What Happened to Roam Research? Magic, Then Friction

Roam Research pioneered networked note-taking, bidirectional links and an automatically generated knowledge graph, and for a stretch around 2020 it felt like magic, as if connected thinking were a feature you could finally install. It built a famously devoted following, the “#roamcult,” and then it faded from the center of the conversation. The reasons are layered: its signature features got copied by cheaper and free competitors, its pricing and rough edges cost it adoption, and, most fundamentally, the tool could not deliver the thinking it seemed to promise. People built sprawling, beautiful graphs that produced no insight, because a connection you do not understand is not understanding. Roam’s real legacy is twofold: it popularized ideas that are now everywhere, and it taught a clarifying lesson about the limits of tools, that no app makes you think, because the structure that matters has to be built in your head.

What was Roam, and why did it feel like magic?

A note-taking app built around connection rather than storage. Where traditional notes live in folders and files, Roam Research was built on bidirectional links: you could link any note to any other, and Roam automatically showed you the reverse, every other note that linked back, plus a visual graph of the whole web. It also worked as an outliner where every bullet was its own addressable block. The effect was that your notes stopped being a filing cabinet and became a network, which mapped onto a powerful intuition, that knowledge is a web of connections, nodes and edges, not a hierarchy of folders.

That is why it felt magical: it seemed to externalize the way good thinking actually works. The promise was seductive, link your notes the way your mind links ideas, and the connections would reveal insights you could not see otherwise, the backlinks surfacing unexpected relationships, the graph showing the shape of your thought. For a certain kind of user, the knowledge worker drowning in disconnected notes, this was intoxicating, and the enthusiasm was real, not just marketing. Roam genuinely did make connection a first-class feature in a way mainstream tools had not, and it deserves credit for that.

Why did the magic fade?

Three forces, in rising order of importance. The first was competition: Roam’s signature features were not as defensible as they looked, and bidirectional linking, graph views, and block references were quickly copied, most consequentially by Obsidian (free for personal use, local files) and by Notion and others, so the thing you needed Roam for became available cheaper, more flexibly, and without lock-in. When the moat is a feature, the moat drains fast.

The second was friction and positioning: Roam’s pricing was high relative to free alternatives, the interface had a steep learning curve, it was cloud-only with data-ownership concerns that the local-first crowd disliked, and its early hype set expectations no tool could meet. None of these alone was fatal, but together they made switching easy and staying expensive.

Factor	What happened	How decisive
Competition	Obsidian, Notion, others copied the key features	Major: eroded the practical reason to choose Roam
Pricing and friction	Higher cost, steep curve, cloud-only, data concerns	Significant: made switching easy
Over-hype	Promised a thinking upgrade no app could deliver	Significant: set up disappointment
The tool-vs-thinking gap	Graphs grew; insight did not	Fundamental: the deepest reason the magic faded

The third and deepest force is the one worth dwelling on, because it explains why even devoted users drifted: the tool could not supply the thinking it implied, and many users discovered that an elaborate graph is not the same as understanding.

Why couldn’t the tool deliver the thinking?

Because connection-making is cognitive work, and Roam automated the recording of connections without performing the work that makes connections meaningful. The app could store that note A links to note B, and draw a line between them, but it could not do the understanding that should have produced the link, the actual reasoning about why these ideas relate and what follows from that. People discovered they could build enormous, gorgeous graphs that were, on inspection, hollow: thousands of links, very little insight, because the links were gestures, not earned connections.

This is the crucial distinction, and it is the lesson of concept-mapping research: as the theory underlying concept maps makes clear, the value of a connected structure comes from the meaningful, effortful act of articulating why two ideas relate, the labeled, understood relationship, not from the mere existence of a line between them. Roam made drawing the line frictionless, which sounds like a benefit but quietly removed the very effort that builds understanding. A graph of unexamined links is the collector’s fallacy in network form: it produces the feeling of building knowledge while building almost none, the same hollowing diagnosed in why the collector’s fallacy ruins personal knowledge management.

What is the real lesson of Roam?

That no tool can install the thinking, because the structure that matters lives in your head, not your app. This is First Brain before Second Brain stated as a postmortem: Roam was a powerful Second Brain that many people hoped would substitute for a First Brain, and it could not, because connections in software are inert unless the corresponding connections exist in your biological knowledge graph. The users who got real value from Roam were the ones already doing the thinking, using it to externalize a structure they understood; the users who hoped the graph would think for them got a beautiful, empty database and eventually quit.

The pattern is general and recurs with every new tool, which is why the lesson outlasts the app. The same arc, magical-seeming knowledge tool, devoted following, disillusionment when it turns out not to do the thinking, plays out repeatedly in personal knowledge management, the broader disillusionment that thoughtful PKM writers like Maggie Appleton captured in reflections on tending a Roam knowledge garden and that the critique of endless organizing named directly: organizing and linking are not thinking, and a system that optimizes the former can quietly let you avoid the latter. Building the understanding that makes any tool useful, rather than hoping the tool supplies it, is exactly the project Building Your First Brain, free for the first 1,000 readers, is built around, and it is why the question to ask of any thinking tool is not “how powerful is its graph” but “does using it make me do the connecting, or let me skip it.”

What are the honest caveats?

Several, to be fair to Roam and accurate about its status. First, Roam is not dead: it still exists, still has committed users, and remains genuinely useful for people whose workflow fits it, so “faded” means “fell from the center of the hype,” not “failed entirely,” and plenty of people still get real value from it. Dismissing it as a flop would be wrong.

Second, the tool-vs-thinking critique is not unique to Roam, and singling it out is slightly unfair: the same point applies to Obsidian, Notion, and every PKM tool, including the ones that displaced it, so the lesson is about the category, not the company. Third, tools genuinely do matter, this is not an argument that software is useless; a good tool reduces friction and supports the thinking, and the right framing is that Roam’s features were real improvements that worked best for people who brought their own thinking discipline, not a scam that promised what it could not give. The balanced verdict: Roam felt magical because it made connection a first-class feature and seemed to promise that connected thinking could be installed; it faded because competitors copied its features, its pricing and friction hurt it, and most fundamentally because the graph could not supply the understanding, an elaborate web of links is not insight. Its lasting contribution is having popularized networked note-taking and having taught, by example, that the thinking has to happen in you, not in the app.

Key takeaways: what happened to Roam Research?

Roam pioneered networked note-taking, bidirectional links and an automatic knowledge graph, and felt magical because it seemed to make connected thinking an installable feature, sparking a devoted following around 2020. It faded for three reasons: competitors like Obsidian and Notion copied its signature features more cheaply and flexibly, its pricing and friction and over-hype hurt adoption, and most fundamentally the tool could not supply the thinking it implied, leaving users with elaborate graphs and little insight because unexamined links are not understanding. The deepest lesson is First Brain before Second Brain: no app installs the structure, which has to be built in your head, and that lesson applies to every PKM tool, not just Roam. Roam is not dead, and tools do matter, but they support thinking rather than replace it.

Frequently asked questions

What happened to Roam Research?

Roam pioneered networked note-taking, bidirectional links and an automatic knowledge graph, and built a cult following around 2020, then faded from the center of the conversation. Three things caused it: competitors like Obsidian and Notion copied its key features more cheaply and flexibly, its higher pricing, steep learning curve, and cloud-only model created friction, and most fundamentally the tool could not deliver the thinking it seemed to promise, users built elaborate graphs that produced little insight. It still exists and has users; it simply lost its hype-cycle dominance.

Why did Roam Research feel so magical at first?

Because it made connection a first-class feature instead of an afterthought. Where traditional notes sit in folders, Roam was built on bidirectional links, link any note to any other and it automatically shows the reverse links plus a visual graph of the whole web. This mapped onto the intuition that knowledge is a network of connected ideas, not a hierarchy of folders, and it seemed to externalize how good thinking actually works, promising that linking notes the way your mind links ideas would surface insights you could not otherwise see.

Why did people stop using Roam?

A mix of practical and fundamental reasons. Practically, free or cheaper tools like Obsidian and Notion copied bidirectional linking and graph views, removing the unique reason to pay for and learn Roam, and Roam’s pricing, learning curve, and cloud-only data model added friction. Fundamentally, many users discovered that building an elaborate graph of links did not produce understanding, the tool recorded connections but could not do the thinking that makes connections meaningful, so the promised insight never materialized and enthusiasm faded.

Is Roam Research dead?

No. Roam still exists, still has a committed user base, and remains genuinely useful for people whose workflow fits its outliner-and-links model. “Faded” means it fell from the center of the personal-knowledge-management hype cycle, not that it failed entirely. It pioneered features, bidirectional linking, graph views, block references, that are now standard across the category, so its influence persists even where its market share does not. Plenty of people still get real value from it.

What is the lesson of Roam’s rise and fall?

That no tool can install thinking for you, because the structure that matters lives in your head, not the app. Roam could record and draw connections but could not perform the understanding that makes a connection meaningful, so users hoping the graph would think for them got a beautiful, empty database. The value came to those who brought their own thinking and used Roam to externalize it. The lesson applies to every knowledge tool: build the understanding yourself; the tool supports it, but cannot replace it.