Todo:

• clean up code/ comment/ create scripts
• work on roam research pipeline
• build app - probably Flutter Framework
• last updated: 11/17/20

An online reading recommender to eliminate noise

How do you decide whether to read an article or not? Do you get sucked in by a catchy headline? Do you read articles sent to you by friends? Do you stumble across them on your social media feeds? Do you have particular writers or authors whose articles you follow?

Given that time and attention are two of our most limited resources and the seemingly unlimited nature of information today, deciding what to read is becoming increasingly challenging. I find myself more and more bookmarking, Evernote clipping, and keeping tabs open of articles that "I think I want to read but just not right now". Generally I never get back to them, and I think part of that procrastination relates to the information overload/ analysis paralysis when it comes to deciding how to spend my attention.

I want to build an app that, based on the text of an article (ideally just a url), will categorize the article in relation to the corpus of articles I have already read and tell me whether it is worth my time to read or not based on my interests or learning objects at that time. Also, whether the article is similar in perpsective to artilces I have already read, or gives different/new perspectives I have not yet exposed myself to. I think it is absolutely critical to try break free of any echo chambers you may be unwittling a part of, so new ideas and perspectives will probably be weighted more heavily.

I am not currently aware of any apps or software that are available whose sole intent is to find the signal in the noise and break you out of echo chambers. My current understanding of social media algorithms is that unless you curate your social media feed by following multiple disparate perspectives, you are basically trapped in an echo chamber. I honestly do not know if this would be a viable product for the masses, given the psychology of human nature, we don't like what we don't like and we generally do not want opposing ideas brought to our attention. So this product would be for myself, and I guess people who are like minded in the sense that the objective is not trying to protect some particular belief/ value system that I hold dear right now but to get as close to base reality and the truth as is humanly possible.

Initially the data will be articles that I and a few friends have read. I will attempt to set up a corpus and user profile for each individual and tailor the recommender to that.