Compare Apples to Apples, not Oranges

This is what Brian Machado is good. When running experiments, we should compare variables in isolation.

This is the most difficult thing to do with robotics. Because there are so many variables.

When you are troubleshooting, you MUST compare apples to apples. Have one thing fixed in place.

For example, when you are comparing two algorithms, you must compare them on the same dataset. Because there are other variables in place.

Annoying thing

The annoying issue is, one algorithm might work better in one dataset, and the other algorithm might work better in the other dataset.

At that point, you need to figure out why that is the case.

Basically The Scientific Method?