You build a code and you run it in order to find the bugs. I’ve been skeptical for a few weeks as the Congressional model was not moving as I thought it should given changes in the environment. I was slow to touch it though, I did not want to introduce my biases.

This morning, however, I decided there had to be a problem, and there was. The generic D v R ballot polls were not updating, so that part of the math for each race had remained stagnate. When I fixed it this morning, a number of races in all the states appeared to change overnight. It was just, however, playing catch up with changes that should have been happening since August.

By noon today, everything should be up to date and reflecting a more up to date view of the races.