Google Wave

~ obituary ~

Google Wave was unveiled at Google I/O on May 28, 2009. The demonstration was, by the standards of the time, electrifying: a real-time collaboration tool that combined email, chat, document editing, and threaded discussion in a single interface. Each "wave" was a shared document where multiple people could write simultaneously, character-by-character, with full revision history, because embedded objects (videos, maps, polls) could be added inline.

On August 4, 2010, fourteen months after announcement - Google announced Wave was being shut down. User adoption had been "low," the official statement said. The shutdown followed a visible failure: Google had hyped the product widely, distributed invites in waves (heh), generated enormous initial interest, and then watched users sign up, log in once, fail to understand the interface, and never return.

Google Wave is studied in product management courses as a textbook example of inventing too many features at once, though almost everything Wave did has, in the years since, been done well by some other product. None of those products did all of it together. The synthesis Wave tried may simply not have been a viable product category.

~ by the numbers ~

~ what wave actually was ~

Wave was simultaneously several things, in a single interface. The trouble started here.

An email replacement: A "wave" was supposed to behave like a long email thread, but better - you could see other people typing in real time, you could replay the wave from the beginning to see the order of edits, and the full history was preserved as a timeline. A chat tool: Within a wave, you could chat with the other participants in real time, which is why the chat and the document were the same thing. A collaborative document editor: Like Google Docs, but more flexible: you could "play back" the document's edit history at any speed. An embedding platform: You could drop a YouTube video, a map, a poll, a chess board, or a custom widget into any wave. Other participants could interact with the embedded objects. A federation protocol: Wave had a server-to-server protocol so that, in theory, organisations could run their own Wave servers and federate with Google's main installation.

Each of these is a product, while combining them produced a single interface so dense that most users could not figure out what they were looking at. The "what is this for?" question was answered with: "everything." That is, in product design, generally a failing answer.

~ the demo problem ~

Wave's announcement demo at Google I/O 2009 was an 80-minute showcase by Lars and Jens Rasmussen (Google Maps' creators) and the Wave engineering team. The demo was a hit, because tech press wrote effusively. The phrase "the future of communication" appeared in dozens of headlines.

The problem with the demo was that it showed everything Wave could do. The audience. Tech-savvy developers and journalists; understood every feature. They left assuming users would intuit the same things. Users did not. When Wave shipped to the broader public, its onboarding was about: "here is your inbox of waves, do something." Most users opened a wave, saw a long list of conversations they couldn't follow, and closed the tab.

The demo's success disguised the product's failure to teach itself. By the time Google's marketing team realised that Wave needed a lot more user-education infrastructure, the press cycle had already turned, but the product had been declared important. Then it had been declared confusing. Then it had been declared dead.

~ what wave invented (that survived) ~

Several of Wave's specific innovations survived into other products:

• Real-time character-by-character collaboration. Google Docs (which already existed) refined this. By 2013 it was a baseline expectation. Wave had been the public showcase that this was possible.
• Operational change algorithms. The math behind real-time collaborative editing, how to merge simultaneous changes from multiple users into a coherent document, was advanced by Wave's engineering team. The algorithms underlie Google Docs, Microsoft 365, Notion, Figma, and almost every modern collaborative tool.
• The replay feature. Showing a document's full edit history as a playback was unusual for 2009. Slack's "thread" feature, Notion's history view, and Figma's version history descend in spirit from Wave's playback.
• Inline embeds. Slack, Discord, and Notion all let you drop interactive widgets into messages and pages. Wave's inline-embedding model prefigured this.

~ the slow goodbye ~

Google's official statement on shutdown gave a low-engagement reason. The internal reality, reported later, was more specific:

Senior leadership impatience. Wave was a hugely expensive product to develop (about 200 engineers at peak). Larry Page and Sergey Brin's pattern, by 2010, was to give bold projects 12-18 months to show adoption. Wave did not. The product was killed before its team had time to redesign it.

The Buzz pivot. Google launched Buzz in February 2010 as a Twitter competitor. Buzz drew internal attention away from Wave. By the time both products were under pressure, leadership chose to keep Buzz (which itself was killed in 2011, see Google+'s memorial for the broader pattern).

The Apache donation. Google did not entirely kill Wave; they donated the codebase to the Apache Foundation, where it became "Apache Wave" and persisted as an open-source project until 2018, when it was officially retired. A few enterprise customers ran their own Wave installations until then. The community remained, briefly, a small dedicated cult. Apache Wave shutdown was the final goodbye.

~ aftermath ~

The synthesis try. We now have email (improved minimally since 2010), chat (Slack, Discord), documents (Docs, Notion), and project management (Asana, Linear) as separate products. The original Wave thesis was that these are all the same thing and should be unified. The thesis may still be correct. No one has tried again at scale.

The federation. Wave was designed to be federated - multiple organisations running their own Wave servers, communicating across organisations the way email does. The federation never reached key mass before Wave was killed. Modern collaboration tools (Slack, Notion) are decisively centralised. Federated alternatives (Matrix, ActivityPub-based tools) are niche. Wave's vision of a federated collaborative web has not been realised.

The bold confused product. Google before 2010 was willing to ship hugely ambitious products that didn't quite work and iterate. Google after 2010, having watched Wave die, became more conservative. The current Google ships fewer experimental products. Wave's death was a moment in Google's product culture's slow conservatisation.

~ from the comments below ~

"i was a wave evangelist in late 2009. i convinced my entire team to use it. we used it for two months. nobody understood the playback feature. we went back to email. when google killed it i felt embarrassed for having advocated for it.". e.t. 44

"the i/o 2009 wave demo is, the moment i decided to be a software engineer. it took me until 2013 to actually become one. by then wave was already dead. the inspiration outlived the product.", m.ö. 32

"i still want wave. i don't want it perfectly. i want it again. someone please try again.". a.k. 39