Youtube "age-gating" more videos, youtube-dl affected

Addressing content concerns (hate, gore, sexual nature, etc.) Youtube is age-restricting, or "age-gating", a larger set of videos. This also includes closing the loophole of viewing such videos from third-party sites. Or, apparently, download tools such as youtube-dl.

The policy change is discussed at The Verge, "YouTube is about to age-restrict way more videos":

YouTube is rolling out more artificial intelligence-powered technology to catch more videos that may require age restrictions, meaning more viewers will be asked to sign into their accounts to verify their age before watching.

Similar to how YouTube used machine learning techniques to try to better catch violent extremism and more of the platform’s most severe content beginning in 2017, and later to find videos that included hateful conduct, the same approach will be used in this case to automatically flag videos YouTube deems not age-appropriate. As a result, YouTube is expecting to see far more videos pop up with age-gated restrictions.

The company is preparing for there to be some mistakes in labeling, as is the case with any rollout of AI moderation tech. And as part of the changes, people watching YouTube videos embedded on third-party sites will be redirected to YouTube to sign in and verify their age. ...

https://www.theverge.com/2020/9/22/21449717/youtube-age-restriction-machine-learning-rollout-kids-content-monetization-creators

Youtube-dl is a command-line utility supporting media download and streaming from multiple websites, including its namesake, and provides functionality in numerous other tools (e.g., mpv). Several recent issues have been opened apparently resulting from the age-gate change.

Issue search: https://github.com/ytdl-org/youtube-dl/issues?q=is%3Aissue+is%3Aopen+age-gate+youtube

#youtube #youtube-dl #AgeGating #mpv #video #moderation #google

Thoughts on conversation-generative online discussion platforms and features

I've become increasily aware of how conversation medium and participants shapes the "quality" of conversation.

Conversation scales poorly.

It's also fragile and very easily destroyed, discouraged, or dissuaded.

The biggest issue I find on Reddit itself is that there's no notion of "thread (or post) as conversation". And absolutely no support for same. Reddit is where interesting conversations go to die.

An item is posted. It's at top-of-page for ... a few minutes or hours, possibly days ... then vanishes And no amount of activity within a thread will boost it, generally. Even those who'd participated in the discussion have no signal of any activity. The best that can happen is that members might subscribe to replies for two days. This is madness.

Put another way, Reddit's post-weighting algorithm is all but entirely determined by posting time, not activity recency. This avoids "necroposting", for both good and bad. For small niche discussion, all but entirely bad.

Problem is that Reddit's scale spans about 6-8 orders of magnitude -- subredduts of < 10 members, to > 10,000,000. One-size-fits all ... wears poorly. Most of the glaring problems are at large scale. The small subs get neglected. Clue flees.

The little-lamented Imzy had the problem of seeing Reddit's problems-at-scale, whilst utterly failing to grasp its own failures-at-inception --- no scale --- and failing to address those. Put another way, how you get to scale, by solving the problems of inception, teaches you nothing about hoe to survive at scale. The problems are entirely different.

As noted at HN, for all its copious faults, Google+ solved this particular problem well. Facebook may also (I don't use it). Microblogging platforms (Twitter, Mastodon, Fediverse) at least present individual posts within a thread well, though they seem to uniformly suck at actual threading (see: Threadreader). Diaspora ... kind of does this but was an immensely clunky slow interface for notifications & response.

But yes, as McLuhan said, "the medium is the message". It has profound impacts and influences, most not immediately apparent -- they're emergent properties.

Independent of medium, scale, expressive richness (e.g., markdown, multimedia), latency, arity, ephemerality / permanence, message size, moderation (leaf-node or trunk), culture, founding cohort, exogenous vs. endogenous motivators and incentives (or demotivators and disincentives), editability/revisability, search, organisation and management tools, protocols and standards, and much more, all matter.

I've discussed some of this at the (rather neglected) discussion of social media types and characteristics at Plexodus Wiki, see especially Platform Types and Features and Capabilities.

---

Adapted from a private Reddit discussion.

#media #conversations #generativity #MarshallMcLuhan #reddit #twitter #hackernews #mastodon #fediverse #diaspora #usenet #moderation #googlplus #gplus #plexodus #plexoduswiki

I am trying to find reliable numbers on how much work in terms of social media moderation is needed if a network becomes at least sort-of-mainstream. If this article is to be believed, you can expect 10 to 20% of all submitted content to be in need of removal.

That seems insane at first glance, but I guess that spammers do deal in volume.

https://digitalsocialcontract.net/what-proportion-of-social-media-posts-get-moderated-and-why-db54bf8b2d4a?gi=cdd73670133d

#plexodus #blocking #moderation

Attention all #newhere!

Thar be #trolls in #Diaspora!

Unlike other social networks like the soon-to-be-deceased #GooglePlus, Diaspora and other federated social networks don't have a central authority to perform #moderation. Mostly that means that there are a few lost souls whose behavior could be described as troll-like. You may find content you don't like, offensive and unwelcome comments on your posts, etc. This is an unpleasant side-effect of Diaspora's design: the platform doesn't need to bow to the rules and regulations of a single government, making it much harder to censor content. (Individual pods are subject to regulations of the government where they are hosted; but the network as a whole isn't.)

I've been trolled! What to do?

You have a few choices.

• You can go to the user's profile page and Ignore them. Ignored users won't be able to follow you or see your posts, comment on your posts, or @mention you in their posts or comments, although they can still comment on posts of others you follow. Like an old flame, there's no way to avoid crossing paths again.

• You can report a post or comment. This is actually a bit tricky to understand. Let's say your account is hosted on diaspora.x.com and the troll has an account hosted on diaspora.y.com. If you report the troll, the message goes to the administrator of diaspora.x.com where they are given the option of removing it from your pod. The content will still exist on other pods (remember the thing about no centralized moderation?).

• You can reach out to the administrator of (in the above example) diaspora.y.com and make the case that the user should be banned. Realize that the troll still may open a new account somewhere else, and several trolls have been known to do this.

This is totally unacceptable!

OK. But that's also the way the system works. You can feel free to make suggestions for improving the system, or better yet, write a fix submit a pull request and make it available to everyone. I've advocated for some kind of machine-learning algorithm to help manage unpleasant content, but I haven't sat down to build one yet.

Thanks for signing up, and I hope you can see past the trolls and make some new friends here!

(Please comment on corrections, opinions, etc.)

We need to talk about "We Need to Talk About TED"

The question: how can one distinguish quality information from entertainment.

There's a fundamental problem with democratic voting processes and voting systems (such as reddit's own post and moderation processes[2] -- which are, in their defense, better than most) in assessing who's qualified to make a judgement -- and then, of course, in determining who's qualified to assess who's qualified.

And there's the question of what to value: originality, relevance, insight, agreement, correctness, humor, entertainment? Slashdot's moderation system allowed for indicating which of these a particular comment fell under, though both the moderation and presentation failed to clearly differentiate and classify among them.

A long-simmering essay finally half-baked.

#TED #filters #noise #moderation #media #infotainment