During Timnit's excellent talk (which you can watch below) the question of whether or not the process of statistical inference, and indeed a scientific method, can be objective and I wanted to mention a few points that came up in discussions I had with students afterwards. https://twitter.com/mpd37/status/1154039192927629312

I believe that statistics provides an objective, mathematical _framework_ for informing the decisions necessary to move the scientific method forwards. It does not, however, tell you how to implement that framework and it does not make those decisions for you.

In other words statistical inference is but a mathematical language for building and communicating inferences, but like any medium it doesn't matter what language we use but rather what we say with that language.

Science cannot be purely objective because nothing informs what questions we answer, or more importantly what questions we _prioritize_ trying to answer over others.

We cannot divorce the questions we ask and the lessons we learn from their consequences, even those due to others. Appeals to objectivity just externalize the cost of unconsidered consequences to others.

Moreover the data itself doesn't exist in a vacuum. Data is collected, and that collection has costs. Who decides or prioritizes what data we collect? Who decides what costs are worth which benefits? Who suffers the costs and who has the opportunity to enjoy those benefits?

Statistical inference and scientific inquiry are just tools. Like any tools they have no intrinsic value in of themselves; their utility depends on how they are used. If we use those tools then we can't ignore that responsibility.