Additional notes for The Luckiest GoldenEye Speedrun Explained

Note 1 (3:30): You might expect this to be 41% faster (√2) but Bond's forward speed and sideways speed aren't the same. In general, his forward speed to sideways speed ratio is roughly 1.25 : 1, but this can vary based on lag and due to a bug in the running animation.

Note 2 (4:20): The 2.4 control style has an added bonus in some levels, the most important being Depot. Normally, the controls are mostly locked during cutscenes, but they aren't properly locked for the second controller. So, not only can Bond accelerate to full speed during the intro, he can also perform actions like shooting, activating objects and reloading if the buttons are pressed on the second controller. On Depot SA and 00A, you need to enter a hangar to pick up a key and destroy stuff, but if you get the cutscene that shows the inside, you can shoot the screen which gets a headstart on the objective and lure the guards out, conveniently opening the roller door for you. This kind of strategy is useful in other levels like Streets SA/00A, Caverns, Bunker 1 SA, Aztec DLTK, and it does have some potential in Frigate for a guard lure, but it isn't commonly used.

Note 3 (4:46): Bond's movement speed is very complex to calculate because it's momentum based and fluctuates depending on his running animation, which has a bug in its loop. Lag plays an important role in the fluctuations and generally tends to make Bond's total average speed worse.

Note 4 (5:06): There is body armor on Agent and Secret Agent, but none in 00 Agent. There's no more body armor in Frigate on the Japanese version regardless of the difficulty. In the speedrun context, body armor is out of the way and health is rarely much of an issue - if it ever is, the run was most likely going very poorly. Speedruns rarely get body armor, unless the time investment is minimal and the level requires taking a lot of damage (Streets, Caverns and Aztec are good examples).

Note 5 (7:20): If Bond is running into an wall at an angle, he will of course still move, but there's no need to dive into how that is processed.

Note 6 (8:06): It's not entirely understood what exactly causes the big lag spike, because switching to unarmed (which has no gun model unless Bond has the sniper) tends to lag the game the most. Also, the movement factor is applied on the frame after the lag spike, not immediately as it's happening, which causes a bit of a stutter where Bond moves a small amount after a big lag spike, then warps ahead significantly.

Note 7 (8:17): The numbers here can vary quite a bit based on a lot of factors. Bond often doesn't have full speed, lag is variable, speed varies based on the running animation, and so on.

Note 8 (8:27): While most warps require some amount of lag, it's possible to make it through without additional lag but with the short burst of speed that R-leaning gives. An R-lean is when you press R + C-left or right to make Bond sidestep. This sidestep is quite short, but it briefly gives Bond a lot of sideways speed. It's often used to propel objects further than a normal throw, and it's used at the start of Bunker 2 to clip the jail door without additional lag.

Note 9 (11:49): In Frigate, the guards' navigation graph is a lot more complete, but hostages only move along a subset of it that's very limited.

Note 10 (12:20): Rear right is technically worse than Mid right for lower deck hostages in theory but not necessarily in practice because of the risk of loading hostages as they're trying to escape. Similarly, Rear left is technically better than Front right in theory for upper deck hostages, but the difference is basically zero and hostages run the risk of loading when trying to escape.

Note 11 (12:52, 13:12): These spheres are purely an incompetent attempt at a visual representation of the trigger areas. In reality, they are higher than that and hostages only need to collide with them, so they can hit them with their heads. This representation was simply eyeballed to give a bit of an idea of how it might look like.

Note 12 (14:12): Because of an error in the loop during a reroll, the worst escape point is skipped in the initial RNG check so the best escape point absorbs its 41/256, bringing it to 81/256. Also, if the RNG is above 220 (35/256), the RNG is rerolled and then goes through its process normally, which gives another possibility to select the best escape point, this time with its normal 40/256 probability. Therefore, the probability is 81/256 + (35/256 * 40/256) = 33.8%. The worst escape point is almost impossible to get on a reroll because its probability is 0 + (35/256 * 41/256) = 2.19%.

Note 13 (14:44): In reality, there are more than 16, but those are the main ones you encounter when shooting guards. There's also one for when a guard dies with his back against a wall, or when they get crushed by the tank, and several different animations for explosions coming from different directions.

Note 14 (14:51): The gut animations are shared with if they die to a hit in the gun. How does that happen? If you're playing on LTK! With 0% health, guards die from literally anything, including a hit that does 0 damage. The only exceptions are the gun doesn't get animation #16, but can get #11.

Note 15 (15:29): I'm aware that odds and probability are not the same thing. Odds refer to the ratio between event A and event B, and probability is the ratio between event A and all possible outcomes. For example, having 2 to 1 odds is the same as having 2/3 probability, because the ratio between success (2/3) and failure (1/3) is 2 to 1. Therefore, calling it the base odds when we're effectively dealing in probabilities is a bit of a misnomer. I decided to use this name simply because it's much easier to say and repeat than "base probability" when I'm already saying the word probability 51 times in this video. Odds are a needlessly confusing roundabout way to express probabilities anyway, so there's no reason to use the word for what it technically means instead of simply using it as a shorthand for probability.

Note 16 (22:15): In reality, hostages escaping are not 100% independent events as they can run into each other, but assuming that they are is the best thing we can do here.

Note 17 (24:41): The chances of Bridget and Oliver choosing the best escape point are 40/256 each, and Chuck has to not choose the worst, which puts his probability at 215/256. The total probability is therefore 40/256 * 40/256 * 215/256 = 0.020504 or 1 in 48.77.

Note 18 (25:35): Sam's incredibly low Front Right escape point total is very peculiar. It could be an issue with the romhack, or it could be that when he chooses that point, he sees you when he's about to escape and redirects (hostages can't escape if you're in their line of sight). It's a bit of a mystery, but thankfully, it doesn't affect the final outcome a whole lot.

Note 19 (28:47): Only the first simulation had a million runs; the others were all done using 500,000 runs instead. Functionally, it's pretty much the same, but more trials means better accuracy, especially with low frequency events.