Operational Transformation

Operational transformation (OT) is a family of algorithms for supporting concurrent editing of a shared document by multiple users without locking. Each user's local change is applied immediately and broadcast to peers; when a remote change arrives that was made against a different version of the document, OT transforms it against the local operations that have happened in the meantime so that all replicas converge on the same final state. The technique was first described in a 1989 paper by C. A. Ellis and S. J. Gibbs at MCC, who used it to build the GROVE group editor, and was refined across the 1990s and 2000s in academic systems such as Jupiter, REDUCE, and SOCT. The core requirement is that the transformation function satisfies a pair of correctness properties (often called TP1 and TP2) which guarantee convergence regardless of the order in which operations arrive. OT entered mainstream software with Google Wave in 2009, where it powered real-time, character-by-character co-editing across distributed clients. When Wave was discontinued the same engine was carried into Google Docs, Google Sheets, and Google Slides, and Apache Wave's public OT whitepaper became one of the most-cited references for the technique. Competing approaches based on CRDTs (conflict-free replicated data types) have since gained ground because they avoid the transformation step, but OT remains in production at very large scale and is generally considered easier to integrate with rich-text and structured documents than early CRDT designs.