My year in review post. 2020 was a terrible year, all things considered, but I found opportunity to rekindle my reading habit and write consistently. Inside are reflections on personal and career growth, the development of my research skills, and some of the best things I read this year.

I did some cleanup of the categories and tag listings. This short post discusses why and how!

Happy new year! This week’s edition is short, featuring a post on Apple’s new M1 chips and why they’re so fast, and another on boundaries within engineering orgs, especially “social good” ones.

In August, I started a reading club with my friend Judah. 17 papers later, this post reflects on some of the things I’ve learned.

How can we estimate suicides in real time when official data sources can have data lags of over a year? This paper, by researchers from Georgia Tech, the CDC, and the National Suicide Prevention Hotline, uses machine learning to model and forecast weekly suicide fatalities in the United States.

Users on online platforms might contribute to a few communities often or many communities infrequently. Can we quantify this dynamic, or quantify how similar communities are? This paper, from WWW 2019, attempts to do this through community embeddings.

How did the early months of the COVID-19 pandemic affect expressions of mental health on Twitter? This paper investigates. Through comparisons between 2020 and 2019, and clinically validated machine learning classifiers, the authors study changes in language on social media.

This study, from the NeurIPS Navigating the Broader Impacts of AI Research workshop, is a field experiment in algorithmic bias. Take 400 programmers and ask them to predict math performance; in what ways are their algorithms biased?