Comment by contravariant | Hacker News Reader

Comment by contravariant | Hacker News Reader

Parent Full thread

I'm not too sure if true neutrality is not the best answer after all. I think these bans are detracting from the real problem, which is that it's Youtube's system for promoting content that is causing problems, not the content on Youtube itself. Youtube has made itself too important a platform to be left in charge of deciding what content people should watch.

Frankly I would feel much more at ease if we could all decide that Youtube doesn't get to decide what content should be deleted, provided they also can't keep recommending videos. Let me explain.

By recommending videos youtube has a huge amount of influence on what content is promoted and what content isn't. However youtube's bottom-line isn't to recommend good content (despite what they claim it is) but it is to maximize ad revenue, which means they need to make people watch as many videos as possible. Superficially this would seem like it should drive Youtube to recommend good content, but it's beginning to look like it is biased towards addictive content, which has questionable quality at best, but might veer towards harmful conspirational theories at worst. It's a classic example of value drift, Youtube has turned into a paperclip maximizer that is beginning to harm humans to produce more paperclips.

The podcast "Your Undivided Attention" gets into the tech and public policy of this. Basically, their suggestion is to modify Section 230 of the Communications Decency Act, so that storage and retrieval is still a safe harbor for providers, but recommendation algorithms should be treated as editorial decisions and be open to the same liabilities that other communications media are exposed to. So a chronological feed of subscriptions to InfoWars, Stormfront, Plandemic, etc would be protected by law, but YouTube, FB, etc could be held liable for harm caused by content that they recommend. People retain their freedom to speak and publish, but they have no right to promotion or amplification.

> recommendation algorithms should be treated as editorial decisions

That's actually a great comparison. I wonder if it will ever gain traction.

That sounds interesting, I'll be sure to check it out, thanks.

But this favors one recommendation algorithm (time based) over others. What makes ordering by time non-editorial?

Laws don't have to be philosophically pure. A law can define or describe what counts as an editorial algorithm and what does not, and it could exclude simple sorting while including the complicated thing Youtube does to keep you watching.

What might be harder is to distinguish sophisticated search algorithms from recommendation algorithms. There's some overlap, but I'm inclined to think that the two are mostly different kinds of things, especially if the search results aren't personalized to the user.

Indeed, but (at least in the US), you can't really do that without running afoul of the first amendment, presenting content is just a form of speech. Government restrictions on speech are the thing that everyone's afraid of, right?

Currently, platform immunity means the platform can't be sued or criminally charged because a user hosts something for which the user can be. The proposed change would expose the operator of an editorial algorithm to liability for the algorithm recommending things that are already illegal.

I don't think the concern with rolling back platform immunity is government restrictions on free speech, exactly. In the US, the first amendment provides the same protections it always has. Instead, concerns include that the risk of liability will hamper innovation, that only companies with a lot of resources will be able to take the risk, and that companies will overreact with their own proactive censorship to stay well clear of areas with potential liability.

We've seen the latter with FOSTA/SESTA weakening platform immunity, which is arguably what its creators intended, but I suspect not what everyone in congress who voted for it expected.

Right, but the "platform immunity" thing applies equally to all platforms. You're instead proposing to extend immunity to some platforms and not others based on how they choose to recommend things. If you (like I do) believe that recommendation is a form of speech, this creates government favored classes of speech.

To word it differently: "Platforms that recommend/aggregate speech in government sanctioned ways are protected from liability, while those that do not are not". That's, while not open and shut perhaps, certainly spooky.

The law already does something like that. If I run the hypothetical blog host conspiracy-blogs.com, and I manually select ten posts I like to put on the front page every day, I can be prosecuted if I select a post for that saying something like:

Tomorrow at 10 AM, join me on the steps of the capitol. Bring your gun and a sign that says "we will shoot you unless you vote no on bill 123.".

The action described is a crime, soliciting others to participate in it is a crime, and my deliberate choice to show it to a larger audience is likely enough to subject me to accomplice liability that virtually any criminal lawyer would recommend against it. If violence actually occurred, there would be an ever lower bar for the victims to clear to sue me. Planning crimes is unambiguously not protected speech.

The proposal described would subject any sufficiently editorial algorithm to similar liability.

> recommendation algorithm (time based)

Can't any choice (not being purely audio, why the title is only in X languages, etc) can be considered editorial, for any TV news show or book? It isn't constructive to construe FIFO as editorial when it is the common way to experience and share events. Since content creators ostensibly retain control of what is published, they could ideally rearrange their content in their feed by re-pub or out of order publishing anyway.

> Can't any choice (not being purely audio, why the title is only in X languages, etc) can be considered editorial, for any TV news show or book?

Yes exactly.

> It isn't constructive to construe FIFO as editorial when it is the common way to experience and share events.

In what context?

> Since content creators ostensibly retain control of what is published, they could ideally rearrange their content in their feed by re-pub or out of order publishing anyway.

This sounds horrendous. You want to encourage content creators to spam you with content so that they're always at the top of your "feed"?

> In what context?

Any episodic context, like 99 percent of the media we consume. TV show episodes happen in a specific order of events, movies are constructed of scenes in a specific order, community growth around a channel is built around common experiences already shared, even looking at history channels, if a story spans multiple videos requiring multiple media releases, when each video relies on information provided from another, the content creator would have to be daft to release them in the reverse, or worse, a random order, when everybody is waiting to find out what happens next.

From my perspective, it seems like the only videos that are viable outside of this context are long-form videos (1hr+) or viral trash.

> Any episodic context, like 99 percent of the media we consume.

But you're suggesting a single "timeline" for all content. So you get one stream of every show on every channel in order. That's not the usual way we watch episodic content. Hell since the invention of the VCR, we mostly just watch the specific stuff we want, not everything in I ur bundle.

You seem to be conflicting the content made by a single entity with the aggregator. For an aggregator, why does time name more sense than any other view (like say clustered by creator, or as is common today ordered based on presumed interest)?

I want to experience media in the format and order in which the creator of the media wants it to be watched. The episodic content that I watch, I watch because I have faith in the creators of it that the next one will be of the same calibre as the previous. Not faith in the publishers. 99% of the time, this is in the order or is released. There are exceptions, which is why we have the concept of creator-defined playlists. To present the content to the user in an order which is not defined by the creator is at the publisher's discretion, and the publisher should therefore take responsibility for presenting it that way.

WRT conflating aggregators and creators, any decision the publisher makes, they should take responsibility for, such as promoting some content and hiding others.

> This sounds horrendous.

Any "editorial algorithm", by the standards you have put forth, can be abused.

I don't disagree. But what you (or the prior posters) are saying is that the law should prevent companies from trying to optimize for the best user experience. If only showing things in time-order is "safe" for the company, but filtering out spam becomes "unsafe", then I as a consumer am forced to endure a crappy experience for solely legal reasons. Neato.

I really love the sound of this idea.

This is a great point and is also true of Facebook and a number of other platforms. Roger McNamee spoke at length about this issue with Sam Harris on his podcast back when it was called Making Sense. His point was essentially that the AI that controls the recommendations is trained to maximize clickthrough rates, and so it learns in its own way to feed us articles that make us angry, because we click and share more that way. Hence the tendency to lead people down rabbit holes of extremism.

Reply on news.ycombinator.com