General

This topic aims to precise general rules regarding runs submission. Please read before submitting your first run. Failure to comply with the rules will result to a rejection of your run.

It is also the place to discuss about them and propose changes. Also tell us if something is unclear or missing! This post will be updated when needed.

Rules changes are in general not retroactive, unless it gives older runs unfair and significant advantages, but one can ask to update a submission to comply with the new rules. A verified run whose proof is no longer accessible will remain, but when possible, the submitter should update the link or reupload the proof.

Also read the speedrun.com rules: https://www.speedrun.com/rules These rules and the ones in this post complement each other.

Specific Rules

Please read the rules of the IL you are running, or the Full Game ones using the "View rules" button. They override the General Rules.

Submissions Rules

Record (1st place) and any run having a time less than 10% worse require a video for submission. Else, runs with times at least 10% worse than the current record don't need any proof (though try to provide a screenshot link in the description if you cannot provide a video). We however reserve the right to reject runs without proof if there are suspicion of abuse of this rule. Ghost or Ghost replays videos are currently accepted as proof of IL runs, but it will no longer be the case once the next major version of SuperTuxKart, expected to be 2.0, is released, and this will apply to all versions (like if a 1.x run is made after 2.0 release). A video is not necessary if a Ghost is provided before 2.0.

There is no strict requirement about that, but please only submit good quality recordings as we may reject low quality videos. Make test recordings before your first run to ensure that there is no major flaws from the recorder.

Please make sure that we can access to your video, and that you linked to the correct video. Check your run infos (correct time, category, etc.) before submitting.

You can always view the already verified runs and their videos to have example of correct submissions.

In case of submissions having issues not covered by the rules, a verifier can either validate the run using a reasonable adaptation like rounding or adding penalty times, or discuss about it. Penalty times can also be applied instead of a rejection in a case-by-case basis and will be clearly shown in the submission's description.

General SuperTuxKart rules

The run must be played on Windows, Linux, or Mac. For IL runs, playing on Android is also permitted. Emulation or virtualization is not allowed.

No Addon Karts are allowed for either the player or AI.

You can use any shortcut/skip as long as it is possible in normal play and not disallowed in the Specific Rules of the level you are running or the Story Mode. So the Big Door Bypass/Gate Skip in 0.9.3 and Grand Prix Skips are perfectly valid in Any% runs, and you can use any known (or new ones!) shortcut/skip in ILs.

A run gameplay must not be influenced by actions done before it (for example, from a previous attempt).

For ILs In Game Times, In 0.9.3 and 1.0, use if possible the Ghost Time shown in the Ghosts List (or equivalently, the one shown in the Ghost's Filename). Otherwise use the time shown at the Finish Screen (not the one next to the character icon); if the timer is only precise to the 1/100 s, put 4 in the milliseconds' last digit (for example 1:23.45 -> 1:23.454). The time shown after replaying a ghost is not valid. If we cannot see a valid time, we will add a penalty time of 0.1 s.

Game alterations and cheating

Allowed mods are:

• Reasonable texture, color, text and music mods. They must not help you, for example showing Checklines is not allowed;
• Numeric speedometer;
• Any mod helping to time your run;
• Some convenience modifications. You can mod the game such that it is faster to reset failed runs, for example adding a feature resetting the profiles every time you quit the Story Mode, so you don't have to create a new one each failure. Such modifications must obviously not help you get better times.

IL Runs (non GP) done on online servers are admissible, but may still be rejected if there were noticeable lags or if we notice that the server made unacceptable modifications (different physics, etc.). Only Live recordings are accepted in these cases. Make sure to pick the "Online" option for the "Subversion" and mention in the description which server was used; failure to do so may result in a rejection of the run.

If unsure, ask us to be sure whether your mods are allowed.

All other mods are disallowed and considered as cheating. This includes but is not limited to Gameplay (allow doing actions impossible to do normally, change the game experience, modify RNG and other game behavior), Physics (modify speeds, etc.) and Geometry (3D models) modifications.

Usage of TAS in any form (for example by modding the game or using other software to help inputting precise actions) is not allowed and considered as cheating.

Note that we rely on trust and assume good faith. With good technical knowledge it is perfectly possible to make cheated runs without any way of getting caught and such runs would pass if the cheater is not too greedy. We count on you to make honest runs and not oblige us to add stricter requirements.

Measures like banning (not accepting any runs anymore from someone) can be taken in case of proven cheating.

Other Rules

Artist's Debug Mode must be disabled.

If compiling the code yourself, be careful to use the code and assets matching the release, and not the latest code from Git, which might have introduced gameplay alterations. Of course the "Game alterations" section applies so you can use the official timer and any legal mod (in this case they must be based on the code corresponding to the release, and not the latest code or anything else).

Similar for the assets, you must use the ones from releases and not for example the newest Svn ones (or things like using 1.0's Candela City in 0.9.3).

For multiple version categories (0.9.0-0.9.2 or 1.x), you are however allowed to use any official code or assets within the version range (unless some code or asset that was quickly fixed later broke something that would give the runner a significant advantage). So you can for example use Git code and features that came after the 1.0 release and until the latest 1.x release, and use Git Commits until the game becomes distinguishable from the latest 1.x release (rules similar to mods apply, for example if a commit changes the kart speed) or explicitly updates the version; you may not use any code that came after such commit for the category.

Tips

Read the various guides in order to learn how the records were made and improve. Watch videos. Or just ask, most speedrunners will be glad to help you :D !

Don't be discouraged if you don't get times as good as the top records nor be shy to submit "horrible" times. As long as you keep trying, you will improve, and often be amazed (or amaze others) how better you are racing after some time! In the other hand, don't force yourself, it is understandable that one may not have enough time to perfect the skills. Just have fun!

We are looking forward to seeing your runs!

One thing that I would like to discuss is unofficial tracks, like Green Valley in 1.0 or Black Forest in 0.9.3

Should they be in the leaderboard? Addons are updated so there can be differences from a run to another depending on when it was done and what addon version was used...

Thanks for this post! I think a detailed explanation of the rules like this is very helpful. A couple of things I'd like to add:

SuperTuxKart is very weird with timing. For time trials (in 1.0), we go by the three-digit time you see when you finish the run. This is different from the two-digit time you see in the top-left corner. This is also different from the time that your replay ghost gets. So if you don't record a live run, show us the screen preceding the replay where it says what time the replay file got (this time is accurate). In general, make sure the video includes the accurate time somewhere.

In Story Mode I believe you don't have to use the timer mod (I think this might be up for debate), but it is in your advantage to use it since it doesn't count loading times. (If you don't use the timer mod, should loading times just be counted?)

Add-on karts are allowed to be downloaded as far as I was aware, since the AI won't use them as long as you select karts from the "standard" category and not the "all" or "add-ons" category?

As for the add-ons, I think it's unlikely that any of the add-ons on the site are going to be updated, considering they almost all of them are former official tracks (with the exception of Black Forest, which won't be updated for other reasons). But if they are updated, at least we'll know, so we can deal with it then. :)

By the way, a lot of the IL rules are out of date (the lap times don't reflect 1.0). Just wanted to bring it to attention ;). I would fix them, but I don't have the power.

Should we clarify the Fort Magma rule somewhere? (If you accidentally restart the Fort Magma challenge, exit to Story Mode. The timer will keep going, but we'll count the run as ending when the loading screen appears.)

GP Rules for 1.0:

Deciding on GP rules for 1.0 is tough for various reasons. Here's what we decided on in the Discord.

Number of AI should reflect number of AI in story mode. Mode should also reflect story mode, so do a normal race for every GP except for To The Moon and Back (do time trial). What place you get doesn't matter, just complete the GP. Giving up isn't allowed, and therefore skips aren't allowed. Use the in-game time.

Ok for the Addons and Gps. I updated the Gps...

For the Timer, it is indeed not mandatory and this is explained in the Specific Rules, but we of course really would like that a runner uses it. Also in this case timing should be only precise to the second rounded up and no milliseconds.

"By the way, a lot of the IL rules are out of date (the lap times don't reflect 1.0). Just wanted to bring it to attention 😉. I would fix them, but I don't have the power." I don't see what is wrong with them. Do you just want to precise which time we need to use? In this case I just added a paragraph about that. I would gladly promote you to Moderator, but don't have the power either Lol

I also added that rules changes are in general not retroactive, but one can ask to update a submission to comply with the new rules. For example some 0.9.3 Full Runs included loading times, so players can ask to retime them. The point is to avoid to reverify and fix everything Lol (like for the 4 ms rule).

The Fort Magma Rule is already written in the Specific Rules of 1.0 Any%.

Sounds good!

Around the Lighthouse says three laps instead of four, for example.

Ah Ok, it is true that this and Minigolf have a different default lap. FIxed. Maybe you can ask a red sword to promote you :p That is also kind of stupid that a verifier cannot correct errors in submissions :p

Thanks!

Luckily verifiers can edit submissions now, though. ;)

Another thing that doesn't matter as much now since there haven't been any 0.9.3 runs recently, how many AI should you race against in 0.9.3 GPs? It was ambiguous on SuperTux mode because there wasn't Story Mode for SuperTux, but the reasonable assumption would be 5 since Novice is 2, Intermediate is 3, Expert is 4, etc. The rules just say "default" though (which, technically the game starts on 3 AI) and all of the GP runs except for At World's End race against 4 AI (the At World's End one does 5, though).

Really? Cool! I once read that you were unable to fix someone's submission but now if they changed that...

I clarified these rules. It should always be based on the Story Mode's counterpart. SuperTux in older than 1.0 is Expert + 1. I never really checked the rules for the Gps and even thought that by default, a particular value was set in each case, but it turns out that it just uses the latest number of laps used.

One thing to clarify: it does matter what place you get in 0.9.3 GPs, since skips are allowed. I updated the rules accordingly (since I'm a regular mod now ;) ).

Added All Tracks GP categories for 0.10 and 0.9.3. I allowed to choose reverse or not for any track (runner has to pick the better one each time), the aim is finally to do all the tracks in any way in a Gp, the fastest possible, without luck influence (Time Trial Only, though in TAS it would be better to allow items so I get tons of Zippers Lol) We use In Game Time in order to not bother with using modded or external timers (and there is no real point in timing how fast you are mashing buttons between tracks...) If there are any objection, I will be glad to hear it Lol

There were already a similar category for 0.8.1, but I don't really know much about that version so I would let someone else adapt the existing rules if needed...

Also, new IL records obtained during running an All Tracks GP are perfectly submittable as ILs as well!

I'm ok with the All tracks GP category. I'm fine with in game time, I'm not a fan of button mashing between tracks either.

Obvious but Lawmaking is what it is... So: "The Nessie's Pond glitch that allows to get a time of 0:00.000 is banned, as well as any similar glitch that gives a very broken time at the end (but you can use the ones in Oliver's Math Class or Volcan Island that save 1/24 s for example)." If that happens, use the 1/100 s time at the end next to the icon and add 4 ms if you want to submit the time.

Btw if you want to have 3 digits everywhere, you can look at this Commit of the Addons code. As it is already established, we should continue to use the times at the results screen, but with 3 digits the time next to the icon is actually the most reliable as it is based on the number of Ticks. https://github.com/Fouks0/stk-code/commit/36188a28ba744636f0cbe3e90cbfd4ea70572811

It is certain if Wax's Snow Peak time is 1:24.548 or 1:24.549? Because if the fourth digit of the ghost replay filename is 5, then the time shown at the end screen is rounded up or down depending on the exact time in the ghost replay file, but after closing the game and reopening it the time is always shown rounded up in the high scores list. For instance, my Fort Magma run shows 1:24.863 in the end screen and indeed the time listed in the replay file is 84.863457 (so it should round down), but when I look at my high scores list now it shows 1:24.864 as the time.

We are using the best valid In Game Time available, which is shown as 1:24.549. If it was 1:24.548 at the live run we would use this, but Wax would need to upload a live recording in this case. In your case if you were unable to submit a live recording of Fort Magma, your time would have been 1:24.864.

Also as far as I know there is no changes from Git and 1.0 that would actually affect the gameplay during IL runs, so maybe we could change the rules to accept compatible runs. After all, there is the 0.9.0-0.9.2 category and the submission would likely have stayed if Wax did not reveal the version (also maybe all his runs so far were also Git? He would have to redo everything Lol)... Edit: His Volcano Island was also done using Git.

The replay file time rounds up to 1:24.549 anyway, so that's fine. A checkline in XR591 was modified in the Git version to make it possible to cut off more of the last corner, so runs in the Git version could be slightly faster. But there are no other gameplay changes between 1.0 and Git that I am aware of, so I think that it would be possible to accept Wax's Snow Peak and Volcan Island runs.

That build might only say "git" due to it being compiled from the git branch, but if it were my build of dark mod, it should have been compiled on the git branch at the commit which was marked 1.0

@Fouks Using elapsed ticks is less accurate than the method used in-game, which is able of sub-ticks precision by interpolating the time of crossing based on the position on track on the tick before and after the crossing.

Usually this estimate and the ghost time are within 2ms of each other, I don't recall the specifics behind this occasional difference.

The main things causing issues is that the distance over track has a "zero band" that can cause some weird things when the kart's center happens to be precisely inside the band during a tick. This needs fixing but would need some serious changes.

As for Wax using the git version : right now, the git version is pretty much gameplay-compatible with 1.0 at least for time-trials (some UI changes affecting story mode make this more debatable there), barring some very small things like the XR591 checkline fix. It'll probably include some story-mode bugfixes for cases where points weren't properly awarded, but those bugs can be avoided already, it's just a random run-killer bug that would go out. Ideally, 1.0 and 1.1 should be a single category for IL and full run purpose, to avoid cluttering the leaderboard for insignificant differences.

Because of this compatibility, I think using current git is not an issue. Though personally, I kept using a fixed 1.0 compile for running purposes.

However, once other things begin to change in preparation for the 2.0 release, there will be a number of compatibility differences that will arise. Hence, git version should be avoided after 1.1. In any case, the version used should be disclosed, because if the verifier catches something amiss that indicates a wrong version (git, gameplay mod, or whatever), the run must be rejected.

So to be clear, is using Git allowed now? If so, I can un-reject Wax's time. :)

Added phrase "Runs that used git code can still be accepted if it is reasonable to think that no gameplay change was introduced in it compared to the release, but again please ensure that you are using the right code." and applied it.

Alistair_Findlay: my Mod code is based on the official Speedrun Timer Mod and saves correctly "1.0" ghosts. You might have only used the Assets and built your binaries with the Git Version, and then Wax used it. Or Wax built himself using the Git Code and the "Dark" Assets

Alayan: yeah, I know that Ticks are only 1/120 s precise and just said that they are more reliable; as we get such timing bugs, the interpolation is definitively broken in some cases while the Ticks based timing does not seem to be glitched. Otherwise having µs precision is certainly neat :p . The idea was to use Ticks times to make all runs consistent and avoid to deal with all these bugs.

Also true, I forgot about the Xr591 and Old Mine Checklines, it was even I who published screenshots to show where the issues were Lol

The difference between the elapsed time and the interpolation should always be less than one tick (although it sometimes isn't, especially if the finish occurs at the edge of a lapline), so it only really makes a difference when more than one run has the same elapsed time. In such a case, I feel like it would be fair to consider the runs tied and it seems a little arbitrary to decide that one run is a few milliseconds faster than the other. But it's very rare that this is the case, so maybe at least for now it isn't much of a problem.

The bigger issue is the glitch that finishing on a checkline sometimes (it doesn't seem to be entirely consistent, it may depend on precisely how the finish is approached) results in the time shown being five ticks faster than the elapsed time. This can only occur in tracks where a checkline is placed on a lapline (in 1.0, the tracks affected by this are Antediluvian Abyss, Around the Lighthouse, Oliver's Math Class, Volcan Island, and Zen Garden). Perhaps it would be possible to just remove such checklines in future versions.

As for compatibility between 1.0 and 1.1, for some reason I was under the impression that the upscaled version of Zen Garden was going to be used in 1.1 (it was intended for 1.0 but was accidentally reverted), in which case runs of that track in 1.1 would of course not be compatible with 1.0. I don't think it matters very much if it's going to be used, though, so if it's considered important to maintain compatibility, I think it shouldn't be changed.