[ColinWright]

  > Where do you get the contents of 'records' from?

I download the "newest" and "news" pages every 15 minutes or so, then collate the data.

  > If we merge bbc.com and bbc.co.uk, we
  > end up with 125 / 10,000.  I suppose
  > that isn't that many compared to others,

It's less than some, but not as many as, say, nytimes.com or techcrunch.com. But it's a major news site, so I'm not surprised people read it and go "Oh, that's interesting, I'll click the HN bookmarklet and submit it."

  > ... still higher than I think it should be.

What do you think it "should be?"

  > A nice solution might be to remind people,
  > ... that it's better to go to the source ...

I personally find it useful to read a non-technical overview, and then if interested, go and find the technical version. Often the article with technical details borders on unreadable.