How should we evaluate progress in AI? David Chapman - Meaningness How Property Taxes Shape Our Cities Connor Nielsen - Strong Towns
This essay is full of questions about the philosophy of science vs engineering vs design. It tries to explain where AI fits in, and discusses the false certainty that powerful tools like these give us. Machine learning/artificial intelligence/deep learning/data science/call it what you want is full of pitfalls because it is so powerful. There’s not much actionable in it, but as someone who’s been pushing the story of automated retraining and continuous deployment to squeeze out performance out of ML models, it certainly gave me a lot to think about.
Complicating the Narratives Amanda Ripley - The Whole Story
In Seeing Like a State, James C. Scott writes: “By a kind of fiscal Heisenberg principle, tax officials transform that which they take note of.” In this essay, Nielsen shows a few examples of this - from the roofs of Paris, to the facades of Amsterdam.
Authority Steve Randy Waldman - interfluidity
This article, recommended by Mike Davidson, discusses many ways we could fix journalism. Basing its recommendations on a bunch of communication and psycology research, the author tears apart a bunch of different situations in which the conversatin could have been nudged elsewhere, allowing for more productive outcomes. There’s a lot to unpack, and I was not expecting it to be so long (should’ve noticed the “39 min read” at the top!), but totally worth it. It reminded me of a tweet by Mason Hartman and the ensuing conversation: “When you ask a really useful question, most people will consciously or unconsciously try to determine if they have a cached answer that roughly fits what you asked. Getting people to consistently forget their lines is a skill worth mastering.”
The Outrage Epidemic Russ Roberts - Medium
This piece reminded me of Nick Szabo’s things as authorities, and how much of our interactions are intermediated by ideas that are tacitly embedded in culture. Society is built on the fact that we can agree without burning mental cycles on things. We encode accepted behavior in culture and technology, relying on those crystals of knowledge to abstract away complexity and layer even more complexity on top. We offload our question of who should go ahead first to traffic lights, and our questions of who must pay back for their actions to courts and judges. If we think of governments as platforms (an analogy from Tim O’Reilly’s WTF), we can see that government’s role is the construction of and upholding of these mechanisms that legitimize sources of authority. The platform coordinates behavior to reliably improve the relation between the individuals who build on top of it, and it does so by setting the rules of what’s admissible and what is not to the point that we don’t need to stop and think about them.
The Harsh Reality Of The Preference Stack Semil Shah - Haystack.vc
Also available as a podcast.This is not a new argument from Russ. He posits that our newfound tribalism isn’t all that new, and that it’s simply been exacerbated due to the incentives of the media industry, the filter bubbles of the internet, and the availability of content. He uses restaurants to illustrate the explosion of available choices, and contrasts how this explosion isn’t all that meaningful when choosing what to eat or shopping for shoes, becoming problematic only when there are high externality costs without a feedback mechanism that makes us pay the price from our mistakes. If we buy an uncomfortable pair of shoes, or order a bad plate of pasta, we suffer. Immediately. If we vote for the wrong candidate, our contribution to their election is minuscule, and their actions are diffused over many years - there’s no clear-cut feedback loop. This dynamic means that the ROI on truth isn’t all that high for any one individual, and in aggregate, we end up with a market failure in which media is louder and angrier, as it sells more than nuanced positions would, and each one of us becomes more entrenched in our beliefs, disconnected from each other. Let’s all try to be more nuanced?
Tech in a time of travesties Jon Evans - TechCrunch
Every time one of my friends has asked me what I think of the startup they’re about to join, I’ve said that they should discount the equity as if it were worth $0. The probabilities work out that way, and given that most of them have other offers on the table, its the rational thing to do. Semil’s post discusses the employee side, as well as the early investor’s side - one I had not considered before. If you want to read more on this, Dan Luu’s piece on startup vs. big co tradeoffs is wonderful, pushing the point that all things being equal taking the megacorp offer is a no-brainer for the employee. As someone who wanted to be in startup land and ended up at HugeCo by accident, I mostly agree with him.
Comparing City Street Orientations Geoff Boeing
Technology does not automatically make the world better. Most tools can be used for both good and bad. The solution to bad outcomes due to technology is not more of the same technology.
Solving Rush Hour, the Puzzle Michael Fogleman
What kind of patterns arise if we look at the street grids in various cities? It’s not all orthogonal cross streets. Open Street Map data, and the tooling to manipulate it, seems to have evolved a lot since I looked at it a couple of years ago. Big props to Boeing for open sourcing OSMnx and sharing so much of his work. Maybe it’s time for a side project…
What determines value? Dietrich Vollrath
Say what you want about 10X engineers being a fantasy, but I’m convinced Michael Fogleman is one of them just by looking at his open source side projects. Here’s yet another super cool project from him that you should check out.
Kelly on Technology and What Technology Wants Russ Roberts - Econtalk (Podcast)
Value is a mindbending idea once you notice that it’s not tied to price, and that prices themselves are just numbers representing the opportunity cost between various goods. I was surprised there was no mention of supply and demand - this is often overlooked when discussing labor markets (ie, bankers don’t make big salaries because they produce that much value for the companies they work for, but because that’s the price the market bears). Culture might be different at facegoopplemazforce, but even if we’re not measured by “butts in seats” time, people value face time, and track who’s in first, and who leaves last. The input/output problem is pernicious in the org behavior experiences of office life, a la Dilbert. Lastly, the third section in the essay made me think of Russ Roberts’ recurring point that GDP per capita is a bad measure of our standard of living, and that we need to figure out another way to measure our progress. What are some realistic alternatives? Are there any?
My Journey in Photography Sam Abell - Jordan Schnitzer Museum of Art
I’m still going strong on my quest to fight recency bias by going back to old EconTalk episodes. I’m doing this partially to get away from the current “everything is politics” environment, but also to unlock knowledge that’s trapped in the recent past.
How Trump’s Trade War Went From 18 Products to 10,000 Keith Collins and Jasmine C. Lee - The New York Times
Most good photographs seem inevitable, almost incidental. Each one takes thought and effort. Hearing Sam Abell walk through his process was wonderful. I’d recommend this whether you do photography or not. What a great recommendation from Chris Michel. I took a few notes:
- Low angle, more impact
- Strong graphics, S curve, powerful diagonals
- Bad weather makes good pictures
- Compose, then wait
- Setting, expression, gesture
- All photographs are about time
- Put your people above the horizon line
- Perhaps most applicable to other aspects of life, on completing projects: you will only see the mistakes - learn from them, and don’t point them out to others.
Trade wars are good and easy to win. Really cool economics data viz work from the NY Times, however sad the topic.