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.

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?)

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?

I recently posted two proof-of-concept pacifist runs through Normal Baal.

Sorceress: https://www.reddit.com/r/speedrun/comments/668zh5/diablo_2_pacifist_74133/ Paladin: https://www.reddit.com/r/speedrun/comments/668zh5/diablo_2_pacifist_74133/dgkofcd/

The response has been fairly positive, with both Yagamoth and BlazerFlamewing attempting runs of their own. Is there enough interest to set up an actual leaderboard for this category here? If so, let's hash out what we think the best ruleset is.

Here are my current thoughts on the rules I originally posted on Reddit.

Rule #1: You may not take aggressive action toward any creature or animated construct, nor may you provoke or summon an entity, or hire a mercenary, to do so on your behalf.

This is the core tenet behind the pacifist concept, and I think it should remain largely unchanged, though there are still a few questions that need to be answered. For example, should any Necromancer curses be allowed? Can anyone use them, or just Necros?

Rule #2: You may enable Ladder-only runewords.

Originally I allowed this because it wasn't known how to get a non-Sorceress past Lord de Seis, and because I thought Edge was necessary for the Sorceress run to be reasonably viable.

My Paladin run shows how to get non-Sorcs past De Seis, and BlazerFlamewing showed () that Edge isn't at all a requirement.

Since enabling Ladder-only runewords requires messing around with text files arguably external to the game, it's likely to be seen as fairly controversial. Let's avoid that controversy by disallowing Ladder-only runewords (or perhaps only allowing them in a miscellaneous category.)

Rule #3: You may freely use /players X.

Yagamoth seemed disinclined to use /players X in his run, perhaps because he's an experienced runner and none of the main categories allow /players X.

But pacifist runs (especially non-Paladin runs) are dominated by long stretches of leveling that are made a ¤lot¤ less tedious by use of /players X. I still think /players X should be allowed, as it makes runs a lot more palatable for viewers and players alike.

Rule #4: You aren't responsible for the behavior of character-independent NPCs.

I think this is fairly obvious and shouldn't be changed.

Rule #5: You may use damaging shrines, since they're "acts of god".

I used a damaging shrine once or twice in my Sorceress run, originally justifying it by saying the in-game character had no way of knowing what was going to happen. Yagamoth didn't want to use damaging shrines in his run.

In retrospect, I'm leaning towards Yagamoth's viewpoint that using damaging shrines isn't pacifistic. We players certainly know what's going to happen when we use them, and it seems against the spirit of the concept to incentivize players to optimize the best way to repeatedly and actively kill Rakanishu with damaging shrines at the start of the run.

Rule #6: If you accidentally do something non-pacifistic, simply apologize and don't allow that enemy to die. If that enemy ¤needs¤ to die, respawn it unprovoked in a new game.

This was a rule I made out of practicality, to deal with a terribly annoying bug that causes your mouse bindings to revert to attack completely unpredictably when any of your equipment changes.

For example, maybe you're fully statted for all of your gear even when naked, and you swap your amulet, or put on a pair of boots. Sometimes (not all the time, it's completely unpredictable!) your mouse bindings will revert to attack. It also happens quite frequently (but not all the time!) when you do the standard speedrunning trick of buying potions quickly by filling up and then picking up your belt.

If you aren't paying SUPER CLOSE ATTENTION, or being completely OCD about fixing your bindings every time you buy potions or switch gear, it's very easy to accidentally take a swing at an enemy, which is a crappy way to lose a multi-hour run.

I think this rule is good because it doesn't senselessly punish meaningless mistakes that largely derive from a glitch in the game. In that way I hope it makes the category more accessible and less annoying. Additionally, as written, I believe it doesn't incentivize players to optimize how to best skirt around the edges of the rule.

I still think actively provoking an enemy in any way and then watching it die (or simply killing it outright with any action) should be completely disallowed, however.

The current rules say timing starts on file select, however, none of the currently listed runs are actually timed in that way. We should thus agree upon a reasonable timing method and the rules should accurately reflect that method.

Timing from file select would mean starting your timer on one of these two screens, depending on how you interpret "file select": http://imgur.com/a/Ofexx

Timing from file select would thus include the intro, as well as entering and exiting Harara Mountains. I don't think the initial menuing and intro/tutorial skipping are worth including in a run for a few reasons:

• They're not interesting.
• Forcing players to start a brand new save file after restarting a run in which they have yet to collect any ball materials makes grinding good starts really annoying.

Anyway, Chumlum and Midboss time their runs from the moment they click "Pinewood Heights" for the first time in the level menu.

I time my runs from the moment I select a ball for the first time in Pinewood Heights, though my times as displayed on this site are altered to conform to the timing method of Chumlum and Midboss. (Though I make no particular effort to select a ball quickly as possible.)

From what I can tell, the general consensus among games with similar situations is to ignore all of the initial menuing, especially when it comes before you start actually playing the game. As a concrete example, consider the way Mario Kart 64 is timed.

I believe my timing method of starting the clock when you select your first ball (and actually start playing the game) makes the most sense.