The Internet No Longer Has an Information Problem. It Has a Memory Problem.
Every day we publish, read, and forget thousands of news stories. The challenge is no longer finding information, but preserving enough context to understand how stories change over time.
Over the last few years, the way I consume news has changed.
Not because I read less. Quite the opposite.
I read more publications, headlines, and sources than ever. Yet I increasingly have a strange feeling: I am better informed, but I find it much harder to understand what is actually happening.
We live surrounded by information. New headlines about politics, technology, economics, science, and culture appear every few minutes. The next day, others replace them. A week later, most have disappeared from our memory.
Then someone asks an apparently simple question:
What exactly has happened with OpenAI?
Or with a company, a conflict, a political party, a technology, or a country.
We remember headlines, statements, or isolated facts, but struggle to reconstruct the complete story.
It is not a lack of information. It is a lack of context.
The Internet remembers documents. We remember stories
The Internet is built around documents. People understand the world through stories.
When we think about the collapse of a company, the evolution of a conflict, or how a technology transformed an industry, we do not remember individual articles. We remember a sequence of connected events.
Yet information online remains fragmented. Each news report explains one small change. Each article spends a few hours on a homepage. Each publication tells a different part of the story. The reader must then reconstruct the context mentally.
That works reasonably well when a story is simple or lasts a few days. It works far less well when it unfolds over months or years. The more important a subject becomes, the more information it generates—and the harder it becomes to understand.
The problem is no longer finding information
For a long time, the Internet’s great challenge was access to information. Today, the opposite is true: we have more than we can process.
Search engines find thousands of results. Aggregators display hundreds of headlines. Social networks mix facts, opinions, reactions, and noise. AI models can generate summaries in seconds.
But I still miss something very simple:
What has changed since the last time I looked at this subject?
I do not need to reread what a company is whenever it appears in the news. I want to understand which decisions it has made, which events truly matter, what their consequences were, and how the story has evolved.
I want to distinguish a genuine development from a repetition; a real change from another headline retelling the same thing.
News is necessary, but it is not enough
A news story explains something that has just happened. That is its job, and it remains essential.
Without journalists, publications, and original sources, there would be nothing to organize, verify, or place in context. The problem comes afterward.
When a story leaves the homepage, it joins a vast, disconnected archive. It still exists, but much of its practical usefulness is lost. The next report reconstructs part of the context. Another source adds a different piece. A third contradicts or qualifies what came before.
The story exists, but it is distributed across dozens or hundreds of documents.
The information is there. The memory is not.
Thinking in entities, not only articles
I began to wonder what would happen if we organized information differently—not only around articles, but around the things those articles describe.
People. Companies. Organizations. Countries. Technologies. Events. In other words: entities.
Each report could stop being an isolated object and become a change in the history of an entity. Articles would remain the original sources, while also contributing to an accumulating view of:
- the current state
- what changed recently
- when it happened
- which sources support it
- how it connects with what came before
The difference appears small, but it changes the central question. Instead of asking “what is the latest news?”, we can ask “what has actually changed?”
From archives to living pages
Most pages on the Internet are snapshots. They are published at a particular moment and remain almost unchanged forever. But many of the things we try to understand are not static.
A company changes. A law evolves. A technology finds new uses. A conflict moves through different phases. A person changes their decisions, responsibilities, or influence.
It therefore makes sense to imagine living pages: places that do not merely describe what something is, but preserve how it has changed.
Not an endless collection of headlines. A structured memory.
An experiment called Atlas
Because I could not find a tool that approached the problem in the way I had in mind, I decided to build a small experiment. I called it Atlas.
Atlas tries to maintain living pages about people, companies, organizations, places, technologies, and events. Each page seeks to answer three questions:
- What is its current state?
- What changed recently?
- Which sources support those changes?
It does not create the original news. It does not seek to replace publications. It depends entirely on them.
Its role is to connect information that already exists, so that context does not disappear whenever a story stops being new. Atlas remains an experiment, not a definitive answer.
But building it has made me think that the future of information may not be only about publishing and searching better. We may also need to remember better.
Artificial intelligence cannot solve this alone
AI models can summarize enormous amounts of information, but the quality of their answers depends on the context they receive. When that context is incomplete, outdated, or disorganized, the answer may be too.
An AI system can reconstruct a story from many sources, but it does not always know which event represents a meaningful change, which information repeats an earlier claim, or which version has become obsolete.
Part of the future may therefore be not only more capable models, but better contexts:
- clear timelines
- identifiable changes
- visible sources
- information that can be updated
- relationships between entities
Before asking for better answers, we may need to prepare better material to reason about.
Information needs memory
For decades, we greatly improved our ability to publish. Then we improved our ability to search. Now we are improving our ability to summarize and generate answers.
But we still struggle to preserve context over time.
Perhaps that is one of the next layers the Internet needs: a layer that does not replace sources, but connects them; that does not eliminate articles, but makes use of what each contributes; that does not treat every news item as an isolated object, but as part of a story that continues.
The Internet’s problem is no longer a lack of information.
It is a lack of memory.
Should we keep organizing the Internet as a collection of isolated documents, or is it time to build pages capable of remembering how the world changes?