1: History/Introduction
I first discovered SDA, and by extension speedrunning in June of 2010 through SirVG's Super Mario Galaxy Uber Challenge videos, as he'd mentioned the full list of challenges was on the SDA forums. At around the same time I'd been replaying old games from my childhood, and had most recently played through Ape Escape. Naturally, the next step was to watch the speedruns of those games, see what people way better than me would do. However, there was no run of Ape Escape at the time. So, not understanding what the forums were for at the time, I made a thread basically asking someone to do a run. I think it was scenic and Caveberry, though, who basically said they were looking forward to my runs. My runs! The thought, unbelievable as it is now, hadn't even crossed my mind. And so I began routing, running, and obtaining capture equipment. I've met a lot of awesome people through speedrunning, but when I look back it was all due to a series of contrived coincidences. Even still, I'm glad it happened, the past 3 years have been amazing and I'm not stopping any time soon.

I started by routing a segmented any% run, but did SS practice runs to get the practice done as efficiently as I could. As I did these practice runs (I think the best was a 1:15), I realized I was fairly consistent at all the strats I was using, and so I started focusing on SS runs. The infinite jump glitch was discovered, the 1 hour mark was broken, people like started running, and I got SSBMstuff to try a full run after watching some of his time attacks. At the time, the goal for those was a total under 4 minutes, and he reached it single-handedly. I wanted to do the best I could, and kept going for good times. I got a 53 in any%, took the record for 100% with a 2:13 and all throughout running the other categories, I kept going back to the time attacks as fun little diversions. I never really took them that seriously, though, until 2013 when I was skilled enough to execute many of the strategies used. My target was the modest goal of sub 5 minutes, a very achievable goal for me. I met it, then got under 4:30, at which point I got frustrated with my lack of progress and stopped playing for 6 months.

When I returned, 3:30 had been broken on the time attacks, any% was a 48 and 100% was down to 2:06. I quickly improved my any% time to a 50 (which I wouldn't submit even if I had the original video, it's just not fast enough...which is ironic considering before I'd have submitted a 52), and improved a few other time attacks as well. In December of 2013 I really started working hard on improving all of my times, taking the 100% record with 2:05:57. This was beaten back by SSBMstuff the next day, only for me to take it back with 2:02:33-again, ironic since I nearly submitted a 2:13 last year and now I'm shooting for sub 2. I also set the goal for myself of taking all of SSBMstuff's time attacks, which would surely be a difficult task. And during the bonus stream of AGDQ 2014, while SSBMstuff was attending, I was playing time attacks, squeezing out those last frames for sub 2, beating the last of the times set before me. The goal had been achieved, my total was 2:05.14. However, my Crabby Beach recording had some dropped frames making it unsubmittable, so I improved it. And some other times as well, knowing that sub 2 would happen. So it did, the time attacks now totalling 1:59.62, with most of that cut coming from re-revisiting Crabby Beach. And then I got beaten on Fossil Field, fonud a new strategy on TV Tower and knocked off another 7 seconds to 1:52.40. It may have taken me more than 3 years, and it might not be a full game run, but I finally submitted an Ape Escape run. Now I just need to get a 47 in any% and a sub 2 in 100% so I can submit runs I'm happy with there...but in the meantime, here's the result of 3 years of effort on playing this game.

2: Techniques
Gadget Cancelling: Normally, when you use a gadget there's a small cooldown period before you can move or use it again. However, switching gadgets eliminates this cooldown period letting you act again immediately. If you're fast enough you switch back to the original gadget and use it again sooner than you normally would be able to. This is most noticeable with the Magic Punch, as you can easily cancel out of it and use it again sooner than the game intended. Finally, Spike picks up the net after swinging it, even after catching a monkey, but switching to another gadget cancels this animation. This saves roughly a half a second each time.

Boost Jumping: While using the Super Hoop, jumping normally kills all momentum. However, by pressing the face button the Super Hoop is on immediately before jumping, you preserve the momentum and can cover longer distances. Double jumping normally resets your momentum to 0, but releasing the left analog stick before jumping the second time will keep most of the momentum.

Infinite Jumping: Exactly what it sounds like, with the Slingback Shooter flicking the stick after jumping tricks the game into thinking you're still on the ground, which lets you jump again. By repeating the jump and flick rhythm you can jump infinitely, covering as much height or distance as you want. This pretty much breaks the game.

Boost to Infinite: This is just an infinite jump chain with the first jump being a boost jump. As a result, you carry a lot more speed through the first few infinite jumps which in turn makes travelling distance faster.

Ghost Car Glitch: This glitch only works in time attacks due to the nature of it. How it works is you place the RC Car where you want the monkeys in the level to act as it is at the start of the time attack, then pause and restart the time attack with the RC Car equipped. This is only useful in a few time attacks, as monkeys being alerted to start their run cycle early is rarely useful. I only used it in Primordial Ooze, though a case could be made for its use in TV Tower.

Lag Reduction: Some sections of levels (most notably Crabby Beach and Hot Springs) lag a lot. Unfortunately (or fortunately, depending on how you look at it), the timer keeps going at its normal rate during this lag. It can be reduced by manipulating the camera to have less things onscreen. This is used in something like half of the time attacks in some form, as Ape Escape is a laggy game and saving those frames really matters.

3: Game Time vs. Real Time
The reason these time attacks have so much optimization to be had, at least to me, is that the energy chips (games name for them) scattered throughout the levels stop the clock. It's stopped for 1 second for every small chip and 5 for every large chip collected, this stacks for as long as you can keep the chain going. Since when you catch a monkey there's a 1.5 second animation during which the timer keeps going (and stockpiled timestop keeps draining!) this is why times such as 1.21 for 4 monkeys and 1.63 for 3 monkeys are possible to obtain. Real time optimizations wouldn't be as interesting, have as varied strategies and would max much faster. With 2 exceptions, however, the fastest real time routes are used for the time attacks in 100% runs, and trust me, those 2 exceptions are overall faster. Like with Marble Blast Gold here on the site, the timer being stopped by pickups is allowed for SDA submissions and it would make game time and real time separate categories. If anyone wishes to make real time runs, best of luck, but I won't be joining you :P

4: Individual Level Comments
Level Name: x.xx seconds, A/B monkeys, X seconds of timestop
Date set: Month Day, Year
Previous records: 13.37 [some other guy], 13.36, 2.93
This is the format I'll be using for the comments. Any of my records will simply have the time stated, any set by other people will have them mentioned next to the time. A/B monkeys means you need to catch A monkeys out of a possible B.

Fossil Field: 5.45 seconds, 4/4 monkeys, 20 seconds of timestop
Date set: February 5th, 2014
Previous records: 7.85 [SSBM], 7.84, 7.57, 6.13 [SSBM], 5.90, 5.62 [SSBM], 5.45
I was content with the 7.57, but SSBMstuff went and got a really lucky run that I had to beat. Then he beat me AGAIN, and I spent 3 hours getting this run. This level was really annoying to do attempts of because in most cases, it didn't matter how I played, the run had no chance to succeed from the beginning because of any of the enemy or monkey positions. This particular attempt was pretty solid the whole way through, nothing special. I could have saved a few frames with a boost jump to infinite jump to the final monkey, but I'd lost a run to that exact mistake before, and it's hard enough to get the required luck and not screw up, despite how short this level is.

Primordial Ooze: 3.10 seconds, 4/6 monkeys, 24 seconds of timestop
Date set: December 15th, 2013
Previous records: 3.97 [SSBM], 3.55, 3.10
This is the ghost car glitch in action. By placing the car next to the stump, the monkey jumps off to the side and is usually in a favorable position to be caught. As a result, I have to be quick about switching to the Super Hoop to boost jump over to the enemies. As much as I'd love to recenter the camera to follow where I'm going, doing so wastes about 0.2 seconds of lag frames. The enemies also like to bounce you away when you hoop into them, so that combined with the awkward camera angle resulted in a lot of slow resets...however, if the chain under the log is successful and the 3rd enemy doesn't bounce Spike away, the run would always be finished. it was just a matter of time until I didn't break the chain, and that's exactly what I did.

Molten Lava: 6.99 seconds, 5/7 monkeys, 44 seconds of timestop
Date set: January 13th, 2014
Previous records: 12.05 [SSBM], 10.xx, 8.89, 7.50, 6.99
This is a bit of a weird time attack in that the Slingback Shooter is never used! As a result, you're completely dependent on the elevator cycle. Everything I do in the first screen up to the elevator doesn't really matter, as long as I collect every chip and get the DBJ onto it. Dropping the car onto the center auto-collects a few extra chips that would otherwise be out of the way. The enemies after the elevator were in a very good position, I was a bit slower than I could have been punching them but I didn't want the chips flying off the edge. The DBJ up the slope is fairly precise, the egg monkey is a pain and the entire lava room is really hard to do right. This one was a matter of grinding out attempts until I didn't choke, and I probably wouldn't have done it if I hadn't needed another 5 seconds to get the total under 2 minutes.

Thick Jungle: 2.63 seconds, 5/14 monkeys, 101 seconds of timestop
Date set: December 11th, 2013
Previous records: 3.03 [SSBM], 2.72, 2.65, 2.63
Holding up on the d-pad to angle the camera saves a bit under a tenth of a second in lag frames, and having the 2nd jump in the DBJ be instant results in a faster bee kill. Funny thing, I thought 2.72 was maxed over a year ago when I first got it, but then the discovery of camera strats resulted in an immediate 2.65, which I also thought was maxed. However, the play during that was very sloppy and I wasted 20 seconds of real time, so I kept going for another few hours to get a cleaner 2.65. These things have a funny way of working out, and this 2.63 happened and was much cleaner than the 3 2.65s I got.

Dark Ruins: 2.18 seconds, 5/13 monkeys, 90 seconds of timestop
Date set: November 6th, 2013
Previous records: 3.89 [SSBM], 2.18
This level intimidated me quite a bit with SSBMstuff's strategy, so I took a more critical look at the level and ended up finding a small improvement to the first area. From there I was able to re-work a few other sections to make it easier to execute and as a result, was able to finish a full chain. Anything under 2.2 is pretty rough to get at the start, so I'm happy with what I ended up with. The flaming enemy after the fan monkey usually doesn't get hit, and in this run I needed those chips too...

Cryptic Relics: 2.82 seconds, 3/8 monkeys, 57 seconds of timestop
Date set: January 13th, 2014
Previous records: 16.97 [SSBM], 13.89, 13.01, 12.18 [SSBM], 11.69, 11.00, 10.69 [SSBM], 10.59, 9.97 [SSBM], 9.82, 6.70 [SSBM], 5.08, 3.94, 3.48, 2.82
The first thing to notice about this time attack is the sheer number of improvements it's had. This is one of 2 stages SSBMstuff and I worked on at the same time, every improvement from 12.18 to 6.70 was in the span of a 6 hour night of grinding this stage. At the time, I'd conceded defeat but when I had only a few levels left to record and needed more timesave, this is where I turned. I doubted I could improve this by more than a few frames with how flawless his run was...so I went looking for alternate methods. The tunnel in the last room was possible to chain through, which meant the first monkey could be skipped. That enabled the chips on the left to be collected, but there was still a missing link. That's where the guided shot comes in, it has to be from a precise spot and lock onto all 3 birds or the higher bird will fall elsewhere and the chips can't be collected. As a result of this, however, I was able to hold the chain all the way from the first barrels to the end of the level! Of course, I was content with the 3.94 until I noticed I still needed a few more seconds to hit sub 2 overall, and this was a likely candidate.
Also, even if the route is outdated, you should all go watch SSBMstuff's 6.70 run. The level is play is insanely high, and I doubt I'd have beaten it without finding a new strat. http://www.youtube.com/watch?v=LhRvSkTluBw for those curious.

Crabby Beach: 16.37 seconds, 6/8 monkeys, 54 seconds of timestop
Date set: January 13th, 2014
Previous records: 23.90 [SSBM], 19.74, 17.75, 17.25, 16.37
Screw this level. Lag plays a huge role in this level, so in the entire first screen I have to play very carefully and make no mistakes as any I would make get multiplied by the lag that would ensue. I have to be moving when I engage the Sky Flyer but neutral on the left analog when I switch to the net or I can't swing it. From there, if the bridge is up and I don't fail the boost jump to catch the UFO monkey, it's an easy finish. The camera trolled pretty hard this run, so it's not as fast as it could have been but I never get all the chips in the water and the bridge was mostly extended which makes it much faster than my other attempts despite the rough ending. I tried to improve further for another couple hours and never even came close to this run, so it's the best I can do right now. Also, this was the run that pushed the total under 2 minutes, and of the last 5 seconds this took out 2/3s of those. It's just a shame the level is such a pain to do...

Coral Cave: 2.69 seconds, 4/8 monkeys, 34 seconds of timestop
Date set: January 12th, 2014
Previous records: 7.87 [SSBM], 6.21, 5.47 [SSBM], 5.33, 3.52 [SSBM], 2.69
This is the other level SSBMstuff and I went back and forth on, when I developed the strategy that is actually still used in this run with no changes. I just wasn't skilled enough to pull it off until I went back to try to improve all of my times. This is actually a pretty fun level to do as well, even if it's really hard to execute. The camera change at the start helped a bit too.

Dexter's Island: 10.87 seconds, 3/11 monkeys, 66 seconds of timestop
Date set: January 12th, 2014
Previous records: 17.52 [SSBM], 14.72, 14.26, 10.82
Another stage where camera manipulation is required to get a good time. I specifically alert the monkey on the boat before I get to him so that he will always be in the same place on the boat. Inside Dexter, I make a break for the bones and then chain through the rest of the level.

Snowy Mammoth: 15.84 seconds, 4/6 monkeys, 34 seconds of timestop
Date set: January 11th, 2014
Previous records: 24.89 [SSBM], 18.xx, 16.92, 15.84
In order to pull this off I had to combine strats from any%, any% no infinite jump and make up a new one for the ending. The jump off the stick saves about a second, the DBJ after the first monkey is from no IJ and the iglooless catch strategy I found while trying to make it more consistent. After the 2nd monkey, the slide down the slope is blind but I still need to get all the chips on it. The snowmen are one of only 2 enemies that take 2 hits from the Magic Punch, which results in semi-tight timing for getting the second hit without the snowman jumping. This is definitely one of my favorite levels to play though!

Frosty Retreat: 1.21 seconds, 4/9 monkeys, 74 seconds of timestop
Date set: January 12th, 2014
Previous records: 1.30 [SSBM], 1.21
The really important thing to note about this level is the use of the RC Car throughout it. Dropping it onto the chips at the start saves about a sixth of a second, but results in a lot of restarts if you lose the timing on the drop. I spent about 3 hours trying to beat or tie SSBMstuff's 1.30 time, and eventually tied it. Unfortunately I was dissatisfied with just a tie, and kept going for 1.26. I ended up losing quite a few 1.28 and 1.26 runs near the end, and I even lost a 1.23 to missing the swing on the last monkey. Another few hours and a dinner break later though, I saw this 1.21 at the beginning. I successfully manipulated the ground based enemies to stay out longer so I can kill them with the RC Car, didn't mess up the gadget switching, manipulated the 2nd set of enemies correctly, didn't get snowballed (happened once), die on that slope (likewise) or miss the last monkey (happened several times). And the first 1.21 I ever saw in over 5 hours of grinding the stage, and probably the only one I'm ever going to see, ended up being my second completion of the strategy. Funny how these things work out.

Hot Springs: 3.67 seconds, 4/9 monkeys, 53 seconds of timestop
Date set: January 13th, 2014
Previous records: 19.47 [SSBM], 5.77, 4.05, 3.67
Even more snowman cancelling, even more camera manipulations, and a completely different route than what SSBMstuff used. Killing the bird and collecting all the chips in the same punch is a bit tricky, but beyond that the rest of the route is pretty easy, contingent on the last monkey not being up the stairs. In this run I thought he would be but then I noticed him on the stairs. I thought I had missed but somehow, somehow the hitboxes had mercy and I saved another 0.38 towards sub 2.

Sushi Temple: 15.45 seconds, 5/12 monkeys, 28 seconds of timestop
Date set: December 13th, 2013
Previous records: 18.05 [SSBM], 15.45
Turning the camera up saves about 2 seconds. It's a really easy and straightforward time attack, but not failing the chain is actually pretty tight.

Wabi Sabi Wall: 3.42 seconds, 3/10 monkeys, 34 seconds of timestop
Date set: December 15th, 2013
Previous records: 3.55 [SSBM], 3.52, 3.42
Even despite the smaller mistakes, I still never broke the chain. I've also never seen a faster start than this.

Crumbling Castle: 1.63 seconds, 3/20 monkeys, 60 seconds of timestop
Date set: December 15th, 2013
Previous records: 1.92 [SSBM], 1.89 [Paine], 1.85, 1.84, 1.77, 1.75, 1.74, 1.69, 1.63
Once I found out that turning the camera could reduce lag, this one fell from 1.77 immediately. From there I just optimized to try to get it to the same level of optimization as the 1.77, and I'm pretty sure the 1.63 opening does just that. I was so shocked when I saw it, though, I started playing really sloppily, but the timestop never broke and I couldn't replicate or beat it.

City Park: 3.05 seconds, 2/13 monkeys, 40 seconds of timestop
Date set: December 16th, 2013
Previous records: 3.35 [SSBM], 3.31, 3.30, 3.26, 3.23, 3.21, 3.10, 3.06 [Mewto], 3.05
I optimized this really early on, but like Crumbling Castle turning the camera helps a lot. This is also the level where camera strats were discovered, when Zorasknight was doing time attacks and was consistently beating my 3.23 on emulator, but failing the chain after. I messed with it on console, squeezed out another frame to get the 3.21 but couldn't get 3.20, let alone 3.1x. Then it hit me: I already knew emulators emulated the lag in this game very poorly, so maybe City Park was lagging imperceptibly? Maybe by doing what Surreal does in AE3 and turning the camera up, said lag would be negated? And that's what happened, and I got an instant 3.10. Unlike Crumbling Castle, someone on my Youtube commented and uploaded a 3.06 after I set my 3.10. After that, I got 3.08, 3.06 and 3.05 on consecutive runs.

Specter's Factory: 2.99 seconds, 3/10 monkeys, 59 seconds of timestop
Date set: September 30th, 2013
Previous records: 7.97 [SSBM], 5.53, 3.90, 2.99
This is probably the level that shows gadget cancelling best, just due to how many times the Magic Punch is used in the first screen. Here I was fortunate enough to not have the UFO monkey jump in his UFO, which saves time on lag, and the last monkey didn't notice me. The first and 3rd monkeys both really like screwing you over and if either one does, the attempt is a wash. To mitigate this I do shoot an explosive at the first monkey.

TV Tower: 4.05 seconds, 4/12 monkeys, 96 seconds of timestop
Date set: February 5th, 2014
Previous records: 12.75 [SSBM], 11.25, 10.82, 4.05
So the 10.82 I had was two tenths of a second off of optimal for the old strategy used. Then yosuke released the 3rd part of his all monkeys and Specter Coins TAS, which caught the first monkey in the basement section through the floor. Zorasknight and I tried for about 20 minutes to replicate it, without success. Early February, SSBMstuff went for the catch in an any%2 run and got it. With a 2nd video to go off of, I was able to get the catch within a few minutes. I had already theorized a 4 second route based off of that catch and the chip duplication I had accidentally discovered before, but never understood why it worked. On the 4th, however, all the pieces fell into place. I was getting the floor catch on the 6th ground pound on average, which isn't bad since it's semi-precise and a 1 frame window to swap (Ape Escape outputs 30 FPS but accepts input on 60). The worst case scenario would be missing several times and having to restart. However, the strategy I had come up with was too hard for me to execute, due to having the chip dupe before the ground pound. On the morning of the 5th, SSBMstuff told me a different strategy that could potentially hit 3.xx, but I was unable to chain it all the way through. From there, I realized that hybridizing the 2 strategies would create one that would, though difficult, be much easier to perform. I began the grind.
This run was a bit rough at the start, bit I got the ground pound 2nd try, with an accidental crawl. I estimated I would need to dupe the 5 chip 6 times, to collect it 7 times. This is a unique case within the game, as the reason it doesn't disappear enabling multiple collections is actually quite simple. It works by making sure the chip still exists as an entity upon hitting the screen transition-this is why I wait for a moment before going back, since if the 'new screen' animation is still playing, Spike won't go into the next screen. I didn't mess that part up despite lots of nerves from the fast 2nd monkey, and also didn't screw up the movement into the final room. I did screw up the drop onto the chips, but thankfully the game didn't punish me for it, and infact I still landed on the 5! As a result, the last monkey was caught immediately and the most involved time attack had one of the biggest cuts observed, from nearly 11 seconds down to 4.

Monkey Madness: 7.99 seconds, 3/24 monkeys, 22 seconds of timestop
Date set: January 13th, 2014
Previous records: 13.31 [SSBM], 12.25, 10.00, 8.13, 7.99
I learned a lot about general movement for this game by doing this time attack. Boost jumping costs a bit of time, and dropping down is faster than boost jumping to infinite jumping over a gap when possible. As a result I've been able to make a lot of smaller optimizations throughout everything, and when I needed a bit more time to hit sub 2, I came here and played for a bit, and managed to nab this small improvement. The time attacks just get less interesting past Snowy Mammoth, mostly due to how simple they all end up being. Still, putting together this table was really fun and has definitely helped my overall play.

Total time: 1:52.40
Total monkeys caught: 74 (animation lengths 1:51)
Total timestop collected: 990 seconds, or 16:30

5: Possible Improvements
Fossil Field could have a half second knocked off with much better luck. Crabby Beach could hit 13 humanly on a perfect luck perfectly played run with a 1/100 crab instant kill, and a TAS could hit 11 or even lower by incorporating the crabs.

6: Shoutouts and Stuff
First and foremost: major shoutouts to SSBMstuff. If he hadn't done the time attacks at all, I wouldn't have gotten him into full game runs, and we wouldn't have anywhere near as many awesome strategies as we do now. The competition and times to beat were also really fun, I just wish we had something else we could go back and forth on instantly like we did on Cryptic Relics. In many cases the routes you came up with were spot on and I could only find small improvements, and almost all of your times were fairly difficult challenges to beat.
Shoutouts to Paine907. We never directly competed, but I know you and SSBMstuff competed on these years ago. That definitely helped push the bar lower, and I'm glad to have done my part.
Shoutouts to Jumpyluff. He beat my old time on Primordial Ooze in August and I came back from my hiatus and beat it back within 5 minutes, then kept going and improved the record...while not recording because my laptop was dead. That's what got me back into running this game.
Shoutouts to Zorasknight for being generally supportive and interested in playing the time attacks. He got me to go back to City Park and as a direct result the discovery of camera strats, and pushed me to improve Snowy Mammoth and Monkey Madness.
Shoutouts to Surreal for using those strats in another Ape Escape game which helped me make the connection.
Shoutouts to Mewto for beating my old City Park, resulting in me cutting off another 3 frames.
SHoutouts to yosuke77 for TASing the game, notably finding the catch used in TV Tower.
Shoutouts to anyone else I might have forgotten, for reasons that I can't remember right now.

I hope you enjoyed watching these as much as I enjoyed making these.
-IHNN