While it’s all technically open data, what we’re seeing is Professor A uses a subset with a model and make a prediction. Professor B use a different subset, use a different model, make a different prediction. Professor A ends up being more correct, but we never know whether it’s because of the data or the code. [laughs]