howell.io

…as unprincipled as the gods, and as much a jack-of-all-trades.

Readings: 22 Mar 2017

Distill is a new machine learning journal, designed to support interactive visualizations that clarify the ideas in their articles – much like how Bret Victor approaches explanations.

Editors Shan Carter and Chris Olah lay out their vision in “Research Debt”:

In research, we often have a group of researchers all trying to understand each other. Just like before, the cost of explaining stays constant as the group grows, but the cost of understanding increases with each new member. At some size, the effort to understand everyone else becomes too much. As a defense mechanism, people specialize, focusing on a narrower area of interest. The maintainable size of the field is controlled by how its members trade off the energy between communicating and understanding.

Of the many factors that led to me bailing out of physics research, I have to put at least some weight on not making it through the transition from textbooks to papers. The former were beautiful, clear, and exciting; the latter required recursively walking the citation graph for three days before you could digest the paper you care about right now.

Even if you are not particularly interested in machine learning, it’s worth browsing through some of the papers in Distill (like “Attention and Augment Recurrent Neural Networks”.) The design and visualizations are beautiful.

Developing good abstractions, notations, visualizations, and so forth, is improving the user interfaces for ideas. This helps both with understanding ideas for the first time and with thinking clearly about them.

The “user interface for ideas” –- what a lovely turn of phrase and goal to pursue.


Over a dozen philosophers came together in Aeon to take the unpopular stance of defending hierarchy.

[…] even the best established practice of medical experts, such as childhood vaccinations, are treated with resistance and disbelief. We live in a time when no distinction is drawn between justified and useful hierarchies on the one hand, and self-interested, exploitative elites on the other.

Equal rights and dignity does not necessitate an equal distribution of power. Certainly well-informed paternalism can shape the common values that help form a civil society.

Good paternalistic interventions, on this view, take two forms. They disseminate knowledge of what is best in forms that are accessible to imperfectly rational agents. And they might habituate individuals’ irrational impulses from an early age such that they later collaborate in the implementation of reason’s prescriptions. Such interventions are justified only to the extent that they ultimately enable us to act more autonomously. That they might is suggested by Aristotle’s theory of habituation, which says that to live well we need to cultivate the habits of living well. Hence, being required habitually to act in certain ways, especially while young, might, paradoxically, enable us to think more rationally for ourselves in the long run.

We are not born good citizens. We are taught, so there must be teachers.