I picked up a cheapo duplicate copy of Poor Economics by Banerjee and Duflo, bought from a bookstore in Kochi when I went to a friend's wedding.It feels like I understand the mechanics and economics of poverty much better now. My favourite chapter was the one about microfinance (aged well, nail on the head), and the one about poor people and their relationship to entrepreneurship.Much of the book is explaining behaviours of the poor that are incomprehensible to people (especially WEIRD people) who haven't actually talked to or known much about poor people. some of this is backed by randomized controlled trials that won them the (economics) Nobel prize.Examples of these behaviours:When the very poor get more money or subsidies, they don't maximise calories per dollar or spend more on necessities, they'd spend on tastier calories and entertainment, and this is a very rational behaviour given their circumstances.The poor would rather spend on cures and injections and expensive healthcare things, rather than cheap preventive things that require follow-ups (like immunization or micronutrient supplement/fortification).In general, I feel like this whole RCT methodology is a narrow frame to look at the problem and selects for a certain way of doing things. Thankfully they do supplement it with ethnography and interviewing of subjects.It's also been ~15 years since this book came out came out so a lot changed. Which also means I could do a lot of digging with my AI to "follow-up" on this book essentially. Few things that stood out to me:The book focused a lot on India, Indonesia and SEA for its research generally, and not too much on Africa. Right now most extreme poverty is basically in Subsaharan Africa and war-torn countries, not so much in India and SEA generally. Unsure how much of the findings will replicate/transfer.India's poverty dropped from ~27% when the book was out to something like ~5-10% (lots of debate over how much reduction, because of measuring methodology changes, etc, but no dispute that it has reduced).The book deliberately champions small changes and microeconomics stuff, and leaves out macro stuff. This is good overall, because it's a lot more actionable and small changes do add up and inform policy. However, from digging up data, it seems like the macro stuff is what really has driven poverty reduction in India rather than the micro-interventions in this book. Specifically from pulling up stuff from microdata.gov.in and other such sources, it appears to me that rise in wages (due to there being more jobs and non-agricultural jobs) and more women in the workforce explains at least three quarters of the reduction. The micro stuff seems to have mattered a lot more in subsaharan Africa with respect to stuff like bednets/malaria etc. Unfortunately, since 2025, foreign aid that made this happen has basically been cut off.The psych/behavioural economics explanations have aged poorly due to the replication crisis, though their research didn't rely too much on these. It was used more to support the explanation for their findings.UPI and cheap mobile money transfers is a thing now, which it wasn't then. So a lot of intervention options have opened due to being able to send money at scale, and "is this better than just wiring in some cash" is a default baseline to measure any kind of policy intervention against.The worst-aged parts are the enthusiasm for Aadhar (always referred to as ~"Nilekani's ID project"). It was not even launched yet, and the authors were hopeful that this would reduce subsidy theft. Extremely strange, given the authors avoid speculation and only speak based on data for just about everything else. Leaving aside all the privacy/surveillance stuff, I looked at whether it actually reduced subsidy theft, and the answer is "kind of." The thing is, it isn't clear that this reduced theft more than what any other normal kind of system of record/non-biometric digital verification might have, and the cost of operationalising Aadhar has been very high.