You can clone with
HTTPS or Subversion.
In the screenshot http://www2.pic-upload.de/11.12.09/cuji68t225hx.png pay attention to the elements marked with number 5.
If I want to do a search with the built-in site search, I have to first delete the useless crud that is prefilled in the input form element. Remove the crud. Make the element descriptive by associating it with a label instead, see http://www.w3.org/TR/html4/interact/forms.html#h-17.9.
I don't see this behaviour. When I click in the search box the "useless crud" is removed automatically.
This is standard behaviour for sites all across the web. I don't think we'll change this.
I like Ovid's idea and will look into implementing it.