SwitzerlandAfter discussing the whole issue with the Mod team, there was a consensus about new Game Boy rules which have been put in place yesterday, so the above is obsolete.
Hi Mod Team,
I wanted to bring something to your attention I noticed while scanning through the GB leaderboard.
Apparently the run from Slamphibian is using save states:
This is about 10 seconds into the video.
Now I'm not completely sure what for example playing through the level and then making a save state (or waiting at the menu screen for a long enough time) does to the game's mechanics.
As it turns out this is not worth investigating, as save states are generally disallowed by the global speedrun.com site rules:
[quote]Tool-assisted emulator features such as save-states, re-recording, or frame advance are not allowed.[/quote]
The run is not complying with this and thus should be removed.
(Side note: regardless of the other issues currently discussed about the Game Boy rules, showing a reset on run start would be good requirement in my opinion.)
Hi all,
Currently, runs in the GB category are not timed exact enough. In this post I'd like to share why I think so and also like to make a suggestion on how I think we could improve timing and the current rules.
(I brought some of the issues up yesterday in the Discord but I'm not sure everyone's aware of the current situation for the GB board.)
Current top runs
Let's have a look for example at the top two runs, the one from poultfloyd and the one from Luzbelheim.
If we define an exact starting frame (which currently does not exist by the rules, I'll get to that later), like the first frame "Bertil Hörberg 2012-2015" is not (completely) visible anymore, we can time those two runs using any video software or online tool like the yt-frame-timer, and get the following:
- poultfloyd: 43s 767ms
- Luzbelheim: 43s 817ms
So there's a small, but clear difference (50ms, or 3 frames @ 60 FPS), and so the runs should not be treated as a tie, poultfloyd's run is faster and therefore should be first on the leaderboard.
If you don't believe me, I made a comparison video of the two runs, you can check it out here:
Both runs start on the exact same frame, and you can switch to the very end of the runs to see if they end at the same time, as the current leaderboard suggests (as they tie for 1st place). On Youtube, you can move between single frames with "," (forward) and "." (back) to get the exact frames.
I'm not sure if the timing was done relying on LiveSplit, which is started by hand and therefore inaccurate? As this is in the "optional" rules, I guess this is currently the case.
The speedrun.com Site rules however clearly states that this would be wrong: "The time of the run is the time in the video from starting condition to ending condition. It is not the time shown on the timer program, which is started and ended with human error."
Optional rules and penalty
There are some more issues to point out which I think are unfair.
My run (https://www.speedrun.com/gunmc/run/zqg0npxm) got a penalty, which is according to the current rules, but in my opinion not fair.
I'm running on original Nintendo hardware (Super Game Boy 2, using an Everdrive X7 to load the ROM), so I honestly don't understand why that should count less than an emulator.
Besides the obvious that I am not using Gambatte, I also can not track my input from the Super Game Boy 2, which would be required by the rules to not get a penalty.
(By the way: It is beyond my understanding how and why the "ss.x99"-penalty got in place, but it messes with timing, which should just not happen, it is the single-most important variable for a run on any SRC board.
Either a run is valid and follows all mandatory rules, or it is not and it should not appear on the leaderboard. Having "optional" rules is not a good idea in my opinion and I've never seen anything like this on another GB game or leaderboard in general. But that's just my personal opinion, see below for a suggestion.)
Suggestion
There would be one solution to all curent problems: Switch to frame-based timing.
- This would be much fairer (and more exact) and a situation with wrong timing like above could not happen again.
- This would need a clear starting frame (like already existing for the end of timing), which makes sense in any case. If there's no defined (frame-precise) starting point, why even bother measuring time in milliseconds?
- This would obsolete all the current "optional" rules and the penalty: LiveSplit is not needed anymore and 30 or 60 FPS doesn't matter (you can use 60 FPS for a chance to win a single frame - I could elaborate on that if needed). The "show input" does not make sense for running on original hardware.
About emulators
There is one additional point I'd like to make (not as important, but still) - some Emulators run far too fast (see here for an extensive comparison of emulators which Legs, me and some other GB runners have done), I think it would make sense to have at least one mandatory rule which emulators are allowed. (Currently, most GB games accept Bizhawk, Gambatte and BGB.)
If this new mandatory rule is not introduced, I could just use a faster emulator, some of which are up to 3.2% faster. For a 43 second run, this would give me nearly 1.4 seconds unfair advantage, and the current penalty would only give me up to 0.1s penalty. So it would only make sense to search for the fastest emulator, which is not a direction a speedrunning Leaderboard should go to. As I said, instead of giving some arbitrary penalty, it is more common (and also makes more sense) to only have mandatory rules, but instead enforce those.
Suggestion summary
- Switch to frame-based timing (and define a start frame)
- Drop the optional rules and penalty
- Introduce a mandatory rule to restrict allowed emulators
Thanks for reading! That was a bit longer than expected, but I think it's worth to have a detailed look at it, as I think the current rules & timing are unfair.
I'd be glad to help out or support transitioning the timing of the GB category to a more accurate, frame-based timing.
Hi, I would like to request leaderboard moderation for https://www.speedrun.com/gbbonk2 I have a pending run, but the game has no moderators at all. Thanks.
Hi, I'd like to request moderation for https://www.speedrun.com/the_ren_and_stimpy_show_veediots I have a pending run and the Mod hasn't been on site for a year. He is inactive on Twitter and Twitch for even longer and didn't react to any messages. Thanks!
@starsmiley Thanks for having a look at my request. The game ( https://www.speedrun.com/daedalian_opus ) still seems to have no Mod, not sure if something went wrong there?
Hi, I'd like to request moderation for https://www.speedrun.com/daedalian_opus I have a pending run and there seems to be no Mod on the board at all. Thanks.
Ah, never mind, I found out via https://web.archive.org/web/*/https://www.speedrun.com/.
It was #118855 but now it seems to be #178C6B.
Hi,
(cross-posting Github issue here - https://github.com/speedruncomorg/api/issues/108 )
I am running a WR bot for certain platforms and until yesterday, I could query newest runs of a platform like this: https://www.speedrun.com/api/v1/runs?status=verified&platform=n5683oev&orderby=verify-date&direction=desc
This gives the following error:
{'status': 400, 'message': 'Please query by at least one of: examiner, game, guest, level, user.', 'links': [{'rel': 'support', 'uri': 'irc://speedrunslive.com#speedrun.com'}, {'rel': 'report-issues', 'uri': 'https://github.com/speedruncom/api/issues'}]}
Do I understand it correctly that the API was changed to require one of the fields mentioned and that it's no longer possible to fetch recent runs by platform? Also, is this permanent or also just temporary due to the increased holiday traffic?
Also the examples on the /runs documentation give the same error ("GET /api/v1/runs?status=verified&orderby=verify-date&direction=desc gets all newly verified runs").
Hi all,
I recently became mod of this game and noticed some inconsistencies regarding timing. With the current rules, some existing runs are timed to fast, while others are timed to slow.
So this is an update for the timing of the NES category. I tried taking the existing rules and specifying them further down, so that it is consistent with the existing runs and for future runs, it's (even more) clear how to time.
Start
Start: "Timing starts when the screen begins to fade after selecting Set Features." Key frame: First frame that shows signs of fading / getting darker.
End
End: "Timing ends when the screen begins to fade at the end of the game." Key frame: First frame that shows signs of fading / getting darker.
Calculating time between start and end frame
Now that you have the two key frames, how to measure the time between them?
- Youtube: You can navigate with "." and "," until you find the right frames and then put them in here: https://mattbraddock.com/yt-frame-timer/ Be sure to get the right FPS and it will calculate the time between the start and end frame.
- Video editing software: Usually some kind of timer is available which can be used to count frames or time directly. There's VSDC for example which is free and has a timer available.
- Twitch: No direct way to do this to my understanding, download the raw video and progress using the method above.
If you're not sure about your time, don't worry, as all times will be checked in the approval process.
Hi, I'd like to request moderation for https://www.speedrun.com/jordan_vs._bird_one_on_one I have a pending run and the Mod hasn't been on site for ~10 months. He is inactive on Twitter and Twitch for even longer and doesn't react to any messages. Thanks.
Hi all, I recently became Mod of this game and I'd like to share a few suggested changes.
As we now have runs on 3 different platforms, categories for the different platforms are added, as the games are not really comparable. New categories will be added if they're needed, this means that so far there are categories for GameBoy, SNES and GameGear.
Second, the rules for all categories were updated to define start and end of timing. I tried to make it as consistent as possible, feel free suggest changes so we can discuss these. For the SNES, there were no rules until now. The character selection makes most sense from my point of view, and the current existing run has been retimed to reflect that.
Thanks for reading!
Hi, I'd like to request moderation for the following two games: https://www.speedrun.com/the_jetsons_robot_panic and https://www.speedrun.com/cutthroat_island I have pending runs and the Mod hasn't been on site for ~10 months. He is inactive on Twitter and Twitch for even longer and doesn't react to any messages. Thanks.
As discussed in the other thread, the rules are now updated to "All Elevators". The existing run doesn't satisfy the new rules (as you stated in your first post), so I will remove it, leaving the category empty for now.
Sounds good. You're right about the 100%, that would also be misleading. I'm going to update the category accordingly.
Hey, thanks for the bump. I saw both threads and think we must first decide on the rules and then decide what to do with the run you mentioned.
The current rules for "Any% Warpless" are misleading, so I also vote for changing them. I like your suggestions, but I think they mean two different things.
- All elevators must be used
- No warps in hidden rooms allowed - this implies that the warp in the first room from 6 to 7 is allowed
What do we want to track? I would say 1) which is kind of like a 100% (we could also name it that way). What are your thoughts on this?
