Reestablishing Reliable Sources: A System for Tagging URLs

post by Riley Mueller (rileymueller) · 2022-12-01T02:27:18.629Z · LW · GW · 1 comments

Contents

1 comment

Before diving in, I will preface saying I have never written a post before, and am in general a lurker. However this idea has been itching on the back of my mind for months and I would love feedback on what I may be missing or if this is something to seriously consider as a future project. I used the new GPT-3 model text-davinci-003 to help with this post, first by writing stream of consciousness style, then asking it to organize my thoughts, then having it write out a post.

Trust in the media has been decreasing for some time now, and it’s becoming increasingly difficult to find reliable sources of information. This has led to people feeling disillusioned with the media and its sources, causing them to question the legitimacy of the news that they’re consuming. From the Sam Bankman-Fried FTX scandal to the Twitter fiasco with Elon Musk, it’s clear that the media can’t always be trusted. However, there may be a solution that could help us regain trust in the media and its sources. This solution involves a system for tagging URLs and subscribing to other users’ authority when it comes to certain topics. 

The idea originated from my want for a better tagging system for online fiction. Often times, authors do a poor job of tagging the content they post, and the search engines of fiction serving websites can be inadequate in helping readers find the stories they’re looking for. With this system, users can add their own tags to URLs, and people who subscribe to the user’s account can search for stories using the tags. The solution could also include a system for aggregating tags into dictionaries. For example, a dictionary of Royal Road Tags could contain all the tags (or a curated selection of only reputable Royal Road fiction taggers) that have been applied to fictions on the site. Additionally, users could subscribe to other users’ authority when it comes to blocking tags on certain URLs. For example, if a user finds that another user, such as fiction_connoisseurXX, does a good job of blocking bad tags on Royal Road fictions, they could subscribe to the other user’s authority for blocking of tags that are on Royal Road URLs. The system could even go a step further and allow users to subscribe to other users’ authority settings. 

For example, if JohnieBoy23 is known for being very careful about who he authorizes to block tags for him (since he is subscribed to other people’s authorities on various topics), people could subscribe to him and check the box for recursive tag banning. This would mean that not only would they get blocks on the tags that JohnieBoy23 has blocked, but also the blocks everyone he has subscribed to has blocked. 

The idea was born from wanting better fiction tagging, but the use case for other domains I hope is self-evident. Lets say a user is subscribed to a general news dictionary that isn't closely inspected and just grabs lots of lots of news article urls and has lots of tags from many different accounts and not all of them trustworthy. When searching for news on this dictionary, if a URL for a story showed up and it was tagged as “misleading”, and the user subscribed to authorities who inspect this dictionary, the user can afford to trust the truth of the "misleading" tag as far as they trust the authority of those they subscribe to. 

Instead of filtering on a mixed bag, exploring what trustworthy sources tag as reliable would also be easy. Bastion_of_Truth has been tagging articles from major news outlets with 'Verified Trustworthy', 'Dubiously Sourced', or 'Has Flat Out Lies' for a while now and has proven themself a tagger worth subscribing to. A dictionary of BoT's 'Verified Trustworthy' sources could be made and searched on and a ban list for articles on BoT's 'Has Flat Out Lies' is also doable. I would expect the system to have deep configurable systems for enabling this kind of thing that most people do not end up touching but established accounts that people trust and subscribe to for sources and banning use extensively.

This system would be a powerful tool for regaining trust in the media (Or at least enable you to choose sources you trust and profit from easily viewing sources that your trusted source trusts). It would provide users with a way to easily and quickly verify the reliability of the sources they’re using, and it would also help to reduce the amount of misinformation that is spread. The idea is still in its early stages, and more research needs to be done before it can be implemented. I am interested in hearing people's thoughts on such a system, obvious holes in the design, etc...

1 comments

Comments sorted by top scores.

comment by the gears to ascension (lahwran) · 2022-12-01T06:33:02.731Z · LW(p) · GW(p)

There are many projects like this. A bunch can be found by pasting a paragraph from the post into metaphor with several different ways of ending the query. unfortunately, there are so many that you'd need to use them on each other to get anywhere! of course, metaphor is quite sensitive to phrasing so it really matters how you frame the query. if you ask, you can even get academic work on the topic! though, it's also always good to ask about drawbacks as well. There's also a bunch of great stuff on semantic scholar.

(my comments have been almost nothing but "hey try dumping your post into metaphor" lately, this search engine is amazing. Seriously, pop open each of those links and see which ones you find worth the time!)