. Upon further review, it was either what I did, a lengthy arrangement to put three yellow panels together in column four, or a difficult two-frame trick to switch the green panel into the third column for a match as it fell and landed, but before it came to a rest. I wish I could have shown those off more. Had I gone for the latter option, I did have a purple panel ready to line up with the one in column four and the one in column two for a potential 6 combo. By the way, the phrase "hindsight is 20/20" is proven to be very accurate for this game. The game's flashy visual presentation loves to overload your senses with information, making it difficult to process it all quickly. In any case, those purple panels came together nicely for the x5.

After that, I saw red panels that I could line up and searched for another one to match. Once in place, I glanced at the garbage display on my opponent's side as a reminder. I only had two rows of combo garbage--the three-wide and four-wide blocks made by my 8 combo, and four rows of chain garbage from my x5 so I let the red panels fall for the x6. With seven rows in place, I dropped my chain and let my garbage fall. In case I miscounted the number of rows I needed, my go-to move is to find the nearest combos available and spam a bunch of those to fill in any gaps. Fortunately, my initial attack was precisely enough to take down Lakitu at 0:14. I am very fortunate that Lakitu only pulled off a 6 combo and did not have enough stop time to move the blue panel down for the garbage clear. I have lost many a good stage on much closer timings than that. A sport can be a game of inches, but Tetris Attack and other Panel de Pon games often become a game of frames. And yes, this is a 60 frames-per-second game.

Stage 2 - Bumpty - 0:12

Fun fact: Bumpty does not appear in the Game Boy version of Tetris Attack. Memory limitations? Hell if I know. I just find it strange that Bumpty is the one that got cut.

Upon further review of the beginning to this stage, I had five red panels staring right at me that I could have lined up in the third column, dropped a blue panel down for the x2, lined up the green and purple panels for a 6 combo x3, and lined up yellow panels for the x4. I could not actually see that during the run, however, so I rose the stack and shuffled things around a bit until green panels lined up for a 4 combo. I immediately had purples ready for the x2 when I saw the reds line up for another match. I had passed by another red on the way and found an opportunity to turn my potential 6 combo into an 8 combo. You should know by now how much I love those. (Ironically, they don't gel so well in combination with 4 combos. Go figure.) I saw the blues in the first column line up for the x3, saw a yellow on top of them, and looked to line up more yellows for the x4. I had multiple options to line up purples for the x5 so I ended up doing a green clear to line up a set of purples in what is called a "time-lag chain". A time-lag chain is done simply by making another chain link while panels are exploding. Itfs not easy to set up, however, but it is essential to building chains quickly.

At this point I have six rows. (Like combo garbage blocks are grouped together during the course of a chain. My 8 combo created a second three-wide block, thus I only had two rows of combo garbage.) I could have shoved a blue panel under a falling blue panel for an x6, but for reasons I have since forgotten, I stopped the chain there. Perhaps I miscounted my garbage due to the unusual 8 combo interaction. Either way, it turned out to be just enough for Bumpty to not handle in time for the 0:12. Sub-0:15s are incredibly luck-based to get in any stage and, for the most part, can only be done with the opponent's cooperation. When it happens, I count my blessings and move on.

Total time so far: 0:26 (For those keeping score with the system that Darkwing frequently uses on his runs where 0:20 on a stage is even par, I am currently -14.)

Stage 3 - Poochy - 0:20

For whatever reason, this stage has proven to be a frequent run-killer for me, which has led to a saying within my livestream chat: when in doubt, blame Poochy.

On this stage, I think my best open would have been to move the red panels from column five next to the blue panels in column two for the easy 6 combo. Instead, I arranged yellow and blue panels into a different 6 combo formation that took longer to make. I had a green panel above and a bunch of green below so I arranged them into a 5 combo for the x2. The choices of colors afterwards were essentially random and I could have arranged them in different ways on my way to the x5. I had made a 4 combo with my x3 so I felt I had done enough to cut the chain off after x5. Unfortunately, Poochy had some green panels ready to clear on my garbage, resulting in several seconds lost. Fortunately, he had no time to even move the cursor and fell for the 0:20.

This stage reveals another quirk about the AI that wasn't even news back in Iron Knuckle's 2007 run: the AI is very simple-minded. In later Panel de Pon games that have AI deal with garbage, the AI will actually pay attention to the panels that come from the transformed garbage and try to chain with them; in this game not so much. I'd ask you to imagine how much harder this speedrun would be if the AI was capable of such a thing, but I don't have to. Just watch Pokemon Puzzle League.

Total time so far: 0:46 (-14)

Stage 4 - Flying Wiggler - 0:20

For those who have only watched runs and do not know/remember the actual story to this game, the plot has Yoshi rescuing his friends from one of Bowser's evil spells. After all of your friends are free, they are playable in the last four stages. If I am not mistaken, Darkwing Duck loves to use Flying Wiggler in the final third of his runs.

As my first instinct at the start of each stage is to search for the quickest 5 combo I can find, my eyes immediately saw the five blue panels in columns two and four. After matching those, I knew that I had a green panel ready to fall so I scrambled for two more greens to make the x2. Since I had three red panels already in place adjacent to the x2, I decided to grab two more red panels to make them fall into a 5 combo for the x3. I thought I had two green panels below it ready for the x4, but one of them was blue so my eyes went towards the yellow panels on the right side to line them up for the x4. This time, I did have two green panels in place for the x5. I was short on combos so I attempted a two-frame trick to match the x6 with the yellow panels on the left side, but I was late so it didn't register. Upon further review, I had a green panel right there to move right for the x6, but I was caught off-guard by my miss.

My attack turned out to be barely short. Wigglerfs stack did reach the top, but he had enough time to match red panels for a garbage clear. In an instance of deja vu from the last stage, Wiggler immediately ran out of time as soon as the garbage finished transforming and lost for the 0:20.

Total time so far: 1:06 (-14)

Stage 5 - Froggy - 0:14

Can someone please tell Big the Cat that Froggy's right over here? Nobody? Ah, oh well.

My 5 combo senses were tingling at the start so I put the green panels into column five. I also had a "T-shaped" 6 combo with the blue panels I could also put on the right side, but, upon further review, making more combos in my chain from that point onward might have been more difficult than in my ultimate course of action. With a purple panel on top of the green panels, I knew I had purples to line up. I also had another purple to shove in front of the falling purple for the 4 combo and the x2. The red panels aligned nicely for the T-shaped 5 combo x3. The yellow panels also lined up nicely for the L-shaped 5 combo x4. I saw on the right that I already had blue panels ready to form the x5 so I shifted attention to the left side to prepare a backup 8+ combo with the green and blue panels, which I decided to make a 9 for reasons I don't recall. I didn't need it, though, as my initial attack was precisely enough to stop Froggy for the 0:14.

Of the five kills so far, I would consider three of them (Lakitu, Bumpty, Froggy) to be "by the book" given what I had to work with. They were also the only ones under 20 seconds. A 20 second average per stage for the 12 stages in this run would constitute a 4:00 time. Considering that getting under 0:20 in the late-game is so luck-based, the early game has to get a bunch of sub-0:20s to compensate. Also, for perspective, my best run prior to this one was a 3:53.

Total time so far: 1:20 (-20)

Stage 6 - Gargantua Blargg - 0:17

Once again, my 5 combo senses were tingling green at the start. With a purple panel to fall, I scrambled to arrange the other purples I had into another 5 combo for the x2. Upon further review, I missed a potential time-lag chain with multiple sets of red, but I still had one with purple and I missed it. I mashed too hard and accidentally did a two-frame trick to screw myself out of the x4. Though, upon further review, I could have done an 8 combo with both the purple and yellow panels that were right there to make up the two rows I was short on, but I was fixated on comboing with the purple panels alone to start another chain quicker. In retrospect, I think it was a good decision because the extra time I gave Blargg let him lower the stack one more row so extra garbage was in order and I had a time-lag chain already in place, allowing for a really quick x3. I suppose I didnft need to do the backup combos at the end since Blargg wasn't doing anything of worth. The only thing I could have possibly produced at that point was lag and, believe me, this game gets FULL of lag when a lot of things are happening at once. All the same, I still ended up with a 0:17. I'm sure I could have gotten 0:14 without the screw-up, but if running this series has taught me anything, it's that I should not try too hard to look a gift-horse in the mouth.

Total time so far: 1:37 (-23)

Stage 7 - Lunge Fish - 0:19

I often say on my stream that I don't even know how I see half of the things I see. The 6 combo with the greens and purples was easy enough, but the follow-up x2 with the yellows was, upon further review, not the first thing I would have used (I would have done something similar with purples), but it turned out to be the quickest means of getting the 5 combo on the chain. My eyes went to the blues in the second row from the bottom, but I had to maneuver a bit to make a proper bridge for the other blue panel to fall in place for the x3. The red panels on the right were in prime position to line up for the x4. Ifm not sure what compelled me to drop the green panel before the x4, but it ended up putting me in perfect position to line up a 4 combo for the x5. At this point, I glanced over at the garbage I had made and realized that I have sent what I usually like to send so I cut off the chain.

Afterwards, I go to my tried-and-true strat of spamming backup combos to fill in any potential remaining space. It turns out that I needed every single one of them to fill Lunge Fish's stack because he decided to go into a panel-clearing spree before surrendering the 0:19.

Like I said before, this game loves to overwhelm you with information. It's hard to focus on what you're doing while still keeping an eye on your opponents actions. Looking back on this stage, I cannot help but wonder what would have happened had I decided to drop my garbage at a different time. Would I have been able to get away with dropping it earlier since Lunge Fish's stack was so high? Would it have helped to keep my chain going longer had I paid closer attention to what Lunge Fish could clear? These aren't questions that one would normally spend too long dwelling over on a sub-0:20 stage, much less in the heat of the moment, but I find it fun to think about in retrospect.

Total time so far: 1:56 (-24)

Stage 8 - Raphael the Raven - 0:18

It is at around this point where the AI tends to get really aggressive with raising its stack. You saw a little bit of it with the Lunge Fish I got, but I find that it more commonly starts with Raphael. If I am cognizant enough of this, I can drop my garbage early and chance getting a kill that is often less than 0:15. Strangely enough, this tactic works for me less often when I am TRYING to make it work. Like a lot of things in life, the fastest stages in this game tend to happen on accident.

It's so hard to look at this opening stack and NOT notice the group of yellow panels on the right side just screaming at you to match them. Upon further review, I could have taken the extra time to set up a T-shaped 5 combo with them, but considering it would have taken some time to move a yellow panel in place for that, I think it was fine to go ahead and do the 4 combo to start. From there, I saw the green panel on top and went to line up the greens underneath for the x2. In the process of setting up the x2, the last color that I saw was purple and went to grab as many purples as I could to set up a 5 combo on the x3. I honestly didn't believe I had enough time for it. I should be fortunate that I had blue panels right there to drop one in for the x4. Afterwards, I feel like I had to tap-dance around all of the red panels I had at my disposal to avoid screwing myself out of the x5. I got so preoccupied with that that I ran out of panels to chain with. Upon further review, I could have moved some purple panels around for the x6, which I would have liked considering I was short one combo. Instead, I used those purples for an 8 combo in my patent-pending combo spam backup strat. I actually found myself underwhelmed with the combos I followed up with and thought that it wasn't going to be enough. To my surprise, Raphael raised his stack close to the top so that the last little insignificant 4 combo I sent (which I did not remember sending at the time because I was in such a rush to raise the stack) ended up being precisely what was needed to fill his stack and kill him for the 0:18.

My luck here was absolutely INSANE, by the way. It is rare for an AI to not make a single combo or chain for an extended period of time. Raphael did not make a single combo or chain for the entire match. I don't care if I did anything horribly wrong; I just count my blessings and move on.

Total time so far: 2:14 (-26)

Stage 9 - Hookbill the Koopa - 0:32

It's hard to go through a run of this game without at least one stage going haywire. It's not a matter of "if", but "when". When it goes bad, especially in the late-game, you just hope and pray to keep the stage under 0:45. Trying to play catch-up in this game is vicious.

Very quickly, I put 2 and 2 together with the 4 combo red panels on the right & the yellow panels on the bottom and made an effort to chain them together. Luckily, I had two more yellow panels in the rows below to make a 6 combo with my x2. The last switch to make the x2 had a purple in it so I grabbed as many purple panels as I could and managed to make a 5 combo with them for the x3. With reds below, I grabbed to set up the x4 and then noticed on top that three greens were going to fall in place for the x5. With all of the combos and chains I like to do in place, I dropped my chain there and proceeded into combo spam mode.

Looking back at what Hookbill was doing, I don't think there was a single thing I could have done differently. Hookbill made enough combos and chains to buy himself quite a bit of stop time to lower his stack. Had I dropped my chain after the x2, it's hard to say if Hookbill would have actually changed his behavior, but with what he was doing in the stage, there was not a single window of opportunity large enough to actually kill him. This is why I say that it is ultimately up to each opponent to determine the fastest you are able to go.

Hookbill proceeds to clear things and lower his stack before lining up green panels to clear on my garbage, a substantial time waste within the context of this run. He mercifully runs out of time after lowering the stack one more time for the 0:32. As much as I wanted a run without a 30-second stage, this one I didn't think was too much under my control and if this was the worst thing to happen in the run, then I would be able to sleep at night.

Total time so far: 2:46 (-14)

Stage 10 - Naval Piranha Plant - 0:15 (!)

Upon further review, if I wasn't tilting so much from the Hookbill fight, I probably should have picked up on the juicy purple and blue 6 combo on the left side, which I could have easily followed with a yellow 4 combo for the x2. Instead, I got fixated on the green panels on the right for the opening 4 combo. With a purple on top, I moved more purples in place for the x2. I saw blue next on the right side for the x3 and also had time to notice purple on the left as an additional match for the 6 combo. I then picked up on the yellow above the purple and decided to line those up for the x4. With a red above and a red below, I just decided to take it easy and fit a red into the falling stack for the x5. It's actually harder to do than it looks because you can mash the button too fast like I did on the Wiggler stage or have lag drop an input and miss the timing. The panels fall pretty quickly, giving you little time to actually switch something in front of them. It's much, much easier to do in Pokemon Puzzle League... or any other later game in the series, for that matter (except maybe Game Boy Tetris Attack, but that has so many other issues).

I could have made an x6 with the falling purple afterwards, but I had a green T-shaped 5 combo ready to set up and wanted to clear all of the blue panels on the left to make it an 8 because I was short on combos and knew that I had to make it up quickly -- that and I love 8 combos to a fault at times. It turns out that I needed every single combo. Piranha went into stack-lowering mode and, despite making an x3 chain beforehand, ran out of time before completely lowering the stack for the 0:15. That certainly does not happen often.

Total time so far: 3:01 (-19)

Stage 11 - Kamek - 0:10 (!!)

Kamek is so strange. He is frequently the worst offender when it comes to aggressively raising his stack. He is fast enough to deal with it and chain a great deal with the extra panels most of the time, but it can backfire on him if he has no stop time. On multiple occasions, I have been able to drop a small amount of garbage on Kamek and finish him in under 0:10, but most of them have been on accident. Also on multiple occasions, I fall behind early and fail to defeat him in a minute. So many different things can happen here.

My 5 combo senses tingled at the start with the red panels and putting them in column three. Upon further review, I could have just as easily made a cross-shaped 5 combo with a yellow panel on the far right for the x2, but I decided to go with a 4 combo with the yellows in the two rows above it to set up the L-shaped 5 combo I knew I had with the purples on the right for the x3. For reasons I cannot comprehend, I thought that I had an easy match with yellow (I actually had it with red), but my setup resulted in an attempted two-frame switch to make a yellow 4 combo for the x4, but I didn't get there in time, putting me instantly into combo spam mode.

Oh, did I mention that Kamek likes to aggressively raise his stack? He had made an x3 and a 6 combo in quick succession, giving him plenty of stop time, but as soon as he noticed that my initial attack was short, he raised the stack to the top. He only had one way to clear the garbage on screen and could not get to it in time for the 0:10, a ridiculously rare time for any stage, much less the final stages. The AI does tend to aim for combos when trying to clear garbage, but in this case it was Kamek's only option and a lengthy one. Thank goodness he likes to raise his stack.

Total time so far: 3:11 (-29)

Stage 12 - Bowser - 0:26

Bowser is the real deal. His speed around the field borders on superhuman. It is almost commonplace for top runners to occasionally spend two minutes on this stage alone. Combine that with the pace I was on and the nerves were definitely present.

This was an unusual case where I was able to think really far ahead. I saw the red panels on the right, the two purple panels on the bottom on either side of the red, the purple panels above the red, the blue panels on the bottom, and the blue panel on top and put a plan to action. Upon further review, I could have made my life a little easier on myself by lining a 5 or 6 combo in column four, but I decided to start with the 4 combo in column five. I lucked into not only an easy 5 combo with the purples for the x2, but also a very convenient lag chain with the greens for the x4. The yellows lined up very nicely for the 4 combo x5. I actually didn't want the reds to match for the x6 since Bowser's stack was so high, but considering I was a combo row short it's more in line with what I usually do.

I wasn't going to be able to do a damn thing to speed up Bowser, though. He led with a 7 combo right into an x5 chain. Then he made another chain in his efforts to lower his stack. He built up so much stop time that there was absolutely no window for him to actually die. There may have been doubts as to whether I could have killed Hookbill quicker, but there was absolutely no question as to how long Bowser was going to last. 0:26 would be the first opportunity for him to die and I had sent more than enough garbage by then. I was actually a little worried about dying at the end (even though I had no reason to be considering Bowser wasn't sending anything else) so I cleared the garbage which created a few lag frames, but that does not impact this particular timing method.

Final time: 3:37 (-23)

"WHAT?! What?!!!" was my initial reaction to finishing this run. The first thing I wanted to do was to make sure that I had actually done the run in Very Hard mode. You have to hold the up and left shoulder buttons while selecting Hard mode to go into Very Hard mode and, unless you are not paying attention to the red backgrounds between stages, you may not realize that you are playing in Hard mode instead (its backgrounds are pink). I was burned by that once before and I know I'm not the only one. It wasn't until I finally saw the red background after Bowser's defeat that it really started to sink in.

With this run, my goal of sub-7:30 real time was achieved! cyghfer did say in 2010 that he would never look at a stage timer ever again once he achieved sub-3:45 (though he never did manage to record any improvement to his 4:19 and left the game soon afterwards) and I must say that I am inclined to feel the same way. Something about these numbers in this particular arrangement just feels right, but I can't quite put my finger on what it is. Oh well. I'm sure I will figure it out someday.

This could not have been possible without the efforts of the people that came before me, cyghfer in particular for his 4:19 run serving as my inspiration to strive for continued improvement in my play. Running this game and this series may not always be the most pleasant experience, but special shoutouts go to the rest of the community that focuses on Pokemon Puzzle League for being there with me for the ride, especially to the likes of:

- bankaipwn, the person I first saw race this series on SpeedRunsLive,
- bbforky, a long-time rival who I consider to be my equal and my friend,
- FFRPro21, the undisputed Stage Clear Champion who still amazes me every time with his play, and
- Montucky, a relative newcomer who has managed to take Pokemon Puzzle League to places it has never been and continues to fight the good fight.

There are plenty of more people that I could specifically thank, but these comments have long since gone on far, far too long for a 20-year-old game with less than four minutes of gameplay.

Until next submission everyone, thanks for watching!