Collaborative Document Editing: How Real-Time Co-Authoring Works
The ability for multiple people to edit a document simultaneously — seeing each other's changes in real time — has transformed how teams write, review, and produce documents. What appears seamless from the user's perspective involves sophisticated distributed computing algorithms operating transparently in the background. Understanding how real-time collaboration works helps users get the most out of collaborative tools and helps developers building such systems choose the right technical approach.
The Core Problem: Concurrent Edits and Conflict Resolution
When two users edit the same document simultaneously on different devices, their clients start from identical states but diverge as each makes independent changes. The fundamental challenge is merging these concurrent edits correctly — ensuring that both users' changes are preserved and the document reaches a consistent final state. Consider the simple case: User A deletes the third character of a word while User B inserts a character before it. When A's deletion reaches B's server, it must be transformed by B's insertion to still refer to the correct character position in the modified document. This transformation — adjusting operation positions to account for concurrent changes — is the heart of collaborative editing algorithms.
Operational Transformation (OT)
Operational Transformation, first described in 1989 and implemented in Google Docs and the now-discontinued Google Wave, is the dominant algorithm for real-time collaborative text editing. OT represents document changes as operations (insert character at position X, delete character at position Y) and defines transformation functions that modify an operation's parameters to account for concurrent operations that have already been applied. The mathematics works correctly for simple text operations but becomes significantly complex when handling rich text with formatting, nested structures like tables, or branching edit histories. The Google Docs implementation reportedly required years of engineering effort to achieve correctness across all edge cases.
CRDTs: A More Recent Approach
Conflict-Free Replicated Data Types (CRDTs) offer an alternative approach to consistency in distributed collaborative systems. Rather than transforming operations relative to each other, CRDTs design data structures such that concurrent modifications always produce the same result regardless of the order in which they are applied — a property called commutativity. For text editing, sequence CRDTs like LSEQ, RGA, and Logoot assign unique, stable identifiers to each character, allowing any two concurrent versions to be merged deterministically. CRDTs are increasingly used in peer-to-peer and offline-first collaborative applications because they do not require a central server to order operations. Automerge and Yjs are popular open-source CRDT libraries used by many document collaboration tools.
Presence, Cursors, and Awareness
The technical challenge of merging edits is only part of what makes a collaborative editor feel usable. Awareness features — showing where each collaborator's cursor is, what they are selecting, who is currently viewing the document — require a separate, lower-latency data channel that broadcasts ephemeral state rather than persistent document changes. These presence indicators are typically implemented using WebSockets for low-latency bidirectional communication, with the server broadcasting each user's cursor position and selection range to all other connected clients. The visual representation (colored cursor bars with author name labels) is a user experience detail that makes a large difference in whether collaboration feels natural or confusing.
LibriDocs is built on open-source collaboration technology. Visit our platform to start collaborating on documents, or contact us to learn more about our technical approach.