The past and potential of Dataview with Michael Brener and Marcus Olsson

Dataview has become a metadata-and-query layer for Obsidian vaults, turning scattered Markdown notes into something closer to a searchable database—fast enough to render results in milliseconds. Michael Brenner, the plugin’s primary maintainer, frames Dataview as a “power user” step beyond links and tables: instead of forcing notes into pre-built structures, it lets users attach metadata (via YAML front matter or inline fields) and then generate lists, tables, calendars, and other custom views from arbitrary queries.

Brenner describes the core workflow as “defer the organization.” Users write notes with minimal overhead—often just a note plus tags and a few metadata fields—then later decide how to view and combine that information. Dataview effectively acts as a search engine or metadata engine across the entire vault, supporting a full query language that can filter and visualize notes by any attribute. Performance is a major selling point: query times are typically under 100 milliseconds on average, with render times often under 20 milliseconds, enabling near-instant feedback when exploring a vault.

Dataview’s influence extends beyond end users. Brenner says other plugin developers can call Dataview directly, running queries and receiving JavaScript or TypeScript results that plugins can render inside their own interfaces. He notes that multiple existing Obsidian plugins use Dataview’s API, including ones that integrate optionally and others that provide folder-based ways to use Dataview.

The plugin’s origins were modest. Dataview began as a roughly 50-line addition to Obsidian’s sample plugin—initially to list files matching a substring. Over time, the idea evolved into a small scripting language for listing notes by tag or folder, with the broader goal of letting people write notes first and query them later in any arrangement they want.

Brenner credits the Obsidian community for Dataview’s rapid growth. After he shared early screenshots in Discord in late February 2021, users and contributors expanded documentation, answered query questions, and helped refine how people learn the query language. He also emphasizes that Dataview is a long-term project he works on in cycles—periods of heavy development followed by pauses to ensure stability.

Looking ahead, Brenner highlights two major directions. First is data security and portability: because Dataview is a query layer rather than a primary storage format, he wants users to be able to export the generated tables and views as real Markdown, so the information remains accessible even if someone leaves Obsidian or stops using the plugin. Second is improving how users manage and discover large amounts of metadata—especially visualization, migrations, and editing workflows.

A recurring theme is lowering the barrier for non-programmers. Dataview queries are written in code blocks, and Brenner acknowledges that the query language has grown complex. He’s working on an alternative query syntax intended to support a bare-bones visual editor—potentially with a GUI that generates queries for common operations, while still allowing advanced users to switch to raw query mode. He also points to the value of examples and documentation, noting that learning Dataview often comes down to “programming by example.”