Gents we have done it again. I caked Bunker 2 Secret Agent 0:37 in a 10 minute session. BELIEVE it or not, this was actually a shocker because:

1) I had a completely swollen hand from arthritis (yikes) I developed a couple of days earlier and was only planning to do a "semi-casual" 30 minute session, and 2) I was in the midst of "working", running an automated test for some software that was going to take 30 minutes, and so I wasn't 100% focused.

I'm disappointed OBS didn't recognize my mic and capture the reaction, because it was pretty cool. But I DID go live immediately on Twitch for some commentary!

I've updated the goals and tier list pages... and I'm still waiting for my hand to heal so I can give S1 SA some proper attempts. We'll just have to wait! 😀

Yup, that's right! Achieved at precisely 11:28pm EST on the 18th of November, 2025. 4 hour session, about 40 hours this season (aka the past couple of weeks).

The full run can be viewed here: https://www.youtube.com/watch?v=bPBwQZDIn-M

(also how's the audio level? I realized the volume was too low in the original video, so I applied gain to fix it).

The full run with bonus webcam & mic (not that I really say anything since I was tired, although funnily enough there was a popoff after I ended the recording): https://www.youtube.com/watch?v=ms_zLxbmA1E

And a silly music vid: https://www.youtube.com/watch?v=psHq9iPMW7A

I've had the itch to play since I came back from a trip to Europe this September after a six month inactivity from the game. Plus there was 700 CAD on the line, so yeah, it made sense to finish this one up after spending about 40 hours last year just after the 1:47 on 00 Agent. 3 122s and a handful of 121 fails this season, most of which had a failed Ouro Shot. So that's about an 80 hour grind combining the 40 or so from last year with the 40 over the past couple of weeks, which is actually quite good for an untied in 2025. More importantly, I think I might be done with Train. 0:55 and 1:21 are about as good as you can pace without wanting to paint the ceiling, and there's no real incentive to effectively dupe 1:21 SA pace on 00 Agent. Note the run features the infamous audio glitch! I'm not sure it actually "helped" me this session, since I had terrible lag throughout, all sorts of dumb guard bullshit (getting backed in the second car multiple times in a row is always fun) and wasn't generally focused on the gameplay.

I don't think I'll make a commentary video of the run, but the tl;dr is this run was "standard 1:21" pace to the 4th car, which is nothing new. That's like once every 15 minutes on a good day. BUT then I got an incredible boost that warped me past the first door in the green car, which PROBABLY made the run barely 1:20 pace at that point. Unfortunately that boost usually comes with lag and pushes you in a way that makes the first of the two double doors tricky to open, which naturally I fucked up. Still, I had the music change at the corner of the brown table in the second room and a clean last brake and Ouro shot. I KNEW it was 1:21 pace after I killed the two ZMG guards. Of course the run doesn't end there since you can get boosted past the Train (which nearly happened on my original 1:22), or bungle the leave, which sounds embarrassing but is very possible and has happened to me under the pressure, especially when every tenth counts for the leave. A 30.63 fade is pretty good but I wasn't sure whether the cinema would pop up since I didn't pay attention to Nat's earlier message, since I was too busy (successfully) trying to do the nifty dual-wielding trick. ;) I was pretty confident this was 1:21 after the cinema popped up. What a nice endscreen.

I haven't felt the urge to play anything else, but with Train out of the way I'll probably play a few levels here and there and see if anything else sticks. Surface 1 Secret Agent 1:46 was on my radar earlier this year, but I lost interest in that one quickly. Might be worth a revisit to break the 24:00 barrier on SA. It is slowly climbing up the LSU rankings, which speaking of...

...18 days until Facility 0:42 ties Dam 00 Agent 1:55 to become the Longest ever LSU... holy smokes! Will it make it? I've been popping into Marc's streams every once in a while and I have seen some close attempts, so we'll see. Regardless, I'll have to think of a new poll for the right side of the site in a few weeks. And on the topic of LSUs again, can we talk about how absurd it is I've had an untied sweep on Control for OVER SIX YEARS now? They're high quality times but even so... I don't think they've been challenged the past few years if at all since I got them. Who knows if or when that'll change.

Ace out!

[disclaimer: I may sound like a sore loser at parts of this post, as part of this subjectively objective ramble]

---

It's joever, ジョ終わりだ, all GoldenEye times are effectively humanly maxed and it's time to move on to another hobby. Even the data says so. It was a good two decades while it lasted.

j/k. 

But we will eventually get there. We are certainly on the downward trend. Then what?

In 2007, during my meteoric rise to Perfect Dark champion, I had my first "I wonder if the games will even be active 5 years from now" moment spurred by reading this Wouter Jansen post on the forums. Bryan Bosshardt had just broken the 20:00 total time barrier on Agent with a glorious Caverns Agent 1:02. Naturally some forum discussion followed, in which Wouter predicted the record would eventually be a multi-way 1:01 tie some years later, followed by a 1:00 in 2020. An incredible post from 18 years ago that aged like wine. I didn't think about the lifespan of the games at the time because I was simply enjoying the present and achieving casual WRs like routine clockwork. And then over the next decade, with the explosion of Twitch, the-elite saw three new waves of eliters, one in 2012-2013, another around 2016, and the last in 2017-2018. More players meant 1) more competition, 2) more strategy discoveries, and 3) more gaming discussion, which led to an surge of untieds across both games. But that bubble eventually burst and there's now been a gradual decline in all three of those domains since.

Up until 2022, the most hardcore untieds would often take me no longer than 60 hours. That number has doubled in the years since. Bunker 1 Secret Agent 0:19 (2022) took 120 hours. Frigate 00 Agent 1:03 (2022-2023) was over 100 hours. Silo 00 Agent 1:20 (2023) and Train 00 Agent 1:47 (2024) took 100 hours each. Caverns Agent 1:00 supposedly took Illu 1000 hours! Granted, these records are so much better in execution than any runs of the past, but grinds feel more exhausting than satisfying these days, and the Excitement of The Endscreen isn't as thrilling when you've failed your goal time countless times and eventually hallucinate what you'd see on the screen. Nothing today quite compares to the feeling of completing a good Train Agent run expecting a 1:03 WR dupe only to see 1:01 unexpectedly appear on the endscreen and then popping the tape back into the VCR to rewatch the run with joy before creating a disclaimer WR post warning of heart attacks for anyone daring to scroll down to the WR post and be equally as stunned because they had no idea you wanted to lower your own record because streaming your attempts wasn't a thing in the good ol' days.

In spite of the increased time investment required for records, there's a strategic art to choosing what to play and when to play it to avoid burnout. If I didn't know this trick I wouldn't be getting WRs every season I boot up the game two decades later. ;) I'm still two seconds away from breaking the 24:00 barrier on Secret Agent, and there's a handful of untieds that I've listed on my goals page still in the cards.

Then there's Perfect Dark speedrunning, which always felt like the cooler, higher IQ, and more rewarding game, until everyone's visual cortex was assaulted with hundreds of pause buffers per speedrun, best exemplified by this GoldenEye X run I watched a few weeks ago that gives you insight into the mentality of your average new wave Perfect Dark runner. Then the game lost its charm with out-of-bounds discoveries on high-skill levels where pause buffers became effectively necessary for optimal execution. In fact, I knowingly Brazen Bull'd myself when I pioneered the out-of-bounds strat on arguably my favourite level (Pelagic II) earlier this year, even predicting such a thing would happen as far back as 2022. It won't be long before we find and use (albeit at a much slower rate) out-of-bounds tricks on the other remaining high-skill ceiling levels, such as Extraction (... and as I wrote that, I realized one already exists one on Special Agent, yet to be used in a run) but that's simply because, at least for me, the pendulum between discovering new timesavers and grinding levels in the traditional speedrunning sense has tilted in favor of min-maxing every exploit in the game's code, which, to be fair, is only natural when you've optimized the levels the way they were played prior to any out-of-bound discoveries anyway. There's still at least room for optimizing full-game runs across both games, and in particular for Perfect Dark since it requires an amalgamation of various skillsets not seen in any individual level and actively punishes pause buffering. I'll probably stream sporadic Perfect Agent full game attempts since everyone always finds them interesting and there's room for improvement there.

But every now and then, as if blessed with a natural drive to maximize the equilibrium between casual play and speedrunning, I also get the urge to pause a grind and randomly casually explore levels, find new tricks, browse Detstar to try some insane glitches I probably haven't tried in a while, play ROM hacks, and theorize ideas for challenges and tournaments (imagine a one or two week long 'GoldenEye Olympics' event every summer with matchcup-like mini tournaments, shooting and strafing challenges, etc. Would be cool, right? Someone needs to workshop this concept into existence). I've also had ideas for interesting ROM hacks that mimic TimeSplitters 2 challenges, such as a 'stealth mode' hack where you'd need to exit a level without being seen by any guard. And then that's a challenge for *sixty* GoldenEye levels. A couple of years ago I also proposed the idea of a GoldenEye randomizer on the-elite forums, the-elite Discord, and the GoldenEye Shooters forum. It's not like other speedrunning games don't risk suffering the same heat death fate, do they? Here's how I outlined my vision for it:

"The GoldenEye randomizer is an innovative mod that transforms the game into a refreshingly dynamic experience by scattering items across levels, requiring players to beat levels in an unorthodox order to complete the game. In this proposed mod, three random levels are unlocked at the very beginning of the game, one of which is Dam, and two others selected at random and spaced a few levels apart (e.g., Dam, B2, and Caverns, but not Dam-B2-Statue, since B2-Statue are adjacent to one another). The goal would be to unlock and complete all 20 levels, but key items needed to progress past parts of a level (such as keycards, the Door Decoder, Flight Recorders, etc.) are placed in random locations on other levels. For example, in this "Dam/B2/Caverns" selection, the keycard at the start of Facility could be placed somewhere on Caverns, so the player would not be able to beat Facility immediately after Dam, and instead have to play Caverns, find the "Facility keycard" there, and then revisit Facility (items would carry over across levels). Then, to continue the previous example, the ignition key on Runway or large key on S1 could be hidden somewhere on Statue, so players would be required to explore the "post-B2" set of levels to complete Runway and S1 in the first row of levels. Such randomization would be designed in a way that requires completing levels in an out-of-order sequence, revisiting levels multiple times, and exploring regions of levels not typically encountered during speedruns. For added unpredictability, the mod would also shuffle guard weapons, guard health, body armor positions, alarm and camera positions, and even mirror the layout of levels (have you tried the Mirror Mode ROM hack?). Logic would have to be designed such that players cannot get softlocked."

I personally feel this would be a fun, timeless experience when there could be so many combinations of logic possible to keep it fresh for years. I've floated this idea to other GoldenEye speedrunners who've similarly express interest in both playing and racing them every other week in the same way Ocarina of Time randomizer runners do. And sooner or later we could see a Perfect Dark equivalent (especially since PD already has a PC port) which would be even more exciting considering how much more vast, nonlinear, and 'broken' that game is. There is a fellow who has been working hard to bring the GoldenEye randomizer to fruition... I'm not sure when it'll be out but rest assured I will probably be the keenest I've ever been about the game in years when it does.

These games will never not be fun for me to play. Freestyling the movement is simply too addictive. But I would love for there to be more ROM hacks that focused on adding gimmicks to and spiking the difficulty of the levels as they exist, TimeSplitters 2 style, for when The Last Untied™ happens.