So the docs say that you cannot create a blank sheet in google docs, that you must upload a sheet to create a new one...not true...On the front of the api page there was a link to a topic in the api discussion group,
create a new spreadsheets on Google Docs . The last entry from Eric gave me the clue or actually the answer. The "helper methods" took a few minutes of debugging to figure out.

In gdata.docs.service, the upload methods for spreadsheet, docs, etc. all call Post like this:

    media_entry = self.Post(media_entry, '/feeds/documents/private/full',
        media_source = media_source,
        extra_headers = {'Slug' : media_source.file_name })

The media_source object identifies the path, length, and title of the file you want to upload. It's an optional parameter on Post, but it's required if you call UploadSpreadsheet. If you call Post directly and leave off media_source, you can create a new doc out on google without the file upload.

Here's the code. The example is in single user mode. When I put it on the requirements page, I'll change it.

    def createBlankSpreadsheet(self):
        client = gdata.docs.service.DocsService()
        gdata.alt.appengine.run_on_appengine(client, store_tokens=False,     single_user_mode=True)
        client.email = 'youremailhere@gmail.com'
        client.password = 'putpasswordhere'
        client.source = 'exampleCo-exampleApp-1'
        client.ProgrammaticLogin()
        
        category = atom.Category(scheme=DATA_KIND_SCHEME,term=SPREADSHEET_KIND_TERM)
        title="HeyMan2"
        media_entry = gdata.GDataEntry()
        media_entry.title = atom.Title(text=title)
        media_entry.category.append(category)
        media_entry = client.Post(media_entry, '/feeds/documents/private/full');
        return media_entry.GetAlternateLink().href

Now this isn't perfect. While the spreadsheet is showing up in google docs, the request is timing out. The app engine has a 5 second timeout on http requests and that doesn't seem nearly long enough for the docs services to do its thing and return a response. I haven't received one response yet. I'll try the test out again tomorrow when I update the site on google.

This goof-up has made it all over the internets. For about an hour this morning google reported all results as distributors of malware. (big QA problem) I kind of freaked out when I began searching for stuff and found that wikipedia was going to "harm my computer". I went straight to Techcrunch and found this story.

Update - From google.