The Order That Fare Gates Bought
Bay Area transit riders know the feeling: a train car that smells of urine, scattered needles on the floor, someone shouting at nobody. For years it was anecdotal. Now it's data. Noah Smith's latest roundup centers on a simple intervention that changed everything at BART — and why that change matters far beyond Oakland and San Francisco.
When Gates Close, Crime Falls
Smith leads with the hard numbers. BART installed fare gates last year at many stations. What followed was not gradual improvement but a collapse in disorder.
"After BART installed fare gates last year at many of its stations, crime on the trains plummeted by 54% in a single year." Smith writes. The time employees spent fixing what riders broke or defiled went from huge amounts to almost nothing.
"It turns out that just a few riders were causing most of the disorder on the BART — and those riders were mostly not paying their fares, since the fare gates were effective in stopping them."
This is broken windows theory in action: visible disorder invites more disorder. Fix the windows, paint the walls, collect the fares — and the whole system stabilizes. Smith names the opposing view anarchyfare — the idea that eliminating rules serves marginalized people. He argues the opposite. Most poor and marginalized people are peace-loving commuters who need the train to get to work. They are the chief victims of the tiny number of chaotic individuals who destroy the commons.
"Restraining a very few uncontrollable and chaotic individuals makes life much better for the poor and working class."
Critics might note that fare gates are expensive to install and maintain, and that they can create new barriers for people experiencing genuine hardship. A gate that stops fare evasion may also stop a mother with no cash from reaching a job interview. The data shows crime fell. It does not show what happened to the people who stopped riding entirely.
AI and the Employment Puzzle
Smith turns to the question haunting every twenty-something with a degree: is AI taking my job? The unemployment data looks alarming for young college graduates. But Smith cautions against reading too much into one metric.
"If you look at employment rates instead, the picture looks very different." Unemployment depends on who says they're looking for work. Employment does not. Recent college grads show constant labor force participation — they're all still trying to find jobs. Meanwhile, some non-college peers have stopped looking entirely. This shows up as higher unemployment for college grads, even though the gap in who actually has a job has widened since ChatGPT's release.
Smith cites research showing that "'AI exposure' and 'interest rate sensitivity' are deeply correlated variables." Jobs most exposed to AI are also most sensitive to macroeconomic swings. The same sectors saw pronounced drops in job postings during the 2020 slowdown — when generative AI could not theoretically be the explanation.
"It still looks to me as if the slowdown in new-grad hiring is not a great example of AI taking jobs."
Critics might note that correlation does not prove causation, but the timing is suspicious. Even if AI is not replacing workers yet, uncertainty about AI's effects could slow hiring. Companies may pause before adding humans to roles they expect machines to handle within two years.
Tariffs and the China Decoupling
On trade, Smith finds mixed results. Tariffs have not created a wave of manufacturing jobs — in fact, manufacturing employment is decreasing. They have not clobbered the overall trade deficit either. November data showed imports soaring and exports falling.
But the bilateral trade deficit with China has come down sharply. "America used to get more than a fifth of its imports from China; now it gets less than a thirteenth." Some claim China simply ships through third countries like Vietnam. Research suggests transshipment is real but limited — perhaps 23 percent of diverted trade at most.
What is more plausible is that China ships intermediate goods — parts, materials — to countries like Vietnam, where they are assembled into consumer goods and sold to the U.S. The manufacturing base is migrating out of China. That helps other countries industrialize. It also reduces America's geopolitical vulnerability.
"While tariffs haven't clobbered the trade deficit or led to a manufacturing renaissance, they do appear to be working to decouple the world's two largest economies."
Critics might note that decoupling carries costs. Higher prices for consumers. Supply chain friction. And the risk that intermediate dependence — buying Vietnamese assemblies built with Chinese parts — is not true independence but a more complex form of it.
Stewart's Economics Misfire
Jon Stewart was Smith's favorite political comedian when Smith was younger. Stewart's heart was in the right place. But in recent years, Smith argues, that desire has morphed into lazy centrist populism. One target: the economics profession.
"Throughout the interview, Stewart seemed to believe that economics is just a sophisticated justification for letting rich people and corporations do whatever they want." Stewart rejected the notion that economists have anything useful to say about climate change, then immediately endorsed a cap-and-trade scheme — something economists invented.
Jason Furman, an incredibly mild-mannered and affable guy, was nevertheless willing to vent about his own interview with Stewart. Stewart has recruited Oren Cass, a Trump supporter and big fan of tariffs, to back his crusade against economics. Smith has written about the vapidity of Cass's critiques. Every time manufacturing employment falls, Smith tweets Cass asking whether he has revised his belief that tariffs help manufacturing. Cass never answers.
"Unfortunately, the 'econ is fake' meme has given a lot of people permission to treat those challenges as if they're a simple matter of common sense. They are not."
Critics might note that economics has its own blind spots — models that ignore power, distribution, and history. Stewart's caricature is uninformed. But the profession's self-defense sometimes sounds like gatekeeping.
Visas and the Offshore Shift
On immigration, Smith warns that restrictions on high-skilled legal immigration push companies overseas. The administration has implemented a huge fee for hiring H-1B visa holders. Most recipients are from nonwhite countries, with India taking the largest share.
"For every visa rejection, [multinational companies] hire 0.4 employees abroad." Those jobs are R&D-intensive. They go to China, India, Canada. Alphabet is plotting to dramatically expand its presence in India. If talent cannot come to America, America's companies will go to talent.
Critics might note that H-1B restrictions reflect genuine political pressure — voters see visa holders as competitors for scarce jobs. The economic case for openness is strong. The political case is harder.
Bottom Line
Fare gates work — they restore order that benefits everyone who depends on public space. But they are not a philosophy, just a tool. On AI, tariffs, economics, and visas, Smith's throughline is skepticism toward simple stories. The data rarely matches the meme. The fix-everything button does not exist. What exists is incremental repair: gates that close, rules that hold, jobs that shift, trade that reroutes. Order is possible. Simplicity is not.