The game-play needs to standout from non-assisted play, and must not be seen as trivial. Note that a game is considered trivial until proven otherwise. If getting perfect times everywhere is not challenging, such a game is considered trivial. If later a technique is found that makes TASing it challenging, that game becomes acceptable
The current rules for beating existing records say
If your tool-assisted movie is slower than the non-tool-assisted world record for the same game, aiming for the same goals, your movie will be rejected.
Given the recent change from "not faster than" to "slower than" for speed rules; these two criteria are now somewhat in conflict.
One criterion (in the rules on triviality) essentially says a run must stand out from human play to be acceptable, while the other criterion (in the rules on speed) essentially says a run simply can't be slower than human play to be acceptable.
This prompted a question in my mind: How do we define a game's triviality?
In my opinion, a game's triviality should be based on the ability of humans to present a perfect (or near perfect) performance, not how challenging it is to make a TAS of that perfect performance...especially when acceptability for Vault is in question.
If the majority of humans can easily beat a game unassisted with perfect or near perfect performance, a game can be argued to be inherently trivial (Desert Bus).
Duck Hunt is a good example of the triviality dilemma.
TASing a perfect time performance in 1-Duck mode is extremely easy to do as it's simply a matter of watching a timer and duck location in RAM then firing on the first possible frame. However, it'd be nigh impossible for a human to accomplish this same feat.
While it doesn't make a very interesting TAS to watch, it would still meet all other criteria for acceptance into Vault as the fastest completion of a game. On a side note, it's also obviously superhuman.Side note #2: I don't believe that a max score run would be trivial as the score per duck can vary (even in 1-Duck mode IIRC) and would require some RNG manipulation to accomplish.
Yet, this mode of the game is currently prohibited from vault based on triviality of how difficult it is to make the TAS.
As vault is meant to be the location of fastest known/possible TASes of games (when entertainment isn't considered), what argument is there to restricting games simply because the act of making the TAS is extremely easy?
Even when a game is extremely easy to TAS perfectly; if it isn't also that easy casually, it's not inherently a trivial game.
TL:DR
How challenging it is to actually make the TAS of a game shouldn't be the determining factor on whether or not a particular game is deemed trivial; triviality should instead be based on how simple the game is to play casually.
Should we modify the Vault rules in the following ways?
1) Eliminate the concept that game-play must visually stand out from human play For Moons, this requirement could be maintained.
2) Clearly define how triviality is determined
The point of our site is to publish impressive movies from a perspective of entertainment (moons/stars), speed only (vault), or both.
Therefore, I simply don't see any value or purpose in banning a game from publication(especially when a TAS shows obvious perfection or superhuman play) simply because the creation of the TAS was itself an easy (or easily repeatable) process.
Given the recent change from "not faster than" to "slower than" for speed rules; these two criteria are now somewhat in conflict.
One criterion (in the rules on triviality) essentially says a run must stand out from human play to be acceptable, while the other criterion (in the rules on speed) essentially says a run simply can't be slower than human play to be acceptable.
I'd like to address this one point. This is actually not a conflict—it is a forward compliance perspective.
When making a new (i.e. first-generation) TAS, it has to do better than what humans are doing at that time. Otherwise there is no point bothering with it when you can just watch the unassisted run and get the same experience.
After the publication, it might so happen that humans find a new strategy that a TAS cannot improve upon.
Because of this, when making an improvement to an existing publication, it has to at least match what it cannot improve. Perhaps it should be clarified that directly comparable sections should be matched, not just final time (e.g. by being faster in one place but slower in another). It should not be slower anywhere, but if it cannot be faster, it doesn't have to be.
Mandating that an improvement outperforms a human run across the board would have been impractical for cases where there are no further improvements to be found with existing knowledge. Enforcing such a rule would make some improvements unpublishable and have us stuck with even worse movies.
Warp wrote:
Edit: I think I understand now: It's my avatar, isn't it? It makes me look angry.
When making a new (i.e. first-generation) TAS, it has to do better than what humans are doing at that time. Otherwise there is no point bothering with it when you can just watch the unassisted run and get the same experience.
...
Mandating that an improvement outperforms a human run across the board would have been impractical for cases where there are no further improvements to be found with existing knowledge. Enforcing such a rule would make some improvements unpublishable and have us stuck with even worse movies.
(rhetorically) So if a TAS already exists as published and humans manage to match it, that's ok; but if the best a TASer can do is to only match the absolute best human out there, it's not ok just because in the grand spectrum of time a human has already achieved that performance once? (end rhetorical question)
It doesn't make sense to hold a 1st generation run to higher standard than improvement runs simply because there happens to be 1 or a small handful of humans who have managed to attain the best possible performance at some point in time prior to the game being first TASed. If the majority of humans can match that performance, then it's a different story: the game is trivial and the restriction is valid.
Assuming that a game isn't trivial for most humans to play, the best possible TAS of that game based on known information is still the best possible result and deserves to be documented on the site as such.
Sure, if a TAS and the best human run are equivalent, someone could absolutely go watch the best human run instead of the TAS. But the fact that someone can see an equivalent performance elsewhere doesn't restrict TAS acceptance otherwise. If there's a TAS video of a game on YouTube or NicoVideo that achieves the best possible performance but that TAS was never submitted to our site, we'd accept a submission from a different author even if it only matched that other run for time. So why does it matter if the other run being compared to was human instead of someone else's TAS? If a submission in question matches the best known existing time (regardless of whether a human or TAS made the run), it should be acceptable. We'd never reject a submission solely because another TAS (which isn't published on the site ) is just as good as the submission in question.
People don't only watch our videos to see only what human's CAN'T do...some (probably quite a few casual watchers) watch to see what's the best that CAN be done. They don't care if a human can match it or not, they are simply trying to see what's the best possible (which, in-theory, is a key part of what TASing is all about).
The above quote specifically points out the dichotomy. In one case, you're arguing that updates to an already published TAS are allowed to be only as good as the best humans, but new TASes aren't allowed to be only as good than the absolute best humans.
We need to stop considering Vault runs from an "experience" standpoint. Vault is supposed to be a place that doesn't consider entertainment value. Thus, whether or not someone can obtain the same experience watching a run equivalent to a vault publication elsewhere is a moot point. The vault is effectively an archive for the fastest completions (or maximum score runs), it's not about the experience.
Joined: 4/17/2010
Posts: 11495
Location: Lake Chargoggagoggmanchauggagoggchaubunagungamaugg
Your whole argument starts from complete misreading.
http://tasvideos.org/MovieRules.html -> Rules for games -> The game must be acceptable -> Vault -> Must be clearly definable as a game, which has achievable goals
That is a requirement for games to be acceptable for Vault.
http://tasvideos.org/MovieRules.html -> Rules for movies -> The movie must be good -> A speed-oriented movie must beat all existing records
That rule is for movies and their optimality. Yet in your wording you treat them both as rules for movies, resulting in moot arguments.
DrD2k9 wrote:
The current rules for a vault run say
The game-play needs to standout from non-assisted play, and must not be seen as trivial. Note that a game is considered trivial until proven otherwise. If getting perfect times everywhere is not challenging, such a game is considered trivial. If later a technique is found that makes TASing it challenging, that game becomes acceptable
The current rules for beating existing records say
If your tool-assisted movie is slower than the non-tool-assisted world record for the same game, aiming for the same goals, your movie will be rejected.
Given the recent change from "not faster than" to "slower than" for speed rules; these two criteria are now somewhat in conflict.
One criterion (in the rules on triviality) essentially says a run must stand out from human play to be acceptable, while the other criterion (in the rules on speed) essentially says a run simply can't be slower than human play to be acceptable.
DrD2k9 wrote:
In my opinion, a game's triviality should be based on the ability of humans to present a perfect (or near perfect) performance, not how challenging it is to make a TAS of that perfect performance...especially when acceptability for Vault is in question.
If the majority of humans can easily beat a game unassisted with perfect or near perfect performance, a game can be argued to be inherently trivial (Desert Bus).
Duck Hunt is a good example of the triviality dilemma.
TASing a perfect time performance in 1-Duck mode is extremely easy to do as it's simply a matter of watching a timer and duck location in RAM then firing on the first possible frame. However, it'd be nigh impossible for a human to accomplish this same feat.
While it doesn't make a very interesting TAS to watch, it would still meet all other criteria for acceptance into Vault as the fastest completion of a game.On a side note, it's also obviously superhuman.Side note #2: I don't believe that a max score run would be trivial as the score per duck can vary (even in 1-Duck mode IIRC) and would require some RNG manipulation to accomplish.
Yet, this mode of the game is currently prohibited from vault based on triviality of how difficult it is to make the TAS.
As vault is meant to be the location of fastest known/possible TASes of games (when entertainment isn't considered), what argument is there to restricting games simply because the act of making the TAS is extremely easy?
Even when a game is extremely easy to TAS perfectly; if it isn't also that easy casually, it's not inherently a trivial game.
TL:DR
How challenging it is to actually make the TAS of a game shouldn't be the determining factor on whether or not a particular game is deemed trivial; triviality should instead be based on how simple the game is to play casually.
Should we modify the Vault rules in the following ways?
1) Eliminate the concept that game-play must visually stand out from human play For Moons, this requirement could be maintained.
2) Clearly define how triviality is determined
The point of our site is to publish impressive movies from a perspective of entertainment (moons/stars), speed only (vault), or both.
Therefore, I simply don't see any value or purpose in banning a game from publication(especially when a TAS shows obvious perfection or superhuman play) simply because the creation of the TAS was itself an easy (or easily repeatable) process.
This tier contains speed-based movies that don't have much entertainment value, but still represent meaningful tool-assisted speedrun records. Game choice is tightly limited. Vault rules filter out games that don't hold much weight when tool-assistance is applied in accordance with the TASing guidelines on optimization. Vault needs clear cuts, so whenever something can not be clearly distinguished, such a movie gets rejected.
If something is trivial to TAS, it won't be a meaningful record, and TASing it would make no sense, because tool-assisted speedrunning is meant to be a competition. Competition involves variety of skills and room to develop them in order to beat the previous record. If the previous record is extremely hard to beat, because research and execution behind it looks exhaustive (SMB), that's one thing. If the previous record involved zero TASing skills and was unbeatable only because the game is trivial once you grab your tools, that's another. We want to encourage competition based on how skilled the TASer is, because that way people push themselves to the limit while pushing the games to the limit. The quality keeps increasing, the records remain meaningful, the site is happy.
How challenging it is to actually make the TAS of a game shouldn't be the determining factor on whether or not a particular game is deemed trivial; triviality should instead be based on how simple the game is to play casually.
All I see in your post is statements, and no real reasons supporting them.
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.
Joined: 4/17/2010
Posts: 11495
Location: Lake Chargoggagoggmanchauggagoggchaubunagungamaugg
DrD2k9 wrote:
(rhetorically) So if a TAS already exists as published and humans manage to match it, that's ok; but if the best a TASer can do is to only match the absolute best human out there, it's not ok just because in the grand spectrum of time a human has already achieved that performance once? (end rhetorical question)
It doesn't make sense to hold a 1st generation run to higher standard than improvement runs simply because there happens to be 1 or a small handful of humans who have managed to attain the best possible performance at some point in time prior to the game being first TASed. If the majority of humans can match that performance, then it's a different story: the game is trivial and the restriction is valid.
Assuming that a game isn't trivial for most humans to play, the best possible TAS of that game based on known information is still the best possible result and deserves to be documented on the site as such.
Sure, if a TAS and the best human run are equivalent, someone could absolutely go watch the best human run instead of the TAS. But the fact that someone can see an equivalent performance elsewhere doesn't restrict TAS acceptance otherwise. If there's a TAS video of a game on YouTube or NicoVideo that achieves the best possible performance but that TAS was never submitted to our site, we'd accept a submission from a different author even if it only matched that other run for time. So why does it matter if the other run being compared to was human instead of someone else's TAS? If a submission in question matches the best known existing time (regardless of whether a human or TAS made the run), it should be acceptable. We'd never reject a submission solely because another TAS (which isn't published on the site ) is just as good as the submission in question.
People don't only watch our videos to see only what human's CAN'T do...some (probably quite a few casual watchers) watch to see what's the best that CAN be done. They don't care if a human can match it or not, they are simply trying to see what's the best possible (which, in-theory, is a key part of what TASing is all about).
The above quote specifically points out the dichotomy. In one case, you're arguing that updates to an already published TAS are allowed to be only as good as the best humans, but new TASes aren't allowed to be only as good than the absolute best humans.
We need to stop considering Vault runs from an "experience" standpoint. Vault is supposed to be a place that doesn't consider entertainment value. Thus, whether or not someone can obtain the same experience watching a run equivalent to a vault publication elsewhere is a moot point. The vault is effectively an archive for the fastest completions (or maximum score runs), it's not about the experience.
Distinction between trivial games and trivial movies is crucial and resolves your questions. If the game in its nature is trivial to TAS, any TAS of it is rejected, because it won't be a meaningful tool-assisted speedrun record. But if the game is complicated and affords serious TAS competition, its movies are accepted, even if human record matches a TAS, as long as there's nothing more to improve with the current knowledge.
#6614: The8bitbeast's SMS Zool: Ninja of the "Nth" Dimension "game end glitch" in 00:21.61
So for non-trivial games, the movie rule that demands beating or at least matching human records simply allows a seemingly-trivial movie to be accepted, to establish the current state of art and to encourage further developments, where a TAS may outperform humans again, or get beaten by them and compete with them again. This is possible because the game is complex enough.
This is the reason we don't particularly care how many humans can match the TAS record. The current Zool record can be matched by infinity of humans, and it won't mean the game has magically become trivial in its nature, all the future movies of it must be rejected, and original acceptance was a mistake. That assumption would be unfair and discouraging to people pushing this game to the limit.
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.
I realize that I'm challenging the status quo and the operating opinion of some (possibly many) in the community. Please don't read any of the following as anger, it's not intended that way.
feos wrote:
Distinction between trivial games and trivial movies is crucial and resolves your questions. If the game in its nature is trivial to TAS, any TAS of it is rejected, because it won't be a meaningful tool-assisted speedrun record. But if the game is complicated and affords serious TAS competition, its movies are accepted, even if human record matches a TAS, as long as there's nothing more to improve with the current knowledge.
You're debating from a perspective that appears to suggest that our site is primarily about competition. You're suggesting that a TAS of any game is meaningless simply because the act of obtaining the optimal method to win is obvious and easy to produce in TAS form and thus wouldn't have room for challenge by another TASer.
Is that all we care about as a site? Do we only care about TASing games that could, in theory, yield a frame-war? Are all other games that can't produce this potential for challenge really meaningless, or is that an arbitrary distinction we have created for some esoteric reason in hopes to spur such challenges?
As I understand things: In it's origin, the site was a showcase of TAS runs that were deemed entertaining to watch, not as a place to hold competition for who could TAS best. Then the Vault was added to archive the fastest known runs of games regardless of whether they were entertaining or not. In other words, the Vault was added to present the best that can be accomplished from a speed perspective even if it was dull to watch.
While this may have allowed opportunity for more competitive back and forth in the community, the results of the Vault TASes being published remained the same....presenting the fastest (known) possible outcome. Then the vault was further expanded to include 100% runs and max score TASes that still were required to do those two things in the fastest possible time, but that's kinda beside the point of my arguments.
Why then are we against presenting the fastest run that can be accomplished from a speed perspective simply because the available tools make it easy to present that particular result and there's no foreseeable way for someone to challenge that known best result?
Why claim that a tool-assisted speedrun record is not meaningful just because it's was easy to accomplish using the available tools?
I realize that competition/challenge can and does happen here, but I've never taken anything on our site to suggest that the purpose of our existence was to only present things that can breed competitive challenge.
Why is one game deemed less meaningful to TAS simply because it's easier to actually create the TAS than it would be for another game?
How much meaning/value any art form holds is not determined by the creators/curators, it's determined by the viewer/consumer.
There's a lot of art in the world that I personally feel is worth less than the individual media components it's created with, yet others would be willing to pay millions of dollars to acquire.
What I or any particular community member may see as a worthless/trash TAS, another viewer may see as one of the most amazing TASes ever created.
Acceptability of TASes that would be archived in the Vault shouldn't be judged based on some undefinable potential meaning that someone may or may not ascribe to the resulting video. Nor should it be determined on how difficult they were or weren't to create. Acceptability should simply be based on if it's the fastest known optimal TAS run (given current knowledge of the game).
Joined: 4/17/2010
Posts: 11495
Location: Lake Chargoggagoggmanchauggagoggchaubunagungamaugg
I don't see you taking any of my arguments and explaining why they are bad or wrong, you simply keep declaring your own statements and posing rhetoric questions that have already been answered.
I pointed out several times that this rule is only for games that may be eligible for Vault, yet you spend paragraphs describing how the whole site isn't meant to only target competition. It doesn't! The site targets both entertainment and speed, one being subjective (artistic merits that can be recognized by viewers), another being objective (time competition). Vault only holds the latter.
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.
I've personally felt conflicted in the past about this rule and its precise necessity. I believe that in an ideal world we would want to publish every game. However, that isn't practical. We have judges and publishers that are required to move the process along, so there is a need to limit what gets sent our way. From that, one of the most obvious subsets to limit is movies that have little effort put into them.
Additionally, a trivial record isn't just meaningless in the sense that there can be no competition. It is meaningless in that who set the record and therefore authorship doesn't matter. Anybody can create the inputs for a trivial movie.
Let's take Barney's Hide & Seek for an example. If left alone, the game will actually play and complete itself. The only required input to beat it is Power On.
Who came up with the TAS for that? The first person to submit? They didn't exactly come up with the inputs, power on is in the emulator by default. The emulator? The first person to ever play the game? The creators of the game?
When a movie requires effort to make, who made the movie becomes much simpler.
If one were to accept trivial records in order to merely document the fastest possible time, one would need to remove authorship or else one would run into a plethora of issues regarding attribution. This site is obviously not set up for that, nor do I know if anybody really wants that at this given time.
[16:36:31] <Mothrayas> I have to say this argument about robot drug usage is a lot more fun than whatever else we have been doing in the past two+ hours
[16:08:10] <BenLubar> a TAS is just the limit of a segmented speedrun as the segment length approaches zero
I've personally felt conflicted in the past about this rule and its precise necessity. I believe that in an ideal world we would want to publish every game. However, that isn't practical. We have judges and publishers that are required to move the process along, so there is a need to limit what gets sent our way. From that, one of the most obvious subsets to limit is movies that have little effort put into them.
Additionally, a trivial record isn't just meaningless in the sense that there can be no competition. It is meaningless in that who set the record and therefore authorship doesn't matter. Anybody can create the inputs for a trivial movie.
Let's take Barney's Hide & Seek for an example. If left alone, the game will actually play and complete itself. The only required input to beat it is Power On.
Who came up with the TAS for that? The first person to submit? They didn't exactly come up with the inputs, power on is in the emulator by default. The emulator? The first person to ever play the game? The creators of the game?
When a movie requires effort to make, who made the movie becomes much simpler.
If one were to accept trivial records in order to merely document the fastest possible time, one would need to remove authorship or else one would run into a plethora of issues regarding attribution. This site is obviously not set up for that, nor do I know if anybody really wants that at this given time.
Games like Barny are trivial-to-play casually, thus TASing isn't even necessary. Other games like Duck Hunt aren't trivial to play (more on this later); and while the act of making the TAS may be trivial, the resulting video still demonstrates superhuman TAS ability.
A game being trivial-to-make, doesn't mean that effort hasn't been exerted in creation of the TAS. This is part of why I have previously emphasized that 'trivial to play casually' and 'trivial to make a TAS' aren't equivalent.
Implementation?
For such trivial-to-make TASes that would still demonstrate super-human ability, would it be impossible to publish them without authorship attribution? Or simply have authorship be attributed to "Many Tasers." Or even "Trivial to TAS"?
With any of those options, no one person has to be identified as the TAS author for attribution reasons. This would eliminate any player points comparison/headache that would result from any one member claiming they were first or that they should get the points for authorship.
The staff could even create a member with the name "Trivial to TAS" for such a purpose. Then when a trivial to TAS game is submitted, the game can be published under this authorship regardless of who actually made the submission. A note could be placed in the judgement notes describing this publication action. This would allow for publication of non-trivial-to-play superhuman runs even if they are trivial-to-make.
Regarding productivity and practicality: if a superhuman TAS is judged to have such simple optimization that it's deemed trivial-to-make, then judgement shouldn't be that hard on whether it's optimized or not (and since we're talking about Vault, there's not reason to consider entertainment). Judgments would essentially be mere formality and shouldn't require much effort on the judge's part.
I've had some private discussion with feos on discord and will be posting some of the thoughts from that discussion here in a moment in a separate post..
feos and I had some back and forth discussion on Discord regarding our respective perspectives and and the purpose of the vault, the following are as concise of a summary as I can provide:
feos: by your logic "as long as it's done optimally, no matter how trivial it is, it must be archived even for such games!!!!!!!"
The following is my response to that summation.
Me:This is close to how I feel, but not exactly.
I'm suggesting that the TASes of games that are not trivial to play casually deserve a place of recognition on our site as examples demonstrating what TASing can do compared to humans (i.e. 'superhuman') regardless of how trivial it is to make that TAS. I have no desire to have TASes of trivial-to-play games like Desert Bus or Color a Dinosaur on the site.
I see 3 levels of triviality. Trivial to play. Triviality of TAS Production. Triviality of resulting video.
feos:I mean you'd be arguing for those types of games I quoted just as well
(Here he was referenceing games such as Dragon's Lair, choose your own adventure stories, color a dionsaur, etc.)
Here's my response which hopefully provides more insight to everyone on why I feel some trivial-to-TAS games which are not trivial-to-play games deserve a place of publication in the Vault.
I can see where someone could use my basic perspective to try and argue for Dragon's Lair or a 'choose an adventure' type game, but the counter argument in many if not most of those cases would be that there's no in-game benefit/value derived by optimizing the input time.
Even in casual play, pressing a direction in Dragon's Lair on the first possible frame vs the last possible frame doesn't change the gameplay or progress; all obstacles to the predefined game progress remain the same regardless of when (in the appropriate time window) that direction is pressed. Time optimization is only affected by the frame delay within that window when the button is actually pressed (a direction press 5 frames later than the earliest possible frame adds exactly 5 frames to the overall time of the run). This isn't the case for games like Duck Hunt.
From a casual play perspective in Duck Hunt, the frame on which a duck is shot does result in changes to gameplay action and therefore adds potential for an in-game benefit/value of time optimization. If not shot immediately, the duck will fly around based on RNG; meaning the gun would need repositioned/aimed. Also the randomized position that the duck flies to before it's shot determines the fall distance and thus introduce another factor in time optimization beyond simply how much time it took the player to pull the trigger; a duck shot higher on the screen falls a further distance and costs extra time beyond the number of frames the player delayed before the shot (shooting 5 frames later than the earliest possible frame may result in greater than a 5 frame delay on the overall run). This means that the both aiming time and firing frame of the gun are actions that impact time optimization from a in-game benefit/value standpoint. This potential in-game variation due to game-play choices provides the in-game benefit/value of optimization: 'Shooting a duck lower on the screen, yields waiting less time before the next duck releases.'
Yes it's trivial to make the Duck Hunt TAS itself because of the tools available. Shots can be made to occur on the first possible frame with the gun pointed in the right place. It's not hard to optimize. But the in-game benefit/value optimization is still present This is the point in optimizing 1-duck mode in Duck Hunt; killing the ducks as quickly as possible to minimize fall distance. It just so happens that it's easy to do this optimization in a TAS environment.
The grey area with the in-game benefit argument is games like Deja Vu or Shadowgate. Speed of input doesn't necessarily affect the gameplay result or create in-game changes. However, those games still at least have the argument of cursor movement optimization beyond general menuing (where Dragon's Lair, choose an adventure, and such games do not even have this optimization challenge).
To put it shortly, there is a reason to optimize the shot frame in Duck Hunt beyond it was the first possible frame to do so.
This reason is the reason the game is not trivial and a speed record of it has value.
EDIT: I want our site to offer even more comprehensive picture of how TASing is superior to human play, but the current approach to triviality prevents some obvious superhuman TASes from being published.
Regarding productivity and practicality: if a superhuman TAS is judged to have such simple optimization that it's deemed trivial-to-make, then judgement shouldn't be that hard on whether it's optimized or not (and since we're talking about Vault, there's not reason to consider entertainment). Judgments would essentially be mere formality and shouldn't require much effort on the judge's part.
Judgment would still be far from a "mere formality".
We don't assume tiers, we consider tiers based on audience feedback and our own experience watching the TAS. Additionally we require some proof of sync (preferably syncing for both a judge and a publisher but this may not always be the case).
Personally I also try to be as personable as possible in my judgment notes. This requires a certain emotional state from me that I cannot just reproduce in mass. I would hate having some dull response prepared for bulk TASes.
The work that goes into publishing with is also nontrivial and not easy to speed up.
[16:36:31] <Mothrayas> I have to say this argument about robot drug usage is a lot more fun than whatever else we have been doing in the past two+ hours
[16:08:10] <BenLubar> a TAS is just the limit of a segmented speedrun as the segment length approaches zero
If the game in its nature is trivial to TAS, any TAS of it is rejected, because it won't be a meaningful tool-assisted speedrun record.
Is it more precise to say that we don't want to publish movies that play a game as the creators' intended if such gameplay is trivial?
In other words, if a trivial game gets broken, it can become non-trivial.
Here is the simplest example I can think of:
Imagine a simple game that you win by walking the character 2 pixels to the right, and the character can move 1 pixel per frame.
The intended gameplay is trivial because someone could do it perfectly on their first playthrough with no knowledge of the game just by holding right for 2 frames.
But then imagine that a TASer discovers that by holding L+R, the character obtains hyperspeed and moves 2 pixels to the right in 1 frame.
Since this trick is unintended by the creators and the TASer has demonstrated special knowledge and/or talent to obtain a faster time, the "game" is no longer trivial, right?
Joined: 4/17/2010
Posts: 11495
Location: Lake Chargoggagoggmanchauggagoggchaubunagungamaugg
DrD2k9 wrote:
Implementation?
For such trivial-to-make TASes that would still demonstrate super-human ability, would it be impossible to publish them without authorship attribution? Or simply have authorship be attributed to "Many Tasers." Or even "Trivial to TAS"?
With any of those options, no one person has to be identified as the TAS author for attribution reasons. This would eliminate any player points comparison/headache that would result from any one member claiming they were first or that they should get the points for authorship.
The staff could even create a member with the name "Trivial to TAS" for such a purpose. Then when a trivial to TAS game is submitted, the game can be published under this authorship regardless of who actually made the submission. A note could be placed in the judgement notes describing this publication action. This would allow for publication of non-trivial-to-play superhuman runs even if they are trivial-to-make.
Regarding productivity and practicality: if a superhuman TAS is judged to have such simple optimization that it's deemed trivial-to-make, then judgement shouldn't be that hard on whether it's optimized or not (and since we're talking about Vault, there's not reason to consider entertainment). Judgments would essentially be mere formality and shouldn't require much effort on the judge's part..
This overthrows the entire purpose of the site as I explained to you thoroughly in private, but it was seemingly not recognized at all.
Allowing trivial-to-make movies will flood us with movies that hold zero speedrun record value, resulting in net drain on movie quality on the site. We spent years of effort to encourage TASers to invest themselves into improving their skills in order to be published, the whole obsoletion system is designed to recognize hard work, and as a result, optimization standards are slowly rising, the movies are steadily getting better in that regard.
We also explicitly require judges to understand the difference between submissions that are easy to improve and the ones that are hard to improve. If something is easy to improve, it means not enough work has been put into the movie, and therefore it's too sloppy to be published. You're suggesting to disregard this whole thing for movies that are trivial to make. And I'm still waiting for a single reason that would imply the site would be better from hosting them.
DrD2k9 wrote:
I can see where someone could use my basic perspective to try and argue for Dragon's Lair or a 'choose an adventure' type game, but the counter argument in many if not most of those cases would be that there's no in-game benefit/value derived by optimizing the input time.
Even in casual play, pressing a direction in Dragon's Lair on the first possible frame vs the last possible frame doesn't change the gameplay or progress; all obstacles to the predefined game progress remain the same regardless of when (in the appropriate time window) that direction is pressed. Time optimization is only affected by the frame delay within that window when the button is actually pressed (a direction press 5 frames later than the earliest possible frame adds exactly 5 frames to the overall time of the run). This isn't the case for games like Duck Hunt.
From a casual play perspective in Duck Hunt, the frame on which a duck is shot does result in changes to gameplay action and therefore adds potential for an in-game benefit/value of time optimization. If not shot immediately, the duck will fly around based on RNG; meaning the gun would need repositioned/aimed. Also the randomized position that the duck flies to before it's shot determines the fall distance and thus introduce another factor in time optimization beyond simply how much time it took the player to pull the trigger; a duck shot higher on the screen falls a further distance and costs extra time beyond the number of frames the player delayed before the shot (shooting 5 frames later than the earliest possible frame may result in greater than a 5 frame delay on the overall run). This means that the both aiming time and firing frame of the gun are actions that impact time optimization from a in-game benefit/value standpoint. This potential in-game variation due to game-play choices provides the in-game benefit/value of optimization: 'Shooting a duck lower on the screen, yields waiting less time before the next duck releases.'
Yes it's trivial to make the Duck Hunt TAS itself because of the tools available. Shots can be made to occur on the first possible frame with the gun pointed in the right place. It's not hard to optimize. But the in-game benefit/value optimization is still present This is the point in optimizing 1-duck mode in Duck Hunt; killing the ducks as quickly as possible to minimize fall distance. It just so happens that it's easy to do this optimization in a TAS environment.
The grey area with the in-game benefit argument is games like Deja Vu or Shadowgate. Speed of input doesn't necessarily affect the gameplay result or create in-game changes. However, those games still at least have the argument of cursor movement optimization beyond general menuing (where Dragon's Lair, choose an adventure, and such games do not even have this optimization challenge).
To put it shortly, there is a reason to optimize the shot frame in Duck Hunt beyond it was the first possible frame to do so.
This reason is the reason the game is not trivial and a speed record of it has value.
EDIT: I want our site to offer even more comprehensive picture of how TASing is superior to human play, but the current approach to triviality prevents some obvious superhuman TASes from being published.
Optimization is not a binary thing, and just having it there is not enough to have a published movie. Optimization is a scale, and we want as much of it as reasonably practicable, because it directly determines the movie quality. We want movie quality to be high. Low-effort movies get rejected because of that. If optimization is not challenging, the movie quality is also low.
You rely so much on what's possible for humans, but humans are evolving as well. Over the course of 10 years I've been watching TAS and RTA scenes, humans were able to pull of TAS strats on nearly daily basis. Exactly because competition is there! They just strive to optimize it, and it gradually gets better for everyone.
Dragster TAS times have been matched by humans, even though we couldn't envision that 8 years ago when Vault was made, to demand that all human records must be beaten and the movie must obviously look better than the best human record. Humans got better and they're catching up with TAS!
Demanding that a movie must be clearly superior is not reasonable anymore, because then we end up with old published TASes that have been beaten by RTA, but no TAS improvement gets accepted because "humans have matched the times".
Relying on "majority of casual players" is even less convincing, because that requires collecting statistics about a sphere that's completely separate from what we deal with here, and also impossible to collect. Vault requires clear cuts, and you suggest to introduce "human element", however vague and subjective it may be. And for what purpose? In order to be able to reduce the overall movie quality. Brilliant.
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.
Joined: 4/17/2010
Posts: 11495
Location: Lake Chargoggagoggmanchauggagoggchaubunagungamaugg
Arc wrote:
Is it more precise to say that we don't want to publish movies that play a game as the creators' intended if such gameplay is trivial?
In other words, if a trivial game gets broken, it can become non-trivial.
Here is the simplest example I can think of:
Imagine a simple game that you win by walking the character 2 pixels to the right, and the character can move 1 pixel per frame.
The intended gameplay is trivial because someone could do it perfectly on their first playthrough with no knowledge of the game just by holding right for 2 frames.
But then imagine that a TASer discovers that by holding L+R, the character obtains hyperspeed and moves 2 pixels to the right in 1 frame.
Since this trick is unintended by the creators and the TASer has demonstrated special knowledge and/or talent to obtain a faster time, the "game" is no longer trivial, right?
We do recognize this in the triviality rule, here's the wording:
The game-play needs to standout from non-assisted play, and must not be seen as trivial. Note that a game is considered trivial until proven otherwise. If getting perfect times everywhere is not challenging, such a game is considered trivial. If later a technique is found that makes TASing it challenging, that game becomes acceptable.
Maybe it's not exactly what you mean, but relying on developer intent alone is shaky ground, so we have to add something more verifiable. But with that requirement we still tried to account for your scenario.
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.
If something is trivial to TAS, it won't be a meaningful record
Of course it will, especially if speedrunning it in real-time cannot achieve the same time.
It's meaningful because it gives an interesting piece of information: How fast can the game be completed theoretically, assuming perfect play?
I think tasvideos should be considered like a library of perfect speedruns. An encyclopedia that collects best possible theoretical completion times of games. If someone would be interested in knowing what this theoretical best completion time is for game X, that information could be found here. Whether the TAS is "trivial" to make is inconsequential. What matters is the piece of information that it gives.
What is the downside you see in having TASes of non-trivially-speedrunnable (by a human) games, even if making the TAS is trivial? What is the harm, loss, or inconvenience caused by this?
Joined: 4/17/2010
Posts: 11495
Location: Lake Chargoggagoggmanchauggagoggchaubunagungamaugg
Maybe if you read more than a single sentence you'll find out that these questions have already been answered?
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.
Having a lot of low quality and not entertaining published runs would diminish the overall worth of the site. Is there something wrong with userfiles? Even they get a place on the front page and can act as a library for tases.
Having a lot of low quality and not entertaining published runs would diminish the overall worth of the site. Is there something wrong with userfiles? Even they get a place on the front page and can act as a library for tases.
How many submissions do you expect to see of TASes for games that are not trivial to speedrun by a human but have a completely trivial TAS?
Having a lot of low quality and not entertaining published runs would diminish the overall worth of the site. Is there something wrong with userfiles? Even they get a place on the front page and can act as a library for tases.
How many submissions do you expect to see of TASes for games that are not trivial to speedrun by a human but have a completely trivial TAS?
[14:15] <feos> WinDOES what DOSn't
12:33:44 PM <Mothrayas> "I got an oof with my game!"
Mothrayas Today at 12:22: <Colin> thank you for supporting noble causes such as my feet
MemoryTAS Today at 11:55 AM: you wouldn't know beauty if it slapped you in the face with a giant fish
[Today at 4:51 PM] Mothrayas: although if you like your own tweets that's the online equivalent of sniffing your own farts and probably tells a lot about you as a person
MemoryTAS Today at 7:01 PM: But I exert big staff energy honestly lol
Samsara Today at 1:20 PM: wouldn't ACE in a real life TAS just stand for Actually Cease Existing
Joined: 4/17/2010
Posts: 11495
Location: Lake Chargoggagoggmanchauggagoggchaubunagungamaugg
Warp wrote:
CoolHandMike wrote:
Having a lot of low quality and not entertaining published runs would diminish the overall worth of the site. Is there something wrong with userfiles? Even they get a place on the front page and can act as a library for tases.
How many submissions do you expect to see of TASes for games that are not trivial to speedrun by a human but have a completely trivial TAS?
Since that was the answer that I've given before, and it now got questioned further, here's what I can reply with.
We shouldn't base our rules on hope that there won't be many such movies, because it's subjective and based on lack of knowledge ("we don't know exactly how many such games are out there, so we presume only few"). If low-effort movies are allowed, it automatically means they are encouraged, because people tend to prefer low-effort work if the outcome is the same.
Now the borderline is high, and people try to work on something good. If the borderline goes lower, in addition to games that fit the suggested rule, people would also submit trivial games that are even simpler than expected, and figuring out the new borderline will take a while. People would be testing "how low you can go", potentially having fights hoping to allow even more trivial games just because they feel the site should blindly be a record database for everything.
Currently I can't say we see serious demand in this area, with lots of TASers sending simplistic games and getting upset over rejections, making this a trend, implying that the rules aren't thorough enough. But in this thread I only see some people expressing feelings, and no one posting convincing reasons pro that rule change. Feels obscure.
It's meaningful because it gives an interesting piece of information: How fast can the game be completed theoretically, assuming perfect play?
Perfect play isn't how we measure movie quality. If you easily achieve perfect time in a trivial game, it's of the same quality as playing SMB in a sloppy, lazy way. I can't consider lazy play a meaningful record that you need TAS tools and skills for.
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.
It's meaningful because it gives an interesting piece of information: How fast can the game be completed theoretically, assuming perfect play?
Perfect play isn't how we measure movie quality. If you easily achieve perfect time in a trivial game, it's of the same quality as playing SMB in a sloppy, lazy way. I can't consider lazy play a meaningful record that you need TAS tools and skills for.
I think you are completely missing my point there.
It's meaningful because it gives an interesting piece of information: How fast can the game be completed theoretically, assuming perfect play?
Perfect play isn't how we measure movie quality. If you easily achieve perfect time in a trivial game, it's of the same quality as playing SMB in a sloppy, lazy way. I can't consider lazy play a meaningful record that you need TAS tools and skills for.
Maybe that's how we measure quality currently but I don't think that's the best approach.
----
One thing that hasn't really been brought up as far as I know is that just because the minimum amount of effort to create an optimal submission is considered low, that doesn't mean it is the maximum amount of effort one could put in. Take the recent Zool run. The author put additional effort into ensuring that the time was as fast as possible, even botting it. It just so happened that that was all there was left.
If people did like proper analysis of these so called "trivial" games, they might find something new. They might not. But I think there's value and meaning in that for sure.
[16:36:31] <Mothrayas> I have to say this argument about robot drug usage is a lot more fun than whatever else we have been doing in the past two+ hours
[16:08:10] <BenLubar> a TAS is just the limit of a segmented speedrun as the segment length approaches zero
Joined: 4/17/2010
Posts: 11495
Location: Lake Chargoggagoggmanchauggagoggchaubunagungamaugg
Memory wrote:
One thing that hasn't really been brought up as far as I know is that just because the minimum amount of effort to create an optimal submission is considered low, that doesn't mean it is the maximum amount of effort one could put in. Take the recent Zool run. The author put additional effort into ensuring that the time was as fast as possible, even botting it. It just so happened that that was all there was left.
If people did like proper analysis of these so called "trivial" games, they might find something new. They might not. But I think there's value and meaning in that for sure.
Research has been put into that movie because the game was already complex enough to be considered non-trivial, so once again, it successfully encouraged competition, which in turn led to those crazy results. I don't understand reasons not to appreciate competition for leading to all the improvements that people keep finding. Of course our whole take on competition is special, as we encourage team work, openness, sharing techniques and developments. But since no movie is absolutely perfect, there's always room for an educated, instrumented "doubt". And since you never know if you'll even find those improvements, finding them also becomes a game on its own!
And yes, if we considered a game trivial, and then someone found a way to make TASing it challenging again, it becomes acceptable, as the rule says.
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.
