What is it?

News Minimalist is an AI-powered news aggregator and curator.

It uses ChatGPT to read the top 1000 news stories every day, and rank them by significance on a scale from 0 to 10. This lets you read only significant news without wasting time on junk.

It also removes clickbait from titles and gives a short summary of each article.

How is significance score calculated?

This is done in two steps:

  1. ChatGPT reads the news story and evaluates seven factors:
    • Scale: how many people the event affected;
    • Magnitude: how big was the effect;
    • Novelty: how unexpected or unique was the event;
    • Immediacy: how close in time is the event;
    • Actionability: how likely it is that a reader can act on the news for personal benefit;
    • Positivity: how positive is the event — used to fix media negativity bias;
    • Probability: how likely it is that events speculated in the news will actually occur;
    • Credibility: how credible is the source.
  2. The final score is calculated as a weighted geometric mean of all the factors. The weights are not constant and I adjust them based on user feedback.

Isn't significance subjective?

I separate significance from importance (or relevance).

Importance is subjective. News about the health of my family members is important to me, but they are not significant to the world.

Significance is objective. It's about how much the event affects the humanity as a whole.

Why is there so much news on {topic}?

I got this message about different topics at different times: COVID, war in Ukraine, climate change, US debt ceiling, etc.

I see several factors that contribute to this:

  • "Hot topics" are more likely to be covered by multiple sources. The more sources are covering the topic, the more variance there is in the description of the events, and the more likely it is that the topic will make the top.
  • Even with the totally random topic distribution, there are going to be stretches where the same topic is covered multiple days in a row. I just used a random number generator to emulate 10 rolls of a dice. The numbers I got are: 6 2 6 4 6 1 2 5 1 2. Notice how number 6 is rolled three times in the first five rolls. Even total randomness does not guarantee variability. And journalists don't write on topics randomly, which means the streches are going to happen more often and will be longer.
  • Some topics are just more significant. ChatGPT doesn't have a topic fatigue, like people do (remember how everyone got tired of COVID news at some point?). If something is more significant, it will give it a higher score consistently and indefinitely.

I understand that this leads to less interesting feeds. But that's the point — we already have algorithmic feeds that entertain us. I want to build a feed that informs us.

Is there an RSS?

Yes. There are two feeds:

  1. Basic feed is free. It is basically a daily newsletter available via RSS. It has news with significance score over 6. That's ~3 stories per day on average.
  2. Personal feed is a part of premium plan. It allows choosing news categories and individual significance thresholds for these categories — for example, technology 5+, sports 6+, etc.