Bunker 1 Secret Agent 0:20 was considered one of the best, if not the best GoldenEye World Record when I achieved it in 2018. It was incidentally also my only PB that year, and a white whale I had been chasing for years, which Goose covers extensively in the Bunker 1 Secret Agent SpeedLore episode. The run is incredibly clean (or "hyper-low in entropy" as I like to measure it on my Run Quality Scale), since it has fine, robotic movement and a nice 1-1-1-3 shot count for the cameras. The thought of 0:19 obviously never crossed my mind at the time; that would be like Boss thinking about Runway Agent 0:21 shortly after he had untied 0:22. Aside from its its strength and flashiness, my 0:20 also precipitated discussion about whether it would eventually surpass Dam 00 Agent 1:55 as the longest standing untied. Funnily enough that thread and this comment below were posted after I achieved 0:19 but hadn't revealed it, so I was smirking in the shadows:

The 0:19 unsurprisingly elicited a lot of reactions and made it big on Speedrunning YouTube. The time was only made possible thanks to a couple of friends who looked into using the 2.x strategy on Secret Agent, followed by a little pace-testing of my own. Interestingly, the 2.x strat was discovered just months after my 0:20, but it wasn’t until a year later that people saw its potential when Joris unhoarded 0:16, the first of its kind with 2.x. Yet nothing came of the strat on Secret Agent. Maybe it was slept on, or maybe the few folks who did look into it earlier didn't think it was any faster than the standard route. It's probably not obvious that the 2.x strategy is only faster if you keep full speed after the 3rd camera.

So then how did the "0:19 strat" come to be on SA in the first place? How was the grind itself? And how does it stack up to 0:20 in terms of pure run-quality? I share some of my personal reflection and technical analysis below.

In April 2022 I joined a voicecall with some friends (RWG, grav) and saw grav stream B1 SA. I was muted and deafened, but I could tell the two were doing some sort of strat-theorizing. The idea was you take out all of the cameras first, and then deal with both pauses at the end. I think the most shocking discovery that proved the strat could work was realizing how far along the corridor you could be and STILL complete the camera objective. I was so intrigued that I decided to temporarily end my miserable Aztec 00 Agent crouch strat grind, which had especially burnt me out since I spent a good 2 weeks playing nothing but Aztec Agent and Secret Agent just prior.

I knew right away that I had to do the trainshot-style 3rd camera for the strat to save any time. It didn't take long for me to get my first 0:19 fail and while the run wasn't legit since I missed three cameras, the pace clearly showed it was possible. 20 minutes later I got my first SUPER legit 0:19 fail, only missing the 3rd cam (which I actually DID hit, just not the lens). At that point I knew I was locked in and genuinely thought I would smash my own untied the same night.

I estimated a Reasonable Grind Time (RGT) of 20-40 hours, which unfortunately was way off since I dealt with a lot more bullshit than I expected in the second half of the level. I failed 0:19s in almost every way possible (missing each camera only, missing various COMBINATIONS of cameras, getting shot the exact moment I pressed A and thus failing the cam snap, failing to pick up the key analyzer because a guard ran into the table, etc.). I was also nervous each time I hit the fourth camera and saw the Objective A: Complete message pop up, which made the easier part of the level (the ending) harder than it should've been. Of the runs I made it that far half of them would instantly die to getting shot into the glass half a second later, something gruesome and mentally exhausting because there was simply nothing you could do to prevent that. The grind ended up taking 120 hours across the span of 3 weeks, with 10 dupes, 20ish overall A completes, and many unfortunate scams. To think I could've gotten it in 30 minutes had I hit the 3rd camera on that early 3rd cam only fail... that's GoldenEye speedrunning for you.

I've posted all of the scams and fails in the B1 section of the GoldenEye vids page, but I'll also do a quick analysis of the 20 dupes:

20 (2): Around 20 hours in. You'll notice the insane flickshot technique I did early on for the 3rd camera compared to my actual 0:19 that has a more gradual turn, which improves consistency. This run also shows the awful key pickup technique I used in the first 60h of gameplay. I also didn't preadjust my angle before the camera pause, so I had to do a horrendous 200+ degree turn after the unpause for the cam snap.

20 (3): Nothing majorly bad, just several little movement errors throughout.

20 (4 ft. swiss): Hilarious commentary. Actually a very good run other than the ugly glass stuck.

20 (5): I got "stuck" on Boris for far too long, but what the hell is that insane nade!

20 (6): Scammed at the glass.

20 (7): Weird movement around a guard by the glass. This run needed a boost after the glass but even then my position after the key analyzer unpause was pretty bad. 

20 (8): Had a shot until I got PUSHED into the side of the wall just after the key analyzer unpause, how tragic. You'll notice this one was achieved on English, while all previous 20s were on JP. I couldn't tell a difference between the two versions, and whenever that's the case I prefer English since the endscreen looks better. I made the switch roughly halfway into the grind, like 60-70 hours in.

You might've noticed, on the early 0:19 fail, that I used a KF7 burst for the last camera. I think it took me less than an hour to realize a precise PP7 shot entirely within my control was much better than being at the mercy of KF7 bullet spread. That's the kind of thinking I develop the deeper I grind a level. Compared to the 0:20 I also do a "sweep shot" for the 2nd cam, which I think is a tenth or two quicker when done nicely. I'm not sure the sweep shot is something I'd have used had I not seen it in an Alex Anderson B1 SA video titled "INSANE 2nd cam" from 2006 that I'm not even sure exists on the Internet anymore. It is some very clever non-obvious movement technique. An underrated aspect of the sweep shot is that the "residual lag" from the camera blowing up makes the gradual turn into the guard headshot a little tricky and inconsistent compared to the classic 0:20 style "backstrafe into left strafe" shot.

Cam shot consistency matters a ton on this level, since most of the grind is just... trying to hit cameras. The better you are at them, the more fun and satisfying the grind becomes (or at least tolerable). That's probably a good reason there's a mix of people who don't mind the level whereas others find it straight up insufferable.

[circa 2013]

InkosiDeVinyl: 21 is luckbox cams

InkosiDeVinyl: 116 is skillz

InkosiDeVinyl: and that's the difference

InkosiDeVinyl: perioud

InkosiDeVinyl: the end.

Perfectace05: with skillz you can get 21/22 with somewaht crap run

On an average day I was probably 4/5 with the first camera, 1/3 with the second camera, and ~1/30 for the third camera, which together meant I'd get a start once an hour. This consistency went up after I developed better muscle memory and realized "easing into" the turn for the 3rd cam instead of doing an abrupt flickshot provided greater control of movement. Contrastingly I was 1/3 with the 20 style C-left shot, so going from having ample time to line up the 3rd cam shot to doing a flickshot decreased consistency almost tenfold!

The guards in the "lobby area" suck and can royally screw you up once you hit the 3rd camera and get a run going. The first is the guard just down the stairs, who can run to the left and block your line to the key analyzer, which is obviously frustrating to see on the one hourly run you make it past the 3rd camera. The other maliciously entities are the three guards who shoot you into the glass door immediately right after the 4th camera. The first of these is the guard right of the table who shoots you after the 4th cam shot (who also RAN into the table and blocked me from getting the key analyzer on an otherwise pure 0:19 run!) and the other two are the ones left of the glass door who shoot you into the right side of the glass or right wall. In hindsight I should've injured the table guard on my way to the key analyzer. Something I'll absolutely do if I decide to dupe 0:19 in the future.

I think I had 5ish solid 0:19 pacers post-glass, which is a little embarrassing because you'd think it shouldn't take more than one or perhaps even two runs to clutch the ending. I somehow missed the cam snap on one run, got shot the exact moment I pressed A for the cam snap on another, got backed by the end guard on a third, got shot into the side wall before the door on a 4th, and had the wrong CC set on a 5th. The 0:19 also cut it close because I was in a pretty bad spot for the key analyzer pause. The run where I fucked up the CC was unbelievably comedic because it was an amateur mistake I saw coming. I had a run earlier in the session where I somehow ended up on the left side of the door and thus temporarily switched to "right strafe CC" to practice the ending if I ever landed in the same spot on a real run. I told myself to re-set left CC just after the run, naturally forgot, and paid the price 30 minutes later. That fail was just a day before my actual 0:19 and mentally fucked with me the most, so I'm happy I was able to persevere through all the nonsense, although I was more relieved than excited when I got 0:19. Unfortunately that's just how it is for most WRs and untieds are these days; they are simply too grindy.

Still, I played the level every day for 5-7 hours for three weeks straight. There's no way I'd have been able to keep up that level of consistency for the 0:20 grind, which took 200 hours across a 4 year period. That's because the 0:19 strat at least has much better flow and is more engaging to play when you're trying to TAS-shot all the cameras out of the way first. It's also more efficient to play the level this way because you're playing out more attempts per unit time. The movement is also more natural since you avoid a pause, two hideous strafe changes right after the 3rd cam and a warp under low lag. And in terms of stream watchability, the level is more suspenseful and hype to watch. You're on autopilot at the start until you hit a trainshot 3rd camera and snap back into focus to get ready for the 4th cam shot just a few seconds later.

The 0:20 strat, in contrast, has more breaks in flow, which ironically makes a case for it being stronger since it becomes a harder level to enjoy speedrunning. The level feels sluggish, harder to stay in rhythm, and less enjoyable over time. Runs were fewer and farther between. Legit fails were hard to come by. Just doing the strafe change after the 3rd camera pause and hoping to avoid a back from the guard down the stairs feels so odd now. Seeing 0:20 fails on the endscreen was also a rare occurrence. 0:19s are way easier to fail now than 0:20s were then (but this doesn't mean completing a 0:19 is easier than completing a 0:20!). A clean 0:19 pacer was usually ~19.7x, so I had 0.3s of leeway on my attempts (which of course, I used entirely to save the terrible leave on my actual 0:19). The end warp also sucked for 0:20; you had to yolo it to eke out the tenths you needed to get a 20.9x, which means you'd miss it over 50% of the time. That's how I lost a 0:20 in 2014. Although to be fair, warping to unnarmed instead of the PP7 on that run was a bad idea. Still, knowing I didn't have to deal with the uncertainty of the warp made the 0:19 grind feel a lot more tolerable. All that said, I'm surprised the strat never "took off" after I posted my 0:19, or after Goose & BK posted their 0:22s. In terms of pace, the strats are break-even at 0:22, but the new strat should be more tolerable to play, and as far as I can tell B1 SA was a widely despised level the old way.

The 0:19 requires greater aiming precision, which you can see in the TAS-like shots, and precision always require you to be in good form. I think it would be a lot harder to return to a 0:19 grind from a break than it would for 0:20. This is also why I was paranoid about my stick snapping mid-grind (something that apparently only happens to me), because your form depends on the sensitivity of the stick you're currently used to. I had thoughts about the stick snapping every other session. It probably wouldn't have been too bad to adjust to a new stick in actuality; just a extra inconvenience for a level like this, since it'd have taken a while to reestablish the muscle memory for the camera shots. That's an interesting phenomenon about this WR, and any others that require similar precision.

I personally like the 0:19 a lot more. It's more my style and looks about what you'd expect from a GE WR in the 2020s. It's a better demonstration of my precise aiming ability, and the level is more fun to both play and watch when you're getting the TAS-like camera shot out of the way first. Although this applies across all levels, I felt there was a greater degree of subtle "140 IQ" things I had to discover along the way to eke out consistency every where I could, which felt rewarding. Ironically, having played the level so much, and looking at my 0:19, I see a lot room for improvement on my 0:19 than the 0:20... though that's probably because when you introduce more refined movement techniques, there's a much wider range between an optimal execution and something that's good enough for the time, whereas with the 0:20, things were more binary.

I've thought about trying to dupe 0:19 on NTSC-J for the fun of it. Even PAL 0:19 is out there since the warp (which is always worse on PAL) is no longer a factor. 0:19s across the board (PAL/ENG/NTSC-J) would be pretty unhinged. I don't think I would have ever had the desire to even dupe 0:20 once.

The real question is though... will we ever see a wacky 2.x 0:18? Heck there's already a TAS of it. Would a "real 0:18" look flashier than the current day 0:19? Almost certainly, but will it ever happen? And will 0:19 one day make it to the longest standing untied ever? Time will tell...