Twenty-Five Years of "Never Forget"

The water never stops at the 9/11 Memorial. It falls along the walls of both pools, gathers, and drops again into a smaller square at the center until it disappears from view. Around the edges, the names of the people killed are cut into bronze parapets, low enough to touch.

The week of the anniversary this year, people moved around the pools the way they always seem to. Some read. Some photographed a name and stepped back. Some simply stood still. The city kept moving on every side of the plaza, and the plaza itself moved at a different speed.

Roy Roberts, a 37-year-old engineer who lives in Midtown, had come down after hearing there would be an anniversary ceremony and wanting to see it for himself.

When I asked what he would tell someone visiting for the first time, he talked about the names. Look at them closely, he said. Check how old they were. Think about how much time has passed, and try to relate to the people behind them.

"Sometimes it's the same as yours," he said.

It is a simple instruction and a hard one to follow. Nearly three thousand people died that morning, a number too large to hold in the mind. A single name works differently, and learning how old that person was works differently still. It returns the event to the scale of one life: someone who had plans for that evening, someone who was about your age, someone who should still be here.

Twenty-five years later, that return takes more effort. Fewer of us carry the event as memory. More of us receive it as history. A question followed me from the Memorial to a panel at the CUNY Graduate Center: once living memory begins to give way, who keeps the record, and what happens when the record itself begins slipping away?

I was eight years old. I usually took the bus to school, but that morning my mom insisted on driving me. Before we left, the living room television was playing something on a loop that looked like an explosion. I didn't understand what I was seeing. I'm not sure I knew what the towers were.

By the time I got to school, everyone was already talking. And I remember, with a clarity I still can't explain, a classmate raising her hands and bringing them together to show the rest of us what the planes had done.

Later, at lunch, the announcement came that the country was under attack and that all planes had been grounded. It was delivered by the lunch lady, the one we all feared and respected in equal measure. What registered most for a lunchroom full of first through sixth graders was that there would be no recess.

I couldn't tell you what our teacher said that day. I couldn't tell you what I ate or what I did when I got home. But that gesture has stayed with me for twenty-five years, along with the lunch lady's voice and the strange fact that my mom chose to drive.

My experience of that day was distant and small, a child in a classroom far from Lower Manhattan. The families who lost someone, the people who escaped the towers, the responders who went toward them and the neighbors who lived for months beside the site carry something of an entirely different order.

Still, the fragment is instructive. Human memory rarely preserves an event whole. It keeps pieces: a screen, a room, a parent's instinct, an announcement, a gesture. Those pieces are vivid, selective and sometimes wrong. That is one reason societies build archives.

The core idea was simple enough to state and unsettling enough to sit with. In 2001, many of the most urgent threats were physical. They were visible. They moved through recognizable structures. Today, the most consequential risks can begin on a keyboard, travel through an encrypted chat, and end in the physical world. The line between online and offline, as Weiner put it, has become "completely collapsed."

The panel at the Graduate Center gathered members of the team that built the September 11 Digital Archive to reflect on what they made and what has happened to it since. It included three of its original builders, Stephen Brier of the Graduate Center, Tom Scheinfeldt of the University of Connecticut and Fritz Umbach of John Jay College, alongside Stefano Morello, who now directs the Graduate Center's American Social History Project and oversees the archive's care.

The evening itself grew out of the doctoral research of a Graduate Center student who is raising children in New York City. Watching them encounter September 11 as history, rather than as something they lived, led her to ask what it would mean to revisit an archive built in the immediate aftermath for a generation meeting the event twenty-five years later.

The moderator, a New Yorker who was in the city that day, opened by reminding the room why any of this mattered. The conversation, she said, was for the people who were lost.

One of the evening's hosts recalled the city in the weeks after the attacks. Commuters passing through Grand Central Terminal and Penn Station were confronted by the faces of the missing, taped to walls and columns by families still searching. He suspected those images shaped the instinct that led historians at CUNY and George Mason University to start collecting.

Before answering his first question, Brier asked for a moment to honor Roy Rosenzweig. Rosenzweig was a historian at George Mason University, and his Center for History and New Media partnered with Brier's team to build the archive. When the work began, the center simply bore its original name. Rosenzweig, Brier's close friend and colleague, died in the years since, and the center now carries his name as the Roy Rosenzweig Center for History and New Media. None of them, Brier said, could have done this work without him.

Brier described September 11 as one of the first historical events of real consequence to be recorded substantially in digital form. It is hard to remember how early that moment was. Email was still new. Digital photography was just spreading. Blogs were still called weblogs.

Scheinfeldt added a detail that is easy to forget. In 2001, the web was mostly a place people went to read. They shopped, checked the news and wrote to friends. Very few people published their own lives online, and persuading someone to upload a photograph or write out their story took real effort. The social web that now records nearly everything had barely begun.

An enormous share of the record was being created in formats nobody had thought to preserve.

About two months after the attacks, the Sloan Foundation called the historians in with a question that shaped everything that followed: what will historians want to know fifty years from now? The answer became one of the earliest large born-digital public history archives, a collection that eventually grew to more than 150,000 stories, photographs, emails, audio recordings and other materials.

The official record of September 11 was always going to survive. Government agencies, the police and fire departments and the major news organizations had their own archives and their own reasons to keep them. The panelists were far less confident about everyone else.

So the archive went looking for ordinary experience, and much of what it holds is ordinary by design. Scheinfeldt described photographs of people walking uptown, of streets being swept, of memorials assembled by strangers: stuffed animals tied to fences, flowers, handmade signs, rose petals arranged in the shape of the towers. One man walked Manhattan collecting flyers and eventually contributed thousands of them. There is a large body of what Scheinfeldt called digital folk art, made on personal computers and passed along by email.

These are the traces people left while trying to get their hands around something too large to hold. They rarely make it into textbooks, yet they are often what makes the past legible.

The archive also became what Scheinfeldt called an archive of last resort. The National September 11 Memorial & Museum would not open for another decade, and in the meantime, collections that had nowhere else to go found their way to the historians. Those included a large body of professional photography of the site and the recovery, whose physical prints are held by the New-York Historical Society; hundreds of video interviews recorded near all three attack sites in Lower Manhattan, at the Pentagon and in Shanksville, Pennsylvania; and the Sonic Memorial Project, which gathered audio including messages people left for loved ones that morning. The Department of Justice handed over roughly twenty thousand public comments submitted about the Victim Compensation Fund. Some of that material, Scheinfeldt said, may now survive only in the archive.

Even the country's most established institutions were unsure what to do with digital material. People who brought emails or digital photographs to the Library of Congress or the Smithsonian, Scheinfeldt said, were often redirected to the historians, who by their own account were figuring it out as they went.

For its first decade, the archive was among the first results anyone found when searching for September 11 online. For ten years, a small team of university historians held one of the most widely consulted records of the day.

We tend to assume that what goes online stays online. The archive's history suggests something closer to the opposite. Of the more than 150,000 items collected, roughly 64,000 were described during the panel as publicly accessible today.

A handful of items were withheld for ethical reasons, privacy or permissions. The largest share, Scheinfeldt estimated, is inaccessible for technical reasons: corrupted files, formats that no longer play in a modern browser, material that depended on software or hardware that no longer exists.

Umbach recalled spending hundreds of hours collecting video that today will not play at all. At the time, he said, he wondered whether they should have transferred it to film, which would have proven far more durable. A digital archive, he observed, gradually becomes a museum of obsolete equipment.

Scheinfeldt described a subtler kind of loss. In a traditional archive, a letter filed in the wrong folder is still sitting there, waiting for a patient researcher. In a digital archive, an item separated from its description can be effectively gone: present on a server and impossible to find.

The institutional story proved just as fragile. The initial grants ran out in 2003. That same year, the archive was handed to the Library of Congress on a hard drive, in a small ceremony that the panelists remembered as the first time the collection had taken any physical form at all. It was also, they said, the first digital collection the Library had ever formally taken in.

The Library sent copies to university library science programs, asking each to develop a preservation strategy. According to the panelists, the long-term preservation the agreement promised never materialized, and the Library never committed the staff or resources to follow through. Scheinfeldt called it a missed opportunity for the entire field. Many smaller digital archives now sit in a similar state: stewarded, but without a real plan for long-term access. Apart from one federal grant for a migration, the September 11 archive has run largely without dedicated funding since.

The names at the Memorial will outlast everyone reading this. The digital record of the same event has needed rescuing more than once in twenty-five years. Bronze ages slowly. Digital memory can disappear unnoticed.

The panelists were unusually candid about their own blind spots. Umbach explained that the first five hundred stories the archive received came from within a few blocks of his Upper West Side apartment, because the team was drawing on its own social networks. Then CNN covered the project, and thousands of submissions arrived almost overnight from a far broader cross-section of the country.

A future historian reading those stories in order, he warned, might conclude that American opinion shifted sharply within a year. What actually shifted was who was being asked.

Archives are constructed. Someone decides what to collect, whom to ask, how to label it and what becomes searchable. Those decisions shape what later generations believe happened.

The team tried to widen the frame. With the Museum of Chinese in America and support from the Rockefeller Foundation, they gathered testimony from Chinatown residents in three Chinese dialects, a community profoundly affected by the attacks whose stories would likely never have arrived on their own. They placed laptops in Latino health clinics so patients could share stories while they waited. Efforts to collect testimony from Arab and Muslim New Yorkers were far less successful. Umbach said those communities were understandably wary in the political climate that followed the attacks, and he described the gap as one of the project's deepest frustrations.

He also named something the team could never have predicted. In 2001, they shared a certain techno-optimism: the belief that digital archives would take history out of the hands of a small professional class and give it to everyone, making every person their own historian. Twenty-five years later, he suggested, that vision has yet to arrive.

When the moderator asked whether archives of everyday life risk losing the larger political story, Scheinfeldt answered with humility. An archive can only do so much, he said. The interpretation belongs to the historians who use it.

An archive, in the end, is a structure through which memory becomes visible. Its shape determines whose memory that is.

If there was a single thread running through the evening, it was work. Unglamorous, repetitive, largely invisible work.

Brier made a point of naming the staff, graduate students and collaborators who built the archive. Morello showed a slide of contributors credited on the current site and noted that more than a thousand people had touched the project in some way. Every item that made it online had to be described, organized and cleared for permission. Every file that survives has, at some point, been checked, moved or repaired by a person.

Morello offered a sense of scale. On another recent project, reviewing model-generated tags for roughly two thousand primary sources took six people about six months. The September 11 archive holds tens of thousands of items.

Brier compared it to the public projects the Works Progress Administration funded in the 1930s: meaningful, labor-intensive and ideal for graduate students, if only someone would pay for it.

Readers of last month's Chronicle will recognize that agency. In the summer of 1936, in the depths of the Depression, WPA funding helped New York open eleven public pools in a single season. Those pools were built from concrete, tile and water. An archive is built from something harder to see. Both depend on a city deciding that shared infrastructure is worth the investment, and both begin to fail when that decision stops being renewed.

Every document a researcher has ever found in an archive was there because someone decided to keep it, describe it and look after it. The labor is invisible precisely when it succeeds.

One of the evening's hosts, introducing the panel, described his own morning. He was near Washington Square Park when the first plane hit. Cell phones weren't working, so he reached his wife through an instant messenger, then walked out of Manhattan. And yet, he said, when he thinks back, he feels certain he watched the towers from the river, from Midtown, from New Jersey, vantage points he never actually occupied. He has seen the footage so many times that he can no longer fully trust his own recollection.

The images meant to preserve the event had, in some small way, rewritten how he remembers it.

Another host recalled his morning with the precision these memories carry: walking up Broadway after voting at his neighborhood school when construction workers shouted that a plane had hit the World Trade Center. Both men were there. One of them no longer fully trusts what he remembers.

Morello described a related pattern emerging from his early analysis of the archive. As his team began reading across the testimonies, he noticed that references to Iraq appear in personal accounts only later in the collection period. He offered it as a question worth studying: how much do official narratives, arriving after the fact, shape the way ordinary people come to describe what they lived through?

When does the documentation of an event begin reshaping the memory of the event itself? The panel left that question open, and it may be unanswerable. But it complicates any easy trust in either form of remembering, and it makes the case for keeping both: a record can steady a memory that drifts, and a memory can give meaning to a record that might otherwise sit unread.

One of the hosts placed September 11 among the handful of moments that generations remember by location: where you were when Kennedy was shot, where you were when the Challenger broke apart. Every generation adds its own. Plenty of people my age can still tell you exactly where they were when they heard Kobe Bryant had died. Over time, the host noted, those moments pass from memory into legend. Events lived by one generation become history for the next.

For twenty-five years, the defining question about September 11 has been where were you. A growing share of Americans reaching adulthood were too young to remember the day or were born after it. For them, the question is becoming how did you learn about it.

My older friends and colleagues still talk about life before and after. They remember walking a relative to the gate at the airport. They remember concerts without metal detectors. For people my age and younger, the security line, the bag check and the posted warnings are simply how things are. We have spent nearly our entire lives in a world shaped by that morning, often without noticing its outline.

The answer to how did you learn about it might be a school lesson, a documentary, a parent's story, a visit to the Memorial or a search that leads to an archive. The quality of what the next generation inherits depends on choices being made now: what gets repaired, what gets described, what gets funded and whose stories are included.

For teachers, Brier encouraged letting students immerse themselves in primary sources, listening as well as reading. Umbach offered a sharper warning. His students, he said, tend to assume that if something doesn't surface in a quick search, it doesn't exist. Archives like this one demand something closer to detective work.

He also offered what may be the archive's most durable lesson. Everyone today knows how the story turned out. In the weeks after the attacks, nobody did. The archive holds that uncertainty in a way retrospective accounts rarely can: people writing, photographing and grieving without knowing what would come next.

Roy Roberts, the engineer I met outside the Memorial the week of the anniversary, remembered that uncertainty too. He was in school that morning, worried because his father worked in Manhattan. His father turned out to be uptown, near Central Park. What Roberts took from the anniversary was practical. He talked about the responders who became ill in the years that followed, including a firefighter he knew who developed cancer, and about preparing so the response is faster if something like this happens again. The long consequences of that morning are still being documented. Roberts framed his hope as learning rather than blame.

Morello closed with a look forward. His team has begun running large language models locally to transcribe handwritten notes, extract text from images, transcribe audio that was never transcribed and fill in missing descriptions, with the goal of making tens of thousands of hard-to-find items discoverable again. He pulled roughly seventy thousand items into the lab for the effort, including nearly two hundred hours of audio awaiting review.

The work creates its own labor. Every output needs a person to check whether the model invented something. The team is also combing the records for personal information that should never have been public, including details about people who were children at the time.

When an audience member pushed the panel to be more optimistic about AI, the panelists agreed the possibilities are real. They also returned to the same point: more data means more review, and review means people.

Another audience member, an archivist at the Center for Puerto Rican Studies, raised a point worth sitting with. People today post everything, he noted, but most of that material lives on platforms owned by private companies. A public archive offers something those platforms never promise: access that belongs to the public.

Scheinfeldt noted that the Internet Archive, working with the Library of Congress, captured September 11 related websites at frequent intervals in the years after the attacks, sometimes almost daily. Brier added that the lesson his team carried forward was to curate collections rather than accumulate everything.

There is a symmetry here. The archive began because an emerging technology was producing records faster than institutions could preserve them. Twenty-five years later, another may help recover some of what was lost. It will still take archivists, funding, judgment and years of patient review. The tools change. Stewardship remains a human job.

The archive has also shaped the field. Dozens of scholarly papers cite it, and later projects documenting the pandemic pointed to it as their model.

What Memory Requires

A memorial preserves names. An archive preserves records. People preserve stories. Institutions preserve context. Each depends on someone doing the work: cataloging, migrating, repairing, funding, teaching, visiting, asking.

Bridges get inspected. Parks get tended. Collective memory rarely has a line in anyone's budget, yet it shapes how a city understands itself. When it goes unmaintained, the loss arrives gradually: one unreadable file, one untold story at a time.

Back at the Memorial, the water keeps falling and the names hold steady in bronze. Roy Roberts's advice still seems like the right place to begin: find a name, learn how old they were, and let them be a person for a moment.

For those who lived through September 11, memory is something carried. For those who come after, it will be something inherited. What they inherit depends on what we choose to keep.