Universal Undo

April 10th, 2012
tech
When you click undo the computer should go back to how it was before you did the most recent thing. This isn't what happens now. Instead each program maintains it's own history, which causes trouble if the sequence of actions you want to reverse includes a program switch. It also increases the mental load: you can't simply have a reflex to trigger 'undo' whenever you do something you didn't mean to. Consistent universal undo would be especially valuable on smartphones, which you're often poking at in awkward places and where limited touch feedback promotes mistakes.

Making a computer act this way doesn't sound hard: when a program receives an action from the user it tells the operating system about it, and then when the user later hits 'undo' the operating system informs the appropriate program. In practice, there are major issues. While almost anything can be reversed, a program might not know how because some pieces are out of its control. For example if I submit a web form and then 'undo', the OS can tell the browser to undo it, but the best the browser can do is hit the back button. There's no way for it to know what command to send to the server to undo the form submission.

Another big issue would be adoption. If you add this to any existing system you're not going to be able to make all programs or apps switch over right away. Until everything is using the OS's undo in place of their own undo it might be confusing and worse than the current system.

(This is a post I wrote a few weeks ago, but now that I work at Google I'm probably going to need to stop posting tech ideas. The culture is very open internally, which means that ideas I have are likely to be dependent on or related to internal projects.)

Comment via: google plus, facebook, substack

Recent posts on blogs I like:

Thing of Things AI use policy

dynomight recently wrote an article calling for bloggers to state publicly whether and how they use AI

via Thing of Things July 6, 2026

Agentic test processes, LLM benchmarks, and other notes on agentic coding from Galapagos Island

I've been using AI fairly heavily since last November and the whole thing is a funny experience. An agent will do something that, if a human did it, you'd immediately fire them. My reaction, of course, is to act as if this is great and spin up a t…

via Posts on July 3, 2026

Variable fonts aren't universally supported

I make a lot of webpages. I also use Lockdown Mode on iOS and MacOS for a bit of extra security. Sometimes I realize that I forgot to test on Safari and it looks like crap, or I test and don’t notice that there’s been a problem for months (as was the case…

via Home June 27, 2026

more     (via openring)