Modern life confronts us with a uniquely large and diverse set of risks.

@renatofrey & Fischer mapped 100 of them using a large Swiss survey, collectively a rich, comprehensive inventory across life domains, superior to conventional selections:

https://journals.sagepub.com/doi/full/10.1177/09567976251384975

People sometimes say they do not have time to do open science. But they can use the time they would have needed 40 years ago to send and request paper copies of articles to read the literature.

Somehow we never acknowledge where we gained time that we can invest in open science.

the outcry on Bluesky over the 1 million Bluesky post Huggingface data set for ML training seems like a perfect illustration of how people’s desires regarding privacy and consent are not just at odds with what platforms give them, but also with *their own actions* (e.g., the consent to terms/conditions they regularly give).

This book makes a strong case for why the current consent-based approach to privacy and data harms is fundamentally inadequate, and we need very different legal tools

last comment on the Bluesky post Hugginface ML training set issue. For those querying whether this has “the consent of the Bluesky team and Jay Garber”, the Bluesky preprint from the beginning of the year explicitly mentioned network wide *sentiment analysis by companies* as a use for Bluesky data…

sentiment analysis *is* machine learning/AI.

…Bluesky is designed for this