Show HN: iffy.app – How (much) to care about politics
iffy.appI built https://iffy.app over the holidays to help navigate (US-centric) political discussions with family.
The key feature is telling you how many hours to spend on a given political topic in a calendar year, based on statistical relevance, potential policy solutions, equity considerations* (see below), and optionally personal relevance.
For each of those categories, a score 0-10 is generated based on how relevant the topic is to the category. Each category will show a description for how the weight was calculated, and 5 relevant facts are linked with references from Google search APIs to get started diving into the topic. You can also ask AI follow up questions about the analysis if something’s not clear.
The weights and textual analysis are generated with OpenAI’s GPT 4o model, but the hours calculation uses a deterministic function based on input weights. There are some other strategies used to make output consistent for a given input, but I don’t want to get into that too much here; see https://iffy.app/#methodology for details, intrinsic biases, and possibilities for improvement!
I’ve already found this useful as a structured approach to political topics ranging from conspiracy theories—weighed low for a left-leaning person like me without much personal relevance—to universal healthcare, which I should allocate a lot of research time for.
Hope this is useful or at least interesting! And that it doesn’t go down (I have hard caps on scale/OpenAI costs/Google API costs for now).
*Equity calculations are there to balance out topics that would otherwise be weighed lower due to limited statistical or policy relevance. However, the equity score is weighed lower for right-leaning users, neutrally for centrists, and high for leftists.