Most of us search Google the same way we always have: type a few words, scroll, click something that looks close enough, and hope. For a while, that worked. Google handed us a list of links and let us take it from there.

What’s happening now is something different. A 2024 study by SparkToro found that nearly 60% of Google searches end without anyone clicking through to a website, and the trend has accelerated since. By February 2026, Ahrefs found that queries triggering AI Overviews now see a 58% reduction in clicks. Google has been systematically inserting itself between you and the original source, answering questions with AI-generated summaries before you ever reach the page those answers came from. The results you do see are filtered through an algorithm that weighs your search history, your location, and the billions of dollars advertisers have spent to appear for particular queries. Two people searching identical phrases on the same day can get meaningfully different results without either of them knowing it. And because Google controls roughly 90% of the world’s search traffic, most people have no frame of reference for what a less mediated search experience would even look like.

The search bar replaced the reference desk without replacing the skills behind it: knowing how to ask a question precisely, understanding how information is organized and who funds it, knowing the difference between a primary source and a summary of one. The assumption was that the technology made all of that unnecessary, which suited Google; a user who can’t navigate information independently is a user who keeps coming back to be guided.

The search bar you already have is more capable than that arrangement requires you to know. With the right syntax, it becomes a precision instrument: narrow by domain, by date, by file type, by exact phrase. We can pull up archived pages, surface open file directories, and even find what people said in forums instead of what brands want us to find. None of it requires a new tool or a paid account. The capability has been there the whole time.

When You’re Not Getting What You Asked For

Google is constantly interpreting you. It swaps in synonyms, personalizes results based on your history, and decides what you probably meant rather than returning what you typed. Most of the time that interpretation is invisible. These tools are how you override it.

site:

limits your search to a single website. Try: site:nytimes.com climate to search only the Times, or site:gov vaccine to pull results exclusively from government domains. It works as a better version of a website’s own search function (most built-in site search is mediocre at best), as a trust filter when you only want results from a specific domain type, and as a research shortcut when you already know which publication or institution you want to pull from. You can also run it in reverse: electric vehicles -site:tesla.com returns coverage that isn’t from Tesla’s own pages.

Number ranges

let you set hard boundaries on any numerical search. Put two periods between two numbers with no spaces: laptop $500..$800 returns results mentioning prices in that range. The same syntax works for years (civil rights legislation 1964..1968) or any other measurement. It eliminates a significant amount of irrelevant results when you’re comparison shopping or trying to find coverage from a specific period.

Verbatim mode

is the most powerful feature most people have never used. After any search, click Tools (just below the search bar), then the “All Results” dropdown, then select “Verbatim.”

Google stops paraphrasing you entirely and returns results for exactly what you typed, stripped of personalization and synonym-swapping. It’s one of the most useful things Google has buried several clicks deep, and the fact that it takes three clicks to reach says something about how much Google wants you to find it.

Quotation marks

work the same way at the phrase level. Try: “the medium is the message”. Wrapping a phrase in quotation marks forces Google to find pages where those exact words appear in that exact order. Unquoted words are treated as suggestions; quoted phrases are treated as requirements. Use this to verify whether a quote is real and trace it to its actual source, to find a specific statistic rather than everything that implies it, or to track down a title you half-remember. It’s also the mechanism behind one of the most useful social search techniques covered below.

The minus sign

removes a word from your results entirely. Put it directly before the word with no space: jaguar -car returns the animal, mercury -planet returns the element or the musician depending on your other terms. Precise, effective, and useful any time a word you’re searching carries more than one meaning.

AROUND(#)

is an undocumented proximity operator that tells Google how many words apart your two search terms can be. Try: climate AROUND(3) policy. The intent is that only pages where those terms appear in genuine proximity show up, rather than a page that mentions “climate” in the introduction and “policy” ten paragraphs later. Google has never officially documented this operator and its behavior is inconsistent, but when it works, it operates closer to how academic databases have functioned for decades. Worth testing, but not something to rely on the way you would a documented operator.

When You Need the Real Source, Not Just a Summary

The difference between finding a blog post about a study and finding the study itself isn’t trivial, and the gap between them is larger than most people expect.

filetype:

returns only a specific kind of file. filetype:pdf remote work productivity returns only PDFs. Swap pdf for ppt to find slide decks, or doc for Word documents. Most research reports, government documents, academic papers, and white papers exist as PDFs and don’t rank highly in regular search results because they weren’t built for traffic. Filetype search gets you past that.

intitle: “index of”

surfaces something most people don’t know exists: open file directories on the internet. Try: intitle: “index of” /pdf “media literacy”

These are servers running with directory listing enabled, a default setting in Apache that displays all files in a directory when no index page exists. Most administrators never turned it off. The result is publicly accessible file systems, packed with documents, datasets, and files that don’t appear in regular search results.

before: and after:

set a date boundary on your results. mental health social media research after:2023 filters out everything published before that year. Use before: to find what was known or written at a particular point in time, useful for confirming a source predates an event or for tracing how a conversation has shifted over time. Combine them with site: for a targeted archive search: site:theatlantic.com AI after:2023 pulls everything The Atlantic has published on the subject in the past two years. This kind of search used to require a library database subscription.

intitle: and inurl:

let you filter by the structure of a page rather than just its content. intitle:”media literacy” returns only pages where that phrase appears in the actual title, not just mentioned once in passing. inurl:gov intitle:”AI policy” finds government pages where AI policy is the stated subject. Combined, they’re considerably more precise than keyword searching alone.

When You Want Real Human Opinions, Not Sponsored Content

SEO has made the first page of Google results increasingly dominated by content written to rank rather than to inform. These techniques route around it.

“can anyone recommend”

exploits a quirk in how people write when they’re asking for help without a commercial motive. Try: “can anyone recommend” noise-canceling headphones under $100. Because the phrase is in quotation marks, Google surfaces only pages where those exact words appear, which means forum threads, community posts, and real conversations where people asked the same question you’re asking. Instead of a sponsored listicle, you get someone’s firsthand experience choosing between two specific products. Swap in “does anyone know a good” or “what’s the best” for variations on the same trick.

@ before a word

surfaces social tags and handles in your results. Try: @reddit home espresso machine. Google officially describes this as a tool for finding social tags, so pairing it with a platform name like @reddit or @twitter alongside your topic pulls community discussions toward the top of your results. It doesn’t filter exclusively to those platforms, but it shifts the result set in that direction. Combine it with the quotation mark technique when you want to narrow things further.

The omitted results link

is easy to miss. When Google adds a note at the bottom of a results page saying some results were hidden because they’re too similar to others, there’s a small link to include them anyway. The results Google omits tend to be less trafficked and less search-optimized, which frequently means they’re more substantive and written for readers rather than algorithms. When doing real research rather than a quick lookup, that’s exactly where to look.

When You Need to Go Deeper

The asterisk *

works as a wildcard for any missing word or phrase. Try: “the * of artificial intelligence”. The asterisk stands in for whatever word you can’t remember or want to explore. It’s invaluable for chasing down half-remembered titles and quotes, and it surfaces the full range of ways a phrase gets used across different contexts, which is useful for research that starts from a concept rather than a specific source.

Stacking operators

is where precision compounds. filetype:pdf “information literacy” site:edu before:2015 finds older academic PDFs on the topic from university domains. site:cdc.gov after:2022 -press release pulls recent CDC content with press releases filtered out. The combinations are where the real power lives, and once you’ve internalized a few operators separately, combining them becomes instinctive.

When You Just Need a Fast Answer

Many of Google’s most useful features are things you’d only find by accident, because nothing in the interface tells you they exist. These all work by typing directly into the search bar.

Paste a flight number

like UA 2157 and Google returns the live gate, departure and arrival times, current delay status, and a real-time position tracker without opening an app or an airline website. This works for any major commercial flight. If you’re picking someone up, it’s considerably faster than anything the airline itself offers.

Paste any package tracking number

and Google recognizes the format automatically, whether it’s UPS, FedEx, or USPS, and shows live delivery status directly on the results page. If you’ve been opening carrier websites every time you get a shipping confirmation, you didn’t need to be.

Type run speed test

and Google measures your download and upload speed directly in the browser, without sending you to a third-party site like Speedtest.net. When you’re troubleshooting a slow connection and don’t want to open another tab, it’s the fastest option.

Type [thing] vs. [thing]

like oat milk vs almond milk, Notion vs Obsidian, ibuprofen vs acetaminophen, and Google pulls a side-by-side comparison panel with key differences. It works for supplements, software, ingredients, and medications. It’s not always exhaustive, but it’s faster than opening five tabs to piece together the same information.

A few more that show up less in guides but earn their place:

• define: [word] returns the full dictionary definition plus etymology

• how to pronounce [word] gives you an audio button and phonetic spelling

• [food] calories brings up nutritional information without leaving the search bar

• sunrise [city] or sunset [city] gives you exact times

• time in [city] shows current local time anywhere in the world

• [amount] [currency] to [currency] pulls a live exchange rate

• stock [ticker] shows a live price chart with trading volume

• tip for $[amount] opens a tip calculator you can adjust by percentage and split by number of people

• translate [phrase] to [language] opens a full translation widget with audio pronunciation

• what is my IP returns your IP address immediately

• random number between [X] and [Y] generates one instantly

• color picker opens an interactive color wheel with hex and RGB codes in the results page itself

• timer 25 minutes starts a countdown without leaving Google

• metronome opens a working, adjustable metronome

• bubble level uses your phone’s gyroscope as an actual level

• breathing exercise guides you through a timed breath pattern

• what sound does a [animal] make plays the actual audio

• flip a coin and roll a die both work exactly as described

• Any math equation typed into the search bar is solved immediately

Google also has a full arcade buried in the results page. Searching solitaire, tic-tac-toe, snake, or pac-man opens a playable game directly, no app or third-party site required. Most people have scrolled past these results for years without realizing they were interactive. And two Easter eggs that have been there since at least 2011 and still work: do a barrel roll spins the entire results page 360 degrees, and askew tilts it just enough that people think something is wrong with their screen.

One more that matters for anyone who makes content: after any image search, click Tools > Usage Rights and filter to show only images licensed for reuse. The feature is two clicks deep, most people who need it regularly don’t know it exists, and using an unlicensed image because you didn’t check is a more common mistake than it should be.

What Not to Do

These are the habits that undermine searches most often, and most of them are so ingrained they feel like standard practice.

Don’t treat the AI Overview as the answer.

The AI-generated summary at the top of many Google results is the feature most likely to be wrong and most likely to present that wrongness with complete confidence. Since Google launched AI Overviews in May 2024, documented errors have included advising users to add glue to pizza, recommending that people eat one small rock per day, producing a response claiming Barack Obama was the United States’ first Muslim president (drawn from an academic book title that Google’s system misread as a factual claim), and, in May 2025, insisting across multiple queries that the current year was 2024. These aren’t edge cases. They reflect a structural problem with how the feature works: it synthesizes answers from sources you can’t always see, using a system that can misread context, miss sarcasm, and draw incorrect conclusions from factually correct sources. If the AI Overview touches anything consequential, check the sources beneath it.

Don’t click the first result without checking whether it’s an ad.

Google labels paid results, but the labels have grown smaller and less visually distinct over time. The first two or three results on many searches are sponsored placements, meaning companies paid to appear there rather than earning their position organically. A business with a large advertising budget can outrank a more authoritative source on nearly any commercial query. Check for the small “Sponsored” label before assuming what’s at the top is what’s most credible.

Don’t assume your results are the same as anyone else’s.

Google personalizes results based on your search history, location, device, and account data. Two people searching the same phrase can get meaningfully different pages in meaningfully different orders without either of them knowing it. When research matters, Verbatim mode or a private/incognito window removes some of that personalization layer.

Don’t use quotation marks on everything.

Quotation marks are precise when you need an exact phrase, but applying them to every search narrows your results so sharply that you’ll miss pages that would have been directly useful. If you’re not searching for a specific verbatim phrase, leave the quotes off.

Don’t add a space after an operator.

Purely mechanical, but it kills the function entirely. site:cdc.gov works; site: cdc.gov does not. The operator and the term have to run together with no space between them.

Don’t just Google it when the stakes are real.

Most people use Google the same way for everything, whether they’re looking for a restaurant or trying to understand a diagnosis, a medication interaction, a contract clause, or a financial decision. That habit works fine for low-stakes questions, but for anything with real consequences, Google’s results, and especially its AI Overviews, are a place to find sources, not a destination. A Guardian investigation in January 2026 found multiple AI-generated health summaries that medical professionals flagged as dangerous, including dietary advice for pancreatic cancer patients that Anna Jewell, director of support, research and influencing at Pancreatic Cancer UK, said could “jeopardize a person’s chances of being well enough to have treatment.” Google is often the fastest way to figure out where to look. Treating it as the place to stop is where the trouble starts.

Beyond Google: You Have Options

Knowing when to use a different tool is part of knowing any tool well. Treating one resource as the default regardless of the question is a habit, and like most habits, it runs below the level of conscious choice.

Google is where most people search, and learning to use it well is worth doing. But Alphabet, Google’s parent company, reported $350 billion in total revenue in 2024, with advertising accounting for more than three-quarters of that, according to the company’s own annual filing. The results Google shows you are shaped by that business model in ways that aren’t always visible. Its algorithm promotes pages built to rank, which isn’t the same as pages built to inform. Its AI summaries synthesize answers from sources you often can’t see, which makes it harder to evaluate whether the underlying information is reliable. And because it personalizes results based on your history, two people searching the same phrase on the same day can land in meaningfully different places. Understanding that context changes what you should reasonably expect from a Google search, and knowing what else is available changes what you do when Google isn’t the right tool for the question.

If the problem is structural — that Google's incentives and your interests don't always point in the same direction — then having alternatives isn't about distrust. It's about knowing which tool fits the question. These eight work differently, in ways that are worth understanding before you need them.

1. Kagi is a paid search engine with no advertising and no sponsored results. Plans start at $5 a month for 300 searches or $10 a month for unlimited. You’re paying directly for the service rather than trading your attention for access, which changes the underlying incentives entirely. Its results tend toward fewer SEO-optimized pages and more original sources, a difference most noticeable when the quality of information matters more than the speed of finding it.

2. DuckDuckGo is free, doesn’t track your searches, and supports all the operators covered above. It also has a feature called !bangs: type !w before any search to go straight to Wikipedia, or !scholar for Google Scholar. It turns the search bar into a shortcut launcher for wherever you want to land, without a company logging where that is.

3. Brave Search is free and privacy-focused, and unlike most alternatives, it runs its own independent search index rather than licensing results from Google or Bing. Most privacy-focused search engines are Bing with a different coat of paint; Brave is the meaningful exception.

4. Startpage is free and returns Google’s actual results without Google’s tracking. It works as a private intermediary, submitting your query to Google anonymously and returning results without storing your IP address, search history, or any identifying data. If you’ve tried the other alternatives and find the results weaker than you want, Startpage resolves that without sending your data to Google directly. One thing worth knowing going in: Startpage is owned by System1, a U.S. advertising company, which it discloses openly and says does not affect its no-tracking policy.

5. Perplexity is AI-powered and built for research questions. It gives you a synthesized answer with sources cited directly alongside it, so you can see exactly where the information came from and evaluate it yourself. For questions where you want a starting point with visible sourcing rather than a list of links to sort through, it’s often faster and more transparent than a traditional search.

6. Bing is Microsoft’s search engine and the second largest in the world by traffic, which makes it the most overlooked real alternative to Google. It’s ad-supported and tracks your searches, so it doesn’t solve the privacy problem — but it runs an entirely different index, which means different results, and that alone is worth knowing. For image search and video it’s often stronger than Google. It’s also the engine powering Microsoft’s Copilot, which gives you AI-generated answers with sourcing in the same way Perplexity does. If a Google search isn’t surfacing what you need, running the same query on Bing takes ten seconds and frequently produces something Google buried or missed entirely.

7. Ecosia is ad-supported and runs on Bing’s index, so the results are comparable to Bing rather than Google. What’s different is what happens to the money: Ecosia is a certified B Corp that directs the majority of its advertising revenue toward reforestation projects and publishes monthly financial reports so you can verify it. It won’t give you stronger results than the alternatives above, but for someone whose searches are already going to generate ad revenue for someone, Ecosia redirects that toward something. It’s a light switch, not a lifestyle change — but it’s a real one.

8. Library databases are the option most people forget they already have. A public library card — free in most cities — gives you access to databases like ProQuest, EBSCOhost, and JSTOR that the open web simply cannot replicate. These index academic journals, historical newspapers, court documents, company filings, and primary sources that were never designed for Google to crawl and never will be. If you’ve been hitting paywalls on research that matters, this is how you get past them without paying. Check your library’s website for remote access instructions; most let you log in from home with your card number.

The Skill Nobody Told You You’d Need

There used to be a professional layer between most people and raw information. Librarians, researchers, editors, fact-checkers: people whose entire job was to understand how information was organized, who produced it, what motivated them, and where the gaps were in any given source. You didn’t need to think much about any of that, because someone else already had.

That layer has largely dissolved. Search engines replaced the card catalog, algorithms replaced the reference interview, and AI summaries are now stepping in where a librarian’s judgment about source quality used to sit. What’s been left in place of all that professional mediation is a search bar and the assumption that you’ll figure it out.

The tools above don't fix that problem, but they change your position within it. Every technique here is a version of the same underlying move: being specific about what you need and deliberate about where to look for it. Most people were never taught to approach search that way, because the assumption has always been that it's simple enough not to need teaching. But the same move works everywhere information is organized: library catalogs, academic databases, legal repositories, government archives.

Search syntax is just the entry point. What's underneath it is a way of thinking about how knowledge is structured and who controls access to it — and that transfers to every tool you'll use after this one.

Share Card Catalog

This is the third article we have released from Issue 23, which print subscribers started receiving last week. Not yet a subscriber? You can sign up for the magazine here.

Japan is the land of the train. 28 percent of passenger kilometers in Japan are travelled by rail, more than anywhere else in the developed world. France achieves 10 percent, Germany 6.4 percent, and the United States just 0.25 percent. Travel in Japan is over a hundred times more likely to be by rail than travel in the United States.

Japan’s vast railway network is divided between dozens of companies, nearly all of them private. The largest of these, JR East, carries more passengers than the entire railway system of every country other than China and India. Each year, JR East carries four times as many passengers as the whole British railway system, even though it has fewer kilometers of track, serves about ten million fewer people, and competes with eight other companies. Japan’s railway system turns a large operating profit and receives far less public subsidy than European and American railways.

In most developed countries, the railways have struggled since the rise of the automobile in the 1950s. From this point on, North America saw the near-total replacement of passenger trains with cars and planes. In Europe, it meant vast government financial support to keep the lines open.

Japan’s different trajectory is often attributed to culture: the Japanese are conformists who are content to take public transport, unlike freedom-loving Americans who prefer to drive everywhere. Europeans are somewhere in between. Culture is also used to explain the incredible punctuality of Japanese railways.

These cultural explanations are wrong. The Japanese love cars, but they take trains because they have the best railway system in the world. And their system excels because of good public policy: business structure, land use rules, driving rules, superior models for privatization, and sound regulation have given Japan its outstanding railways.

This is good news for friends of rail. Culture is built over centuries, and replicating it is hard. But successful public policies can be emulated by one good government. Much about Japan’s railway system could be replicable around the world.

Japan’s railway companies

Today, the most striking institutional feature of Japanese rail is that it is privately owned by a throng of competing companies.

The railway arrived in Japan in 1872, during the Meiji Restoration, which opened the country up to foreign trade, ideas, and technologies. Like most Western countries, Japan nationalized its railways in the early twentieth century, creating what became known as Japanese National Railways (JNR). But it did not nationalize all of the lines, focusing only on mainline railways of national importance, and new private railways were still permitted.

Between 1907 and World War II, Japan saw a boom in new private electric railways, coinciding with rapid urbanization. Technologically, most of these private railways were similar to the famous interurbans in the United States: they were basically electric trams, but running between cities as well as within them. The American network eventually withered, and almost nothing of it survives today. In Japan, however, the network consolidated, and the light tramlines gradually evolved into heavy-rail intercity connections.

These companies are today known as ‘legacy private railways’ on account of their having been private since their inception. There are eight legacy private railways in the Tokyo metropolitan area, five in the Osaka–Kobe–Kyoto megalopolis, two in Nagoya, and one in the fourth city of Fukuoka. There are also dozens of smaller ones elsewhere. In the three largest urban areas, these operators account for nearly half of railway track and stations, as well as a plurality of ridership. The largest, Kintetsu, not only operates urban services, but a whole intercity network stretching from Osaka to Nagoya.

These companies often compete head-to-head. At its most extreme, three separate commuter lines compete for the traffic between Osaka and the port city of Kobe, running in parallel, sometimes fewer than 500 meters apart.

Meanwhile, the nationalized railways were managed by JNR. In the postwar era, JNR was responsible for building the famous Shinkansen system, as well as running commuter and long-distance lines throughout Japan. But in 1988, it was largely privatized, broken into six regional monopolies for passenger services together with a single national freight operator. These are collectively known as the Japan Railways Group (JR).

This means that Japan has ended up with six railway companies that trace their descent to the nationalized railways, the sixteen big legacy companies that have always been private, and a host of minor legacy railways, as well as numerous underground metros (some private, some municipally owned), monorails, and tram systems. This institutional diversity is striking enough. But equally striking is the consistent business model that has evolved amidst this pluralism: the railway that builds a city.

Railway-led urbanism

If I take a train to go for a solitary walk in the countryside, the railway company can capture some of the value it creates by charging me for the journey, just as other companies capture the value of the goods and services they provide by charging for them. However, if I take a train to visit family, clients, a theater, or a shop, an important difference appears. The railway can capture the value it creates for me by charging me a fare, but it cannot capture the value it creates for those at my destination. As transport infrastructure creates benefits that produce no revenue for providers, free markets rarely build enough of it.

Japan has partly solved this problem by enabling railway companies to do a great deal beside running railways. Take the example of the Tokyu corporation, one of the legacy private railways in southern Tokyo. You can not only travel on its trains, but also ride a Tokyu bus, live in a Tokyu-built house, work in a Tokyu office complex, see a doctor in a Tokyu hospital, buy groceries in a Tokyu supermarket, spend an afternoon at a Tokyu museum-theater-cinema complex, take your children to their amusement park, and even die in a Tokyu retirement home. The positive spillover effects of the railway on these things are captured by Tokyu because it owns them. The president of Tokyu has said:

I think that though we are a railway company, we consider ourselves a city-shaping company. In Europe for instance, railway companies simply connect cities through their terminals. That is a pretty normal way of operating in this industry, whereas what we do is completely different: we create cities and then, as a utility facility, we add the stations and the railways to connect them one with another.

This model was pioneered in the 1950s by what became Hankyu Railways. Hankyu’s network connects central Osaka to its northern suburbs, as well as Kyoto and Kobe. Its innovative founder Kobayashi Ichizo first built suburban housing, then a department store at the terminal station; he then created a hot spring resort, a zoo, and his own distinctive brand of all-women musical theater, the Takarazuka Revue. He also began to run bus services to and from his stations. Other companies emulated Hankyu’s example: Tokyo Disneyland is a collaboration between Disney and the Keisei Railway, while Hanshin in Osaka owns the Hanshin Tigers baseball team.

Core rail operations are profitable for every Japanese private railway company, but they usually only account for a plurality or a small majority of revenue. The rest is contributed by their portfolio of side businesses. There is a natural financial synergy between the reliable but unremarkable cash flow of train fares and the profitable but riskier real estate and commercial side of the business. Railway companies’ side businesses also attract people to live and work on their rail corridor, reinforcing the customer base for the railway services themselves.

This virtuous circle is enabled by transit-oriented development. Japan’s liberal land use regulation makes it straightforward to build new neighborhoods next to railway lines, giving commuters easy access to city centers. It also enables the densification of these centers, which means that commuters have more places they want to go.

Railways cost a lot to build, but once they are built, they can move enormous numbers of people, far more than a road of similar size. This means that they work best in cities with a high density of people, jobs, and other activities. In 2019, New York City was the only American city where rail had a higher modal share than cars, in part because Manhattan has 2.5 million jobs, two million residents, and 50 million tourist visits crammed into 59 square kilometers.

This does not mean that rail-oriented cities must be structured like Chinese cities: islands of high-rise apartments connected by metros and separated by motorways. Japanese cities have the lowest residential density in Asia, and a plurality of the Japanese live in houses, usually detached ones. The urban area of Tokyo, the densest Japanese city, has a weighted population density less than that of many European cities, including Paris, Madrid, or Athens. Japanese cities have vast low-rise, predominantly residential suburbs, built at densities that might be higher than what is typical in the United States, but that would be quite normal in Northern Europe.

What makes Japan’s cities particularly suited to rail is thus not their residential districts, but their huge and hyperdense centers. These really are special: the cores of Tokyo or Osaka are unlike anything that exists in Europe or North America. Many of their features are famous worldwide: the vertical street zakkyo buildings, underground streets, shopping streets under rail tracks, covered arcades, elevated station squares, and vertical cities. Getting millions of commuters and shoppers into these downtowns is where rail excels because its extreme spatial efficiency means that infrastructure with a relatively modest footprint can transport vast numbers of people into a small area.

None of this emerged from a coherent masterplan of transit-oriented development like Copenhagen’s Finger Plan or Curitiba’s Trinary System. Postwar Japanese opinion was committed to decentralization both to rural peripheries and to the suburbs through greenbelts, motorways, and new towns.

Instead, this variety and adaptability around railways is possible because of the way Japanese urban planning works. Since 1919, Japan has had a standardized national zoning system, but it is much more liberal than development control systems in Western countries. The Japanese authorities did not intend or even desire dense urban centers, but they did not prevent them, rather like nineteenth-century governments in the West.

This liberal zoning system is reinforced by private access to city planning powers. Thirty percent of Japan’s urban land has been subject to land readjustment, where agreement among two thirds of residents and landowners in an area is enough to allow its replanning, including compulsorily taking and demolishing land for amenities and infrastructure. Initially land readjustment was used only to assemble rural land for urbanization, but over time it was increasingly used to redevelop already urbanized areas, and new variants were created to build the skyscrapers that surround the major stations of central Tokyo.

The history of the private railway companies could be written as a story of land readjustment projects: the initial building of the lines in the interwar years proceeded through one land readjustment project after another. Postwar improvements such as double-tracking, platform lengthening, and constant redevelopment of stations and their immediate thresholds were only possible because the railways could secure land takings cooperatively with local businesses and landowners.

Perhaps the greatest example of this phenomenon involved Tokyu. In 1953 the company decided to build the Den’en Toshi Line, or Garden City Line, to serve a rural area southwest of Tokyo. This would be enabled by a series of land readjustment projects collectively among the largest in Japanese history.

Over 30 years, 3,100 hectares were covered, of which only 36 percent was devoted to residential and commercial development, with 20 percent for forest and parks, 17 percent for roads, and much of the rest for watercourses. The population of the land readjustment zone would rise from 42,000 in 1954 to over 500,000 in 2003.

By connecting the affluent southwestern suburbs to Tokyu’s main real estate hub next to Shibuya station, now the second busiest in the world, the Den’en Toshi Line allowed Tokyu to become the largest private railway by revenue and ridership. The Japanese government and academics generally consider the Den’en Toshi Line to be the best corridor of transit-oriented development in Japan.

But the railway-as-city-builder model is not the only reason Japanese railways have been able to thrive. European countries usually prohibited railways from running real estate side businesses, but in the United States and Canada the practice was extremely widespread in the nineteenth and early twentieth centuries, and many famous railway suburbs were developed this way. Despite this, passenger rail in these countries collapsed in the mid-twentieth century. Part of the difference was that Japan did not extend the same implicit subsidies to cars as Western governments did.

Pricing driving

The land of Toyota, Nissan, and Honda is not an anti-car nirvana. In fact, Japan has excellent motorways, and across the country as a whole a small majority of journeys are made by car. But Japan is a place where cars and car-oriented lifestyles compete on a level playing field.

Japan is one of the only countries to have privatized parking. In Europe and North America, vast quantities of parking space is socialized: municipalities own the streets and allow people to park on them at low or zero cost. Initially with the intention of encouraging the provision of more parking spaces, Japan made it illegal to park on public roads or pavements without special permission. Before someone buys a car, they must prove that they have a reserved night-time space on private land, either owned or leased.

Since parking on public land is banned, municipalities are not worried about overspill parking from developments with inadequate private parking. They therefore have no reason to impose parking minimums on developments: the market is left to decide whether parking is the most valuable use of private land. Where land is abundant, as in rural areas, suburbs, or small towns, private parking is plentiful. But in city centers, it is outcompeted by other land uses. According to the late Donald Shoup, central Tokyo has 23 parking spaces per hectare and 0.04 parking spaces per job, compared with 263 and 0.52 for Los Angeles. Even Manhattan, the densest urban area in North America with the lowest levels of car ownership, has about 60 parking spaces per hectare.

Japanese roads are expected to be self-financing. Motorways are run by self-contained public cooperatives, very similar to the statutory authorities that ran English roads and canals between 1660 and the late 1800s, and funded by tolls on their users. Vehicle registration taxes, which are allocated to localities for road construction and maintenance, are worth three percent of the Japanese government budget.

These measures, adopted in the 1950s, were not intended to suppress car use – the point was to fund a massive road expansion – but they have forced private vehicles to internalize many of their hidden costs. In the Tokyo urban area, the average household spends 71,000 yen ($450) each year on public transport fares and 210,000 yen ($1,350) on car purchase and maintenance costs.

But the private car was not the only competitor faced by the private railways. For eight decades in the twentieth century, they also had to face the juggernaut of Japanese National Railways. Its privatization in 1988 removed the final obstacle to creating the world’s best railway system.

Privatization

Railway privatization in Britain, New Zealand, Argentina, and Sweden has had a mixed reception, and all of those countries, apart from Sweden, have taken steps to reverse it. In Japan, it has been so successful that the government subsequently privatized the metro systems in Tokyo and Osaka.

In the postwar period, JNR enjoyed real successes. It built the revolutionary Shinkansen, the first high-speed railway in the world. It also aggressively electrified and double-tracked major trunk lines, quadruple-tracked lines into and out of major cities, and added city-center loops and freight bypasses. But these achievements were overshadowed by two problems.

The first was politics. Many countries adapted to the rise of the car by closing the least profitable parts of their passenger rail network, like the consolidation of American freight rail into the Class I operators or the Beeching Axe in Britain. In Japan, however, the ruling Liberal Democratic Party drew its support from rural constituencies, whose support it retained with pork-barrel politics. Its ‘rail tribe’ group, led by rural MPs, prevented JNR from adapting itself to mass motorization.

JNR therefore did not amputate gangrenous rural and freight services that imposed heavy costs with few benefits. Worse, it continued to build new loss-making rural railway lines, known in Japanese as Gaden-intetsu, or railways pulled into the rice field.

The second problem was organized labor. In general, Japanese trade unions are known for their moderation and responsibility, a generalisation that also held true for the unions at the legacy private railways. The JNR unions, however, became highly militant, secure in the knowledge that their nationalized employers could never go bankrupt. Their largest series of strikes in 1973 provoked riots from commuters.

The railway unions imposed overstaffing on revenue-generating urban services, at a time when both international and private domestic operators were reducing staffing requirements against a backdrop of higher wages and the growing automation of signaling and ticketing. As a result, 78 percent of JNR’s costs were related to labor, compared to 40 percent for other Japanese railways. The average worker at a private railway was 121 percent more productive than their JNR counterpart.

By the early 1980s, only seven out of 200 JNR lines made a profit. Successive governments deferred serious reform, running up debt, cutting down investments in new urban lines, raising ticket prices to twice those of comparable private railways, and increasing subsidies – which rose until annual subsidies equaled the total cost of the Shinkansen.

In 1982, Prime Minister Yasuhiro Nakasone started to privatize the railways. Unlike other countries, Japan simply returned to the traditional private railway model of the nineteenth and early twentieth centuries: tracks, trains, stations, and yards were owned by vertically integrated regional conglomerates.

There are substantial advantages to vertical integration. Railways are a closed system that has to be planned as a single unit. Changing the timetable at station A can affect the timetable at station Z; buying new trains that can travel faster might require changes to the infrastructure so they can reach their top speed, which in turn requires rewriting the timetables. This becomes especially complicated if different services share tracks. To prevent delays from propagating from one service to another, the timetable needs to be carefully designed to make best use of the available infrastructure.

The starkest effect of privatization was a massive and immediate increase in labor productivity and profitability relative to the legacy private railways. In fact, this began before privatization: its mere threat strengthened the government’s hand when bargaining with the unions and forced JNR to begin closing rural lines.

Privatization saw a general trend of productivity improvements, following a big one-time improvement between 1982 and 1990, when the workforce was cut by more than half, 83 loss-making lines were removed, and JNR’s debts were transferred to a holding company.

The second great advantage of privatization was to allow the JR companies to emulate the railway-as-city-builder model of the legacy private railways: for instance, JR East owns two shopping center brands, a ski resort, a coffee chain, and even a vending machine drink company. The JR companies have not ignored their rail business: they have continued to build new high-speed lines and urban tunnels, upgrade stations, and implement a host of other improvements such as the introduction in the 1990s of smart cards that allow passengers to pay their fare with a tap.

Regulation

This does not mean that the Japanese railway industry is a pure creature of free enterprise. No railway system ever has been. The Japanese system has found an equilibrium that makes rail policy explicit and limited. Leaving aside railway safety and business regulation, there are two main policy levers: fare maximums and capital expansion subsidies.

Price controls are often cited as a classic example of misguided government intervention, whether through rent controls, caps on the price of gasoline, wage freezes, or minimum agricultural prices. Tokyo’s infamously crammed trains are a symptom of underpriced rush hour traffic.

Railways have market power because the substitutes for railway trips – coaches, cars and planes – are quite a different product. This monopolistic position has historically meant trouble: monopoly systems, whether private or public, have a tendency to abuse their position to charge higher prices and run bad services. For this reason, the private monopolies that were common in the Western world before World War I often had price controls imposed on them. For example, most of the American streetcar networks were operated as long-term, price-controlled franchises granted by the city.

Price maximums, if set too low, could have ruined Japan’s railways. This is exactly what happened to many Western transit services after the First World War. But the postwar Japanese practice has capped fares generously. The system is explicitly designed to maintain profitability per rider, which in turn incentivizes the companies to maximize ridership. That buys political legitimacy for the privatized system, which is necessary for the continued provision of capital expansion subsidies. Indeed, during the long deflation era between 1992 and 2022, it was common for operators to charge below the maximum, and the real value of railway fares continued to rise. Fare maximums are set on the basis of the average cost structures of all railway operators in a region, so companies with below-average costs like Tokyu would often charge below the cap to maintain a competitive edge, prevent public backlash, and maximize traffic to their side-businesses.

Other than the fare maximums, the railways are free to make their own decisions about timetables, service patterns and day-to-day operations, a highly specialized and technical task which requires deep expertise. This contrasts with the government meddling with, say, Amtrak’s routes.

Carefully designed public subsidies also play a useful role. Although Japanese railways do not receive subsidies for day-to-day operations, they do receive government loans and grants for capital investments. These are typically tied to public priorities, such as disability access or earthquake-proofing, or to projects that have large spillovers that the railway company would be unable to internalize, like removing level crossings, or elevating at-grade railways or trams in order to reduce road congestion and accident risk. Generally, the local prefectural government will match the contribution of the national government. Larger new build projects are subject to lease back or debt-payment conditions that fare revenue is expected to pay back.

Subscribe now

The recipe for successful railways

Railway companies invested heavily in real estate businesses, often funding lines through selling land for housing around new stations. Liberal spatial policy meant that such development happened easily, even as it enabled dense development in urban cores where radial rail lines converged. Rail companies were generally vertically integrated regional monopolies, owning the land, track, and rolling stock, setting their own timetables, and employing their staff. The state imposed controls to stop them exploiting their monopoly position, but it did so cautiously, allowing them to make sufficient profit that incentives to invest were preserved. Capital subsidies were targeted at providing specific public goods that normal commercial operations overlooked.

The above paragraph could be written by a historian of the future about contemporary Japan. But every word in it could also be written by a historian today about the United States in the nineteenth century – usually seen as the epitome of capitalist individualism. This striking fact contradicts the idea that America’s supposed individualism foreordains it to be the land of the car, or that Japan’s supposed communitarianism foreordained it to be the land of rail.

It also puts pressure on the idea that the demise of rail is the inevitable consequence of cars. All countries saw some shift to cars in the twentieth century, and all rail industries had to respond to that. But public policy had an enormous effect on how successfully they did so. The rise of zoning restrictions on density, excessive price controls, nationalization, and vertically disintegrated privatization have hampered Western rail in remaining competitive against cars since the 1920s. By maintaining and restoring the institutions that built the first railway systems in the nineteenth century, the Japanese have created the mightiest railway system of the twenty-first.

Matthew Bornholt is an urban planner and transport researcher. You can follow him on Bluesky.

Benedict Springbett is a writer and Bar student. You can follow him on Twitter.

A friend of mine bought a Tesla Roadster back in the early 2010s. At the time, spotting a Tesla on the road was a rare event. Maybe even occasion enough to stop and take a picture. I never got the chance to photograph one, let alone drive one, until I met this new friend recently. This was my chance to experience the car firsthand.

We walked to the parking structure to see it. As soon as he opened the door, something looked... off. On the outside, it was a pristine, six-figure roadster. But the inside looked completely custom. Not "custom" in the sense of a professional shop install, but more like the driver himself grabbed a hammer and chisel and made it his own.

First, the driver's seat had been altered. It was much lower than usual and didn't match the passenger seat. My friend stands 6'7", and the Roadster is a tiny car. He physically couldn't fit, so he modified the seat rails to lower it. But that fix created a new problem: the door armrest now dug into his hip. So, he took a file to the interior panel, shaved it down, and 3D printed a smaller, ergonomic armrest. He even 3D printed a cup holder for the passenger side so his coffee was within reach.

To me, the idea of taking a Dremel or a file to a $100,000+ car was unimaginable. You must be crazy to do it.

He caught the look on my face and shrugged. "Hey, it's my car. I paid for it. I intend to be comfortable in it."

I never thought of it like this. That sentiment stuck with me. Recently when I read an article by Kent Walters about filing the corners of his MacBook, those same feelings resurfaced. My work MacBook has edges so sharp that I've often felt like I was slicing my wrist on the chassis. I treated this as a design flaw I had to endure. But not Kent. He treated it as an obstacle to be removed. He literally filed down the corners of his laptop to ensure the machine he uses every day was comfortable.

I may not have the guts to file my work issued MacBook, but I'm no stranger to customization... in software. I modify my tools constantly. I spend days tweaking my IDE, remapping keyboard shortcuts, and writing custom scripts until the software is unrecognizable to anyone else on my team. I don't think twice about rewriting a config file to make the tool fit my brain.

When I was a kid, I always had a screw driver around, fixing a device that wasn't really broken. On the home computer, I modified everything. I once deleted all .ini files to improve performance. It didn't work, but it led to a fruitful career.

But somehow, when it comes to expensive hardware now, I freeze. I treat the physical object as a museum piece to be preserved. I bought a docking station to banish the laptop to a shelf, using an external mouse and keyboard to avoid touching the sharp chassis. I built a complex workaround to accommodate the tool, rather than performing the simple, brutal act of modifying the tool to accommodate me.

We treat our physical tools as if they are on loan from the manufacturer.

You'll see a musician buying a vintage guitar but refuses to adjust the action, terrified of ruining the "collector's value." Meanwhile, the working guitarist has sanded down the neck and covered it in stickers because it feels better in their hand. The software engineer accepts the default keybindings to avoid "bad habits," while the power user creates a layout that doubles their speed.

If you own a tool, whether it's a car, a computer, or a line of code, you own the right to change it. The manufacturer designed it for the "average" user, but you are a specific human with specific needs.

Remember grandma's couch in the living room? It had that plastic cover on it. It was so uncomfortable, but no one dared to remove it. The plastic was to preserve the sofa. No one got to enjoy it, instead everyone accommodated the couch only to preserve its value. A value that one ever benefits from. Don't let the perceived value of an object stop you from making it truly yours. A tool with battle scars is a tool that is loved.

I’ve been teaching college Earth science courses as a part-time faculty member for a long time now, all while juggling other jobs. I started because it was enjoyable; no one gets into this line of work for the famously poor pay or complete lack of job security. Working with students is just one of those genuinely fulfilling experiences that is addictive enough that they ought to warn people about it.

But thanks to generative AI, it has become mostly miserable―at least in certain settings.

For the last few years, I’ve been exclusively teaching asynchronous online courses, meaning recorded videos rather than live sessions. These have always been a bit more challenging than face-to-face classes, where you have a greater ability to keep the students on track. If a student doesn’t have to show up in a room for an hour at a scheduled time and no one can see their involuntary facial expressions when they don’t understand something, the probability increases greatly that they’ll just… fall off.

Read full article

Comments

1. Ask the AI-in-education enthusiast to clarify their premise.

(Slide created by Jane Rosenzweig, Director of Harvard College Writing Center)

2. Ask the AI-in-education enthusiast if they are familiar with recent research indicating that generative AI leads to widespread “cognitive surrender.”

Shot:

“Conceptually distinct from cognitive offloading, which involves strategically outsourcing a discrete task to an external tool (e.g., using a calculator), cognitive surrender represents a deeper abdication of critical evaluation, where the user relinquishes cognitive control and adopts the AI’s judgment as their own….

Across our studies, we observe that when System 3 [i.e., AI] was available, people readily engaged it and frequently adopted its answers. This shift reflects a reallocation of cognitive control rather than mere effort saving. System 3’s fluent, confident outputs are treated as epistemically authoritative, lowering the threshold for scrutiny and attenuating the metacognitive signals that would ordinarily route a response to deliberation. In the case of cognitive surrender, there is a shift in the locus of control, with an external system (System 3) occupying the default position.”

(Paper here https://papers.ssrn.com/sol3/papers.cfm?abstract_id=6097646, with my emphasis)

Chaser:

(Paper https://arxiv.org/abs/2505.01106, BlueSky thread here)

3. If you feel the need to pile on with research, consider citing to this recent report from Stanford showing the complete lack of empirical research to support the use of AI in education.

“Research on how AI impacts K-12 students and educators is still extremely limited.

“As of October 2025 the AI Hub for Education Research Repository contained over 800 academic papers relevant to AI in K-12 education. Our review found that only a small subset (20 papers) produce strong causal evidence. Causal evidence provides the strongest basis for estimating how a tool impacts students and educators. The current causal research is still very limited: we did not identify any high-quality causal studies in K-12 settings in the U.S. for students and very few for teachers.

Students

• Immediate gains with access: AI tools significantly improve student performance on math practice, programming projects, and writing tasks while students have active access to the technology.

• Short-term boost, uncertain transfer: AI improves performance with access but when assessed independently without AI support, effects are mixed.

• Easier doesn’t mean better: AI tools can alleviate students’ cognitive burden and foster positive experiences in learning, but can be at the expense of deeper thinking.

• Pedagogical design matters: Tools designed with pedagogical guardrails (such as AI chatbots for tutoring that provide step-by-step reasoning instead of direct answers) show more promise than general purpose AI tools.

(source, my emphasis added re lack of causal evidence)

4. Ask the AI-in-education enthusiast if they are familiar with any of the recent efforts led by students pushing back hard on the intrusion of AI into their education.

a. In Pennsylvania

“Through its countless new programs and AI-centered events, Penn has positioned AI as an inescapable future that we all must accept in order to achieve success. There is no doubt that AI is part of the current occupational landscape, and we will certainly encounter it long after we graduate. Nevertheless, we attend this institution to develop hard skills, question the world around us, solve problems, produce new ideas, and the ability to think for ourselves. With the University forcing AI into our learning every chance it gets, do we end up gaining knowledge or cheat codes?

“The irony is that as Penn pours endless money and energy into AI advancement in its attempt to get ahead, the University is only quickening its own demise. AI cannot coexist with education — it can only degrade it. As technology advances and workers are replaced by machines, schools are some of the only places we have left to explore and wrestle with human thought. With our own university leading the charge, AI is now corrupting those few sacred spaces and leaving us with nowhere to engage in true scholarship.”

(Source, with my emphasis added)

b. In Colorado

“In an online survey conducted by Flynn Zook, a CU Denver student, nearly 300 respondents weighed in on the agreement. Fewer than 10 expressed clear support, while a small number said they were undecided. Many respondents raised concerns about environmental impact, intellectual property and the potential use of tuition dollars to fund the initiative.” (Story link)

c. In Ohio

(Source)

5. Politely point out that Sal Khan, perhaps the most prominent advocate for the capacity of AI to “revolutionize“ education, has recently changed his tune.

“‘For a lot of students, it was a non-event,’ Khan told me recently about his eponymous chatbot, Khanmigo. ‘They just didn’t use it much.’”

“Kristen Musall, a geometry teacher at Hobart High, gave Khanmigo a try when it first rolled out. Musall appreciated its encouraging, teacher-like tone, but she found that students didn’t really care for the bot….Musall no longer uses Khanmigo in her class. She says there’s been more enthusiasm for the product among administrators than teachers in her school.

“Kristen DiCerbo, the organization’s chief learning officer, said AI can only respond to students based on what they ask. And it turns out, she said, ‘Students aren’t great at asking questions well.’”

(From Chalkbeat’s Matt Barnum)

6. Direct the AI-in-education enthusiast to the PureGenius website to see if they get the joke.

(source; feel the future)

7. Ask the AI-in-education enthusiast if they are familiar with the broader pushback against the intrusion of education technology into schools led by educators and parents.

“McPherson Middle School, about an hour’s drive from Wichita, is at the forefront of a new tech backlash spreading in education: Chromebook remorse.

“Schools in North Carolina, Virginia, Maryland and Michigan that once bought devices for each student are now re-evaluating heavy classroom technology use.

“Now children’s groups and educators concerned about screen time are turning their attention to school-issued laptops and learning apps. Parents are flocking to support efforts, like Schools Beyond Screens and the Distraction-Free Schools Policy Project, to vet and limit school tech.

“Sarah Garcia, also 13, said spending less time online had prompted students to talk more. ‘Since we don’t have our Chromebooks in front of our face,” she said, “most people now interact with their, like, peers and stuff.’”

(source)

8. Gently remind the AI-in-education enthusiast that we have evidence in our own lifetime that highly addictive products marketed to children that cause serious harm are something we can address through policy and norms.

(source)

9. If the AI-in-education enthusiast has the audacity to cite f***ing AlphaSchool as counterexample and “proof of what’s possible,” liberally reference any or all the myriad reasons this is one of the most embarrassing possible arguments they could make.

“Former Alpha School employees told me that the company’s increasing reliance on generative AI in every aspect of its operation, as well as the constant monitoring and tracking of every student’s mouse movements, is making students anxious and does not always provide the quality of education Alpha School advertises to parents.

““All educational content is obsolete. Every textbook, every lesson plan, every test, all of it is obsolete because gen AI is going to be able to deliver a personalized lesson just for you,’ Joe Liemandt, Alpha School’s ‘principal’ and the founder of Trilogy, the company that owns many of the apps used by Alpha School, said in a podcast interview published last year.

“When a student requires help with additional questions, the chatbot fails to identify which specific question is being addressed,’ an internal Alpha School document outlining issues with AlphaRead says. ‘Accuracy of the content provided by the [AI] tutor is a concern. There are instances where it not only delivers incorrect answers but also provides convincing yet flawed justifications. Despite raising multiple queries about a particular answer, the chatbot erroneously confirmed an incorrect option as correct.”

(source)

From teacher Michael Pershan:

“Alpha is not trying to provide the best, most ambitious math or ELA education possible according to conventional understandings of that term. If they were, they’d keep studying ELA/math in the afternoon. Instead, their goal is to minimize the time spent on core academics while maximizing skills.

“This is unusual! This is not what most schools are trying to do!

“What’s most novel about Alpha School and Math Academy is their fundamental orientation towards K-12 schooling. The goal, quite expressly, is to minimize it and move on. Move on to what?”

(source)

From teacher Dylan Kane:

[Dylan explains how AlphaSchool measures its “success” by demonstrating how much faster students improve on the NWEA MAP, a mediocre-to-shitty assessment of student learning, compared to national average of students. In 2024-2025, using mediocre-to-shitty ed-tech product iXL, AlphaSchool students outgained the national average by 2.6x. Dylan reports on how students are doing this year using AI: on math, 2.5x, and on reading 2.8x]

“Again, let’s take these numbers at face value for a minute. Last year, the program was mostly iXL + a culture laser-focused on motivating students. There was little to no generative AI involved. This year, Alpha School has overhauled their academics, released their own platform, and incorporated generative AI throughout. They are finally doing what they say they are doing: AI-driven schooling. And the results are…more or less the same?

“This is completely fascinating to me. 100 million dollars, tons of hype on the internet, grand claims about the future of education. And the results haven’t budged from bribing kids to try hard on iXL?”

(source)

10. If all else fails, try appealing to the poetry of human existence. But don’t hold your breath.

(Source)

Our days are indeed precious on this earth. Today, the The New York Times published a long story about what happened to my father due to his reliance on AI for medical guidance. I am very grateful to reporter Teddy Rosenbluth for sharing his story with the world, and to all of you for your enduring support.

Subscribe now