I really dislike that replying or commenting on things is seen as a positive indicator. This comment would be seen as positive engagement, I think the "algorithm" shown is fairly useless.

Especially since the weights are tied entirely to how well calibrated the individual predictive models are.