Lights Out
Note: ASD will be off next week, returning the week after from a new home, explained below.
Substack
Starting off a newsletter talking about issues within the newsletterer community may be a little navel-gazey, but it’s a controversy that’s come to a head and is worth addressing:
Under pressure from critics who say Substack is profiting from newsletters that promote hate speech and racism, the company’s founders said Thursday that they would not ban Nazi symbols and extremist rhetoric from the platform.
Not great! The Atlantic found dozens of newsletters openly displaying Nazi imagery and white supremacist rhetoric, and the response from Substack has been to…defend their right to use a free newsletter platform to spread their hatred and profit off it.
It is being framed as a free speech debate, but it really isn’t. As anyone subjected to the endless rhetoric on social media knows by now, a newsletter company is not the government, and the First Amendment does not apply. Substack has no problem moderating and censoring other kinds of speech, as it does not allow adult content (porn) on its platform. Sex workers can’t use Substack, but Nazis can, and are tacitly supported and encouraged by the company’s leaders, even. Gross.
Two things are at play here - one is the VC-backed, ‘free’ nature of Substack, which has played a large part in its rise to dominate the digital newsletter space. When I was toying with the idea of starting a newsletter, Substack’s simple, no-cost tools made it easier to give it a whirl. This frictionless adoption is precisely their business model, because the founders and their investors are focused on growth and scale, not nickel and diming everyone who wants to use their platform.
The other, more insidious narrative at play is the fetid swamp of tech founder groupthink that weaves capitalism, libertarianism, and white male supremacy into an ideology that has become irresistible to many of the country’s rising entrepreneurs. Watching Elon Musk turn Twitter into Stormfront is not exactly discouraging the deplatforming of Nazis, as racist authoritarians are given bylines elsewhere in prominent mainstream publications. Quite a few tech bros seem to have the same toxic politics in part because they all talk to one another, and when your social circles go fash, it takes a certain strength of character to speak out, which Substack’s founders do not possess.
Nor is this the first time Substack has had criticism aimed its way for publishing vile bigotry - it has platformed anti-trans crusaders for years. It makes sense that, as Nazis are being mainstreamed by the GOP and the world’s richest man, companies with loose ethics and a strong dose of Founder Mindset would follow suit.
All this is to say I will be moving this newsletter off Substack in the coming weeks, giving it its own domain, and hopefully ending up in everyone’s inboxes with little disruption. This will cost me a modest amount of money, which is fine, but is also illustrative of a lot of what we talk about around here - platforms lure users in with a ‘free’ product, and in that way become the product. Maybe Substack will weather the storm of authors departing over its policies - a different group of its right-wing writers penned a letter in support of publishing Nazis - but for many of us, it’s worth paying for the right to have a say in what sort of stuff our work appears next to.
AI
There are two conflicting narratives in AI right now - it’s either about to learn geometry and become so powerful it takes over the planet and eradicates humanity, or it’s bad at pretty much every task it does, and a liar to boot.
There is evidence for the latter claim, with more arriving on a near daily basis. Google couldn’t even create a sizzle reel for its fancy new AI without heavy editing. The problem Google had, and which much of the AI hype papers over, is that it’s difficult to create chatbots that convincingly imitate human conversation, much less a human with superpowers. Which begs the question: do we need chatbots to become our friends? What problem are we solving for here?
We’ve talked about how Google’s search results are already becoming polluted with AI-generated nonsense. Decades ago, Google’s search algorithms seemed like magic to the Internet’s early denizens. You suddenly had what felt like an impossibly large amount of data at your fingertips. Finally, the computer was able to tell you something new and interesting, rather than passively accepting your commands. Then, once the ad people got their grubby mitts on it, Google became a dusty highway of sketchy billboards, gradually shoving the Internet’s treasured wisdom and weirdness further down below the fold to sell clicks.
Perhaps it is our desire to have the computer tell us new things that now drives publishers to team up with AI companies so their bots can deliver a version of the news:
Earlier today, OpenAI, the maker of ChatGPT, announced a partnership with the media conglomerate Axel Springer that seems to get us closer to an answer. Under the arrangement, ChatGPT will gain the capacity to present its users with “summaries of selected global news content” published by the news organizations in Axel Springer’s portfolio, which includes Politico and Business Insider.
You may not have realized that bots like ChatGPT are rolled out like standard software and therefore contain a snapshot of the Internet that ends on the day they’re built - ChatGPT’s current version contains no new scraped information since April. Deals like Springer could feed ‘trusted’ sources of data into the models, allowing them to act as virtual newsboys which doesn’t sound totally awful, assuming they report the stories truthfully, which is no guarantee. In exchange, Springer is being paid by OpenAI and providing access to its archives.
In a happy bit of coincidence, as I was writing this piece I received an alert (from Google news, natch) about a lawsuit the New York Times has filed against OpenAI and Microsoft. Battle lines are being drawn between media companies who see AI as an opportunity and others who see it as a blight.
As someone who obsessively reads news and news alerts, a chatbot that could accurately summarize the day’s events and highlight certain stories that fit my interests would indeed be a useful tool. Unfortunately, even the most powerful models have a long way to go before I’d trust any of them to accurately regurgitate a few articles.
And, again, what is going on with our obsession with talking computers? It feels a little like weird authoritarian fantasy to force a piece of software to tell you your ideas are good, or find things you’d find interesting like a captive librarian. Are we certain that if we did create a digital superintelligence it would want to listen to all of our bullshit? Maybe that’s why OpenAI has a team of engineers coming up with ways to control a hypothetical smart AI by putting dumber AIs in charge of it. Sure, why not.
While investors are pouring incredible amounts of money into building talking computers, scientists are finding useful applications for what we’re now calling AI. Unsurprisingly, fields of study with giant datasets can take advantage of systems with very powerful computing tools:
In only the past few months, AI has appeared to predict tropical storms with similar accuracy and much more speed than conventional models
[…]
Figuring out a single protein structure from a sequence of amino acids used to take years. But in 2022, DeepMind’s flagship scientific model, AlphaFold, found the most likely structure of almost every protein known to science—some 200 million of them.
There are some very cool applications for LLMs and the kind of deep learning and neural nets that power our extremely expensive chatbots. None of them involve telling you which Insider article got the most eyeballs yesterday, but they could be helping cure diseases, diagnose conditions, or predict disasters, which feels significantly more useful for humanity writ large.
But! Some have pointed out that while AlphaFold has potentially disrupted an entire branch of science, no one is quite sure exactly how it came to its conclusions. Which, obviously, is a problem:
An AI model might predict a thunderstorm’s arrival but struggle to explain the underlying physics and atmospheric changes that triggered it, analyze an X-ray without showing how it arrived at its diagnosis, or propose abstract mathematical conjectures without proving them.
There is precedent for trusting models that humans can’t fully understand - it’s literally the origin story of quantum mechanics, which even Einstein called ‘spooky’. In sum total, powerful computational models will undoubtedly be a boon for mankind, and lead to many scientific breakthroughs. Whether we’ll ever get over our obsession with forcing them to be our friends is another matter.
Pedestrians
We have talked about pedestrian and road deaths in the US, and how our country is a far more deadly place to drive and walk than any of our peers. Unlike deaths of despair - a major driver of mortality in the US that continues to climb - slamming your car into someone walking or riding a bike in or near a road is almost always unintentional and tragic. What gives?
It would be easy to blame our giant cars, high speed limits, and poorly designed roadways with little or no space for humans on foot. But it turns out there’s an even more insidious culprit: darkness.
It turns out that three quarters of our pedestrian casualties happen at night. Even when adjusted for (godforsaken) Daylight Savings and winter darkness hours, the death rate skyrockets in the hours between sunset and rise.
Obviously, darkness is not the actual culprit, but the way it interacts with American transportation infrastructure is - our roads are more likely to lack protected sidewalks, and the suburbanization of poor families means more carless households are living along dangerous stretches of roads:
Nationwide, the suburbanization of poverty in the 21st century has meant that more lower-income Americans who rely on shift work or public transit have moved to communities built around the deadliest kinds of roads: those with multiple lanes and higher speed limits but few crosswalks or sidewalks. The rise in pedestrian fatalities has been most pronounced on these arterials, which can combine highway speeds with the cross traffic of more local roads.
This is most pronounced in Southern and Sun Belt states, many of which lack the basic urban and suburban pedestrian infrastructure of older cities that existed before cars:
The number of pedestrian fatalities in Florida has increased 75 percent since 2009, while the population has increased around 17 percent.
In addition to geographical factors, the way Americans drive has changed more dramatically than other countries. We’ve got smartphone addiction and distracting screens in our cars, combined with the preponderance of automatic transmissions enabling distracted driving:
It’s perhaps not surprising then that Americans spend nearly three times as much time interacting with their phones while driving as drivers in Britain…
[…]
In the U.S., that distracted driving — detected when phones are tapped or in motion in vehicles traveling faster than 9 miles per hour — also typically peaks in the evening hours…
And while drunk driving actually declined in the most recent survey, more drivers in violent accidents now test positive for other drugs, which coincides with the country’s opioid crisis.
Basically, the pedestrian death rate in this country is a grim mélange of all things American - poorly designed, car-centric neighborhoods with dangerous crossings and walkways, poorly enforced traffic laws and high speed limits, distracted, drugged drivers with multiple screens pulling their eyes away from the road for the split second required to plow through a crosswalk at deadly speed.
The good (?) news from the research is that child fatalities are down, and cities with proper pedestrian infrastructure have not gotten any less safe. But without major changes in places they’re extremely unlikely to happen, walking near or along a road in much of the country is a dangerous proposition, getting deadlier each year.
Buses
While we’re discussing poor transit infrastructure, allow me to climb atop my soapbox for a moment.
I currently live three blocks from an intercity bus station. Two months ago, I did not live three blocks from an intercity bus station. The mundane explanation for this change in status would be that someone built a bus station nearby my house, but that would be wrong. What the city of Philadelphia has done is move the curbside bus service stop from a location in the city’s center to a curb near my house. It was forced to do so because Greyhound vacated its full-service bus station, and the added number of buses to the already inadequate location made the traffic situation untenable.
So now, on a lonely stretch of curb near a train station, a car rental lot, and a strip club, hundreds of people a day stand around in the elements with no bathrooms or shelter. The dry cleaner I use in the adjoining strip mall has had to add a full time security guard to dissuade people from taking shelter inside.
The guard isn’t the only new fixture in the neighborhood - at least one cop car is permanently idling with lights flashing, for whatever reason. Two or three ‘ambassadors’ in vests help answer questions and keep people organized as the buses come and go. I have no idea who pays for all of this, and it is not particularly disruptive to either traffic or pedestrians and commuters in the area - the sidewalks are quite wide - but the entire thing feels sad and dehumanizing.
It is, as it turns out, a trend across much of the country:
Houston, Philadelphia, Cincinnati, Tampa, Louisville, Charlottesville, Portland, Oregon, and other downtown bus depots have shuttered in recent years. Bus terminals in major hubs like Chicago and Dallas are also set to close. Greyhound and other companies have relocated their stops far away from city centers, which are often inaccessible by public transit, switched to curbside service or eliminated routes altogether.
Some of the budget bus lines like Megabus use curbside pickup, but Greyhound had long operated out of proper bus terminals. You’ll never guess what has changed:
Greyhound bus stations nationwide are closing and relocating outside central business districts after being acquired by an investment firm that rose to infamy for its acquisition and gutting of American newspapers.
Greyhound has long been the brand most closely associated with intercity bus travel in the United States. Its stations, often architecturally and culturally significant, occupy prime downtown real estate that is considered ripe for commercial and residential development.
Ahhhh, yes, of course. Greyhound was sold in 2021 to a UK conglomerate, who then sold the bus service itself off to a German company, but kept the stations. Then, late last year, it sold the stations to Alden Global Capital, a notorious PE firm who set about selling them off to developers.
Intercity bus service is used disproportionately by poor and minority communities, many of the same non-car families in danger of being wiped out by speeding motorists:
Roughly three-quarters of intercity bus riders have annual incomes of less than $40,000. More than a quarter would not make their trip if bus service was not available, according to surveys by Midwestern governments reviewed by DePaul University.
As we see with so many things, our government’s failure to properly fund public infrastructure creates opportunities for private vultures to come in and dismantle it. The ability to board a bus between two cities is now susceptible to the whims of a bunch of business executives dispassionately scrutinizing whether it would be more operationally efficient to sell the station and cancel service.
I confess that, despite occasionally taking a Megabus to and from New York when convenient, I previously gave little thought to how unpleasant it was to wait on a sidewalk with a bunch of strangers, exposed to the elements. Now I see people suffering this indignity every day, in an area of my neighborhood surrounded by empty parking lots and self storage buildings, as developers build high-rise condos and apartments nearby, and wonder why the city (and/or the bus companies) won’t put up even a token sum to build a decent shelter and a couple bathrooms for these people, who are just trying to get where they need to go.
Short Cons
NYT - “That was how he became one of hundreds of thousands of people who have been trafficked into criminal gangs and trapped in what one research group has called a “criminal cancer” of exploitation, violence and fraud that has taken root in Southeast Asia’s poorest nations.”
The Conversation - “According to a survey of U.S. adults, Americans in October 2023 were less likely to view approved vaccines as safe than they were in April 2021. As vaccine confidence falls, health misinformation continues to spread like wildfire on social media and in real life.”
CNN - “Navy Federal Credit Union, which lends to military servicemembers and veterans, approved more than 75% of the White borrowers who applied for a new conventional home purchase mortgage in 2022... But less than 50% of Black borrowers who applied for the same type of loan were approved.”
AP - “The students in Louisiana's off-the-grid school system aren't missing. But there's no way to tell what kind of education they're getting, or whether they're getting one at all.”
NY Post - “The bloated band of vaguely titled aides, accountable only to Adams, comprised roughly one-third of the Mayor’s Office staff during the yearly period ending June 30 and cost taxpayers $24.3 million, according to payroll records.”
WIRED - “A little-known surveillance program tracks more than a trillion domestic phone records within the United States each year, according to a letter WIRED obtained…”
Bloomberg - “About 48 million Americans experience some degree of hearing loss, according to the National Council on Aging, but only about 20% of them do anything about it. Among those with mild hearing loss—the largest group—the percentage is even lower.”
Know someone thinking of befriending a chatbot? Send them this newsletter!