Double trouble -- our peer-reviewed machine learning/transient paper has also just been published. This is where we clean up the sample and see that the fewer contaminants there are, the stronger the correlation with nukes and the larger the shadow deficit. It also shows that the
