The argument for using In Game Time (video to come)

[RWG]

1. the psychology of the game itself telling you your time (example: super metroid runners go apeshit over IGT despite being real time)
2. the timing every single run problem
3. encourages strategy development more than normal because you cant always out-execute a run by 1 whole second, whereas RTA you can just out execute by 1 frame which requires no innovation
4. much stronger storylines, drama, etc surrounding the levels and their history because of the high rate of tying times and the contrast between those and untieds

oh point 5, IGT cant be manipulated outside of the game via hardware bs unless the IGT counts loads and shit
obviously like circuit board shit is not included as a general rule but yes i know it can affect
i mean like accepted practices in communities
ssds, newer console versions etc
memory cards
RWhiteGoose - Today at 7:47 PM
yes
that's my main fucking argument
your computer can run fucking wsplit slower/faster than other computers
like these people are fucking retarded honestly(edited)
grav - Today at 7:48 PM
yea id lump that into that point as #5

grav - Today at 7:45 PM
then after elucidating these points, i would say how like every speedrun community would be using IGT if it was convenient to use and splits/rta was created as a solution to the problem of games not having a precise/accurate IGT
like theres a lot of games that have a sick IGT that people just use for leaderboards
despite them being twitch/srl communities that are born in rta era
RWhiteGoose - Today at 7:45 PM
Any examples of those games?
grav - Today at 7:45 PM
streemerz
tons of misc pc games

first 0:28 in SM https://www.youtube.com/watch?v=dNL8rdn00IU&t=15m25s (to show keenness/how much other communities care about IGT)

Grav to post more arguments/videos/examples

[Grav]

https://www.youtube.com/watch?v=fKbYek1GTg8 first 0:30 reaction

[flicker]

https://www.youtube.com/watch?v=xxh8yeLyjxc&t=3m26s (pokemon yellow 1:59 igt)
https://www.youtube.com/watch?v=9kVG3dr6Dx0&t=4m (pokemon yellow 1:55 igt)

[Grav]

another broader point:

there is a spectrum of acceptability of IGT in both the IL and single segment domains. If every GE IL had whole minute IGT, it would be absurd to use because almost all levels would be perma tied to the point of complete irrelevance... compared to whole second IGT where only like 10 levels have lots of ties out of 60 and all but 1 are not even maxed. conversely, if every single nintendo game had a built in IGT precise and accurate to the whole second/tenth/millisecond, everyone would be using them over splits because of the convenience (and psychology as stated before).

people  ALWAYS ALWAYS bring up metroid prime single segment as the reason why IGT is literally hitler. but they fail to consider the spectrum of acceptability and just take this single case in fucking SS that IGT for everything including ILs should be banned. a 13 hour speedrun with a 1 minute IGT precision wouldnt be bad at all, for instance. but yeah ILs are so different than RTA usually its stupid as fuck to compare it or to make universal claims. ILs will be much more case by case basis on whether a GE style timer is good or not. a game with only 5 ILs total and they are really easy to optimize would be bad, but 60 levels with 50+ being either not maxed or flat out miles away from human perfection are totally okay.

look at this pic to see what i mean:

Spoiler

also here is werster explaining why IGT should be a case by case basis (For single segment):
https://www.youtube.com/watch?v=RaMGf1V3MMM

[Grav]

https://www.youtube.com/watch?v=xxh8yeLyjxc&t=3m26s (pokemon yellow 1:59 igt)
https://www.youtube.com/watch?v=9kVG3dr6Dx0&t=4m (pokemon yellow 1:55 igt)

these are good but i dont think using werster (a known champion for IGT) is good to drive the point home that IGT is psychologically way more rewarding. he was using IGT officially in that run anyway. garrison on the other hand was the person who crusaded to end the IGT movement in super metroid and was the first to use RTA seriously... but even him still took way more satisfaction from IGT in the end despite the runs being RTA

[Grav]

the final closing point should be like IGT is basically better in EVERY conceivable way except for when the precision or accuracy of the timer is such that it hampers the community activity and competition at the top level.

[Grav]

[8:38 PM] Vital Greens: Ill do my notes tonight is that too late?
[8:38 PM] Vital Greens: Well goose ur smart you can extrapolate
[8:38 PM] Vital Greens: Ill just jot them here
[8:39 PM] Vital Greens: 1. Igt is accessible, easy etc for new players
[8:39 PM] Vital Greens: 2. Igt is easier for short stufg i.e il
[8:40 PM] Vital Greens: 3. Pls talk about total points, cause noobs joke about ties, but there like no ties in the too 100 for points
[8:41 PM] Vital Greens: Not that i checked
[8:41 PM] Vital Greens: But for ils the game is meant to be played as a whole
[8:42 PM] RWhiteGoose: Good arguments, yeah post these in the thread later
[8:43 PM] Vital Greens: 4. Its a unique experience where you dont know what the time is gonna be, so its chocking when you get a pb sometimes
[8:43 PM] Vital Greens: Like its a mystery box
[8:43 PM] grav: oh yea thats 1 thing i forgot
[8:43 PM] Vital Greens: If you have a timer you already know
[8:43 PM] Andy Brennan: the reveal rush
[8:43 PM] Vital Greens: Thats why some ge reactions are so good
[8:43 PM] flicker: hide your splits
[8:43 PM] alkamaass: have done that
[8:44 PM] alkamaass: helped with nerves
[8:44 PM] grav: that ties into my better drama point nicely
[8:44 PM] Andy Brennan: real time always feels like adding an extra layer of arbitrary
[8:45 PM] Andy Brennan: speed running already is an arbitrary challenge
[8:45 PM] Andy Brennan: but if it's timing and saving your times at least it's helping to facilitate it
[8:45 PM] PartyBoy, Live, TV, Spec: we'll arbitrarily make a frame that runs only 30 frames per second and arbitrarily tie it to each one of these
[8:45 PM] Andy Brennan: talk about saving the time too
[8:45 PM] Andy Brennan: shame Pilotwings etc doesn't save your best times
[8:45 PM] grav: ah yeah thats important
[8:46 PM] alkamaass: tbh goldeneye would've had a much much smaller speedrunning community if no igt
[8:46 PM] Vital Greens: It wouldnt exist imo
[8:46 PM] Vital Greens: Only rta
[8:46 PM] PartyBoy, Live, TV, Spec: it would be as popular as SM64 single stars
[8:46 PM] Andy Brennan: ^
[8:46 PM] grav: yes
[8:46 PM] PartyBoy, Live, TV, Spec: which is = zero popularity
[8:47 PM] Vital Greens: Well sm64 is even the biggest game and hardly has an il scene
[8:47 PM] Vital Greens: So ge would have no hope
[8:47 PM] grav: yep perfect example
[8:48 PM] Vital Greens: Oi funny example
[8:49 PM] Vital Greens: Imagine if olympics sprinters carried stop watches and had to start/stop their own timer lol
[8:49 PM] Andy Brennan: lol

[wheatrich]

fwiw IGT for GE gives us all the great reactions.  1:12.9x or whatever at the time wouldn't have been close to the same.

[Aztec Exemplar]

is there that henrik vs ace train vid around?

the random difference in real time for some runs is pretty dumb

[Justin Thorogood]

With RTA you would also factor in time it takes to scroll through menus which is cancerous for ILs, especially when a majority of the levels are under or close to 1 minute long

[Retrix]

I'm for IGT. It's classic. It works well with GE, which still has a solid playerbase.

I'd post more but lunch is over.

[Grav]

bump

[deletedprofile.u]

Wow wtf just discovered this thread. Double bump

[Grav]

[8:41 AM] grav: have to add dam 52 to the IGT video when you make it too
[8:41 AM] grav: dam 53 was the PRIMARY argument for shitters about why igt is bad
[8:42 AM] grav: ill post to remind

[Grav]

Arch 15 too

[RWG]

posting a few thoughts

--

Goldeneye's speedrunning is different from wider "speedrunning", in that, GE is more similar to "getting a high score in the game, within the confines of the game."  All that matters is what the endscreen displays, just like this would be all that matters in a game of Donkey Kong or whatever.  Yeah, we lump in GE with "speedrunning" because it's similar in a lot of ways, and easier to widely promote that way, but the reality is that, in practice, it plays more like going for a high score.  The outside time does not matter... what matters is what the "score" shows on the endscreen.

--

Also, here's a post about how the rankings system is effective with in-game time, and wouldn't be any more effective with "real time": https://www.reddit.com/r/speedrun/comments/9fonoa/real_time_vs_ingame_time/e5yz4zl/

--

I haven't read this thread in a while, but if anyone wants to chime in with more points, partial scripting for a potential video, etc, that'd be great.  I guess as GE vids get more popular it's time to get this one fully debunked & understood, once and for all.

[Retrix]

I think it'd be interesting to have a chart for PD total time based on what people's in game files read. Not sure of an application for it.

[RWG]

Has this ever been explained?

Henrik's Train Agent 1:04 appears to be faster than Ace's Train Agent 1:03.

What's the modern understanding on why this happened?  Was it caused by in-game lag?  Or was it an external rendering/encoding issue?

I'm working on the script for this In Game Time video finally, and understanding or being able to explain this anomaly (or why this happens regularly... are there any other examples?) would be super useful, if not necessary.

Thanks

[Icy]

It's generally assumed that lag is what affects the difference. I don't know the technical details, but this testing by Henrik and these video comparisons by Karl show some more results.

[Wyst3r]

We know that the game calculates the number of (integer) frames passed between each update. However, I can't quite remember what happens to the leftover decimals, if they are discarded or reused the next update? If they are discarded (which I would guess)  then that might account for real-time differences, since runs where the lag is exactly at or close to an integer threshold would be fairly accurate compared to realtime (assuming the n64 clock frequency used in-game is correct), and runs that are between two tresholds would be slower or faster, depending on which threshold they were closer to (since the value is rounded up or down to the nearest integer).

Would have to be confirmed though.

https://www.youtube.com/watch?v=4_L8H9jDkF8

[mw]

To give a layman's explanation that I believe is correct enough:

Ace's run is less laggy (from killing guards and such), so his real time moves closer to the in game time. Henrik's is a little more laggy, so his real time is more slow than the in game time. Henrik's run completes first, but the time that is added from lag is great enough that he gets a 1:04, while Ace gets a 1:03.

So think of it like this: if you scale their bond speeds to be equal, it's like Henrik is racing against an in-game clock that is moving 1.0X times as fast as Ace's. This very minute difference at the edge of a second barrier was enough to make the difference.

[Wyst3r]

That might be true in this particular case, but as Karl’s S2 video showed, it’s not necessarily true in general. More lag does not guarantee a shorter realtime, in fact it’s sometimes the opposite. The explanation in my previous post would predict a more random (or at least a level-specific) relation between lag and realtime. It’s not the absolute amount of lag that matters, but rather the lag relative to the thresholds (the amount of which would be impossible to tell just by looking).

Edit: After looking at the assembly code again, I can confirm that the decimals mentioned in my first post are indeed discarded. The pseudocode looks something like this:

Code: [Select]

cycles_per_second = 46552500; // 46 MHz
frames_per_second = 60;
cycles_per_frame = (cycles_per_second / frames_per_second);
cycles_passed = (get_cycle_counter() - prev_cycle_count);
frames_passed = round(cycles_passed  / cycles_per_frame)

...

prev_cycle_count = get_cycle_counter()

As an example, let's look at 3 different scenarios, one where the game updates at exactly 60 Hz, one where it's 39Hz, and one at 41 Hz, and see how long it takes (in realtime) for the in-game timer to reach 10 seconds:

Update frequency	Frames per update	Realtime (10s IGT)
60Hz	1 (1.00)	10.0
41Hz	1 (1.46)	14.6
39Hz	2 (1.54)	7.6

number_of_updates = ((10 * frames_per_second) / frames_per_update)
cycles_per_update = (46552500 / update_frequency)
realtime = ((number_of_updates  * cycles_per_update) / 46552500)

[RWG]

https://docs.google.com/document/d/18BXByHdECj2w6gWtIVz7ld4RiRrXjNAX6sLzTAJEK2E/edit?usp=sharing

Here's an early script for the IGT video... I don't know if I will even mention the Ace/Henrik Train 103/104 thing, as it might just overly complicate the matter.  I'm just looking for some feedback on the overall cohesiveness, (do the arguments make sense, the flow, etc).  Feel free to share ANY thoughts or anything you think I should add or remove... roughly 3 pages on a google doc = 10 minutes for me, so I don't want to extend it much longer if it can be avoided.  Again though, if I cut out the Train story, there is probably room to add something else.

One problem I'm having atm, is how do I reconcile ridiculing the idea of "using an external scoring system and ignoring the point score in a game like Donkey Kong", while still vouching for our own rankings system, which is just as arbitrarily created?

[RWG]

Here's the vid: https://www.youtube.com/watch?v=CqW-Z8BWsD8