hughsie / oars Goto Github PK

I noted that the generated XML is indented with two extra spaces, it's not a big deal, but it felt a bit weird. I don't know if it's intentional, since the code looks like the indentation was added intentionally.

Recently a software mining cryptocurrencies appeared in Snap store. Most people will probably agree cryptocurrency mining is not much worse than advertising, and that the main issue was that the action was not indicated to the user. As we already have advertising attribute, it might be useful to have “mining” or “resource utilization for monetary gain” in the Money category as well. It might have “None”, “Opt-out” and “Mandatory” options.

Games are a really important use case for OARS, however it'd be awesome if we could cover more categories of apps. Even if it's just to make creators of those apps feel more 'welcome' in using the interface. This is feedback we've heard a lot on Flathub as we're encouraging app developers to add OARS metadata to their appdata.

Since OARS is used for deriving a local age ratings system, chances are that local developers will have local ratings in mind when using OARS, so it'd be nice if in the XML tab we could see a resulting local ratings system.

Since there are many of these, we could show the user agent's corresponding ones + a dropdown with locales/territories for users to quickly see what it'd be in other systems.

I think this is a super useful feature, taking into account that there's no quick alternative to getting those.

If the application being rated is a programming language or programming environment, then users could implement their own "Online Text-only Messaging" and similar things asked under "Social".

Our interpretation for GNU Octave is that the answer is None in this case.
We think the question means "as provided by the application out-of-the-box."

Would it be useful to have some text in the webapp to help developers from bumping into this?

"If your game/application allows users to write their own scripts or code, then these questions would typically pertain to the out-of-the-box experience. For example, if a game does not provide Online Audio and Video Messaging, but has an embedded Python console that could theoretically be used by end-users to implement video chat, the answer is probably None."

I noticed this when trying out https://hughsie.github.io/oars/generate.html for an upcoming app of mine.

It is a frontend for GitHub/BugZilla (letting novice users send bug reports to projects easily).

The Online Text-only Messaging question does not have any option for public, semi-moderated (by the project owner and maybe, rarely, GitHub) discussion. It only lists "between users" and "no chat" options.

Also, Online Audio and Video Messaging: GitHub and BugZilla allow users to upload images and video (though semi-moderated). None of the available options is particularly relevant.

Also, Information Sharing: the app itself does not collect telemetry, but is a frontend for GitHub/BugZilla, which may themselves collect such data from user interactions.

So, perhaps this project needs a wider range of options?

I could've sworn it was in the specifications, can someone help inform me?

Since #40 has been locked due to my admittedly poor choice in wording the suggestion I've voiced there, unfortunately, remains undiscussed.

Any separation of heterosexual relationships from homosexual ones in a project like this is deeply problematic and is incompatible with the declared intention of OARS as a project does not agree with discrimination based on sexual orientation.

Thus I suggest getting rid of any reference to homosexuality and to only flag romance in general. In countries where homosexual content causes any issues, romantic content can simply not be shown at all. This would allow the project to comply with local laws without compromising on its stated goals.

I just generated a rather minimal OARS rating for an application and it turned out, that the advertisement rating in the Money section is not included in the output.

Is that intentional?

Currently, the generator offers four classes of apps:

• Game with networking
• Game that does not use any kind of network access
• Application that can connect to the Internet
• Application that does not use the Internet

If you choose “Application […]” rather than “Game […]”, you don't get any questions about content. But there are apps other than games that contain potentially-objectionable content:

• Endless ships encyclopedias and other “knowledge” apps that contain lots of built-in content. History apps, for example, may well contain references to the Crusades – the exemplar for violence-worship mild.
• Flathub has apps which distribute religious texts. I am not an expert, but I'm pretty sure there are references to historical slavery in the Bible.
• (I'm sure there are other examples.)

I think the wording on the generator should be changed to steer authors of apps like these through some (but perhaps not all) of the content questions.

(This was also raised here: flathub/flathub#1081 (comment))

In order to allow OARS to be useful for a wider range of users, I would propose the introduction of a new "mood" category of classifiers, consisting of an "intensity" and a "fear" classifier to begin with; these apply more to games than applications but could also be applied to other media content.

My rationale is this:

• Content can be frightening or unsettling without being expressly violent. PEGI acknowledges this by having a separate "Fear" content descriptor (and an elevated version thereof, "Horror"). This doesn't really apply to applications but would allow developers of games with such content to flag it.
• "Intensity" is a factor in how media content impacts viewers. The German USK takes this into consideration in its age ratings for videogames and I believe to some degree so does the German FSK for films. Low overall age ratings by the USK generally indicate that the content is simple enough in its structure and slow-paced enough to not overwhelm younger children, e.g. by making sure not to put too much pressure to act on the player or not exposing them to constant rapid sequences of exciting content. As I've personally grown older I've come to more appreciate less intense films and games for myself so having this as an explicit descriptor feels to me like it could be useful. Additionally I know a number of people with sensory processing issues or other conditions that means they can easily get overwhelmed by media that's too "intense", which again feels like user groups that would benefit from the information conveyed by a general rating there.

I've come up with roughly this breakdown:

General intensity

A summary of the general intensity of the content, e.g. the pacing of a piece of media or pressure to act on the player in a videogame. Heated arguments between characters being a major part of the work, or rapid sequences of tense scenes increase the perceived "intensity" of a work, and conflict-oriented content should be considered more intense than non-competitive (e.g. solo versus competitive sports).

This is taken into consideration by e.g. the German USK video games rating authority in their age ratings. The classifications here roughly coincide with the USK 0, 12, 16 and 18 ratings.

• None: No significant pressure and a complexity and tone of content that would be suitable for young children.
• Mild: Faster pace and more pressure; conflict presented is non-violent or staged to be clearly recognisable as fiction.
• Moderate: Rapid pacing and aggressive conflict.
• Intense: High-pressure pacing; games forcing players to put themselves in danger.

Fear

Whether the content includes elements frightening or threatening to the person using/viewing/playing the content.

• None
• Mild: Content that might unsettle or frighten people with particular sensitivities or young children.
• Moderate: Moderate use of content specifically intended to create a frightening mood.
• Intense: Heavy use of content specifically intended to frighten the viewer/user/player.

KDE Discover supports OARS since 5.26: https://invent.kde.org/plasma/discover/-/merge_requests/357

Hi,

While I acknowledge that you feel like it is reasonable to include sex-homosexuality due to it being outlawed in certain jurisdictions.

I believe it is also important to note that in your jurisdiction it is not only legal but also protected by the Equality Act 2010 (along with many other European countries with similar laws).

And while I am not a lawyer I believe this qualifies as harassment due to this having the effect of violating the dignity of people purely on the aspect of them being homosexual.

And aside from the legal aspects, this is despicable on moral grounds.

And you mention the following:

However, in other countries (for example, the EU), discrimination laws explicitly prohibit discrimination on the basis of gender or sexuality. So while LGBT tagging may be available in OARS data, consumers of that data must only apply it in countries where the law requires that.

This is not adequate, you are still providing tools and services to facilitate discrimination of people based on homosexuality.

And once again while I am not a lawyer, I believe this probably counts as illegal discrimination according to UK law.

To reproduce

Using the Open Age Ratings Service (https://hughsie.github.io/oars/generate.html):

1. Select “Application that does not use the Internet”

2. Select either OARS version 1.0 or 1.1 (same result in each)

3. For every other question, leave the None setting.

   At the end you will get:

   <content_rating type="oars-1.1" />

4. Add this to your appdata.xml and validate it using appstreamcli validate …

What should happen

It should validate.

What actually happens

Validation fails with the following warning:

W - ind.ie.Gnomit.appdata.xml.in:ind.ie.Gnomit.desktop:47
    Found empty 'content_rating' tag.

Note: Validation passes with appstream-util validate-relax

Coming here from https://gitlab.gnome.org/GNOME/gimp/merge_requests/25

When going through the questionnaire to create a suitable rating, I noticed that there are some questions concerned with sharing personal data.

With most application which create content, this can also happen by adding meta data to the files involved. Should this be covered by OARS as well?

As an example, sharing images with your past GPS locations embedded can be as good as real-time location data if it enables others to predict your location at a specific time.

The bot created this issue to inform you that pyup.io has been set up on this repo.
Once you have closed it, the bot will open pull requests for updates as soon as they are available.

It would make it easier to refer to specific items in the generator if each item in dropdowns included the level; i.e. instead of just:

Using an online API, e.g. a user-counter

…it could read:

Mild: Using an online API, e.g. a user counter

This makes the relationship between the dropdowns more clear and helps the make the generated XML less surprising to developers. It also makes it easier for elementary documentation as we can refer to the dropdowns by their level instead of the descriptions themselves (which could be more liable to change).

Trying to keep the header and footer in sync between changes is slightly annoying, and will only become more annoying if/when more pages are added. Since you're hosting on GitHub Pages, it would be fairly simple to pull the shared elements out into a Jekyll layout to keep them in one place.

I'm happy to tackle that for you (and add instructions to the README for local development) if you'd like.

There are "Online Text-only Messaging" and "Online Audio and Video Messaging", which seem to make some amount of sense. However, where would that leave photo messaging? Would a new "Online Photo Messaging" section make sense to cover user-to-user interactions where sending photos is possible? Or should this just be grouped under "Audio and Video" since it's similar?

flathub/flathub#1081 (comment) raised a question about what ratings are sensible for browser-like apps:

FeedReader itself has no advertising or adult content, but you could easily subscribe to gambling, porn and ad-heavy RSS feeds. I'm asking because I can find a generator but I can't actually find a spec for what the OARS options are.

I think it would be a bad idea to rate browser-like apps as "extreme" in all current content categories – the app itself doesn't ship offensive content – but I do think “can access arbitrary web content” is a meaningful piece of information. The only other app store I have at my fingertips – Google Play Store – has something like this. Feedly is an RSS reader. It is rated PEGI 3 (which seems correct) but there is an additional section:

Interactive Elements
Users Interact, Unrestricted Internet

"Users Interact" is analogous to social-chat / social-audio I think; perhaps OARS 1.2 could grow social-internet, where extreme means "can access arbitrary internet stuff"?

Whilst looking to add lbry to flathub, there were questions about how to make OARS better able to cover user generated content applications:

Should OARS tags be added to the appdata file and if so how? Content on LBRY is created by users so it can contain any of the content warnings found in OARS.

What isn't subject to the content are these settings found in the app:

"Share Diagnostic Data" is enabled by default
"Show NSFW content" is disabled by default
Maybe OARS needs a new tag like social-content with None, Regulated, Unregulated? This might be better discussed on the the OARS bugtracker (if they have one).

There are already tags for nudity, sexual appearance, and sexual themes. It seems to me that the only reason to specifically call out homosexuality is for the purposes of discrimination.

I understand that the current position is that without this rating it wouldn't be possible to distribute some software in some countries. But, I strongly believe this stance should be reconsidered.

There is an opportunity here to stand up for what is right and refuse to comply with those who would encourage this kind of discrimination. Yes that would make OARS unusable in certain countries, but how is that any different from the common software licenses we use which enforce the essential freedoms we believe in at the expense of making our software unusable for some?

The alternative is knowing releasing software that will be used for the express and sole purpose of discriminating against people. This tag's only purpose is to suppress human rights.

I had previously opened an issue on the GNOME Software bug tracker sharing some concerns I had about two of the cultural sensitivity questions. When I wrote it, I assumed it was a GNOME Software thing. Someone then pointed that OARS is responsible for the rating system.

Below is everything I wrote there, unchanged. I'd like to know your opinion on this matter.

The age rating system contains "Without reference to homosexuality" and "Without reference to adultery" as justification for certain ratings.

I propose removing "Without reference to homosexuality" for the following reasons:

• Homosexuality is a sexual orientation just like heterosexuality. Since there is no equivalent justification for heterosexuality or any other sexual orientation in the age rating system, this may be considered homophobia.

• There is already a line saying (translating from Portuguese, may be a little off) "No reference to or any representation of sexual nature". It already covers "any sexual content", so mentioning homosexuality there is not only homophobic, but also redundant.

I propose removing "Without reference to adultery" for the following reason:

• Adultery historically comes from religious grounds. It may be a crime in some places, but mostly it falls down to cultural, religious or personal beliefs. For a software meant to be used all around the world, it is best not to be partial in this matter.

• Either way, it is redundant. The line "No reference to or any representation of sexual nature" should cover it.

There are many other redundant justifications in there. It would be best to carefully review them all.

Much like the linked page to the TIGRS spec, it would be useful to have a page where we can see the whole spec in one place. At the moment it seems like you have to use the generator to see what is included in OARS.

It would be nice to have this in a human read-able format somewhere in this repo. Especially it would make it easy to open new issues in GitHub linking to specific lines or to be able to submit pull requests

Similar to (but more specific than) #15. Several apps submitted to AppCenter use online APIs. The applications themselves may not collect any data, but the fact that they ping an online service at all (potentially revealing IP address and any other data sent/received) has privacy implications.

For AppCenter, we've interpreted this case as at social-info: mild or above (depending on data sent). However, it's not clear that using an online API at all would require this rating from the generator. Perhaps this could be clarified in the generator text?

This is more of a question; does it make sense to add Endless OS—or more broadly, individual OSes—to the distributors section? We use OARS for built-in parental controls, but that might also apply to other GNOME distros these days as well.

My instinct is that it does make sense, since it's a feature of the OS, and it better reflects the reality that OARS is actually widely used by distros.

Alternatively, we could add "GNOME Parental Controls" as its own item but 🤷🏻

If it does make sense to add individual OSes, I'd be happy to submit a PR with a handful of distros that I've confirmed ship the OARS-powered parental controls ootb.

It might lend more confidence in OARS if we list projects that are using it, like GNOME Software, elementary AppCenter, and anyone else. This could be as simple as adding it to the homepage for now (unless/until it gets large enough to warrant its own page).

In feedback for a PR Vasily Galkin noted

As far as I understand the version "oars-1.1" standardize the list of expected ids. Is there some syntax to indicate that all ids included in standard except explicitly specified are set to none?

Such syntax would make such sections much more readable, since even chats/games/shopping-apps/etc would only set relevant tags without listing all that .

For apps like meld it would be extremely more readable: having a single line indication that "all content attributes for oars version X.Y were checked" (so all of them are none) - for me it looks more readable than explicitly listing them.

This seemed like interesting feedback, especially as the range of properties is expected to grow.

It was suggested that we provide an OARS rating for our game project, which is a sci-fi empire-building / war game.

I'm having difficulty interpreting several of the content-related OARS questions, particularly about what is a "depiction" of violence or similar concepts.

FreeOrion has, arguably, no "characters", fantasy, realistic, or otherwise. There are space and ground battles in which intelligent beings are implied to be killed, but there is no visual representation of this besides some high-level text and a few numbers about the results of a battle that says some ships were destroyed or damaged, or that a planet was captured after a ground battle. Is this a "depiction" of violence or bloodshed at all? If so, is it "realistic"?

Additionally, there are various different intelligent species that can be on planets or ships. Is a "species" equivalent to a "race" for purposes of discrimination? (There are also "humans" as a species, but there is no distinction about different types of them.) This question may be moot in this case, as there is content in the game called a "Concentration Camp" that is used for removing a particular species from a planet, which I assume is enough to warrant the "intense" rating regardless of the details. But, I'm wondering how I would answer that question if that content wasn't in the game... There would still be a strong distinction between populations on planets or ships based on their species, and just one species can be on each planet, which is arguably a version of "speciesism" built into the game mechanics, and players make gameplay decisions about what species to colonize on planets.

In the current specification, all violence categories (maybe except the first - violence-cartoon - one) seem to be directed towards humans or "characters that look human". It is unclear whether any of these categories could be applied to violence against (non-human) animals or realistic-looking fantasy characters. It would be great to either add animal (and realistic non-human characters) specific violence category or to make the current categories more generic.