><i>This is an important question. This is perhaps the most important question for society to answer today. Attention translates to influence. What people pay attention to influences what they believe. What they buy. How they vote. What they fight for. That’s why autocrats try to control the media. It’s why companies pay billions in advertising.</i><p>Good luck. When I have talked to people about advertising in general, I've found that <i>they just don't care</i>. Apparently it is a very special mode of thinking where you consider your own attention valuable, attach anything more than 0 importance or cost to watching a message, worry about getting influenced (people will quickly think you are crazy when you choose this tack)... For sure I have found some people here and there who are sympathetic to this way of thinking, but if you want a wider debate about these sorts of issues I fear the first step is let go of your assumption that you can get people to consider your arguments at all if you'd just explain your reasoning clearly enough.
Paying attention to the same things as everyone else seems like a bad idea? It means a bunch of non-experts paying shallow attention to things they probably aren't going to go into any depth on and will forget about within days, or it can result in unproductive doomscrolling where everyone worries fruitlessly about the same things.<p>Modern civilization is built on specialization. We don't have to all worry about the same things. We can pick and choose, and the other stuff is Someone Else's Problem.<p>This is why there are specialty sites, like forums and subreddits and Hacker News. However, even Hacker News is awfully broad, showing people lots of stuff they probably don't care about and aren't going to do anything about.<p>RSS feeds and subscription-based following can help if you're careful about curation, but following <i>people</i> has the downside that many people don't stick to the subjects they're experts in and will often repost stuff you don't like because they don't have similar tastes.<p>And this means having custom filtering would not be a bad thing if the providers of the algorithms could be trusted.
Interesting little read.<p>"Engagement-based ranking, use by many social networks, is even worse. It concentrates attention on controversy, outrage, and out-group animosity."<p>It does concentrate on those things!<p>Perhaps a sad fact for humanity is that "our collective participation in public discourse produces the shared values and narrative that enables collective action toward common goals" - and maybe unconsciously, those common goals are controversy, outrage and out-group animosity.<p>Maybe social media really just exemplifies a really ugly side to human's that has been enabled by the internet. Perhaps, through the evolution of our species, we have been hardwired for negativity, and the internet just makes it a lot worse.<p>Just a thought.
> Social networks and online forums, as the public square where public discourse takes place, are public goods that should be independent of the whims of politics.<p>This is possible if everyone would agree to some reasonable ground rules… which they won’t.<p>For one thing, the public square is <i>about</i> politics. No matter what your rules of discourse are, they will allow some narratives to flourish, and since the goal is to dominate, this is necessarily at the expense of other narratives. Which means your rules are inherently political (and therefore, subject to debate), whether you want them to be or not.
The marketplace of ideas relies on everyone in it being some kind of purely rational actor who seeks truth. In reality, it can be to your benefit to believe something that isn't true, to get an out-group punished and yourself elevated. People will change what they believe or what they say the believe based on who is in the room with them. You have incentives to believe in certain things. This is going to affect what you choose to post yourself as well as what you upvote.<p>Attempting to measure, in real time, the validity or the popularity of an idea provides a perverse incentive. I honestly believe there is simply no way to do this.<p>Maybe the closest is by returning comments in random order on every load, and hiding all likes/upvotes on any comment for a set period of time. Even returning comments chronologically gives precedence to ideas that occur often, even if they are posted by the same user. This is however not a "fun game" and unlikely to be successful as a site.<p>Even with this, I think you would still end up with sites having a definite lean one way or another on certain issues, as users mentally total up they kinds of comments that reached the top after the results were revealed and users self-select into or out of the site.<p>So probably the "best" way is just random, always, and never show any rankings at all. People are free to rank the comments mentally based on their own incomplete information, potentially poor reasoning skills, and their own incentives. But that's the best you can do.
What deserves our attention is a deep question that is important personally and societally.<p>I believe that it can help to use the global ethics "capabilities approach" by Martha Nussbaum:<p><a href="https://iep.utm.edu/ge-capab/" rel="nofollow">https://iep.utm.edu/ge-capab/</a>
Recently I read somewhere that people prefer "potential" to "factual value". For example some prefer a potential Oscar film rather than a film awarded an Oscar. So perhaps the new algorithm for fostering true telling and honestity is base on the potential of people to construct the new rules not to adhere to them. Perhaps this is the NIH syndrome at work, or just that people prefer rediscovering facts than learning facts. But rediscovering requires a very good teacher and a lot of time.
<i>> For example in an online forum, the most up-voted posts may be shown on at the top of the page. This rule concentrates attention on popular content. But this is a terrible rule</i><p>It is a terrible rule indeed. It also serves mob censorship of opinions, especially when there is "downvote" option.