This is the method I thought was a best practice.
They look for inconsistencies along axes in the mathematical space that the simulation would be taking place in. If in fact we are living in a simulation, there would be computational error associated with calculations, leading to less or more error of certain variables along certain axes. If we found larger or smaller standard distributions of observables along particular axes, lines or curves in space, this would suggest computational effects. So far, nothing fishy detected. So either we aren't in a simulation, or they don't have these errors in their computations. We also have a lack of data as we have just started science.