JARED KNOWLES.

I am so interested in attribution science and its applicability to social science as a complement (and perhaps corrective) to the obsession with isolating causality in a complex and multi-causal world. Great rundown by Ars Technica.

Pretty in-depth reporting here on what the near-term incentives are for Iran in the war with the US. Not as cut and dry as I thought and I appreciated the depth and breadth of sources in this one.

This is a really cool explanation of Pareto efficiency (which sounds boring I know) using interactive data viz and clear storytelling. It’s a great example of how a concept becomes more approachable when transported into a familiar domain in a friendly and thoughtful manner.

I really enjoyed this entry inHadley Wickham’s series on LLM coding tools. It is a level-headed look beyond the hype and hate at what works, what doesn’t, and working through conflicted feelings about how LLMs are changing statistical software practice.

This is a detailed look into the computation behind weather forecasting and how climate science is being integrated. The layering of models and the choice between ML and equation-based modeling always fascinates me. Social science has a lot to learn from these approaches.

This is bad. I use open models all the time - for simple coding, chores, and sensitive data. The open models are really good and super efficient. This move to squash them with regulation instead of competition is the proof. Innovation should be done in the open, like with OSS.

My old graduate school roommate, Daniel Walters a professor at Texas A&M, is co-author on this excellent article about the eroding independence of federal agencies and the opportunity it creates for more executive overreach and erosion of trust in the government.

A really excellent essay about how high-risk, low expected value acts like gambling, trying to go viral, etc. are a symptom of how the incentives have completely changed in the K shaped economy.

I have been interested in this idea and this article introduced me to the concept of “Answer Engine Optimization” or AEO. Unlike SEO, right now at least, it seems very easy to move and adjust. And signs are this is rapidly becoming the way to reach people…

Brian Phillips nails it. From an entertainment perspective, VAR is an abomination in almost every sport. It has gone too far. I hate waiting 20 seconds after any meaningful action to see if the referee will make or reverse a call that changes the game.

The term “understanding” does a lot of work in thinking about code and data science. All code is an abstraction. And, abstractions embody a balance of tradeoffs between competing concerns like readability, performance, and maintenance costs. This post lays that out really thoughtfully!

Consumer sentiment at an all time low. Corporate profits at an all time high 15% share of GDP. Employee compensation below 10% of GDP. Prices soaring, choice falling. Something has to change. Freedom and liberty can still be restored.