Mistakes and Corrections
An inexhaustible mine of content.
As much as I hate mistakes, I still make them all the time. You’d think I’d have learned by now.
Not all authors are afforded the ability to correct the record, but through this medium, I can. And so I shall lean into it, and celebrate each of my most (and least) spectacular failures in turn, as they occur.
I’ll keep adding to this collection over time until it gives me more pride than shame. But for now, and without further ado, here are the horrors that have me tossing and turning at night, the terrors I dread whenever I start writing, the glimpses into how I can do better — starting with the most recent.
11/28: Not so much a mistake as an oversight. I omitted the credit for the idea that we should to separate the question of AGI safety from the question of its humanity. It’s Sasha who opened my eyes to this wonderful and hard to deny take: that making an AI more like us does not rhyme with making it safer. (She also put me onto how dolphins communicate holographically)
11/7: My model predicted the Republicans would win a massive landslide. The code was, upon revision, full of bugs and mistakes, and thoroughly unscientific. Needless to say, I got it wrong.
10/16: Finding galaxyhenges is not quite as simple as I put it. Cosmic distances make it a lot harder to estimate the relative sizes and masses of objects from their angular size (and redshift) alone. The equations I shared were inaccurate.
7/26: I wrote that Juneteenth (June 19th, 1866) marked the true end of slavery. In fact, many slaves weren’t freed until much later.
I’m sure I’m missing tons. I will offer a bounty (in the form of a $25 Wikipedia donation and a coffee or drink of your choice) for any additional ones y’all find!