The Elite Forum
The Big Three Plus One => GoldenEye 007 => Topic started by: RWG on February 08, 2017, 07:11:41 am
-
This thread will be stickied and serve as a repository for any Frigate 00 Agent runs where the hostages escaped in 1:06 or faster. This is something that none of Ace, Luke or myself, three of the greatest players of all time, have ever seen happen. However, this seems to be happening at an alarming rate these days. This is shocking because a Frigate hostage completion of this speed is perhaps more significant in the community than an Untied WR. However, these have always been extremely poorly documented.
In an effort to better document all Frigate 00 Agent 1:06 and Lower Hostage Completions, please post them in this thread for future reference.
-------------------
1:04 Speed
Luke Pettit (1:06 completed) https://www.youtube.com/watch?v=8BwXzsQesjU
Luke Pettit (1:05 completed) https://www.youtube.com/watch?v=Pe1038OXkwU
Dan Parker (1:04 failed) https://www.youtube.com/watch?v=frhqpCtN7xQ
1:05 Speed
Chebon Shayen (1:05 failed) https://www.youtube.com/watch?v=Dqbs3Pv_boc
David Clemens (1:05 completed) https://www.youtube.com/watch?v=s3Vn3WHNJno
Oscar Pleininger (1:08 quitout with Obj A on screen) https://www.youtube.com/watch?v=ULBpa3ddQkI
Jonathan Hauptman (1:05 completed) https://www.youtube.com/watch?v=f96RyeOyrSk
1:06 Speed
David Clemens (1:06 completed) https://www.youtube.com/watch?v=di0_jxYGICo
Dusky Yeudall (1:08 completed) https://www.youtube.com/watch?v=rCDFWILa4PA
Uncertain Speed
Jim Barrett (1:10 completed; 1:06 or 1:05 speed) https://www.youtube.com/watch?v=aSSy9uFH_UA
David Clemens (1:07 completed; 1:06 or maybe 1:05 speed) https://www.youtube.com/watch?v=5rUXXQwgP2E
As of September 3, 2018; we have seen 11 Total 1:06 or Faster Completions
-----------------
Please add to the list if I've missed any of these historic completions that need to be documented for future's sake.
-
Does my 1:10 count if Obj A was up for five seconds?
-
All likely everdrived
Convinced more than ever a completion with this strat is a 1/10,000+ chance event, and a good case could be made to remove the level off the ranks.
-
Is it not extremely dubious that two of the three 1:04 completions EVER belong to Pettit? Swiss was the rightful man who was owed 1:04. Clemens is sensible because Clemens finds all these "mystery strategy myths" that improve hostage odds, knowing exactly when to or when not to shoot, for example. Jimbo and Twii we can discount as one off megatroll runs where neither of them could fully capitalize on their lottery luck.
How did Pettit get two 1:04 completions? And within days? And then never again?
The chances of getting TWO 1:04 completions are 1/10,000*1/10,000 or roughly 1 in 100,000,000. You read that right, one in a HUNDRED million.
Did Pettit really use up all of the luck in Frigate 00A history? Or is there something sinister going on here?
-
Nice rare-completion wiki, I like it! That Pettit mystery :grin:
runs where the hostages escaped in 1:06 or faster.
Does my 1:10 count if Obj A was up for five seconds?
Don't you have a college degree? :nesquik:
-
I forgot to mention I also saved six hostages on my run.
-
isn't the flight recorder on Statue also everdriveable? Better remove that stage also. Maybe even facility because of Doak location? Or how about Egypt?
This topic seems like it has the secret purpose to make people doubt Luke Petits WRs.
lets face the fact that both GE and PD are games where you can't guarantee a 100% cheat free enviroment. There are many ways to influence either the game or the video afterwards. I can have 10 unspottable fake vids up by the end of the month, no problem. Your best chance at getting the truth about these Petit runs is just to straight ask for an explanation. Ask him if he actually owns an everdrive to begin with. Maybe he can share more details about the runs and build some credibility.
Also, if the chances of 1:04/5 hostage completions really is 1:10000, then it's in no way strange that three (even top) players haven't seen it happen once.
-
Was Everdrive even a thing back when Pettit got 1:05?
-
yes
-
Guess it's fake then :kappa:
-
loooool
-
Its probabillity.. even though its 1/10,000 or whatever doesnt mean you cant have 2 or 3 in a row.. its just randomness.
-
Yeah I'd like to point out the strat I got my 108 and 109s with has even worse odds. And I got both of those records within 1 try from each other. That is ridiculous lottery luck. It might honestly be the luckiest thing that's happened to anyone in this game.
-
Let me calculate the odds of that. Now this current WR strat has about 1 in 10,707 odds, so I will assume that one has something like 1 in 12,000. Could be worse, so this is a pretty conservative calculation.
P(X) = 3C2 * (1/12000)^2 * (1 - 1/12000)^1
= 0.00000624947%
or 1 in 16,001,333
To compare, the odds of winning the jackpot is 1 in 14 million. This ridiculous luck is probably why I've had no success on the level in the past 9.5 years. Complete joke.
-
Yeah I'd like to point out the strat I got my 108 and 109s with has even worse odds. And I got both of those records within 1 try from each other. That is ridiculous lottery luck. It might honestly be the luckiest thing that's happened to anyone in this game.
nahh thats 206 ;)
-
2:06 on what? I said 'this game' so if you're referring to Attack Ship Agent you didn't read my post properly.
BTW the 2:06 is fake. :nesquik:
Even if it isn't fake, prove it has 1 in 16 million odds.
-
Also, if the chances of 1:04/5 hostage completions really is 1:10000, then it's in no way strange that three (even top) players haven't seen it happen once.
But it is strange in a way that a non-top player has seen it happen multiple times.
-
Yeah I'd like to point out the strat I got my 108 and 109s with has even worse odds. And I got both of those records within 1 try from each other. That is ridiculous lottery luck. It might honestly be the luckiest thing that's happened to anyone in this game.
Surely the odds should be same or better since the engine host has more time to escape, and might be able to get to a secondary escape point (so 2/6 odds instead of 1/6). The rest is identical to WR strat.
Some guesstimates on hostage odds (assuming no choke/groin deaths):
1st hostage: 6/6 probably (He is fairly quick and is released very early)
2nd hostage: 1/6 odds (There's no way he can get to a secondary escape point, but he seems fast enough for the first one even with some hinderance)
3rd hostage: 3-4/6 maybe? This guy is fast and can probably get to at least the 3rd point on the front of the boat.
4th hostage: 2-3/6? He can escape on Agent, and here we do the bridge as well. So he should be able to get to at least the 2nd point.
5th hostage: 1/6
So roughly 1 in 216 runs assuming the worst.
-
If he had access to an everdrive and was able to make the hostages complete at 1:04 speed every time, why wouldn't he have posted a 1:04 by now?
-
I'll be the one to ask the same recurring question when talking about Frig Ag 23:
Have people actually played the level enough?
-
If he had access to an everdrive and was able to make the hostages complete at 1:04 speed every time, why wouldn't he have posted a 1:04 by now?
Who is he? Luke P? If so, to avoid suspicion from the proof moderators. Which is what would make him so clever - 105/106 a week apart, 100 a year later and now he would have presumably lost interest or lacks the time or motivation to work on improving them any further. Which makes sense. What do you think, posting 100/104 2 days apart would have been smart? That is exactly what Henning did and look where he ended up!
For what it's worth, I am NOT saying Luke has faked his records, which some imbecile will accuse me of saying. I'm just stating the likely course of action a cheater would take with his Everdrive.
I'll be the one to ask the same recurring question when talking about Frig Ag 23:
Have people actually played the level enough?
Huh? What is the point you are making here? By the way, people are talking about 00 Agent here.
Yeah I'd like to point out the strat I got my 108 and 109s with has even worse odds. And I got both of those records within 1 try from each other. That is ridiculous lottery luck. It might honestly be the luckiest thing that's happened to anyone in this game.
Surely the odds should be same or better since the engine host has more time to escape, and might be able to get to a secondary escape point (so 2/6 odds instead of 1/6). The rest is identical to WR strat.
Some guesstimates on hostage odds (assuming no choke/groin deaths):
1st hostage: 6/6 probably (He is fairly quick and is released very early)
2nd hostage: 1/6 odds (There's no way he can get to a secondary escape point, but he seems fast enough for the first one even with some hinderance)
3rd hostage: 3-4/6 maybe? This guy is fast and can probably get to at least the 3rd point on the front of the boat.
4th hostage: 2-3/6? He can escape on Agent, and here we do the bridge as well. So he should be able to get to at least the 2nd point.
5th hostage: 1/6
So roughly 1 in 216 runs assuming the worst.
Your "worst" case scenario, which isn't even accurate (some are clearly wrong and 5th hostage is 1/7, not 1/6) doesn't account for guard death animations, random nades and consoles blowing up thus killing hostages, hostage takers getting stuck on doors/chairs, etc. which believe it or not, is a bigger factor than the hostage escape points themselves.
-
Think henrik ment perfect scenario reguardless of stucks/deaths etc..
We need more people playing frig 00, period. I know ogran and i have have played a fair bit of frig 00 recently (maybe 20 hours or so) and got nothing to show for it.. not even a bug fail A complete... ZERO.. i dont play out runs slower then 108 so obviously this has something to do with it. Not sure if OG has had any fast hostages..
-
Think henrik ment perfect scenario reguardless of stucks/deaths etc..
Yes, but we don't play in an ideal world. All of those mishaps can happen during a run. Which eventually puts the odds to 1 in 10k or worse.
-
I agree Ace.. were all arguing about odds and such but the bottom line is its very very rare. And the only way were guna nail this is if more people play it. We need more data
-
I don't disagree. Experimental data does go hand-in-hand with analytical data to back it up. But that has already happened dude. Henrik calculated the odds 4 years ago. I've refined them to be approximately 1 in 10k, which is what players who have a good idea of the level agree with. The sheer number of WR pace runs with hostage fails are in the hundreds of thousands. There have only been about 10-15 WR hostages. The result is the same, +/- 10%. That backs up the odds we mathematically deduced.
What 'more data' are you looking for here?
-
1 in 216, lmfao, who exactly from the Oldboy Dinosaur Illuminati is paying you to trick us into wasting our lives away playing Frigate 00A in hopes of "reasonable" 1 in 216 hostage odds?
We don't care about math or statistics. We have actually PLAYED the level and seen with our own eyes how it works. I think we can trust four of the top players with the most time on Frigate far more than we can trust some INTJ statistician who probably hasn't seen the sunshine in weeks.
Here is a look at some more reasonable odds that I created after discussing personal experiences with Luke, Ace and Alex:
(https://i.imgur.com/CMl0gv7.png)
-
I get the 1/10,000 number cause it makes more sense.. but were going on runs conpleted/attempted.
But its like FR odds.. just cause its 1/9 or what ever doesnt mean you see one every 9 runs.
If you have a bag of 10 pool balls... 9 were red and one was blue.. statisically your guna eventually pick that blue ball.. if some one never chooses the blue ball from random then thier lead to beleive there could be more then 10 balls in total or not there at all..
I know tou know WAY more then me about ge in general and frig overall.. im just trying to put some perspective back in.
-
the thing with Statue is, for example, the odds of the best FR is 2 in 17 (the worst FR has half the odds as the others)
so on a 217 pace run, the odds are 15/17 that you WON'T complete the FR.
This means that after say, 20 Statue 217 runs, the chances of you not having completed the FR once are (15/17)^20 which is 0.08181760331 or about 8%. In other words, if you were to get 20 Statue 00A 2:17 pace runs, you then have a 92% chance of completing at least one with the FR.
The same math would indicate that on Frigate 00A, every failed run only had a 215/216 chance of failing, with a 1/216 chance of completing (again, assuming no hostage deaths, slow taker deaths, nades going off, etc.)
So here are some interesting points in that math where Frigate 00A 1:05 or lower hostages get interesting:
In 11 runs, there is a (215/216)^11 chance they all fail, which is 95% (5% chance of completion)
In 149 runs, there is a (215/216)^149 chance they all fail, which is 50% (50% chance of completion)
In 497 runs, there is a (215/216)^497 chance they all fail, which is 10% (90% chance of completion)
In 993 runs, there is a (215/216)^993 chance they all fail, which is 1% (99% chance of completion)
So assuming the 1 in 216 odds are correct, you need about 149 runs to the boat in 1:05 or lower to have had a 50/50 chance at getting the WR.
-
Your "worst" case scenario, which isn't even accurate (some are clearly wrong and 5th hostage is 1/7, not 1/6) doesn't account for guard death animations, random nades and consoles blowing up thus killing hostages, hostage takers getting stuck on doors/chairs, etc. which believe it or not, is a bigger factor than the hostage escape points themselves.
1. I did mention death animations.
2. The bridge hostage odds IS 1/6, not 1/7, as stated in the facts topic:
Since the bridge hostage's unique (and optimal) escape point happens to replace #2, the above order is true for the bridge hostage as well.
3. My calculation assumes fully legit fails where all hostages survive. A nade going off during the run might make the run non-legit, and thus shouldn't be counted.
4. I do account for stucks. Remember that these only happen when the hostage is loaded. Most of the time the hostages won't be. The exception to this is when they encounter other guards, which mostly affects the slowest hostage, and potentially the 4th hostage. I did some test runs with the slowest hostage, and he was able to get to the escape point in time, even when getting blocked by other guards in the engine room. Not sure about the 4th hostage, but even if his odds were 1.X/6 instead of 2 that wouldn't make a huge diffeence in the result.
Henrik calculated the odds 4 years ago.
The odds of all hostages getting their optimal escape point yes. But that's clearly not necessary (1st hostage is guaranteed to escape for starters). I did make one mistake this time though, since the true odds are not exactly 1/6, but rather 40/256. With this in mind, we can make a new calculation:
(40/256) * (121/256) * (81/256) * (40/256) = 0.003651157 = 1 in 274
Let's say you're right and the Agent hostages are much slower than i thought. This would mean:
(40/256) * (40/256) * (81/256) * (40/256) = 1 in 833
So somewhere in that range would be a fairly safe bet.
-
the math is getting closer to the 1/10,000 number
-
If the odds are 1 in 833 of completing, the odds are 832/833 of not completing. This means you need 577 legit 1:05 runs before you can expect a 50% chance that one of those 577 runs will have completed.
Good thing I've got 3 1:04s now, only 574 more 1:04 fails to have a coinflip at the untied!!!
-
Look, don't try to be smart and talk about 'statistics'. From experience we know what the odds hostage completions are and they are not even close to your numbers you have thrown around.
In the future please only post statistics if you believe they are true. I mean you obviously aren't confident since every Frigate post you keep changing your odds.
-
Goose was merely presenting some alternative facts
-
He is talking to Henrik, who always changes his odds in every post. I only do the math with what I am given.
-
I've only changed the odds once, and it was based on a simple mistake (1/6 instead of 40/256). The 1/833 was a hypothetical response to Rayan's criticism and not intended to be a correction of my initial post. And as stated before, the 1/10k from 4 years ago was not a calculation of the actual odds at all, but rather, the odds of getting "perfect" hostages.
From experience we know what the odds hostage completions are and they are not even close to your numbers you have thrown around.
Right, 1 in 45 billion or 1 in 10000 or whatever number have been thrown around this topic. That would mean 180+ hours of consecutive legit 1:05 fails. Sounds way more realistic.
-
So Henrik, just to confirm; your OFFICIAL decided odds are 1 in 274?
-
I have a theoretical experiment request (might be able to be scriptified).
1. Kill a hostage taker
2. Mark the time which it says hostage released
3. Mark the time which hostage escaped
Repeat with different RNG,
Do this for every hostage taker.
Results:
Plot the distribution of times for each hostage taker and normalize it vs number of escapes.
Approximate the time which each hostage normally gets freed on an "normal" run, and add the time for each hostage escape time.
Do some math and integrate the possibilities for each integer second approximately.
(Possible contaminants to the run)
1. Guards blocking hostages
2. Death animation times and probabilities should be included
3. Lag :clemensmyth:
Would cream my pants if I saw this. If anyone could help me set this up myself like lua scripts to set enemy health to 0, loading save states, triggering data export upon memory addresses being a particular value that would be super super sweet.
-
I am willing to help with this. Henrik, I'm sure you would be of great help as well since you know a lot of the lua stuff, assuming you are interested
-
so f keen for this!!!!
-
Sounds cool and would be fairly straightforward to script, but in order for it to produce realistic results, you'd need to play out the runs manually after killing the hostage taker (if you just stand still, you'll keep the hostage loaded which will produce different results, also you'll avoid alerting guards that might block the hostage later on).
Dunno how many runs you need to get good estimates of the odds. An option might be to teleport away after the kill by manipulating Bond's position coordinates, but i'm not sure what the side effects of that might be (or if it's even possible).
-
Yea I thought that the loading and unloading of hostages would influence realistic probability, but this might be a good way to get a lowerbound on theoretical hostage escapes to start.
As for the number of runs 100 or so for each hostage is probably enough to fit a curve or something. depends on variance and stuff but ya thats just pulled out of my ass estimate
As for playing out runs henrik, I thought that you could maybe record bond's inputs to be the same, invincibility and no actor collision on, so you can always "go through the motions". If that starts being done then you might want to start monitoring every hostage as to when they escape, and probably just end up waiting at the boat. Although that gets more messy I guess and would require much more data to see what the best time could be.
-
The problem with pre-recorded input is that even if you enable invincibility/zero collision radius, any change to the RNG will desynch the input. Essentially you'll quickly start to get different amounts of lag which will make Bond go off course. This is one of the reasons why TAS'ing this game is so difficult and time consuming, since you can't reuse input.
-
I would be interested in instead of getting a better odds calculation if we could figure out what strategy in terms of noise gives optimal odds. I personally think being quiet is better because I feel luring particular guards out earlier raises odds they block hostages. This is definitely true on sa anyway but on 00 not sure yet.
Any way this could be tested? I counted and I shot only 8 bullets before first pause and I think a lot of other people shoot like 12 by then. You can see how many more guards are in the halls on their runs and I also kill the first guard on bottom of stairs at the start of the level which I thknk helps a bit.
-
The problem with pre-recorded input is that even if you enable invincibility/zero collision radius, any change to the RNG will desynch the input. Essentially you'll quickly start to get different amounts of lag which will make Bond go off course. This is one of the reasons why TAS'ing this game is so difficult and time consuming, since you can't reuse input.
Maybe a scripted bot based on waypoints could be used like the one I mentioned before in one of the Dam topics. The source (MHS script) for my Dam bot can be downloaded here (https://1drv.ms/u/s!AnIIXfP1PGtGgp5LdxcaniH6h3aIPA). It is free for anyone to do whatever they want with, but there are no instructions at this point. I imagine it would be a lot of work, but if anyone has the time and wants to try and use this for something like timing hostage escapes I could help getting it up and running over Discord or Skype if needed. There is already some code there for logging the position of Bond and the gate guard which could easily be modified to log hostage coordinates instead. By the way I used MHS 6.1 and Mupen64Plus 2.5.0. Also here (https://www.youtube.com/watch?v=XvvXVu5DijU&feature=youtu.be) is a video I just recorded with two example runs. It works reasonably well, but as you can see it's not quite perfect.
-
Oscar Pleininger's 1:08 quitout wth A up has been added to the repository!
https://www.youtube.com/watch?v=ULBpa3ddQkI
This makes 9 runs in elite history, out of roughly 20,000,000+ with 1:06 hostages or faster!
-
1:04 Speed
Luke Pettit (1:06 completed) https://www.youtube.com/watch?v=8BwXzsQesjU
...
Black and white video and obscure music from a relatively unknown Eliter. Hmm I wonder what this reminds me of. :nesquik: :thinking:
...just kidding, Luke P is legit. I'm just trolling. I mean this video is actually incredible: https://www.youtube.com/watch?v=Cujw-vc0Ywo
-
bumping this thread in hopes of someone completing the project of "following around all the hostages and timing each of their possible completions and seeing how often they pick certain despawn points."
-
Rare things happen in the world al the time. Someone wins the lottery every week. Being lucky is not proof of cheating
How often do top players play frigate? I barely ever see it streamed
-
I had around 500 105 or lower fails in the 120 hours or so I put in a couple years ago. Shit level. About 100 104s
-
I played close to 700 hours of Frigate 00 Agent in 2010. At one point I was failing 30 107s or lower an hour. My quickest complete was a 1:12.
Since 2010 I have played 100-200 more hours. Again, no completions.
The strat is a scam. How is it possible that all of Ace, Alex, Goose & Luke have spent a total of 2,000 hours (which at 20 fails an hour is equivalent to 40,000 legit failed runs) with no objective A completions?
-
Another 1:06 speed:
Dusky Frigate 1:08 (https://www.youtube.com/watch?v=rCDFWILa4PA)
Watched this run and remembered this thread
-
Has anyone ever made a mod on an everydrive so that the 5 hostages always run straight to the closest despawn spots and played it out so they dont get stuck to see whats the fastest they can escape? Is 104 maxed or is it in theory able to go lower? Can you like tas a moded file so you release them as fast as possible and they escape as early as possible?
-
Has anyone ever made a mod on an everydrive so that the 5 hostages always run straight to the closest despawn spots and played it out so they dont get stuck to see whats the fastest they can escape? Is 104 maxed or is it in theory able to go lower? Can you like tas a moded file so you release them as fast as possible and they escape as early as possible?
I think doing this and playing the level manually, counting number of runs between completions, then assuming every run where they don't choose the best despawns would fail, and simply dividing the probability by 6 or 7 or whatever the number of spawn points are per hostage would be the most efficient human-reliable way to test this. It would at least give a rough idea of the probability, and it would be quick to see who's correct. If it still takes 10+ runs to get a completion, the figure would be closer to Ace's, but if it completes almost every time, it'd be closer to Wyst3r's.
Because of death speed randomness and the other chaos leading to hostages' deaths, I would assume it's closer to Ace's figure.
-
Has anyone ever made a mod on an everydrive so that the 5 hostages always run straight to the closest despawn spots
I made this today:
https://www.dropbox.com/s/9wd0yk70y4oohs9/Frigate%20Optimal%20Escape%20Points.rom?dl=0
This could probably tell us alot about how much escape points matter versus death animations/obstacles.
-
I made this today:
Would it be realistic to have a general rule to only post patches rather than linking directly to ROMs?