twitter / the-algorithm

Source code for Twitter's Recommendation Algorithm
https://blog.twitter.com/engineering/en_us/topics/open-source/2023/twitter-recommendation-algorithm
GNU Affero General Public License v3.0
61.99k stars 12.15k forks source link

Free Speech and Hidden Content, Improving Moderation and Scoring Systems on Twitter X #1839

Open mikedsayre opened 1 year ago

mikedsayre commented 1 year ago

Is your feature request related to a problem? Please describe. Users interactions and engagements are being "managed" automatically by the Twitter Algorithm or by an individual, moderator, and in extreme cases of content or information (like pornography or illegal matters), which is required. With the exception of extreme case users, the visibility of a User pertaining to their content is not being handled as "Free Speech" yet. A Users ranking in tweets, replies, interactions, and engagements are hidden, or downgraded by a scoring system by Twitter. This scoring system is not sustainable, constant management and tweaking of this system will be needed and the owner(s) of Twitter will decide this. Elon Musk publicly defends free speech, and claims Twitter will be a free speech platform for the masses. This will cause constant user frustration over an ambiguous system and eventually push users away to find other platforms to spend their time on, or spend their time building a following. When will this Twitter free speech platform be completed?

Describe the solution you'd like I would propose a public scoring system of some kind similar to Community Notes, but much quicker in nature. Maybe a hybrid upvote, downvote system with Tweets and replies, one score or vote per verified user. Some users prefer to ghost content and information but would participate if their scoring actions weren't public, and one vote per user could remedy this and be easily managed with verified accounts. Think Reddit voting and scoring of posts and replies, but with verified accounts only. This would eliminate manipulation and hogging of scoring and sentiment by unverified accounts, and shill accounts as it happens now on Reddit. Advertisers (and news or information outlets) could be managed this way as well, and gather important data regarding their product or message without force feeding the timeline with unwanted ads and better align their message and product with specific users in a granular fashion. As it is now, ads are force fed based on a users interactions and content consumed, which isn't entirely accurate. Most users go to Twitter to find entertainment of some sort, let them eat cake if they need it, feed them their medicine when they want it. You can't force an action or ads on users, build the desired product or service and let it be what it is, let the truth be recognized immediately. As it should be for all information and content on Twitter, if it is to be truly Free Speech. Government requirements would be met, and they would also be able to gather vital intelligence of the public sentiment and beliefs and better align their messages and information to educate or disseminate those beliefs as a countermeasure with advertising or campaigns in entertainment or news.

Describe alternatives you've considered Facebook and Reddit have public scoring and sentiment systems, as do other social media platforms. I would hope Twitter could be the platform of choice for the masses in this regard, and build a easy-to-use public interface which can alleviate the constant algorithm updating, and hidden rankings of a users content eliminating mistrust and frustration built around free speech. if this continues as is, the future of Twitter will be labeled as a managed platform by Elon Musk and drive away users.

Additional context The heart button could be revamped as a popup scoring system, similar to Facebook's system, however this is just an initial proposal to get the ball rolling on better ideas. Considering Twitter's future as an everything app called X, I would presume artificial intelligence could be a key component of all this as well, although this idea could be an entirely new thread in itself.

TolstoyDotCom commented 1 year ago

Most people realized direct democracy doesn't work centuries ago: it becomes ochlocracy. Not too long ago Occupy Wall Street tried direct democracy and we see how that worked out.

For a tangible example, go into a pro-X community on Reddit and, when they have a deceptive post about X, post how it's wrong and see what happens. You'll get downvoted to oblivion for telling the truth: that's direct democracy.

The only solution is a judging system where there are a limited number of judges, their votes are public, and everyone can comment on their votes. If Simon Cowell were tempted to vote up someone who sings off-key, he wouldn't because he knows he'd face public criticism.

Of course, Twitter couldn't be trusted with such a system. Twitter employees - from the top to the bottom - are grossly incompetent and have no moral character.

mikedsayre commented 1 year ago

You are probably correct regarding direct democracy, even with verified accounts, I could see some piling on and manipulation occurring.

I'm trying to post my concerns because of the hidden content and engagement factors currently which are extremely frustrating, and I'm sure for others trying to build a following or simply interact. Currently there isn't any reason or idea of what caused the downgrade, or any instructions for the guard-rails to follow with posting content or replies. It's a crap shoot, and Twitter will lose users with this oversight, or maybe this is their intent?

Would you have a better sentiment or scoring system than a public voting panel?

I ran this question through Bard AI and here's what I received:

And the most efficient methods: