![[personal profile]](https://www.dreamwidth.org/img/silk/identity/user.png){kind=small}
David Sedaris, Calypso: Essays mostly centering on his family, with whom relations are both fraught and close, including the death by suicide of his sister and his adventures in housesharing on vacation with them.
Claudia Rankine, Just Us: Poems and prose about whiteness and related topics. “notes on the state of whiteness” reproduces portions of an edition of Jefferson’s Notes on the State of Virginia, with text not about Blacks removed. It’s powerful. A lot of the prose touches on various ways in which whiteness enables not having to see, and thus not seeing or remembering, both overt racist violence and racist structures. Rankine deeply interrogates her own reactions as a means of interrogating the world’s.
Janelle Shane, You Look Like a Thing and I Love You: I sort of thought by the title it would be about anthropomorphism, but the subtitle is “How Artificial Intelligence Works and Why It’s Making the World a Weirder Place.” Along with cute illustrations, the book explains pretty much that, at a high level, along with what AI is and isn’t good for. AI is particularly not good at resisting targeted attacks; stickers or changes in pixels can change “gun” to “toaster” in AI vision, and—attention, fic writers—“a low-security fingerprint reader can be fooled 77 percent of the time with a single master fingerprint.” Like people, AI is lazy, so when you try to train it to recognize skin diseases, it will instead learn the easier trick of recognizing the rulers that are often in the picture with actual cancers, which is clever from its perspective but not from ours. That’s also how it learns to replicate human biases (in favor of men, against people who went to HBCUs, etc.). If you use the same camera to take pictures of the training set of “right” answers, then it may learn to use the camera metadata instead, although my favorite examples were the AIs that learned to exploit features of the training system, e.g. anomalies in the modeling of physics that allowed them to accumulate infinite force or crash the system when they were about to lose or even hack into the answer key and award themselves right answers. AI, that is, is very much like Captain Kirk and the Kobayashi Maru. Also for authors: there are certain training datasets that lots of systems use. If you managed to submit enough samples to those databases to corrupt them—maybe 5%--you could make your adversarial attacks on the system succeed. This vulnerability also suggests big problems of overfitting—the training set matters too much and the algorithm too little. Also: you could hack voice-to-AI systems so that a human would hear one thing and the AI system would hear something very different. Fic possibilities, worrying realities. One last tidbit to ponder: we so much don’t know how algorithms learn that when people are assigned to circle the part of a picture that helps them figure out what’s in the picture, algorithmic performance goes down, so either people are wrong about how they recognize dogs (etc.) or something even stranger is going on. On the plus side, made me kind of want to write Leverage fic?
Amy Stanley, Stranger in the Shogun’s City: A Japanese Woman and Her World: A relatively well-educated but economically marginal woman in the last years of the shogunate, after several failed marriages, headed to Edo and sought fortune. She found sexual abuse, exploitation, and also a freedom she didn’t want to give up, including in her tumultuous marriage with a warrior/occasional ronin. I quite enjoyed Stanley’s story of an ordinary woman who defied convention, with ambiguous results—the past is a foreign country, but people lived there too.
Sarah A. Seo, Policing the Open Road: History of how the automobile helped create modern policing and modern criminal procedure law. The perceived uncontrollability of auto-enabled drivers—including a lot of white people, though of course race was always a factor in who got policed—drove an expansion in police forces, expansion in the number of laws governing cars, and expansion in the things that police were allowed to do to drivers. A little repetitive, and would have been aided by more comparisons with other countries to get a sense of how uniquely American the story was, but also full of connections that, while often obvious once stated, need to be stated. A particular favorite: the early argument that autos were so heavily regulated, and of such great importance to the public, that they could be forfeited freely for auto-related crime because they were essentially already the state’s property.
Alexis Coe, You Never Forget Your First: A biography of George Washington emphasizing his political/spy skills during the War and also how much male biographers have really, really liked his mighty thighs (the mightiest thighs I ever have theen!), which they focus on to show how manly he was despite apparently being sterile. Coe also brings forward his treatment of the people he enslaved, including his use of their teeth to use in his dentures (he paid, but below-market rates).
J.D. Dickey, Empire of Mud: History of the early years of the city that became the District of Columbia, patched together out of several cities, with emphasis on what it was like to live there, both for Black and white inhabitants. Bad, unrepresentative government (including a bunch of spending on ultimately useless canals) was a hallmark even before it was set in stone after the Civil War for exactly the reason you think it was: to ensure that the large Black population wouldn’t control the local affairs of the seat of government.
Slavery’s Capitalism, ed. Sven Beckert: Essays about how slavery built American capitalism as it was, including by funding Northern development and trans-Atlantic networks of wealth. Many of the essays deliberately challenge the distinction between a “slave society” and a “society with slaves” by highlighting how the latter often benefited from interactions with the former, going well beyond a conspiracy between the “lords of the lash and the lords of the loom.” Includes a very interesting analysis of the “limestone” South—portions of upper Kentucky, Virginia, etc. with uncharacteristically rich soils and uncharacteristically well-developed cities—before the Civil War, they were more urbanized (and productive) than the midwest. Nonetheless, the argument is, population growth and public education lagged, dragging down innovation. Even where there was a well-developed white middle class, slave economies, the theory of government they relied upon, and the white men who were powerful in them opposed widespread public education, which capped how much development could occur in the long run. (Fun fact: southerners allocated the funding they did for public education by how many children were in a district, but public schools charged tuition, so the slave states set high tuitions that only wealthy parents could pay, making the publicly subsidized schools into private academies for elites. Tuition existed in the North too, but was much lower and so attendance was far more widespread.) The conclusion: there were no environmental barriers to slavery thriving in the North, and while Northerners may have been wrong about the reasons that slavery slowed growth, they were correct about the depressive economic effects of its expansion on overall economic growth.
Claudia Rankine, Just Us: Poems and prose about whiteness and related topics. “notes on the state of whiteness” reproduces portions of an edition of Jefferson’s Notes on the State of Virginia, with text not about Blacks removed. It’s powerful. A lot of the prose touches on various ways in which whiteness enables not having to see, and thus not seeing or remembering, both overt racist violence and racist structures. Rankine deeply interrogates her own reactions as a means of interrogating the world’s.
Janelle Shane, You Look Like a Thing and I Love You: I sort of thought by the title it would be about anthropomorphism, but the subtitle is “How Artificial Intelligence Works and Why It’s Making the World a Weirder Place.” Along with cute illustrations, the book explains pretty much that, at a high level, along with what AI is and isn’t good for. AI is particularly not good at resisting targeted attacks; stickers or changes in pixels can change “gun” to “toaster” in AI vision, and—attention, fic writers—“a low-security fingerprint reader can be fooled 77 percent of the time with a single master fingerprint.” Like people, AI is lazy, so when you try to train it to recognize skin diseases, it will instead learn the easier trick of recognizing the rulers that are often in the picture with actual cancers, which is clever from its perspective but not from ours. That’s also how it learns to replicate human biases (in favor of men, against people who went to HBCUs, etc.). If you use the same camera to take pictures of the training set of “right” answers, then it may learn to use the camera metadata instead, although my favorite examples were the AIs that learned to exploit features of the training system, e.g. anomalies in the modeling of physics that allowed them to accumulate infinite force or crash the system when they were about to lose or even hack into the answer key and award themselves right answers. AI, that is, is very much like Captain Kirk and the Kobayashi Maru. Also for authors: there are certain training datasets that lots of systems use. If you managed to submit enough samples to those databases to corrupt them—maybe 5%--you could make your adversarial attacks on the system succeed. This vulnerability also suggests big problems of overfitting—the training set matters too much and the algorithm too little. Also: you could hack voice-to-AI systems so that a human would hear one thing and the AI system would hear something very different. Fic possibilities, worrying realities. One last tidbit to ponder: we so much don’t know how algorithms learn that when people are assigned to circle the part of a picture that helps them figure out what’s in the picture, algorithmic performance goes down, so either people are wrong about how they recognize dogs (etc.) or something even stranger is going on. On the plus side, made me kind of want to write Leverage fic?
Amy Stanley, Stranger in the Shogun’s City: A Japanese Woman and Her World: A relatively well-educated but economically marginal woman in the last years of the shogunate, after several failed marriages, headed to Edo and sought fortune. She found sexual abuse, exploitation, and also a freedom she didn’t want to give up, including in her tumultuous marriage with a warrior/occasional ronin. I quite enjoyed Stanley’s story of an ordinary woman who defied convention, with ambiguous results—the past is a foreign country, but people lived there too.
Sarah A. Seo, Policing the Open Road: History of how the automobile helped create modern policing and modern criminal procedure law. The perceived uncontrollability of auto-enabled drivers—including a lot of white people, though of course race was always a factor in who got policed—drove an expansion in police forces, expansion in the number of laws governing cars, and expansion in the things that police were allowed to do to drivers. A little repetitive, and would have been aided by more comparisons with other countries to get a sense of how uniquely American the story was, but also full of connections that, while often obvious once stated, need to be stated. A particular favorite: the early argument that autos were so heavily regulated, and of such great importance to the public, that they could be forfeited freely for auto-related crime because they were essentially already the state’s property.
Alexis Coe, You Never Forget Your First: A biography of George Washington emphasizing his political/spy skills during the War and also how much male biographers have really, really liked his mighty thighs (the mightiest thighs I ever have theen!), which they focus on to show how manly he was despite apparently being sterile. Coe also brings forward his treatment of the people he enslaved, including his use of their teeth to use in his dentures (he paid, but below-market rates).
J.D. Dickey, Empire of Mud: History of the early years of the city that became the District of Columbia, patched together out of several cities, with emphasis on what it was like to live there, both for Black and white inhabitants. Bad, unrepresentative government (including a bunch of spending on ultimately useless canals) was a hallmark even before it was set in stone after the Civil War for exactly the reason you think it was: to ensure that the large Black population wouldn’t control the local affairs of the seat of government.
Slavery’s Capitalism, ed. Sven Beckert: Essays about how slavery built American capitalism as it was, including by funding Northern development and trans-Atlantic networks of wealth. Many of the essays deliberately challenge the distinction between a “slave society” and a “society with slaves” by highlighting how the latter often benefited from interactions with the former, going well beyond a conspiracy between the “lords of the lash and the lords of the loom.” Includes a very interesting analysis of the “limestone” South—portions of upper Kentucky, Virginia, etc. with uncharacteristically rich soils and uncharacteristically well-developed cities—before the Civil War, they were more urbanized (and productive) than the midwest. Nonetheless, the argument is, population growth and public education lagged, dragging down innovation. Even where there was a well-developed white middle class, slave economies, the theory of government they relied upon, and the white men who were powerful in them opposed widespread public education, which capped how much development could occur in the long run. (Fun fact: southerners allocated the funding they did for public education by how many children were in a district, but public schools charged tuition, so the slave states set high tuitions that only wealthy parents could pay, making the publicly subsidized schools into private academies for elites. Tuition existed in the North too, but was much lower and so attendance was far more widespread.) The conclusion: there were no environmental barriers to slavery thriving in the North, and while Northerners may have been wrong about the reasons that slavery slowed growth, they were correct about the depressive economic effects of its expansion on overall economic growth.
Tags: