Post subject: TAS and scoring.
Editor, Skilled player (1157)
Joined: 9/27/2008
Posts: 1084
When TASes are submitted and published, their times are shown rather prominently. Time is an easy standard to measure by, as it's automatically measured by the length of the input file. A lot of games have a clear endpoint, and reaching that endpoint often only has time as a measurement stick for how good we did. Though, what about score? It's not as easy a thing to automatically read, yet it is still a clear and obvious value of measurement, should one want to run by it. Many older games have a scoring system. Many score to infinity, or whatever integer limit they programmed, and some would allow endless scoring if deaths aren't restricted by some conduct, and yet a few others have their scoring, while otherwise finite, are broken by some trick, like the Super Mario Bros. "turtle on the stairs" trick even when restricting deaths. Some games do have an interesting limit to reach, though. Whether it be a hard-coded limit or based on what's available to score from, reaching that limit itself can be an interesting process. I feel as though the impression of TASes to be accepted here is strongly for shortest time, with no exceptions to be made for score. Of course, this feeling isn't exactly true, as we have score-based TASes published here, but the recent Track & Field submission has definitely elected to take time over score. An example where it is perceived we take time over score, and so the submission. Whether this perception is accurate is not for me to answer, but it is one I feel. Regardless, if there is a way to promote creation of TASes of games that have an interesting scoring system, I'm for it. Discovering which games have an interesting scoring system, fitting under the criteria where it doesn't have boring "score grinding" tricks or endless gameplay, is generally up to the runners to find, much like how we need to pick non-trivial games to speedrun. I see the goal in a similar way as 100% runs, as you are trying to max something out on the way to clearing the game. So Link picks up this heart piece, so too will I shoot this enemy for score. In any case, I frequently see bonus rounds skipped or failed instantly just so we end faster.
Patashu
He/Him
Joined: 10/2/2005
Posts: 3999
Maximum score being an interesting goal is mostly only true for arcade games (and even then only for the subset of arcade games that can't be counterstopped, that is, the limit for how big of a score the game can store is reached or even overflowed). I'm mostly only aware of shmups being interesting to play for max score, and it's extremely hard and requires a lot of esoteric knowledge to do this, so it's unsurprising to me that max score TASes aren't as thoroughly explored as max speed TASes. At least there's [2836] GC Ikaruga "2 players, maximum score" by keylie in 21:29.40 and [3138] Arcade Osman "maximum score, deathless" by xy2_ in 09:33.18.
My Chiptune music, made in Famitracker: http://soundcloud.com/patashu My twitch. I stream mostly shmups & rhythm games http://twitch.tv/patashu My youtube, again shmups and rhythm games and misc stuff: http://youtube.com/user/patashu
Post subject: Re: TAS and scoring.
DrD2k9
He/Him
Editor, Expert player, Judge (2036)
Joined: 8/21/2016
Posts: 1009
Location: US
FatRatKnight wrote:
...I see the goal in a similar way as 100% runs, as you are trying to max something out on the way to clearing the game...
I agree that achieving a maximum score (on games with a definable maximum) seems comparable to "100%" or "100+%" run. In those "100%" type runs, the goal is collecting all possible items and/or completing all dungeons/levels/etc. in the fastest time. However, that still acknowledges that time is sacrificed when compared to absolute fastest "any %" endgame runs. Setting a goal of "maximum score" is similarly acknowledging that you're (most likely) going to sacrifice time to attain the highest score. In that way, I can see "maximum score" being feasibly considered a publishable category like "100%," "any %," etc. That being said, the primary difference between most "any %" runs and "100%" runs is that "100%" runs show more of the game than the "any %" runs. Unfortunately, setting a goal of "maximum score" for a run doesn't guarantee it's going to show any more of the game than any other (lower score) run. The only assumption that can be immediately made is that the run will be longer because extra time will be spent to achieve a higher score. In my opinion, a higher score run based simply on score value that doesn't also show some new game content is unnecessary for this site. Watching such a TAS would essentially be watching a longer version of a shorter TAS with the only satisfaction for watching the extra time being a difference in a displayed digit sprite. If new worthwhile game content is also showcased while achieving the (theoretical) maximum score, I'm for acceptance as a separate category.
Post subject: Further thoughts on TAS and Scoring
DrD2k9
He/Him
Editor, Expert player, Judge (2036)
Joined: 8/21/2016
Posts: 1009
Location: US
Some further thoughts regarding this concept. 1) If "max score" becomes a valid publishable category, timed sports games could theoretically become a valid game choice with a goal of maximum score. Examples: Baseball, Basketball, Soccer, Football, etc. for any system. Though there is no maximum score in many of these sports, the video game mechanics would determine a maximum potential score in the set playtime. I can imagine there's a subset of TASers who would be interested in some of these type of games. 2) There are games with theoretical maximum scores that would end in longer runs and showcase more game (even if it's somewhat repetitive). Examples: Many athletic and non-timed performance based games. (Track & Field type games, Mario Party type games, gambling games) 3) There are some games where the endgame/"game over" animations vary depending on final score (whether it is a maximum or not). Examples : Gameboy Tetris, NES Track & Filed II 4) There are some games where the endgame credits are only shown if a certain score level is achieved even if it's not the theoretical maximum. Example: NES Open Golf 5) If this becomes a publishable category, I feel sorry for the judges who have to sit through these potentially very long, potentially very repetitive videos necessary to achieve some of these score based goals. 6) To my understanding, TAS originally meant "Tool-Assisted Speedrun;" with the original intent of TASing games being strictly speed-running. TAS has since evolved to also mean "Tool-Assisted Superplay" as in the title of this website. This evolution allows for speed to not be the only consideration of what makes a great TAS video. The simple fact that we have additional categories beyond the absolute fastest "any%" run demonstrates this idea. It seems to me that effective definition of "Superplay" for the purposes of this site is the following: A run that accomplishes X goal(s) as fast as possible while also attempting to provide some degree of entertainment value. Some speed/entertainment trade-offs are acceptable. If that is an acceptable definition, "maximum score" runs should qualify. It also, unfortunately, means that goals such as "fastest death" or "fastest 'Game Over'" screen should also be acceptable. Which I guess brings us back to the crux of this discussion...is "maximum score" a category we want to accept for publication on this site?...And more generally, which goals are worthy of publication and which ones aren't?
Editor, Skilled player (1157)
Joined: 9/27/2008
Posts: 1084
So, stuff happened in this submission. (edit:fixed) It brought to light a detail I wish to contest with the Vault rules as written.
Maximum points or score is allowed as a full-completion category, provided that: * There is no better way of defining full completion in the game. * The maximum score is easily defined and absolute - it must not be possible to gain a higher score, even theoretically. It must be definable without being dependent on precise time, speed, or similar requirements. * The maximum score is limited by not being able to gain any more points, not by hitting a score cap or overflow. If it is possible to score points infinitely, score cannot be used to define full completion.
If it can not be proven you have hit the highest possible score, yet multiple people can't improve upon the attained score value, is this still grounds for rejection? By the definition of the question I put forward here, we have not hit a theoretical limit. However, we did hit a practical limit, and have not been able to improve on it. I wish I could cite on the Ikaruga example, as the scoring is rather complex (is it possible to have absorbed just one more bullet in that mess?), but that run was accepted in another tier. It therefore did not have to deal with Vault definitions. I also have a single level Krazy Kreatures TAS. Granted, it's some side project and not actually published here, but would it be allowed in Vault, assuming the played mode was acceptable? It aims for maximum score. I do not know if I have gotten the highest possible score. I can tell you that there is a finite amount of time before the control will be locked and the level ends. I can tell you the strategy I've used is probably a very good scoring one, but I can't tell you if there exists a better strategy than the one I used. Is this scoring TAS to be rejected for being unable to prove I have achieved the highest possible score? I can only argue the strategy I used reached its limit, but not that the strategy has reached the absolute limit. We commonly use time as a criteria for obsoletion. Is there a reason we can't use a score value as a criteria for obsoletion as well?
DrD2k9
He/Him
Editor, Expert player, Judge (2036)
Joined: 8/21/2016
Posts: 1009
Location: US
FatRatKnight wrote:
We commonly use time as a criteria for obsoletion. Is there a reason we can't use a score value as a criteria for obsoletion as well?
Does the following hypothetical example explain what you mean? Current publication of game X = 11270 frames with final score of 148765 New submission of game = 11270 frames with final score of 153972 Should the new submission with better score obsolete the old publication even though they are equal on time?
FatRatKnight wrote:
...but that run was accepted in another tier. It therefore did not have to deal with Vault definitions.
Though I'm sure it would be a pain to implement, perhaps a completely new tier or vault sub-tier needs to be considered for runs of maximum known achievable score (assuming the score can't be run to infinity). This would require modification to the current rules. Specifically the following: * The maximum score is easily defined and absolute - it must not be possible to gain a higher score, even theoretically. It must be definable without being dependent on precise time, speed, or similar requirements. This would need modified to allow games where a maximum score isn't as easily defined. Also the requirement to not have a theoretical improvement to score would need eliminated. Honestly, this site accepts speed runs as the best known that have theoretical (yet unproven) speed improvements (many submissions even detail these potential improvement possibilities in the submission text!). If unproven theoretical speed improvements don't restrict a know submission from being accepted as fastest, why should an unproven theoretical score improvement restrict a run of highest known achieved score from being accepted? (Assuming there's a tier/sub-tier for it). * The maximum score is limited by not being able to gain any more points, not by hitting a score cap or overflow. If it is possible to score points infinitely, score cannot be used to define full completion. The restrictions on cap/overflow could be changed to accept reaching the cap/overflow point as the maximum attainable score. Also this rule suggests a question on when a game is considered completed. Currently for endless games, a point of no-new-content is considered completion for TAS purposes. If a maximum score is attainable but long after the game has reached a point of no-new-content, which of these is considered the actual completion point? This question also supports a reason to have a separate tier/sub-tier from speed based runs. Again, I anticipate that implementation of these ideas would likely be a significant amount of work for the operators of the website. Even if the general community accepted these ideas, It would ultimately fall on the site admin to determine whether or not new tier/sub-tier system would be created and implemented. If these "maximum score" ideas could somehow be implemented as simply new branches, implementation would be easier. But I think that would be a harder sell to the community. Perhaps the best solution would be a completely separate website (which could be linked as a sister site) which contains only TAS runs of known highest scores. This would again be more work to implement, but would eliminate the congestion of current tiers and movie lists. "TAS" would then gain yet another evolution to it's definition as "Tool-Assisted Score-run." ____________________________________________________________________ I recently read Mothrayas's judging comments of Submission 5560 which addresses arbitrary goals. Perhaps it would be good to get his perspective on these "maximum score" ideas.
Noxxa
They/Them
Expert player, Moderator (4131)
Joined: 8/14/2009
Posts: 4083
Location: The Netherlands
DrD2k9 wrote:
FatRatKnight wrote:
We commonly use time as a criteria for obsoletion. Is there a reason we can't use a score value as a criteria for obsoletion as well?
Does the following hypothetical example explain what you mean? Current publication of game X = 11270 frames with final score of 148765 New submission of game = 11270 frames with final score of 153972 Should the new submission with better score obsolete the old publication even though they are equal on time?
This is already a thing. When a category such as low% or max score gets a new movie that achieves its goal better, it obsoletes it regardless of any time change. For instance, the last Super Metroid low% publication is a few minutes slower than the one it obsoletes, but it gets 1% less items so it's considered an improvement. In general, questions like this are already answered in the history of site publications for these types of goals. The only issue here is that max score is not a Vaultable goal in many of these cases. The Vault is a speedrun archive - TASVideos is still primarily a site for tool-assisted speedrun records, after all - and it demands either any% fastest completion, or full completion, as qualifiers. If a max score cannot logically prove itself to be full completion, it is not full completion. Now, it could be possible to think up a separate category system or whatever (not "tier", as it would contradict the definition of tier) where we could accept score runs, but I don't really see it being in the interest of the site. Score runs are not our primary focus, never have been, and since we're talking about runs that aren't accepted in our current framework (in which they would be allowed if they qualified for Moons given sufficient entertainment votes), we're talking about runs that are typically boring, repetitive, and long - amplifying publishing work, for relatively little gain.
DrD2k9 wrote:
* The maximum score is easily defined and absolute - it must not be possible to gain a higher score, even theoretically. It must be definable without being dependent on precise time, speed, or similar requirements. This would need modified to allow games where a maximum score isn't as easily defined. Also the requirement to not have a theoretical improvement to score would need eliminated. Honestly, this site accepts speed runs as the best known that have theoretical (yet unproven) speed improvements (many submissions even detail these potential improvement possibilities in the submission text!). If unproven theoretical speed improvements don't restrict a know submission from being accepted as fastest, why should an unproven theoretical score improvement restrict a run of highest known achieved score from being accepted? (Assuming there's a tier/sub-tier for it).
This was answered in the Punch-Out submission thread - you're comparing apples and oranges, or to be precise, you're comparing optimization level (which cannot be precisely defined) with completion criteria (which can be precisely defined, and which have to be precisely defined for any movie to be publishable - any movie that does not reach a clearly defined goal or end-point is rejected). Remember that the Vault only allows any% and full completion runs - if a max score run isn't full completion nor can prove itself to be such, it does not count as full completion, nor any%, so it cannot be accepted for the Vault.
DrD2k9 wrote:
* The maximum score is limited by not being able to gain any more points, not by hitting a score cap or overflow. If it is possible to score points infinitely, score cannot be used to define full completion. The restrictions on cap/overflow could be changed to accept reaching the cap/overflow point as the maximum attainable score.
Sorry, but I am going to instantly reject any proposal that would force us to publish a 74-hour pinball run, or a 792-hour long Desert Bus run. This is one of the main reasons we require entertainment voting support for publishing movies with goals like this.
DrD2k9 wrote:
I recently read Mothrayas's judging comments of Submission 5560 which addresses arbitrary goals. Perhaps it would be good to get his perspective on these "maximum score" ideas.
That is entirely unrelated to this discussion, which is about the limits of acceptability of score runs for the Vault. That judging comment is about arbitrary branches that can be accepted to Moons or Stars. Max score runs already can get published to Moons or Stars easily, provided they have the audience support - and when they do, they can make for some of the greatest movies on the site. But they have to be considered interesting and entertaining enough.
http://www.youtube.com/Noxxa <dwangoAC> This is a TAS (...). Not suitable for all audiences. May cause undesirable side-effects. May contain emulator abuse. Emulator may be abusive. This product contains glitches known to the state of California to cause egg defects. <Masterjun> I'm just a guy arranging bits in a sequence which could potentially amuse other people looking at these bits <adelikat> In Oregon Trail, I sacrificed my own family to save time. In Star trek, I killed helpless comrades in escape pods to save time. Here, I kill my allies to save time. I think I need help.
DrD2k9
He/Him
Editor, Expert player, Judge (2036)
Joined: 8/21/2016
Posts: 1009
Location: US
Thank you very much for the clarifications and perspective. My own comments weren't meant as advocating for the inclusion of Score-based TAS videos on the current site, just voicing thoughts/concerns that I (and likely others have) had regarding the concept. Your arguments regarding the current approach to such submissions are exactly why I suggested that a completely separate website would likely be the best option for score based TASing instead of speed based TASing.
Mothrayas wrote:
Sorry, but I am going to instantly reject any proposal that would force us to publish a 74-hour pinball run, or a 792-hour long Desert Bus run.
LOL! I had thought about Desert Bus myself before I posted the list of game examples above, but forgot to include it. I fully expected it to be used as a key argument against ungodly long score based TAS videos. Thanks again!
Editor, Skilled player (1157)
Joined: 9/27/2008
Posts: 1084
Mothrayas wrote:
Sorry, but I am going to instantly reject any proposal that would force us to publish a 74-hour pinball run, or a 792-hour long Desert Bus run. This is one of the main reasons we require entertainment voting support for publishing movies with goals like this.
Alright, it is clear we shouldn't accept any run with rather repetitive scoring systems. I will not press for the acceptance of such runs to Vault. This then leaves a problem on my side I have to address: What sorts of scoring systems should we accept to start with? The definition in the rules already cover for games with a decisive upper limit. Meeting the goal, then finishing the movie as soon as possible is the standard. I do not want to include runs with a practically boundless upper limit. I want to include runs with an indeterminate upper limit, but decidedly finite. Proposing a complex definition to get the subset of runs I want to see published into Vault will put undue strain on the judging side of things. If the definition is to work, it must remain a simple one. Super Mario World has a scoring system. I wish to exclude it. Mario can repeat stages endlessly. Alien Hominid also has a scoring system. I want this one excluded as well. Find a safe spot and shoot back and forth in an endless loop. Krazy Kreatures has one as well. This one I would like included. Fields have a finite size, and you can only clear a finite number of lines before it moves on to another phase. Various arcade games like Pac-Man, Galaga, etc. have scoring systems. Once the difficulty maxes out, and no new content is seen after one loop at maximum difficulty (not counting unintended things like the level counter looping around or glitching out some versions of the game), perhaps that should be the stopping point. And if we determine that as our goal, there should also be an easily provable maximum for this goal, as we go through a finite number of stages. Of course, I'm twisting some aspects and attempting to squish this type of run to fit within the current Vault definition, by selecting a particular point of view, but I need to consider these thoughts when trying to come up with an appropriate proposal. ... I'm posting these thoughts now. It's going to take hours before I can think through my intent properly. When I can come up with something reasonable, some way to readily and somewhat systematically determine if the scoring system is "interesting", I'll finish my train of thought there. Of course, I realize the problem of Vault for uninteresting stuff, yet I'm asking for interesting scoring criteria, something which seems decidedly at odds against each other. My intent is to reduce some perceived fear of rejection of score runs of an otherwise passable goal just because the audience didn't find it interesting enough, to help get a few more score runs submitted to the site. Placing a "safety net", so to speak, in hopes that runs that really are Moons quality can get some security of publication.
Post subject: I surrender.
Editor, Skilled player (1157)
Joined: 9/27/2008
Posts: 1084
Had a good discussion with Mothrayas a few hours back. Now that I had the time for the thoughts to sit and reform in my head, it's time I put down a few clear examples. Some ground rules with the Vault has been clarified in the discussion, as well. Any memory corruptions or other tricks that directly modify score, bypassing normal means of attaining score, may be banned and such a run is still Vault-allowed, as they'd otherwise make the 100%-like completion definition rather fuzzy. Any tricks that, while they don't modify score itself, but modify a limited resource that would otherwise stop score from rising endlessly (such as time) must be allowed, or else we leave the parameters of the Vault by excluding more than the bare minimum restrictions. After further thought, this one fact of allowing corruption of a secondary resource that can lead to endless scoring greatly restricts any arguments I may have on widening Vault allowance on score runs. I do not have an answer. I will explore this reasoning later in this post. Regardless, I want to build on what I can elsewhere. I'll fetch a few examples. The scoring systems I seek must not reward repetitive behavior, and most certainly not those allowed without bounds. Super Mario Bros. can get a bunch of points, fetch a 1-up, die, and repeat a level forever. Restricting deaths would move toward finite resources, but then we're applying a restriction beyond the bare essentials the Vault seeks. It's not even that strange a trick. Of course, there is the turtle on the stairs trick which can get a particularly generous number of points, so even restricting deaths wouldn't end all the degenerative behavior I would like. This is one example I wouldn't mind rejection for Vault. Super Monkey Ball has most of its stages with less than 100 bananas in it. However, there is one I can think of where there are more, and even though it is clear the designers intended it to be impossible to collect 100+ in that stage, TAS can do it, getting points and lives from the bananas, die, and repeat forever. We can apply no deaths, putting a complete stop to this repetitive behavior, and probably get the run published anyway as the game lends itself very much more than Vault in its appeals, and the fact it uses time remaining as a primary scoring factor helps it further. But then there's no Vault "safety net" I'm looking for. Krazy Kreatures I've mentioned before. The game works in phases for each stage. First, stuff floods the stage, and if there's no room left, game over. After some determined number of line clears, it moves on to the timed phase where you are no longer in any danger of game over, as the flood of stuff stopped, so the point of phase 2 is to score what you can with whatever is left on the field. Once time runs out, the game runs its bonus count-up, where you maintain control for a short time more before it locks and you move on to the next stage. I'm aware of tricks to extend control during the count-up, but not endlessly, and I know of some scoring abuse in there, but it's not infinite. This is an example I want allowed for Vault, as I consider it having an interesting scoring mechanism that can be bested by a theoretical better strategy the original run didn't consider. There are many auto-scrollers where time to completion doesn't significantly change based on player's skill, outside of boss battles. Rather, whether the player can survive, and how many points they get, depend on skill. Survival is usually trivial in TAS conditions. How easy it is to get points depends on the game. If the enemies are sparse enough, maximizing score becomes trivial, and can be rejected on the basis of triviality. If scoring is difficult like Ikaruga, then there is no clear maximum, only that there must be a maximum from the lack of any loops or other infinite score grinding techniques. Infinite score grinding techniques. Oh, I hate the possibility of something undiscovered. I'm just listing out examples of things I'd like Vaultable or otherwise, but if we always assume there's some hidden infinite scoring trick somewhere, and that we never allow ourselves to restrict those, my rules completely break down. If we do have a game where there's no known score grinding technique, and we have some upper value we can't get past regardless of proof, such a run would be accepted to Vault in my imaginary world. However, the crux of the problem is that, if some technique is discovered later that breaks what would otherwise be a restricting resource (such as the clock stop in Punch-Out!! making the time resource endless), the published run will be in some mysterious grey area where it shouldn't have been published because the scoring system now allows for infinite scoring with this trick, and it is impossible to obsolete because for any run that gets a higher score, that run could simply be beaten by a trick we can't restrict by Vault rules. But such a run using the trick can't be published as it fails to meet the ideals I'm looking for in my imaginary world, where the trick allows for endless scoring. This is problematic. Because of this sort of definition where corrupting secondary scoring resources must be allowed under Vault rules, I am unable to provide a useful proposal for a rules change to better fit my views. Therefore, I am unable to suggest anything beyond the status quo as it stands. I find this somewhat upsetting, I feel like there should be a change. Everywhere I push, I just can't find the foundation to set my cards. I concede further attempts at changing site rules, should the rule involving what I outlined as a "corruption of secondary scoring resource" be considered an absolute, unchanging standard for Vault. However, having reached a conclusion, even one I dislike, there is some closure I can feel comfortable with. Unable to come up with a proposal that can satisfactorily handle the problematic case I pointed out, I have given myself a type of paradox I can't defeat. Until this demon is slain, I can't see any better way to redefine rules. That still leaves a nice score listing and Tool-Assisted Scoreruns for fun, at least. Perhaps not changing the site, but no one minds a spare topic or page, even if the runs aren't going into the submission queue.
Site Admin, Skilled player (1234)
Joined: 4/17/2010
Posts: 11251
Location: RU
FatRatKnight wrote:
So, stuff happened in this submission.
Indeed.
Warning: When making decisions, I try to collect as much data as possible before actually deciding. I try to abstract away and see the principles behind real world events and people's opinions. I try to generalize them and turn into something clear and reusable. I hate depending on unpredictable and having to make lottery guesses. Any problem can be solved by systems thinking and acting.
Noxxa
They/Them
Expert player, Moderator (4131)
Joined: 8/14/2009
Posts: 4083
Location: The Netherlands
feos wrote:
FatRatKnight wrote:
So, stuff happened in this submission.
Indeed. https://i.imgur.com/k6r2Gdn.png
It links to 5861S. It's supposed to link to 5681S - #5681: mPap's NES Mike Tyson's Punch-Out!! "high score" in 1:15:08.46
http://www.youtube.com/Noxxa <dwangoAC> This is a TAS (...). Not suitable for all audiences. May cause undesirable side-effects. May contain emulator abuse. Emulator may be abusive. This product contains glitches known to the state of California to cause egg defects. <Masterjun> I'm just a guy arranging bits in a sequence which could potentially amuse other people looking at these bits <adelikat> In Oregon Trail, I sacrificed my own family to save time. In Star trek, I killed helpless comrades in escape pods to save time. Here, I kill my allies to save time. I think I need help.
Editor, Skilled player (1157)
Joined: 9/27/2008
Posts: 1084
Well, that was a mistake. Fixed. I no longer wish to change the Vault. However, I can still design a system of rules for fun. As I'm no longer constraining myself to a system present on this site, I won't have to worry about conforming to what is there. With no goal to change this site, I do not have to answer to anyone. So I will have freedom to imagine what I wish and share what I come up with. So, what's the main goal for these rules I'm making? Record keeping for scoring. That 792-hour Desert Bus run just to reach 99 points is a record. However, not all records are practical to encode into a full video. That 792-hour Desert Bus run is going to be a nightmare for encodes. If I want the rules to include accepting this as a record, such rules should not include compulsion for publishing a full proper video. And especially not for infinite movies. A number of games allow for infinite scoring. Many others have limited scoring. There's a few that are apparently intended to have limited scoring, but a trick or glitch breaks its system and allows for infinite scoring instead. Infinite scoring should not consider a programmed upper boundary as a stopping point. Something like Tetris has infinite scoring. Several sorts of Mario games also fall under this category. Limited scoring is defined simply that the scoring system does not allow one to get points endlessly. Again, this should not consider a programmed upper boundary as a stopping point, as this has to do with the resources provided by the game with which to score, not the upper limit to where the score itself can go. Restrictions on tricks or glitches are encouraged to provide a limited scoring environment where such tricks or glitches would switch it over to infinite scoring. Targeted scoring is to pick some interesting spot, such as the programmed upper boundary, as a stopping point for a run. If a case can be given for some spot other than the upper boundary, it should be considered accordingly. The standard spot would be some upper boundary where either the score freezes, wraps around, or starts displaying nonsensically. So, to give a few published examples from this list... Infinite: Frankly, it's infinite. We don't have a lot of those. #3912: adelikat's FDS Super Mario Bros. 2 in 115:17:46:40.00 Limited: [530] N64 Tetrisphere "Time Trial" by Acmlm in 07:00.00 [886] N64 Wetrix "1 minute challenge" by Deign in 00:49.72 [970] NES Track & Field "max score, playaround" by Phil in 12:55.43 [2836] GC Ikaruga "2 players, maximum score" by keylie in 21:29.40 Targeted: [1596] NES Tetris "maximum score" by Acmlm in 03:11.78 [1875] SNES Tetris Attack "maximum score" by zvsp in 01:15.38 [1904] SNES Panel de Pon "maximum score" by zvsp in 00:39.20 [2691] N64 Magical Tetris Challenge "maximum score" by PoochyEXE in 01:57.38 [2759] C64 C64anabalt by dwangoAC in 21:25.22 Whether I missed a few cases isn't particularly important, I wanted to select out a few examples that struck out at me. I haven't fine-tuned any rules or descriptions as of yet to put out there, I'm simply trying to come up with useful definitions and terms to start from.
Active player (324)
Joined: 2/23/2005
Posts: 786
Mothrayas wrote:
This was answered in the Punch-Out submission thread - you're comparing apples and oranges, or to be precise, you're comparing optimization level (which cannot be precisely defined) with completion criteria (which can be precisely defined, and which have to be precisely defined for any movie to be publishable - any movie that does not reach a clearly defined goal or end-point is rejected). Remember that the Vault only allows any% and full completion runs - if a max score run isn't full completion nor can prove itself to be such, it does not count as full completion, nor any%, so it cannot be accepted for the Vault.
I don't think this is an unfair comparison. What if a vault submission aims for 100% completion, but it is later discovered that 101% completion is possible? It implies that the first submission's goal wasn't precisely defined. I seem to remember a "105%" Donkey Kong run recently which begged this question. I think that "full completion" is just as subjective and fluid to define as any other goal -- for instance, Zelda 1 "full completion" doesn't toggle every save flag such as bombing open all the secret doors. The community came to a consensus that collecting all items counts as 100% for the game. Every game is going to involve this kind of consensus at some level. Even "beating the game" can be controversial, such as in the SMB -3 stage ending, which "beats" the game simply by setting a flag (Beaten castle level in world 8 or higher) rather than beating Bowser. This could be considered the same kind of memory corruption we're trying to avoid in score-attack runs.
Vault rules wrote:
Maximum points or score is allowed as a full-completion category, provided that: There is no better way of defining full completion in the game. The maximum score is easily defined and absolute - it must not be possible to gain a higher score, even theoretically. It must be definable without being dependent on precise time, speed, or similar requirements. The maximum score is limited by not being able to gain any more points, not by hitting a score cap or overflow. If it is possible to score points infinitely, score cannot be used to define full completion.
Can you give an example of a published movie on this site which falls under these rules? Personally, I feel that this wording is very poor and does not convey the spirit of what it was intended to mean. If we have no movies which fall under these rules, I'd advocate changing this wording to something more clear, such as "Score attacks are not allowed in the vault"
Mothrayas wrote:
Sorry, but I am going to instantly reject any proposal that would force us to publish a 74-hour pinball run, or a 792-hour long Desert Bus run. This is one of the main reasons we require entertainment voting support for publishing movies with goals like this.
Add a disclaimer: "We reserve the right to reject any movie which is unreasonably long, even if it meets all other requirements for publication, on the basis of reducing stress on our server and our publishers. Such movies generally include those which are over ten hours long and largely consist of no meaningful content."
Noxxa
They/Them
Expert player, Moderator (4131)
Joined: 8/14/2009
Posts: 4083
Location: The Netherlands
CtrlAltDestroy wrote:
Mothrayas wrote:
This was answered in the Punch-Out submission thread - you're comparing apples and oranges, or to be precise, you're comparing optimization level (which cannot be precisely defined) with completion criteria (which can be precisely defined, and which have to be precisely defined for any movie to be publishable - any movie that does not reach a clearly defined goal or end-point is rejected). Remember that the Vault only allows any% and full completion runs - if a max score run isn't full completion nor can prove itself to be such, it does not count as full completion, nor any%, so it cannot be accepted for the Vault.
I don't think this is an unfair comparison. What if a vault submission aims for 100% completion, but it is later discovered that 101% completion is possible? It implies that the first submission's goal wasn't precisely defined. I seem to remember a "105%" Donkey Kong run recently which begged this question. I think that "full completion" is just as subjective and fluid to define as any other goal -- for instance, Zelda 1 "full completion" doesn't toggle every save flag such as bombing open all the secret doors. The community came to a consensus that collecting all items counts as 100% for the game. Every game is going to involve this kind of consensus at some level. Even "beating the game" can be controversial, such as in the SMB -3 stage ending, which "beats" the game simply by setting a flag (Beaten castle level in world 8 or higher) rather than beating Bowser. This could be considered the same kind of memory corruption we're trying to avoid in score-attack runs.
This whole argument rests on the idea that because categories are humanly defined, they're subject to change. In practice though, that doesn't really happen - at least not to any relevant capacity. Name a single publication on this site for a Vaultable category that ended up having its goalposts moved for what constitutes any% completion or full completion. Also, games like the Donkey Kong Country series with completion percentages like 101%, 102% or 105% are still following clearly defined in-game maximums, the only unusual part being they don't add up to "100". But that's still a clear and fixed game design choice, not an evolving definition.
CtrlAltDestroy wrote:
Vault rules wrote:
Maximum points or score is allowed as a full-completion category, provided that: There is no better way of defining full completion in the game. The maximum score is easily defined and absolute - it must not be possible to gain a higher score, even theoretically. It must be definable without being dependent on precise time, speed, or similar requirements. The maximum score is limited by not being able to gain any more points, not by hitting a score cap or overflow. If it is possible to score points infinitely, score cannot be used to define full completion.
Can you give an example of a published movie on this site which falls under these rules? Personally, I feel that this wording is very poor and does not convey the spirit of what it was intended to mean. If we have no movies which fall under these rules, I'd advocate changing this wording to something more clear, such as "Score attacks are not allowed in the vault"
[3245] A2600 Pitfall II: Lost Caverns "maximum score" by Alyosha in 09:30.32 (it was accepted to Moons before this rule, but qualifies) [3386] PSX TOCA Touring Car Championship "max points" by Noxxa in 10:37:18.72 But regardless of examples, I don't see how these definitions are unclear or poorly worded. You may not like that it excludes score attacks like the recent Punch-Out one, but that's because those, as discussed, fail to qualify for full completion, and the Vault rules have always made clear that any% and full completion are the only allowed categories there.
CtrlAltDestroy wrote:
Mothrayas wrote:
Sorry, but I am going to instantly reject any proposal that would force us to publish a 74-hour pinball run, or a 792-hour long Desert Bus run. This is one of the main reasons we require entertainment voting support for publishing movies with goals like this.
Add a disclaimer: "We reserve the right to reject any movie which is unreasonably long, even if it meets all other requirements for publication, on the basis of reducing stress on our server and our publishers. Such movies generally include those which are over ten hours long and largely consist of no meaningful content."
This violates our Vault rule directive of being minimally subjective, on top of being practically impossible to define. Where do you put the limit on what's acceptable or not? Also, I don't think "we reserve the right to randomly reject movies from the Vault that meet all requirements for publication" is a very favorable rule.
http://www.youtube.com/Noxxa <dwangoAC> This is a TAS (...). Not suitable for all audiences. May cause undesirable side-effects. May contain emulator abuse. Emulator may be abusive. This product contains glitches known to the state of California to cause egg defects. <Masterjun> I'm just a guy arranging bits in a sequence which could potentially amuse other people looking at these bits <adelikat> In Oregon Trail, I sacrificed my own family to save time. In Star trek, I killed helpless comrades in escape pods to save time. Here, I kill my allies to save time. I think I need help.
Active player (324)
Joined: 2/23/2005
Posts: 786
Mothrayas wrote:
This whole argument rests on the idea that because categories are humanly defined, they're subject to change. In practice though, that doesn't really happen - at least not to any relevant capacity. Name a single publication on this site for a Vaultable category that ended up having its goalposts moved for what constitutes any% completion or full completion.
The stage -3 ending SMB run. It was first submitted as an any% run, then there was some drama over whether it should have obsoleted the warps (any%) run, and in the end it was put in its own category. Note that you said "vaultable category", not vaulted movie, and any% is a vaultable category. Besides, I don't think that "any%" and "100%" should change definitions depending on whether the run is being considered for vault or not.
Mothrayas wrote:
This violates our Vault rule directive of being minimally subjective, on top of being practically impossible to define. Where do you put the limit on what's acceptable or not? Also, I don't think "we reserve the right to randomly reject movies from the Vault that meet all requirements for publication" is a very favorable rule.
Sometimes, there is little difference between a rule, and a principle on which the rule is based. Let me ask you this: what, exactly, makes Pinball and Desert Bus so bad? What constitutes an undesirable game, worthy of molding the vault rules specifically to keep them out? I'm not saying I want to see these movies published; I agree that publishing these movies would be a ridiculous violation of common sense. I'm saying that every objective rule has a subjective sentiment behind it, and it's better to simply describe the sentiment, rather than building rules to enforce and obfuscate it. This is transparency. Is Desert Bus unpublishable because it is too long? Then why not state that outright? I like the vault and I'm a huge proponent for its existence, but edge cases with the vault bring out some of the worst of bureaucracy of this site. It's so much simpler when we can just collectively say something is fun to watch or not.
Noxxa
They/Them
Expert player, Moderator (4131)
Joined: 8/14/2009
Posts: 4083
Location: The Netherlands
CtrlAltDestroy wrote:
The stage -3 ending SMB run. It was first submitted as an any% run, then there was some drama over whether it should have obsoleted the warps (any%) run, and in the end it was put in its own category.
Well, in the end, the definition did not change, as -3 ending had to be published to a separate branch instead. The SMB ending definition always has been to save Princess Toadstool, and that has never changed. The definition was tested and clarified a bit, but not changed.
CtrlAltDestroy wrote:
Note that you said "vaultable category", not vaulted movie, and any% is a vaultable category. Besides, I don't think that "any%" and "100%" should change definitions depending on whether the run is being considered for vault or not.
I agree, but my point on specifying vaultable categories was that those categories are supposed to be unchanging. Non-vaultable categories may have their conditions change, like low% taking a lower percentage, and the like.
CtrlAltDestroy wrote:
Let me ask you this: what, exactly, makes Pinball and Desert Bus so bad? What constitutes an undesirable game, worthy of molding the vault rules specifically to keep them out? I'm not saying I want to see these movies published; I agree that publishing these movies would be a ridiculous violation of common sense. I'm saying that every objective rule has a subjective sentiment behind it, and it's better to simply describe the sentiment, rather than building rules to enforce and obfuscate it. This is transparency. Is Desert Bus unpublishable because it is too long? Then why not state that outright? I like the vault and I'm a huge proponent for its existence, but edge cases with the vault bring out some of the worst of bureaucracy of this site.
What makes Virtual Pinball max score and Desert Bus bad are repetitiveness to enormous ends, and the fact that practically nobody is seriously interested in watching playthroughs of their nature in their entirety. But those are still subjective qualifiers in the end, and creating a speedrun archive without subjective qualifiers was the whole point of the Vault. If we introduce qualifiers of things like "repetitiveness" or "watchability", we ruin the purpose of the Vault as an speedrun archive. It just becomes yet another tier of subjective content, where content would need to be rejected and accepted on people's whims rather than on a sizable but consistent set of rules. So, instead of putting up such qualifiers, score capping or non-full-completion score maxing are just considered not full completion and considered outside the scope of the Vault.
CtrlAltDestroy wrote:
It's so much simpler when we can just collectively say something is fun to watch or not.
I would say I agree, but implementing that would mean removing the Vault entirely, and I imagine that's not what you want either. That also would still mean Punch-Out max score gets rejected for the same reason it is now - poor audience response. That's a common pattern with these sorts of discussions, by the way - people often want to expand the Vault rules because a recently rejected or submitted Vault-ineligible is "fun" or "interesting", even though if that were really the case for the broader audience, then they would not need to qualify for the Vault anyway. Vault rules always have only applied to runs that fail to get the audience's support.
http://www.youtube.com/Noxxa <dwangoAC> This is a TAS (...). Not suitable for all audiences. May cause undesirable side-effects. May contain emulator abuse. Emulator may be abusive. This product contains glitches known to the state of California to cause egg defects. <Masterjun> I'm just a guy arranging bits in a sequence which could potentially amuse other people looking at these bits <adelikat> In Oregon Trail, I sacrificed my own family to save time. In Star trek, I killed helpless comrades in escape pods to save time. Here, I kill my allies to save time. I think I need help.
Post subject: Beetle Mania added (targeted score)
Editor, Skilled player (1157)
Joined: 9/27/2008
Posts: 1084
If a scoring action is repeatable, then it should not be publishable in Vault. I'd argue that Desert Bus has a repeatable action for a fraction of its point that can be used to get an arbitrarily high value of points (for a ridiculously long time per point, but said time can be extended infinitely). Pinball happens to have an infinite loop that works with zero input once it is set up, but once again, the action can be repeated for an infinity of times, to get any arbitrarily high score value you wish. Personally, I would not allow such infinity to be publishable. Even if the score caps out, it doesn't restrict the scoring action itself, so there's little of interest in just repeating that action over and over. And targeting some convenient score value, such as the 99 cap of Desert Bus, is not a valid reason for a Vault safety net. Basically repeatable actions have no measurement of completion. So I won't mind keeping such runs out. Finite scoring actions is the main target for what I'm trying to achieve. Infinite, unbounded: #3912: adelikat's FDS Super Mario Bros. 2 in 115:17:46:40.00 #4168: Aqfaq's Genesis Virtual Pinball "maximum score" in 00:05.46 These runs end up repeating some sequence of events without bound. Even when the score hits the maximum, it continues to repeat the scoring events. These are examples of runs that don't try to limit scoring in any fashion, although what their goals may be isn't strictly for fastest score/time in such a loop. Still, the game's score cap is not to be considered. Infinite, targeted score: [1596] NES Tetris "maximum score" by Acmlm in 03:11.78 [1875] SNES Tetris Attack "maximum score" by zvsp in 01:15.38 [1904] SNES Panel de Pon "maximum score" by zvsp in 00:39.20 [2691] N64 Magical Tetris Challenge "maximum score" by PoochyEXE in 01:57.38 [2759] C64 C64anabalt by dwangoAC in 21:25.22 #2211: alden's SegaCD Desert Bus in 41:17:15:05.68 #3523: Bobo_the_King's SNES Super Mario RPG: Legend of the Seven Stars "Beetle Mania" in 45:35.38 Thanks, ThunderAxe31 All these runs allow for infinite scoring, under the definition that you can repeat the scoring actions endlessly as opposed to respecting whatever programmed limit the game has. However, a particular point total is selected for one reason or another, and the results are published as above. On a side note, I disagree (weakly) with C64anabalt's acceptance into Vault. The fact the score display glitches at a certain point isn't really any more special an endpoint as Desert Bus' 99 point limit. If it is a commonly accepted endpoint in the community, then we have a goal to aim for in a game where we otherwise wouldn't have a potential Vault goal. I'm well aware I want more scoring runs published, yet by the rules I'm trying to design, this one wouldn't fit nicely. Limited, firm maximum by single or targeted cycle count, infinite otherwise: [970] NES Track & Field "max score, playaround" by Phil in 12:55.43 [2189] NES Gumshoe "maximum score" by Highness in 14:43.98 There are various arcade games that repeat a sequence of stages endlessly. At some point, the changes in gameplay or difficulty stops, and effectively the same sequence as the last cycle starts replaying. As with time-based goals that pick this sort of endpoint to stop at, we can also potentially calculate an absolute maximum possible score for such an end-point, and use that as a 100% criteria. Potentially. ... Okay, I'm proposing a possible interpretation of the rules to leverage my vision, but it's a point of view to consider. In any case, so long as the stages themselves don't allow for infinite scoring without clearing them, such as death abuse to go back to a checkpoint, or infinite time and scoring resources are provided within a stage, then limiting the stage count to a convenient point where gameplay or difficulty no longer changes is a sensible way to give a useful limit. Limited, firm maximum: [3245] A2600 Pitfall II: Lost Caverns "maximum score" by Alyosha in 09:30.32 [3386] PSX TOCA Touring Car Championship "max points" by Noxxa in 10:37:18.72 [3461] Vboy Nester's Funky Bowling by Spikestuff, MESHUGGAH in 01:02.70 These games have a readily determined maximum to calculate or discover. There is a decisive limit based on the resources the game allows, and the goal is to use those resources to their limit, then clear the game as soon as possible. Even if there is a programmed limit that is beneath the total of scoring resources (and this recent Quest for Glory post gives a good example), simply selecting that limit means we allow ourselves to pick and choose which resources to arbitrarily ignore. Then all of a sudden it becomes really murky what it means to max scoring. Personally, I'd say ignore the programmed limit, and assume all scoring resources will add up to a higher total. Anything less, and we can't really say we've completed everything in the game. I will have to admit the possibility of mutually exclusive scoring resources, in that using one denies the other. Whatever this implies, I'm not willing to go into details there as of yet. Limited, soft maximum: [530] N64 Tetrisphere "Time Trial" by Acmlm in 07:00.00 [886] N64 Wetrix "1 minute challenge" by Deign in 00:49.72 [2246] Genesis Olympic Gold: Barcelona '92 "playaround" by Toothache in 10:32.23 [2621] GC Resident Evil 4 "The Mercenaries: Castle" by Ubercapitalist in 13:44.33 [2836] GC Ikaruga "2 players, maximum score" by keylie in 21:29.40 #5681: mPap's NES Mike Tyson's Punch-Out!! "high score" in 1:15:08.46 Some of these are actually rather fancy to watch. Personally, I want to see more runs like these. Part of a fear I perceive is that there is no Vault "safety net" to guarantee a publication of a high scoring run, and the site is primarily built around speed records. Part of my goals is to help promote such runs, where feasible. Included in those goals is to allow restriction of glitches or tricks that would turn a limited scoring environment into an infinite scoring one. Olympic Gold was originally published into Vault, but then it was determined this was a mistake. I am curious as to the original line of thinking that got it into Vault to begin with. Limited, low score desired: [1633] SNES Kirby's Dream Course "maximum score" by snc76976 in 1:09:13.63 [2718] Arcade Neo Turf Masters "maximum score" by mamuuuut in 09:17.00 [3150] NES Lee Trevino's Fighting Golf by Acmlm in 08:26.00 [3194] NES Mini Putt by TaoTao in 06:54.17 #2908: HappyLee's NES Super Mario Bros. "lowest score" in 14:32.80 Okay, so I listed a bunch of golf games. I suspect some of them are in Vault not because of an optimal known score, but rather that a low score happens to be conducive to fastest time. The Kirby one there isn't necessarily faster with a low score in this case, as presumably some of the shots could be ended earlier rather than extending it just to keep the shot count low, but it isn't Vault so it's published anyway. The point is that not all games use a high value for making you look awesome. For this post, I haven't tracked down the Sokoban TASes that aims for fewest moves, but those decisively lend themselves to fastest times as well. It isn't necessary that a low score means a faster time, and I did track down a submission that does exactly that to avoid points from time countdown. Limited, other/undecided category: [814] SNES Star Fox "maximum score" by YtterbiJum in 19:47.37 Haven't looked over this one. It has emulation issues, back in those old days? I get the impression the highest score possible isn't infinite, at least. [2310] PSX Bishi Bashi Special "Marathon Challenge, maximum score" by Spikestuff in 33:53.43 It is clear this one is limited. What's not clear is the fact each stage has its own scoring criteria, and another for overall ranking. The overall ranking has a clear limit of 10 per stage, giving a firm scoring limit there, but the individual stages are softer in that you try to get the most you can within a time limit for some of them. [2852] PSX Bishi Bashi Special 3: Step Champ "Track & Field, best performance" by Spikestuff in 02:41.27 Apologies. The first Bishi Bashi Special TAS has me scared now, so I have yet to put in effort here. In any case, the high score here is the farthest the long jump or javelin throw can get, which has a clear limit, but an unclear maximum. I'm going to leave this here as a limited, not categorized for now. [3023] PSX Bishi Bashi Special "Time Trial, maximum score" by Spikestuff in 05:37.75 Another Bishi Bashi Special. The goal is to get the fastest in-game times, at sacrifice to the overall input file length for the necessary manipulation. I don't know how to really categorize what seems like a speed oriented goal, except for in-game times. Kind of reminds me of the various Sonic runs where the bonus countdown is tolerated for the sake of faster in-stage times. [3508] PSX Metal Gear Solid: VR Missions by theenglishman in 1:35:46.46 There's a variety of max scores in here. The in-game times are also preferred over input file length. In some cases, the in-game time would be negative if the time bonuses for ammo/sneaking wasn't capped, making the 0"00 times more of a targeted goal due to a programmed limit. The run is basically a variety pack of scores. Perhaps like the Track & Field runs. Not categorized: [3513] NES Crash 'n' the Boys: Street Challenge "best ending" by Inzult in 08:10.86 I'm pretty sure this run aims for getting 1st place in every event. It also tries to do so as quickly as possible in many cases. I'm not sure what counts as better scoring in most of the individual events. There is one event where the TAS crosses all the buildings nice and fast, and I'm not sure if failing that quickly, even with the 100% restriction of all 1st places, can make for a faster run, but in any case, it does cross those buildings fast. Score not apparently a primary goal: These runs get a nice score, but time is put ahead of score. While other goals may be put ahead of time, a high or targeted score is not one of them, and time is aimed for immediately after the completion of this specific goal. [1290] NES Dr. Mario by CtrlAltDestroy in 01:12.83 The score is crazy high. The goal is to complete one level of Dr. Mario, the highest allowed difficulty from the starting settings, and to complete this level as quickly as possible. The fact it has an amazing score as a side-effect of this time goal, even though this one-level score is not likely to be beaten any time soon, doesn't mean score should be used as a measurement to obsolete this movie. If a higher score is achieved at a longer time, then this would be a record that can stand side-by-side with the existing fastest run, although not necessarily published together by TASVideos standard. [1634] DOS Epic Pinball "Super Android" by dwangoAC in 05:07.53 The goal here was to max out various counters as quickly as possible. While the process of getting them maxed awards a lot of score, it doesn't actually end the run itself, and higher scores are readily possible. The goal is a specific completion point that in itself isn't targeting a particular score, and therefore score has a lesser priority to time where it is to get to this goal as quickly as possible. [2091] GB Super Mario Land by MUGG in 12:08.75 The run aims for fastest time. Without impacting time, there is a secondary goal to collect every coin and defeat every enemy it can to maximize score, but this is beneath fastest completion. Unless the time remaining bonus is worth more than what can be collected in the meantime, this isn't a top scoring run. [2113] SNES Super Putty "all stages" by Dooty in 16:53.17 The aim is to clear all stages. It feels like an appropriate 100% goal. Based on the submission text, time is prioritized after the goal to see through every stage. Score does not appear to be a significant factor in routing. [3146] GBA Sonic Advance 2 by Dashjump in 17:41.10 The standard in going fast. Wait, is maximum score an appropriate tag? I'm confused... Please clarify the line of thinking here. In-game time is a standard among Sonic runs, particularly as waiting around intentionally at the goal just to avoid a longer score countdown is uncool, and plagued earlier Sonic games (this particular title lets you skip the countdown time). I'm not entirely sure how I feel about in-game time as scoring, but this game has an actual scoring mechanic in it, and if score was considered, it certainly wasn't considered ahead of time. [3509] Arcade Jungle Hunt by £e Nécroyeur in 08:43.33 The movie runs through enough cycles to reach max difficulty of the game. It doesn't help the description indicates going through the game in record time, meaning time is likely considered ahead of score. It was accepted into Vault, and Vault can only use time as comparison point for obsolete runs, or possibly a correction of a prior goal that missed a point of completion for 100%-like goals.
CtrlAltDestroy wrote:
I think that "full completion" is just as subjective and fluid to define as any other goal -- for instance, Zelda 1 "full completion" doesn't toggle every save flag such as bombing open all the secret doors. The community came to a consensus that collecting all items counts as 100% for the game. Every game is going to involve this kind of consensus at some level.
Tangentially, my belief on what counts toward full completion is any event that is tracked by the game in its most persistent manner available, done with non-repeating events. Defeating some random low-end enemy is generally not tracked by the game, and almost always repeatable even if it is tracked by a counter. Using a bomb is tracked, by the fact you now have one fewer to use and is saved across sessions, but this is by way of a repeating event. Marking a room on the map as explored is tracked by the game, and is a non-repeating event. Even if the event in question is not a "major" event anyone cares about, like opening this small chest and only finding a trivial amount of money, so long as this chest can't be re-opened, why should we exclude it for any reason other than "it's not important"? Oh, and don't try to twist things by saying "everything is repeatable by deleting the save." The whole argument is based on the lifetime of the most persistent medium by which the game tracks things, and deleting a save is decidedly the end of such a lifetime. I am trying to make an argument that also fits within the definition of games that do not have a persistent save file, ones that drop all tracking once the power is turned off. There are also games that do their tracking based on a password system as well, and that can be counted just fine as a persistent medium. If the run is entertaining enough, and the community is enough in agreement on the current run being "complete enough," I do not wish to impose on the extreme of getting every tiny side event done just because it's non-repeating and tracked by the save file. It's a good run already, and enough percentage of everyone is fine with what was done. I have my own beliefs, and do not wish to spoil a collective belief using one that only one person has. On a side note, I did present some difficult questions in The Guardian Legend. Start your read here and work your way down a few posts. In any case, I see a Vault valid 100% goal as being one where you complete a game as far as it will go. I wish to tie the max scoring definitions to be scoring as high as the game will allow, on whatever strategy you use to get there. This is as "100%" as such a scoring run will go. A repeating action that gives infinite score means we lose meaning of 100% on score. Much like Super Metriod's Time/Space Beam allowing essentially infinite% collection means we lose what 100% really means. ... I'm hoping my memory is on target with this glitch.
DrD2k9
He/Him
Editor, Expert player, Judge (2036)
Joined: 8/21/2016
Posts: 1009
Location: US
For my own understanding/enlightenment, I'm going to throw a wrench into the engine and see if it causes problems...
Mothrayas wrote:
What makes Virtual Pinball max score and Desert Bus bad are repetitiveness to enormous ends, and the fact that practically nobody is seriously interested in watching playthroughs of their nature in their entirety. But those are still subjective qualifiers in the end, and creating a speedrun archive without subjective qualifiers was the whole point of the Vault.
^emphasis mine. I recognize that no one (or very very few people) will be even remotely interested in watching one of these runs in its entirety. However, for Desert Bus, using score as a 'point of completion' was not the reason for rejection. Similarly, length of the run was also not the stated reason for rejection. It was rejected from Vault publication because it wasn't distinguishable from a human speedrun. Isn't determination of whether or not a run appears to be different than a human speedrun a subjective qualifier? If we use this subjective reason for rejecting the Desert Bus submission, perhaps all the other Vault-tier publications (as well as any submissions rejected for this same reason) should be reassessed, since "...creating a speedrun archive without subjective qualifiers was the whole point of the Vault" As an example (and at risk of having one of my own publications un-accepted), NES Castelian is perhaps technically distinguishable from a human run (menu management). But, regarding game-play itself, the Castelian TAS is hardly distinguishable from what a human speed-run would look like if the human player knew the enemy spawn positions/movement speeds and the path to take to the goal. These are things many human speedrunners do know about the games they run. Should this Castelian run then have been rejected for the same stated reason as Desert Bus? If so, I am in a position to potentially lose a published movie. And maybe others are too, as there may be other current vault publications whose game-play isn't necessarily distinguishable from human speedruns other than the technical management of arbitrary things like menus. If Castelian was acceptable without distinct evidence indicating it wasn't a human speedrun (other than arbitrary stuff like menu management), then acceptance/rejection of Desert Bus unfortunately needs reassessed as well as any other submissions rejected for the same reason. Please understand, I'm NOT trying to advocate for Desert Bus to be accepted. In fact, the rule for judges regarding triviality may be an even better rejection reason for both Desert Bus and Virtual Pinball.
The run must not be seen as trivial. If a run consists of doing a trivial strategy, made only nontrivial by having to do it over the course of several rounds, it will still be judged as trivial. For example, bowling games where the player gets a strike with ten pins every time.
Then again, one could argue that 'triviality' could be considered a subjective assessment also. I guess what I question about something being 'distinguishable from a human speedrun' centers on whether or not arbitrary technical actions (pre-game menu management, frame perfect clearing of level intro-screens, frame perfect skipping of cutscenes which are intended to be skipable, selection of action icons in an adventure game, etc.) are enough to deem a TAS run distinct from a human run when all other aspects of game-play aren't distinguishable; especially when considering that many human runs don't even consider pre-game menu management as part of the timed events, and options are usually selected before time starts--which is usually upon pressing start or sometimes at the first control opportunity after an initial cutscene when starting the game. Human timing also often ends on certain in-game events (i.e. boss death animation, credits roll, "the end"), not on button presses that initiate that event, so this could also be considered a technical distinction not a game-play distinction. As a side note: A2600 Dragster is a game currently published in the Vault tier even though the judgement was for Moons. This should probably be fixed. Anyway, Dragster is an example where the in-game time is considered over total game time--this is fine for vault purposes--but the best in-game time of 5.57 achieved by the TAS has been matched in real-time by a human. Should this disqualify the TAS from vault tier publication (had it not been accepted as a moons quality)? Please don't make the argument that it wasn't accepted as a Vault run to disregard this hypothetical question. I guess my perspective is that the fastest/best TAS is exactly that, the fastest/best TAS. It should be acceptable as long as it isn't sub-optimal or trivial, regardless of whether or not it can be distinguished from a human speedrun. Example: A2600 Adventure is a mere .4 seconds faster than the current real-time record. The only distinguishing characteristics of the TAS is the wobbly movement. But if you showed both videos to someone who didn't know one was done by TAS, they wouldn't necessarily assume it was. Or they would assume both were. A TAS shouldn't necessarily be rejected just because a human is able to match the game-play aspects of the run once technical arbitrary details are disregarded (i.e. Dragster, Adventure). If that were the case, then TAS runs that only appear to showcase faster speed than a human due to non-gameplay technical aspects should be rejected and/or removed if already published (i.e. Castelian). These thoughts aren't meant to be restricting what runs get accepted, nor are they meant to have current publications removed. If anything, it's advocating for a slightly broader definition of runs deemed Vault worthy by softening/eliminating the 'distinguishable from human speed run' requirement so that more games can be showcased. Side note #2: If this post would be better suited to a different area of the forum, please let me know and I'll copy it to that area for discussion.
Post subject: Limited scoring vs. Endless scoring
Editor, Skilled player (1157)
Joined: 9/27/2008
Posts: 1084
Aside from having a rejected TAS based on triviality, I don't have much to add to your post. That's for our site staff to determine, I believe. I'm an Editor, so technically, I'm part of that staff? Just use common sense to get my meaning, alright? I keep talking about infinite scoring and limited scoring. It's time I define more precisely what these are. Score is... Score In general, there is a clear delineation of what is score and what isn't score, as depicted by the game. Destroying an enemy in some arcade style game is worth score. Destroying an enemy in Dragon Quest is worth experience. If it is pretty clear the intent of the value is score, then it is score. ... I'm now expecting some kind of debate over Super Paper Mario's "score" value. To which I have no good answer for as of yet. If there are minigames or similar such events within a game, each with their individual scoring, then they are to be considered separately. Ignore score limit The programmed score limit, be it an upper limit that freezes the score once it gets there, or is unchecked and causes some kind of glitching once it is there, should not be considered as a theoretical stopping point. After all, such a limit has no impact on whether a player can continue to use scoring actions. If such a limit does not exist, can the player continue indefinitely to get a higher score? If gameplay stops after hitting a score threshold (example), it should refer to Vault rules for publication. If it can be determined as an appropriate ending for Vault, particularly in cases where there is no other ending criteria, it can be published according to Vault rules. Encourage limited scoring environments Many games are inherently endless scoring. Many others are limited, and do not allow repeating scoring actions without bound. There are some that provide limited scoring options, but due to a few tricks or glitches, it turns into an endless scoring game. To keep the scoring mechanic in such games interesting, any restrictions that prohibit such actions, in order to create a limited scoring environment, are allowed. This rule must be allowed to deal with the possibility of this: * A limited high-score run is published * A trick is discovered that removes a resource restriction (like time), enabling endless scoring * Without restrictions, . o A run that doesn't use the trick can't obsolete the current run since a higher score with the trick is possible . o A run that uses the trick can't obsolete the current run with a higher score because it is in an endless scoring environment . o Essentially, the current run is in violation because of a then-undiscovered trick, and is now impossible to obsolete To deal with this "demon," restrictions of certain glitches, tricks, or actions must be allowed so that a later discovery does not break the format set out here. Endless Scoring I called this "infinite scoring" before. An endless scoring situation is one where the game allows the player to score as much as they like, given time. The game never restricts the players capability to do actions that would result in more score. The game's score cap is not to be considered. Just ask yourself, if you are approaching this limit, and you cheat with a memory edit to reduce your score, would you be able to keep increasing score in this loop? If this is "yes", we're in an endless scoring environment. Runs of this sort are not compelled for publication. Only audience reception should be considered. Guidelines on what constitutes an endless scoring environment: * An endlessly spawning source of scoring enemies or other resources are available * You can repeat a scoring action without bound * Able to revisit stages, or at least parts of one by a checkpoint system * Time is not used to force progress toward an ending, or can be extended without bound Targeted Scoring A subset of Endless Scoring. As such, it should require good audience reception for publication. Most games with Endless Scoring are likely to be submitted in this sort of category, as otherwise you have infinitely long runs. Technically, it can also be done in a Limited Scoring environment, by simply selecting a score other than the maximum achievable with the resources provided. The game could have a ranking system, and awards maximum rank for a particular threshold, making the margin above unnecessary to reach this rank. Such runs will not be compelled for publication under these scoring rules, as it falls in the category of Targeted Scoring, but might be publishable anyway depending on Vault or popularity. Guidelines to expect from targeted scoring: * Some programmed score maximum * Enough score to reach top ranking * A convenient, although arbitrary, stopping point of score (like 10,000,000) Limited Scoring This type of environment puts a limit on what scoring is available to the player. At some point, the player can gain no further reward in trying these actions. Again, the upper programmed limit should not be considered, only the limit of actions the player has access to. Guidelines for limited scoring: * Enemies or other resources used for scoring are strictly finite * Any object used for scoring does not return to a state where a prior used scoring action is possible * Does not allow revisiting old areas for the life of gameplay * Game uses a time limit, with no boundless ways to extend it There are also rules of endless arcade-style games elsewhere. If there is a point where the difficulty maxes out, and the gameplay is effectively a loop of stages at maximum difficulty at that point, it can be considered an appropriate endpoint to do exactly one loop at that maximum difficulty. This is true for time-based goals of Vault. As it should be true for score-based goals of these rules as well, so long as it isn't possible to endlessly score within a stage. Turning endless to limited Due to the encouragement of these very rules in promoting TASes of limited scoring games, Not all games necessarily have a type of gameplay that can lend itself to restrictions to become a limited scoring environment. Guidelines that allow limited environments: * If a trick or glitch is discovered that makes scoring endless, ban it. * "Never look back." If a run can turn finite by restricting revisits to earlier areas, such as by death in a checkpoint system or by warps, then restrict that. There are still probably a few holes to plug even here. Particularly since I mention guidelines rather than rules. We have to start somewhere, and I'm hoping to get things together. One issue of note is what to do when there are different restrictions that separately allow limited scoring environments. My initial thinking was prefer the one that ends in higher score, but that may lead to some ridiculous definitions. A second thought is to prefer the simpler or more intuitive restriction set. No deaths allowed is a simpler rule than "no more than 2 deaths per stage." If an action is to be restricted, it has to be a complete restriction. "No 1-ups from jumping on the same enemy" allows for jumping on the same enemy a finite number of times, and is therefore not a complete restriction on a particular action. "No jumping on enemies" is a complete restriction on such an action. This will take further thought, of course.
Noxxa
They/Them
Expert player, Moderator (4131)
Joined: 8/14/2009
Posts: 4083
Location: The Netherlands
DrD2k9 wrote:
Desert Bus
It was rejected for being trivial, in the sense of that it is impossible to have a different outcome in terms of being a speed record. The game runs on a fixed time, so there's not much to speedrun, or any way for a TAS to get a faster time than a speedrun or regular playthrough. That's what that rule means. By the way, no movie is going to be unpublished. We've never done that, nor do we have any intent to start doing it.
DrD2k9 wrote:
As a side note: A2600 Dragster is a game currently published in the Vault tier even though the judgement was for Moons. This should probably be fixed.
It was moved to the Vault post-publication after ratings turned out to be poor. In any case, tier judgments in a submission have no long-term say on what tier a movie is in - publications can be moved across tiers at all times whenever deemed appropriate. In fact, there's a topic for that.
http://www.youtube.com/Noxxa <dwangoAC> This is a TAS (...). Not suitable for all audiences. May cause undesirable side-effects. May contain emulator abuse. Emulator may be abusive. This product contains glitches known to the state of California to cause egg defects. <Masterjun> I'm just a guy arranging bits in a sequence which could potentially amuse other people looking at these bits <adelikat> In Oregon Trail, I sacrificed my own family to save time. In Star trek, I killed helpless comrades in escape pods to save time. Here, I kill my allies to save time. I think I need help.
Post subject: This post attempts to define exactly what score is.
Editor, Skilled player (1157)
Joined: 9/27/2008
Posts: 1084
I want to minimize interpretation. Vault-style is to minimize subjectivity. I still have definitions to build, so this is to be considered a WIP. Have I defined well enough what score is? Is the length of text sufficiently short? Does it sufficiently follow some intent I've apparently been pushing for? Any gaps in explanation? (SMB low score probably exposes one) Defining what score is In general, use common sense. Games often clearly indicate a number used for scoring, and this is to be used for comparisons to determine whether to obsolete a prior run. The score number may include a ranking, does not necessarily have to benefit the player, and may have a table of high scores in the game itself this number goes to. Even if common sense fails, and we somehow determine the Experience stat in Dragon Warrior is to be used as a measurement of score, it still has to follow the rules of Limited Scoring. If the goal does have a finite quantity to collect, consider as well whether it can fit the 100% definition of Vault, and if so, it's a Vault TAS, not a scoring TAS. In some cases, a low score is desired. In general, low in-game times and low move counts also lend themselves toward low realtime, but whether because of long bonus countdowns (This one avoids that), roundabout actions needed to keep the value low, or require extra manipulation at places that aren't scored, they do not always have the shortest input files. Also note that games with in-game times can also give time bonuses to this scoring that usually stop at 0.00 seconds and go no further negative. This cap is to be ignored like any other cap, and the time score measured accordingly, where feasible. In rare cases, there is encouragement for both a high score or a low score. Super Monkey Ball gives unique rankings in its credits game for rather positive scores as well as particularly negative scores. It takes perceived skill either way you choose to get as far from zero as you can, so either goal is viable. Ranking and score Imagine you just got S++ after a stage. That's your rank. Rankings are outside the scope of these rules, and are not to be considered for scoring. Ranks are a step function that takes in your score and spits out a result, usually a letter grade or some words of how awesome you are. In a particular example, Bishi Bashi Special has a scoring for each individual stage, then awards up to 10 points for a total score across all stages. These 10 points are based on how well you did on the stage, as a step function to the stage's actual score, and therefore are to be considered a rank. ((Yes! I figured out how to categorize that TAS! Satisfaction.)) If the goal is nothing more than to achieve the highest rank, it is, at best, a Targeted Scoring run, and therefore should not be published under scoring rules, requiring other criteria such as the Vault or popularity for publication. Obsolete criteria Under construction... Score is to be used as top priority. Time is the secondary measurement by which a run can obsolete another. A run that takes 2 more minutes but scores another 10 points is a superior movie. Another TAS that gets exactly the same score, but does it 30 seconds faster is also a superior run. For games with multiple scoring criteria, like minigames with a variety of scoring, (under construction)
DrD2k9
He/Him
Editor, Expert player, Judge (2036)
Joined: 8/21/2016
Posts: 1009
Location: US
Mothrayas wrote:
It was moved to the Vault post-publication after ratings turned out to be poor.
That mostly makes sense.
Mothrayas wrote:
In fact, there's a topic for that.
I'll be the first to admit that I don't fully explore the forums (even if I probably should to a greater degree) unless there's a particular topic I'm seeking. So I did (mostly) read through the linked topic and I want to know if I understand correctly: When it comes to tier-shifting a movie, the site considers post-publication ratings as more important than pre-publication voting. Correct?
Mothrayas wrote:
...publications can be moved across tiers at all times whenever deemed appropriate.
When this kind of tier shifting occurs, should a note be made in the submission notes to prevent others from being confused the way I was?
DrD2k9
He/Him
Editor, Expert player, Judge (2036)
Joined: 8/21/2016
Posts: 1009
Location: US
A comment on Score Based TAS after reading FatRatKnight's categorizations and preliminary rule guidelines It seems to me that determining in which scoring category a particular game fits, will be the most the most difficult aspect of implementing any score focused system.
Player (142)
Joined: 7/16/2009
Posts: 686
I'd just like to point out that I don't like the argument about highest/maximum score not being precisely defined for many games. Ignoring games where infinite scoring is possible, "maximum score" is a perfectly precise definition. Whether we know what that score is or not doesn't really matter. "Fastest speed" is subject to the exact same issues: there's always the possibility that better is possible, but unless such is known now, we still accept the movies. Given BizHawk's option for infinite movies, even the problem of infinite scoring could be worked around: an infinitely scoring movie beats a finite one, and infinitely scoring movies can be compared by seeing whichever approaches infinity faster. In the end I don't really have a horse in this race, though. I'm quite content with the current rules, as all of these movies are eligible for moons. I'm more disappointed in the audience response than the rules, but I suppose that just illustrates that "score movies should be excepted" is not a sentiment shared by the majority of TASVideos.