1. Genomic Prediction of Social Mobility. New paper using 20k individuals predicts SES mobility with genomic scoring. In English this means that by looking at your genes, you can predict, at least to some extent, who is going to rise and fall in socioeconomic status. The study included predictions of siblings (since children have different mixes of their parent’s genes) and they were still able to do predictions. This is a powerful test, as siblings share home environment. Stephen Hsu is saying “game over“. But it’s probably more accurate to say game started. See for example Eric Turkheimer pointing out the fraction of change predicted is small. But these kinds of studies will get more powerful as sample sizes go from tens of thousands, to hundreds of thousands, to millions. The train has left the station. Though it will take time for all the implications to sink in.
2. More Links Between Alzheimer’s and Herpes Virus. Old theory: amyloid beta plaques in the brain cause Alzheimer’s. New theory: amyloid beta is the body’s response to viral infection, and the real cause of Alzheimer’s is a virus, in particular Herpes simplex virus. Worth noting nearly everyone carries some form of the Herpes simplex virus, for example HSV-1 is the variant that causes cold sores. So just as not everyone who smokes gets lung cancer, not everyone who has HSV gets Alzheimer’s. But that doesn’t mean it’s not the cause. The battle over whether Alzheimer’s is a viral disease is not over by the way. It’s still being actively argued. But at this point the argument reminds me of how for decades doctors fought the idea H. pylori bacteria caused ulcers, because they were so invested in the old paradigm of spicy food and stress. Good article by Ed Yong on the latest.
3. Apple combines machine learning and Siri teams under John Giannandrea. Here’s the story in TechCrunch. And why would anyone care about this particular Apple re-org? Because it means Apple is moving away from a purely functional organizational structure, since Giannandrea reports directly to CEO Tim Cook. Moving to a more divisional structure is a huge change for Apple. And a necessary step if Apple is ever to produce services (maps, Siri) which can compete with Google. Ben Thompson argued back in 2016 why Apple needed to change their org structure, and his article holds up well. It’s worth re-reading Thompson to understand why this is important.
4. Mind as prediction. It is a trope to say nobody understands how the brain works. This is true of course. But I would argue that before you can understand in detail how the brain works, you need some kind of overarching theory of what a mind is and what is does. One framework is the mind constantly predicts what happens next, updating its model as sensory feedback floods in. So mind is constant prediction, updated against reality. And not just passive prediction, but actively predicting what will happen if you do particular actions. This is a fairly old idea. I first came across it in Jeff Hawkins’ 2007 book On Intelligence, and it was a revelation (at least to me). Because it seems so obviously correct. And over the past decade I’ve seen this framework slowly gaining ground. In particular machine learning has hastened the trend. To that end, here’s a new article arguing for the mind as prediction framework: To Make Sense of the Present, Brains May Predict the Future. If you like it, you should also consider Scott Alexander’s post on the same idea It’s Bayes All the Way Up.
5. African multiregionalism. Good piece by Ed Yong. Here’s one bit:
She and others argue that humans originated from several diverse populations that lived across Africa. Separated from each other by geographical barriers, they mostly evolved in isolation, and each group developed some of our hallmark traits, but not others. But their separation wasn’t constant: As a changing climate remodeled the African landscape, greening deserts and drying out forests, those early humans were repeatedly drawn together and pulled apart. Whenever they met, they mated and mingled, exchanging genes and ideas in a continent-wide melting pot that eventually coalesced into the full bingo of features that you or I might recognize.
This theory, known as “African multiregionalism,” is a fundamentally different view of how we came to be. It’s saying that no single place or population gave rise to us. It’s saying that the cradle of humankind was the entirety of Africa.
6. Teardown of Sonos One compared to Amazon Echo. Ben Einstein tears apart a Sonos One and an Amazon Echo Plus, comparing the two. No surprise, Einstein predicts Sonos is in deep trouble. Lots of good pictures in the teardown, and Einstein clearly knows his way around a soldering iron. Here’s the key bit:
It is always tricky to estimate BOM cost without diligently researching each custom part and purchased component, but my suspicion is that despite the 25% lower price tag, the Echo Plus is about 15–20% more expensive than the more premium Sonos One.
Because of Amazon’s platform advantage, it’s selling more expensive hardware at a lower price. Hard to compete with that. link
And that’s all my links for this week. Hope you have a good weekend!
Praxtime by Nathan Taylor 