We don’t forecast elections on vibes. We do deep reporting on the ground, we have long conversations with ordinary people, we don’t just ask them all the same 25 questions. We are right as often as you are, and even when we are wrong, readers gain insight, unlike with a bad poll.
It's funny how people who "forecast" elections on vibes can't comprehend using an algorithm. Like maybe it's wrong; lots of design choices in building a model. But we do our thinking when building the model. Once it's built there aren't many decisions to be made from day to day.