As I said here, much as I love ManyEyes, it would be drastically better if it accepted data in the same format (tab-separated columns of values) that it expects to be submitted as a form from a URL, so your visualisation could be based on a source of regularly updated information.
If you can parse columns of tab-separated text provided from within a browser, there’s no reason why you couldn’t parse JSON or even heavier XML formats (especially GeoRSS), and then you could build a visualisation on a dynamic flow of data. For example, the Viktorfeed, or Google’s elections JSON thingy. I suspect that if ManyEyes was a Google project it would have done this first, before fiddling with files or clipboards.
Meanwhile, how much does the WordPress.com scriptmangling piss me off? A lot.