AMV Content Rating System

[Phade]

Arguably, the math proposed would average this out into a number that is somewhat meaningful

The averaging is why the system is likely to work. It is super-easy for the user to provide one number. The question boils down to this: "Give a 1-to-100 value between 'wholesome' and 'Satan' for this video." All the rest of the text is guidance/noise.

You would need to give a lot of really concrete guidelines and specific examples of what kinds of things would place a video in a "higher" numerical category

We can come up with a decent-sized group of some trusted volunteers (from Discord and the forum here) and a set of 20-to-30-or-so highly known AMVs to run initial tests with. We get everyone to give feedback on those videos and then look at the results. We then present the results as examples. "This Totoro AMV scored a 7." "This DBZ video scored a 41." "This Evangelion AMV scored a 66." "This Helsing AMV scored an 82" and so on.

If an AMV has been sufficiently scored with a high number of values and consistent results, that AMV could be used as an example for the rating system.

Over time, we can manually or mathematically select specific members as "curators": people who seem to have a good eye for giving honest and consistent results. The curator score could then be weighted within the averaging system.

I think a tag system in lieu of a numerical rating would solve a lot of this -- or a tag system in addition to a much more simplified content rating system like is proposed here. Have 4-5 content categories and then a robust tag system that allows users to specify what specific content is in a video -- violence, language, nudity, sexual content, etc. -- so that users can skim the tags and decide if that "R"-rated video is actually rated that way for things that they personally find distasteful. Consider adding a(n optional) weighting system to each tag as well to help identify how much of a factor each element is in a video. For a good example of how this might work, check out what AniDB does -- an example of their tag system can be seen here (scroll down a bit to see the tags).

Yes, this is exactly what I am proposing.

I really hope you and your team are considering a generalized tag system

This will definitely be included and is in the current plan.

Phade.

[Phade]

Why are you so concentrated on numbers?

Numbers are easy to average and filter: “I want a video that scores less than 75 on the content rating scale.” The rest of the values are non-numbers but correspond to a number range

If you are still going to display the other things (general subjects, specific warnings), then why bother with the numerical score?

It is easy to provide a single number between 1 and 100. The other more specific ratings require more effort.

So there's no functional difference between those ratings and they can be combined into one rating which wipes out literally a quarter of your numerical scale.

The functional difference is the severity of each rating. “I gave a score of 26 which is PG, but it’s just barely PG; other people might consider it G.” “I gave a score of 58 that's technically PG13, but just barely below an R rating; other people might consider it a low R rating.”

In what world is this frictionless?

“Type a single number,” has much less friction (effort) than, “Provide a 1-10 rating for these 7 aspects: language, violence, nudity, etc.”

Even then, it is mathematically possible to take the 7 or so 1-10 values and convert them into a single 1-100 score if need be.

Not to mention, you're asking people to numerically score something that is not normally numerically scored

The same would apply to scoring the more specific sensitive aspects (language, violence, nudity, etc.)

...expecting everyone to read the guidance on it is a pipe dream.

The guidance is ultimately not the focus. The question is, “Give this AMV a score between 1 and 100 where 1 is ‘wholesome’ and 100 is ‘hellish’.” While the individual score value is subjective, the average over time becomes relevant.

It would be more intuitive if you changed the scale to be something like "How old do you think someone should be in order to watch your video?"

This is exactly part of the guidance. If you as a person more easly relate to age as an internal guide, the “age” column will point you towards a corresponding numeric value. Other people may be more description-based (criteria guidance). Others are visually-based (symbol). If your personal answer for a particular value for an AMV is “13 years old”, pick a corresponding numeric value between 50 and 60. We can easily instruct people not to take too much time on, “Well, should it be 56 or 57?” The correct answer is always, “go with your gut”; that amount of hair-splitting will not make any significant difference either way. If even this narrowed value range is concerning, our advice would be, "Just hit the mid value," and everything will still work.

But, overall, I think a numerical score for this content is unnecessary as a whole.

My two counterpoints are: It’s really, really easy to do. The average of values over time becomes meaningful relevant data.

The low amount of effort (give one number) vs the high amount of return (I can more easily filter and get a quick impresion of video content) makes it an easy win.

[SQ]

In what world is this frictionless?

“Type a single number,” has much less friction (effort) than, “Provide a 1-10 rating for these 7 aspects: language, violence, nudity, etc.”

I am arguing for not using any numerical score at all.

Not to mention, you're asking people to numerically score something that is not normally numerically scored

The same would apply to scoring the more specific sensitive aspects (language, violence, nudity, etc.)

Sounds like we're in agreement that a numerical scale isn't the best method.
Again, I am arguing for not using any numerical score at all.

My two counterpoints are: It’s really, really easy to do. The average of values over time becomes meaningful relevant data.

The low amount of effort (give one number) vs the high amount of return (I can more easily filter and get a quick impresion of video content) makes it an easy win.

Again, as a viewer, this numerical scale does not help me in any way that is not better served by the more specific tags.

[KeiichiFace]

I don't think this is it. Someone telling me that a video is a "78" doesn't tell me what kind of content is actually in the video. And then if someone is telling me the content that is then making it a "78 video", why have the number in the first place? I also don't know why we need to have a scale be 1-100 when it could easily be 1-10 (and even then it feels like overkill to me).

Couldn't we just put tags for specific content warnings?

Also, this is based on the notion that graphic violence is on the same level as sexual content. In some spaces, one is more accepted over the other. And there's tasteful sexual content in some things that would not make it immediately inappropriate.

[Phade]

Again, as a viewer, this numerical scale does not help me in any way that is not better served by the more specific tags.

There will still be a specific tags system that you and everyone else can use. This proposed numerical scale is not a replacement for a tag-based system.

However, I know that I will use and would greatly benefit from a 1-100 "wholesome vs hellish" score, and my guess is that there are many others like me. The amount of effort to implement a single number feedback system is extremely low, while the return on that feedback would be extremely helpful to me and persons like me.

[Falconone]

How would it be done, how would it be calculated, what do people need to do? Only the uploader or also the viewer?
What is someone seeing as moderate violence? what would count as mild violence? I dont know, for me its totally different then maybe others would see it.

[KeiichiFace]

But what does the score actually tell you that the tags don't?

Being told something is a 78...that doesn't tell me the content of a video.

An Adolescence of Utena video could get a high score for the tasteful nudity at the end, despite being tasteful and not really showing anything.

These are the pitfalls I foresee

[Phade]

I don't think this is it. Someone telling me that a video is a "78" doesn't tell me what kind of content is actually in the video.

A score of 78 is not intended to tell you the content of the video. A score of 78 will tell you the AMV is likely not wholesome.

CrackTheSky previously gave the example of the AniDB ratings of anime, which happened to be Ranma 1/2 (one of my favorite anime).



The review rating of 8.16 tells us nothing of the content. However, it tells us that it's probably a good anime compared to others. To know if it's a good anime for you in particular, you may need to look at the tags (action, comedy, contemporary fantasy, fantasy, manga, martial arts, romance, school life). Both the rating number and tags are relevant and useful in their own ways.

The same goes for the content rating score (but in reverse). A low content rating score is likely "safe" to watch, while a high score may need more investigation before choosing to view (check the tags, content warnings, specific warnings).

And then if someone is telling me the content that is then making it a "78 video", why have the number in the first place? I also don't know why we need to have a scale be 1-100 when it could easily be 1-10 (and even then it feels like overkill to me).

If you use a 1-10 scale and allow one decimal place (ex., 7.8), it is exactly the same as a 100-point scale, meaning a value of 7.8 on a 10-point scale is the same as a 78 on a 100-point scale with no decimal.

Couldn't we just put tags for specific content warnings?

We will have specific content warnings along with the numeric score, just like the above AniDB example has a rating score and tags.

Also, this is based on the notion that graphic violence is on the same level as sexual content. In some spaces, one is more accepted over the other. And there's tasteful sexual content in some things that would not make it immediately inappropriate.

You are absolutely correct and I am in no way implying that one type of objectionable material is more or less severe than another. If I find a video that has a 78 content rating score, I will be sure to check the tags, aspect warning scores, and specific warnings before viewing that AMV.

But, let's flip this the other direction for a moment. An AMV has a content rating of 12. To me and to a large number of persons, this AMV is fine to watch no matter what. If I am sitting down with my nephew and he wants to stream Pokémon AMVs but I haven't seen all the AMVs he wants to watch, a random AMV that 87 people have given it an average of 12 on the wholesome scale, I'm going to have no problem hitting "play" and not being worried that someone created a horror-based Pokémon AMV that I missed.

For another example, I could tell the system to stream 10 random AMVs that I have not seen yet that have a star score greater than 4 and a content rating score less than 34 with more than 42 content rating scores given. Boom! I'm in for a good time without worry of general objectionable material showing up.

To summarize: A low content score implies that there is a low likelihood of objectionable material in the AMV. A high content score does not tell you what objectionable material may be in the AMV, just the likelihood that the AMV is objectionable in some way. To know what the objectionable material may be and the severity of that material, you would need to review the tags and aspect warning scores for that AMV before viewing.

I hope this clarifies things (at least some).

[SQ]

Again, as a viewer, this numerical scale does not help me in any way that is not better served by the more specific tags.

There will still be a specific tags system that you and everyone else can use. This proposed numerical scale is not a replacement for a tag-based system.

However, I know that I will use and would greatly benefit from a 1-100 "wholesome vs hellish" score, and my guess is that there are many others like me. The amount of effort to implement a single number feedback system is extremely low, while the return on that feedback would be extremely helpful to me and persons like me.

Can you explain why you need 100 levels of nuance?
What situation would a video at 0 - 10 be okay but 11 - 25 wouldn't?

I can understand the desire to look for family-friendly videos, but I cannot think of a situation where 1 - 25 would need to be more than one category.
If I was so concerned about the presence of ANY type of potentially objectionable content, I would not be trusting the number, I would be pre-screening all videos for my panel or whatever I was using it for to make sure they were safe.


What situations would a video at 50 - 70 be okay but not 70 - 90?

When you think "I am in a very specific mood and do not want to watch anything that might offend me," do you not generally have some thought about what things offend you more than others?
Even in porn-friendly spaces, some types are okay but others aren't. What exactly makes a 100? Maybe I'm ok with scat and diapers but I don't want a snuff film. How do I know where that falls by looking at the number?

I think the colors and "age ratings" have _some_ merit, but I fail to see how such a large scale is helpful.
Even you yourself say large swaths of the numbers are functionally useless until they are averaged out.

What does that mean, exactly? That everyone watching the video is told to also rate the scale 1 - 100?
Is this not an extra level of fiction?
At that point, why not assign number values to tag groups and have the score automatically applied based on the amount and types of tags added?
We could leave the user out of it completely and have no pain points at all.

I believe this is what you mentioned in your first post, but you've contradicted yourself since then.

If you use a 1-10 scale and allow one decimal place (ex., 7.8), it is exactly the same as a 100-point scale, meaning a value of 7.8 on a 10-point scale is the same as a 78 on a 100-point scale with no decimal.

So why not just use 1 - 10?

[Phade]

How would it be done, how would it be calculated, what do people need to do?

Question: How objectionable is this video? 1 = low; 100 = high. Type number here -->

Need guidance? Here is a sheet to help you... (see OP sheet)

Only the uploader or also the viewer?

The uploader and viewer can give scores.

What is someone seeing as moderate violence? what would count as mild violence? I dont know, for me its totally different then maybe others would see it.

It is a general guideline for people who like written guidelines. The interpretation of that guideline is up to the individual. However, statistical averages will give the score most people would agree upon.