This is a fantastic and approachable article written about a common financial product! If this isn't for you, we have different tastes. The well-defended essay by Byrne Hobart, The 30-Year Mortgage is an Intrinsically Toxic Product.

Also, if you've taken a course on statistics, that's not enough to do much useful stuff. But it'll let you get the full value from this explanation of why deep learning works better with more layers even without more data. It also generally explains the notion behind why deep learning works the way it does – which it doesn't explain at all, really, so I'll tell you: it makes a random guess about how a neural net might be weighted, and then changes the guess in all the directions that make it suck less. Imagine wanting to chart a best-fit line, so you draw a random line, and see, when you nudge it up or down, which one makes it better or worse. Then you do that nudge, and the same for the tilt. Over and over. Makes sense for lines, but the idea that it can be used to do speech recognition is a tad weird. True, though.

If you think Trump is merely a sign instead of a cause of the collapse of American hegemony, consider COVID-19 infection could cause an enduring mental degradation equivalent to about 8.5 IQ points. I was living in Taiwan when this all started, and the State Department ordered me back (because I couldn't guarantee an arbitrarily long visa). I'm pretty concerned about this.

As election predictions go, an important thing to pay attention to is how older voters disfavor Trump much, much more now than before. I heard a pollster who did many interviews say this was largely because they perceive Trump as being indifferent to whether they live or die. That's an absolutely insane way to feel about people. Extremely aggressive to express. I don't think they're wrong.

A number of news sources wrote a story about a COVID vaccine trial continuing after the death of one of the participants. Many forgot to mention the person who died was in the control group, and didn't get anything.

China is threatening to kill visitors to their country (or, Hong Kong, the coolest place they've had a chance to ruin) if other countries don't play ball. What issue is so important to them they'd do this? Allowing refugees from their disasters.

I think this is an important thing to consider when we talk about American greatness. Regulations of all types are punishing, frequently more so for poorer people. I'm all for regulation of negative externalities, but the whole subject has been approached with no humility or caution. The correct level of wisdom is to say that unintended consequences will dominate all political decisions.

I find this blog entry, telling people to be more prolific, to be sort of funny. I don't disagree, but it doesn't say much, and his post frequency is less than mine. So am I doing better than him on his own metrics? I suspect not. No, the problem is that his advice isn't diagnostic. Perhaps, he thinks there is some ratio of craft to marketing that's ideal, so he produces less so he can get his writing more widely shared. That's how I found it! But that's the opposite of his advice. I don't think this is even well thought through enough to be hypocritical. Advice is challenging. Put more effort into it.

A prominent physicist argues that bigger particle colliders are unlikely to make new discoveries, and we ought to investigate existing conflicts or inconsistencies in known physics. I think this is necessary, but also pretty guessable from a public policy perspective. Reality will have hundreds of most-fruitful avenues to explore, but "more science funding" is a public choice mess. Clear, concrete projects always get disproportionately more funding, which is why people think infrastructure means roads and bridges when, last I checked, something like 90% of American infrastructure is underground pipes.