Teeejj:

It's surely true that not asking for positive change is unlikely to achieve it. So far I've been nothing but polite and matter-of-fact in this discussion. I fail to see how anyone could construe my behavior as "hassling", though in my estimation that could easily describe your own comment. What could you have hoped to accomplish with your post other than to silence reasonable discussion?

BenInSweden:

It's obviously false that nobody wants to record flaps, as this very thread was created because someone wants to record flaps. Moreover, the game itself already reports all of the relevant information! (As do all games in this genre that I'm aware of.) That functionality didn't just materialize out of the ether. The game's developers believed that enough potential customers would be interested in flaps that they spent development resources implementing a feature to report individual lap times.

It's also historically false that nobody "has had the desire" to record flaps, as The Sonic Center kept and ostensibly still keeps track of them. Even though the site is now essentially defunct, beginning in 2003 and for over a decade, it appears to have been the premier destination for the Sonic speedrunning community, and was quite popular by the standards of the early internet.

Based on forum and leaderboard participation, Sonic 3 speedrunning was vastly more popular under The Sonic Center's stewardship than under yours. Note that, for example, at least 78 people submitted AL flap times to The Sonic Center between 2003 and 2016: https://www.soniccenter.org/rankings/sonic_3/time_attack/azure_lake/single_lap . That's a significant desire to record and compete in the flap category!

But even absent this overwhelming historical evidence, simply the fact that virtually every other game in this genre -- on SR.C or elsewhere -- organizes their leaderboards more or less in the manner that I'm suggesting, should be a strong indicator that perhaps this leaderboard is serving its community poorly.

It's hardly persuasive to claim support from "a few runners" -- who for whatever reason can't be bothered to publicly express their opinion (much less actually play the game!) -- confirming your personal bias that SR.C "isn't the place" to record flaps because they are "so short". I'm not aware of any other game on SR.C or anywhere else that has ever made such a distinction. (And even if an obscure one exists, none of the ~25 most popular racing games appear to do so.) All other racing game communities appear perfectly happy to record arbitrarily short flap times, and their members appear perfectly happy to compete in achieving them: https://www.mariokart64.com/mk64/course.cgi?c=3 . (This is just one example among dozens.)

Anyway, it's difficult to imagine what potential harm there could be in reorganizing these leaderboards to conform to the consensus standards of the communities of every other racing game ever, and I've soundly refuted your every excuse for why such a reorganization shouldn't take place, with exhaustive documentation and references to boot.

Admittedly, I did speculate that the derelict state of these leaderboards was the reason why Werster (who you brought up) chose not to participate in them. This speculation was apparently incorrect, which apparently inspired you to categorically assert that "[I] go off and make claims that have no factual basis for [my] arguments", though in truth that seems to be a more apt description of your own posts. Regardless, my speculation about Werster never had any particular bearing on my line of reasoning.

That being said, whatever reason Werster has for not participating here, it remains mystifying and problematic that you don't think these leaderboards should adapt to include someone like him, if you believe his runs to be legitimate and verifiable. I don't know anything about him other than that he appears to be the world record holder in the categories this leaderboard purports to track (and more), and that he's incredibly prolific, as at the moment he has 963 other runs listed on SR.C, most of which appear to have been done on emulators and verified by their respective communities, including many emulated Genesis runs.

It seems quite unlikely that he uses substantially different setups for different Genesis games, so it's peculiar to say the least that his S3 Competition Mode runs somehow don't comply with your emulator verification rules, despite his standards of proof apparently being sufficient for so many (all?) other communities. I also note that, weirdly, he now even has a run on these very leaderboards, added by you three months ago: https://www.speedrun.com/s3k_competition_mode/full_game#S3 , though at first glance nothing appears to distinguish it from any of his other S3 Competition Mode runs.

Anyway, this was never supposed to be about Werster, and as I already mentioned, my speculation about him has no particular bearing on the topic at hand beyond the general observation that if you're somehow excluding runners like him, perhaps these leaderboards aren't serving the community very well. But given that you linked one of his videos, and given that as a result I then looked at his channel, I can't help but note that it seems peculiar to me that his runs wouldn't be listed here, and that you would claim the reason is because he doesn't comply with emulator verification rules despite doing so on hundreds and hundreds and hundreds of other runs, many of which are emulated Genesis runs, and some of which are even emulated Genesis runs of other Sonic games!

Maybe, just maybe, these leaderboards aren't adequately serving the community, and maybe, just maybe, that has at least something to do with the sad state of S3 Competition Mode speedrunning today!

I suspect it was incorrect to branch your roads at 1:30, as it was already clear that Longest Road was part of your win condition. It was definitely incorrect to miss the win outright at 2:37! (With 2xWood 5xOre in hand, you can trade 4xOre for 2xBrick via the Ore port you control, build two roads, and retake Longest Road for the win.)

[quote=BenInSweden]I'm not on a pc at the moment, but from a quick glance over these, the flaps are generally longer than a single race here on comp mode.[/quote]

Of what significance is the length of the flaps? You said that you couldn't find any games that tracked flaps. I showed convincingly that, in fact, multilap racing games here and elsewhere overwhelmingly default to tracking flaps. It seems to be interesting information that most games not only display and record, but that most players also seem interested in tracking.

In other words, ok, S3TA has shorter races than most multilap racing games. So what?

[quote=BenInSweden]Where they are in the single digit seconds like here (in the case of dfr and mrm at least) glitches or oob are performed to get the time and aren't separate categories.[/quote]

Of course you know that the 800-pound gorilla in this space -- MK64, which has far more players and runs than every SR.C game listed above combined, and three times as many players as any other MK game -- keeps leaderboards for both shortcut and non-shortcut runs. (As do SMK, Super Circuit, MKDS, MKWii, MK7.)

Yes, you can find games that don't do this, as well. But it's starting to seem like you're simply cherry picking information that confirms your existing bias, in order to reach a predetermined conclusion. That is, you don't appear to be arguing in good faith, and it appears that you have no intentions of ever changing anything, no matter what, and there's nothing that can convince you otherwise.

I believe that approach is unwise and ill-serves this community. You should try to determine how to make the most interesting and accessible leaderboard for this game. Yes, we can look to other leaderboards and similar games for guidance, but we need not be bound by them. (But even if you want to slavishly adhere to prevailing trends, as we've seen, MK64 -- the biggest game in this genre by far -- organizes its leaderboards essentially the way I'm asking you to do here, as do most other MK games, all of which are also much bigger than any multilap racing games on SR.C.)

Consider that running this game might be more attractive to more people if the leaderboards tracked all interesting and distinct ways of playing the game. (I can only imagine that Werster, whose video you linked before, and who I now note has times on his youtube channel for every single category I'm suggesting, hasn't submitted here because these leaderboards are such a disaster. The dude has 900+ runs on SR.C and for some reason hasn't bothered to submit his world record times to this S3TA leaderboard ... 🤔)

[quote=BenInSweden]By combining S3 and s3k, the glitched category end up requiring S3 to compete, having a board that separates them means they can be competed on separately, hence why they are separate.[/quote]

You don't need S3 to compete in the glitch category. You need S3 to do BP glitch runs specifically. This fact remains true no matter how you organize the leaderboards.

Separating leaderboards by game means that you're tracking S3 and S3K times for AL, CG, DP, and EM, even though those runs are all mechanically identical regardless of your choice of game. You're keeping two separate leaderboards for each of those tracks that should logically be merged. Furthermore, as currently organized, both your S3 and S3K leaderboards "should" be nothing but glitch runs. Non-glitch runs aren't competitive on those four tracks, regardless of your choice of game.

Separating leaderboards by glitch status means every single leaderboard is tracking an interesting and mechanically distinct run, and you're tracking all interesting and mechanically distinct runs (ignoring character choice.) Nobody will be excluded that isn't already excluded under your current system. You will need S3 to run BP glitch, which is true now, and will always be true, regardless.

In fact, separating based on glitch status is more inclusive because then everyone can compete together on unified leaderboards for 9 of the 10 races, regardless of which version of the game they own. Right now the S3 category excludes all S3K players from competing on all five tracks, instead of just the single mechanically distinct category on a single track. And obviously the reverse is true: S3 players are currently excluded from competing on the S3K half of the leaderboard, even though only one of the races is mechanically distinct.

[quote=BenInSweden]So if we did something with splitting this off it would likely be using a different terminology like "Shortcuts"/"No Shortcuts".[/quote]

Ok? It doesn't matter what you call it, as long as the chosen names are meaningful and relevant.

I'm not aware that there's any difference between the two games when it comes to the DP glitch, and the video that you linked doesn't appear to express an opinion one way or the other. If I'm correct, then simply separating runs by glitch status is sufficient, and seems like the more sensible option.

Either way, you would need a copy of S3 to run the BP glitch. Every other category could be run on either game. Nobody would be "walled off" that isn't already being excluded under your current system.

The point is that by separating runs by glitch status, you would allow people to actually run and compete in the non-glitch category. As it stands, all of the records here "should" be nothing but glitch times -- except S3K BP, obviously -- though that's not the case because no expert players have bothered to submit times. Regardless, only supporting glitch runs is uninteresting.

Note that on SR.C, most multi-lap racing games with IL leaderboards do, in fact, track flaps:

https://www.speedrun.com/dkr/levels https://www.speedrun.com/mrm/levels https://www.speedrun.com/bfr/levels https://www.speedrun.com/looney_tunes_racing/levels https://www.speedrun.com/pmwr/levels https://www.speedrun.com/mrc_multi-racing_championship/levels https://www.speedrun.com/rblxsbb/levels https://www.speedrun.com/roblox_karts/levels https://www.speedrun.com/mxvsatvreflex/levels https://www.speedrun.com/mx_vs_atv_untamed/levels etc.

Additionally, many popular old-school racing franchises like Mario Kart and F-Zero have community websites predating SR.C where they track flaps:

https://www.mariokart64.com/mk64/wrc.cgi https://www.mkwrs.com/smk/ https://www.mkwrs.com/mksc/ https://www.mkwrs.com/mkdd/ https://www.mkwrs.com/mkds/ https://www.fzerowrs.com/snes/ https://www.fzerowrs.com/x/ https://www.fzerowrs.com/mv/ etc.

From what I can tell, tracking flaps is overwhelmingly the default behavior in multi-lap racing games and their communities.

Also note that very nearly 100% of games on SR.C separate leaderboards based on glitch status -- i.e. they don't put glitch runs and non-glitch runs on the same leaderboard -- because people enjoy competing on both styles of run.

I was going through a very old hard drive and happily recovered some pictures (and a single video) from 2003, when I played this game. Most of them would (apparently¹) be world records by some margin if they could be verified. I thought that the SR.C S3TA community might like to have these times to refer to as baseline times to beat.

I also suggest that the S3TA boards here be expanded to record the fastest single lap, to separate glitched times from non-glitched times, and to have a leaderboard for each of Sonic, Tails, and Knuckles, as the three characters are mechanically distinct.

Notably, you shouldn't need separate leaderboards for S3 and S3K. The only difference between the two games that I'm aware of is that the hole in the ceiling in Balloon Park was patched in S3K. If you have separate boards for glitched and non-glitched, that will automatically be taken care of.

Currently, the SR.C leaderboards have a mixture of glitched times (Azure Lake S3K, Balloon Park S3, Endless Mine S3, Endless Mine S3K) and non-glitched times (all other tracks.)

Anyway, here are the times that I recovered:

Azure Lake - 17'31" - (5lap, Non-glitched, Sonic) Azure Lake - 02'81" - (Flap, Non-glitched, Sonic) Balloon Park - 21'81" - (5lap, Non-glitched, Tails) Balloon Park - 03'35" - (Flap, Non-glitched, Tails) Chrome Gadget - 28'76" - (5lap, Non-glitched, Sonic) Desert Palace - 26'38" - (5lap, Non-glitched, Sonic) Desert Palace - 04'86" - (Flap, Non-glitched, Sonic) Desert Palace - 04'46" - (Flap, Non-glitched, Tails) Endless Mine - 35'90" - (5lap, Non-glitched, Knuckles) Endless Mine - 06'98" - (Flap, Non-glitched, Knuckles) Endless Mine - 25'60" - (5lap, Glitched, Tails) Endless Mine - 04'73" - (Flap, Glitched, Tails)

Note that The Sonic Center has a leaderboard for S3TA, but from what I can tell, none of the claimed records have even a picture for verification (much less a video), and the site seems mostly defunct. They also appear to have a "freestyle" section intended for glitched strategies, but users have clearly posted glitched times in the normal (non-glitched) leaderboard.

¹ I haven't made an exhaustive effort in search of better times, there may be other sites, communities, or channels on youtube with more up-to-date information and better times.

Despite having 19 people with the verifier role (or above) here, I have observed on many, many occasions that runs here languish unverified for weeks.

This is especially egregious for top 10 runs and world record runs. At this very moment there are at least three 1:17 NMG runs (i.e. runs that are faster than the listed WR) awaiting verification.

Please grow the corpus of verifiers and consider removing the role from anyone who hasn't verified a run in the past two weeks, as it seems those users are simply wearing the role as a badge and not taking their responsibilities seriously.

@Wafu Yes, my intent is to write this script. I think it's ridiculous to just throw away a decade of historical runs when we can so easily convert the times for most of them. Anyway, so far my understanding of loads is as follows:

1. Any frame of the "door opening" loading animation is a loading frame, except for at the start of the animation when the door is still fully closed.

2. As long as "the pentagrams" show that "Save And Exit Game" is highlighted, any frame in which those pentagrams' animation is in the same state that it was in the previous frame is a loading frame.

Are there other situations or exceptions to these rules that you can think of? Does this cover local games, LAN games, TCP/IP games, battle.net games, etc.? Are there differences between Windows and Mac? Don't some of those options have an additional loading bar under the (static?) door picture? Are there situations in which the player can save and exit and induce loading frames without the pentagrams highlighting "Save And Exit Game"? Does the use of a glide wrapper affect the appearance or behavior of any of these UI elements?

To anyone who has ever submitted a pacifist run, and other interested parties, do we want to make an exception here to the recent base-game rule that forbids starting a character in Classic and converting it to Expansion?

MacroBioBoi: Allow conversion DrCliche: Allow conversion blazer-flamewing: Forbid conversion Zutatensuppe: Allow conversion Blixoxo: ?

Obviously we can't read a byte in memory for a VOD, but if we can describe how the game looks and/or behaves shortly before, during, and after the autosplitter pauses or unpauses the timer, we should be able to write a script that will very closely mimic that behavior for re-timing old runs.

Unfortunately I still don't understand what constitutes a "load". A good explanation would look something like this:

There are exactly two situations where the game is "loading", and thus the autosplitter pauses the timer:

1. The player clicks "Save And Exit Game". From the moment of that click until the character selection screen appears, the game is loading and the timer should be paused.

2. The entire duration of what is commonly known as the "opening door" loading screen, which occurs after you attempt to enter a game with a character. If this loading screen is visible, the game is loading and the timer should be paused.

Please note that the above is just an example of the type of explanation that would be helpful. As I personally don't know what situations are considered "loads", I just wrote something that sounded reasonable to me, but might be incomplete or incorrect in various ways.

If someone with this knowledge can precisely enumerate every situation that constitutes a load, taking care to account for differences that may exist between different game modes, different networks, and anything else that could potentially be relevant, that would be great.

Pretend you are talking to a malevolent AI that is simulating a perfect and catastrophic idiot, who knows nothing but what you tell him, except when the opportunity exists to misunderstand anything, in which case that will be exploited to maximum effect. That is, please make every effort to be as complete, consistent, explicit, and precise as possible. Someone has to turn your explanation into a computer program.

What do you mean by "loading the game"? What player actions and game states trigger the starting and stopping of the timer? Exactly when and why does the timer start and stop? What is displayed on screen at the time?

What do you mean by "entering a new map"? What player actions and game states trigger the starting and stopping of the timer? Exactly when and why does the timer start and stop? What is displayed on screen at the time?

What do you mean by "clicking ingame items on the map"? What player actions and game states trigger the starting and stopping of the timer? Exactly when and why does the timer start and stop? What is displayed on screen at the time?

Are there any other instances where the timer stops? What player actions and game states trigger the starting and stopping of the timer? Exactly when and why does the timer start and stop? What is displayed on screen at the time?

An autosplitter for Diablo 2 which removes "loads" exists. If we are going to write a script to automatically process old videos that were run without the autosplitter, with the intent to replicate the behavior of the autosplitter, we must know the what, when, where, why, and how of whatever the autosplitter is doing.

During the rules discussion today (2020-05-30) I attempted to get anyone in Teo-'s chat to elucidate what constitutes a "load", but no one could tell me a definition that wasn't obviously inconsistent and/or incomplete.

If we want to automatically re-time old runs (that have videos) via a script, this information needs to be known. (And it is known, at least to whoever wrote the autosplitter.)

When and why does the autosplitter stop the timer to remove loads? (What player actions and/or what game states trigger the starting and stopping of the timer?)

It wouldn't be terribly complicated to develop a python/opencv script to automatically re-time all old runs that have video. For other runs, just worst-case convert missing times using the method I outlined earlier, agree upon a default sorting method, and list both timing methods so interested parties can re-sort the leaderboards if they desire. Talk of retiring any runs from the leaderboards is definitely barking up the wrong tree. It's both wrong-spirited and entirely unnecessary.

My suggestion:

• Write an autosplitter that removes loading times.
• List both RTA and NoLoads timing simultaneously on the leaderboards.
• Re-time the top X runs for most categories. (Pain in the butt, I know, unless someone can automate this, too.)
• Determine the maximum and minimum ratio between RTA and NoLoads for all runs that have this information available.
• From now on, force all competitive runs to be submitted with both timings. (i.e. Make a rule that if you submit a top X run in any category, you must time your run using both methods.)
• Recalculate these ratios every time a new run is submitted which lists both times.
• Automatically worst-case-convert runs that list one time but not the other. (i.e. If RTA only is submitted, multiply by 1 / Min[ {RTA_i/NoLoads_i} ]. If NoLoads only is submitted, multiply by Max[ {RTA_i/NoLoads_i} ].)

This will allow new runners to submit runs using either timing method without hassle, and they will be approximately well-ordered on the leaderboards, while also ensuring competitive runners aren't hindered by inconsistent loading times and hardware differences. Furthermore, viewer quality-of-life won't be negatively impacted by competitive runners switching en masse to a sterile, soundless game.

Which timing method you want to give preference to on the leaderboard is a question, but not that big of a deal, in my opinion. In games (that I'm aware of) that already do this, they order the leaderboard by time without loads, and all competitive speedrunners use LiveSplit and the appropriate autosplitter without complaint.

If you're going to make it even more of a pain in the rear to log in to the site, can you PRETTY PLEASE let us stay logged in forever? It really grinds my gears that the dang site logs me out every couple days.

You can easily make a mistake and not even notice, or fail to notice until it's too late to "fix" it, especially if you happen to be streaming and interacting with chat. If we're going to allow an exception, I think it makes sense just to flag runs with errors of any kind -- as long as the player isn't being intentionally exploitative -- rather than impose a weird penalty on the player that covers some situations but not others, or weirdly distinguishes between an accidental hit that kills and one that doesn't.

That way the purists can turn up their noses at runs with errors and filter them out, while people who are more interested in seeing how far the concept can be pushed (regardless of perfection in execution in the face of random BS) have a larger pool of entertaining runs to draw from.

At the end of the day, do we want runs like blazer-flamewing's and MrLlamaSC's to be on the leaderboard, available as a resource for people interested in D2 pacifist runs? (I do.) Do we want more people to try the run and participate in the competition, and not be turned off by the fact that one of the best players in the world tried it and "failed"? (I do.)

MrLlamaSC did a run today, and the beetle lightning wasn't triggering on Thorns damage for him, either. I must have just never noticed before, maybe due to having lots of Nova Shield gear by that point in previous runs.

What do you folks think of adding a flag in the pacifist category to indicate minor rules violations, like those commonly caused by the bindings glitch? MrLlamaSC is doing a (probably WR) pacifist run right now that I personally would like to see on the leaderboards, but he accidentally swung at a few things, and also encountered a very improbable situation where he couldn't TP out of Diablo's bone prisons, so had to attack one of them.

I personally prefer being soft on the bindings glitch, but I got such opposition to the idea here on account of it being ambiguous and potentially exploitable and impossible to enforce that I just stopped arguing for the exception. Perhaps an "Accidental Hits" flag could be a good compromise to still allow those runs on the leaderboard?

I'm not sure ... I didn't feel like I was taking any damage from hidden charged bolts, but I also had more lightning resistance than I normally do throughout the run. I'm honestly just not sure.

In my most recent run, at some point all of the lightning enchanted enemies stopped releasing charged bolts when struck, and switched to only releasing them when they died.

Rakanishu's LE processed every time he took damage, as usual, but by the time I got to Beetleburst, LE was only processing on death, and it persisted at least through Stormtree.

Has anyone ever seen this glitch before? I sometimes use a character editor called "Hero Editor" to test things, is it possible that has somehow caused unintentional changes to the game that affect new, unedited characters?

The "set your hotkeys" guidance isn't supposed to be a rule you must follow, it's just advice how to not accidentally attack things (which ¤is¤ forbidden). If you think that's confusing, perhaps you could put it in parentheses or add something like "to avoid this glitch, it helps to ..."

The skill listings aren't meant to be comprehensive, you could add "e.g." to the examples if you think that reads better. I think the rules as written are pretty straightforward, but examples are provided just in case people aren't quite sure about intent or interpretation.

Like, I can sort of understand someone asking "but can I use Holy Fire"? But once you say "no Holy Fire" I don't think any reasonable person will then wonder whether they can use Holy Freeze.