OpenID is an excellent case study of a promising technology killed by being incubated in the techy echo chamber. For example, the spec (and the process to create it) spent a mindbending amount of effort on getting delegation right, so that people can port their identity around by the simple expedient of putting references to their delegated openid providers in their user-supplied identifiers, and no visible effort on making the end-user experience less painful than an IRS audit administered during a root canal.<p>Supporting email addresses is where we should have been several years ago. (Take a look at who actually gets used as OpenID providers: the gigantic free mail folks who hold 50%+ of the market and had the instant scale to make things work. That shouldn't have come as a bolt out of the blue! Now the best OpenID implementations resemble "Pick which of the following you have an email address at"!)
Identity and data access are a key part of simplifying UI on the web but also on the desktop. I should be able to log and access my data on my netbook, desktop computer, and the web through the same identity.<p>This would also favor open APIs to manage/move the actual data more easily as they would seam so natural then. This would be a shield against people wanting to control our identity on the web (Facebook) and people wanting to disempower us of our data (Google's Chromium OS).