### more\/what code does and doesn’t do Code here is short for algorithmic rationality: gathering data and making judgments by following strict rules. The appeal of "data-driven" and algorithms is that they promise fact-based, objective, and fair decision-making by removing the human factor. Unfortunately, this is true only in very limited circumstances because data does not grow in the wild, waiting to be gathered. It takes its meaning from theory, and theory contains judgments about what matters and what doesn't. In that sense, it is theory that generates the data. This is well-understood in science which is a process of distinguishing between assumptions (existing theories that are built into the meaning assigned to the collected data) and the narrower questions of a new theory tested against that data. This is why every scientific theory comes with qualifiers, why choices and judgements are subject to much debate, and why even the best theories are eventually superseded. Algorithms are a specific type of theory that depends on standardized data and hence captures only those aspects of the world that have been rendered algorithmic. This process of datafication of the real world obscures the qualifiers of the theory. Far from the promise of being fair by removing the human factor, algorithms are powerful tools that give good information only under strict oversight that requires skills usually cultivated through craft.