Regularisation and logistic regression

What happens when you add more features than the data can support, how a penalty on coefficient size turns an exploding model into a stable one, and how one...