Fluffaway II

Almost any comment section on a polarizing topic includes a huge number of people supporting or decrying the issue in question. These comments can largely be ignored, as they're guaranteed to have been said before elsewhere and are often redundant. There are interesting arguments to be read, it's just that they're buried in annoying fluff. This algorithm uses an AI to read a whole comment section. It identifies highly similar comments and compresses them into the bare statements.

Let's take as an example an argument about glass vs ceramic bowls, in a world where that's controversial. You'll have 800 people saying glass is better and 500 people saying ceramic is better. This part of the conversation can be easily compressed into those two statements. 200 of those will add information, such as the material properties of glass or the aesthetic properties of ceramic. These can be compressed into a bullet point list of the properties in question. (truth or fact of statements is a different question, we're summarizing here, not trying to extract facts) Finally 10 or 20 comments will include unique arguments and perspectives, or more extensive analysis not seen elsewhere. The large number of personal insults and references to mothers and ancestry can be omitted.

So for the above comment stream the output would be: 800 people say glass is better. 500 people say ceramic is better. Here is a bullet point list of the properties of each. Here are the comments with more useful information. Then the summary of the conversation can be read without the annoying fluff and bias in a fraction of the time.

Modern AI is fully capable of doing this, but if you don't believe me you may envision it as a website with human editors doing the same. Or aliens, whatever tickles your pickle.