I have my own theory on this, I don't know how similar it is to Krauss'. I also don't know how I'm going to explain..
To make it abstract, perhaps think of a number with a value of 0 on a piece of paper (what that value means is irrelevant. You 'name' it 0, even if it could be 5). There's a simple 'law of nature' that your piece of paper has to follow: the net value of all values on it is 0.
The number tends not to change on average, but occasionally, it can get a little higher or a little lower. Say a 1 is generated. That can't just happen without anything else happening, because now the law of nature of your piece of paper is violated. Now you start to wonder about what 0 is. 0=-1+1 among other things. What this means is that you only need to take -1 out of the 0 to turn it into 1.
So that's the only action that happens on your piece of paper: -1 is taken out of 0. As a result, the 0 just became a 1. That minus one you took out though, that's still on your piece of paper! So now you suddenly have 2 'things' out of 'nothing': 'something' and the exact opposite of that 'something'. In a way though, you still have only one number that really matters to you, because who cares about anti-matter?
It's not over though: 1 is suddenly taken out of -1! Now you have this on your piece of paper: 1, 1, -2.
Suddenly you have 2 'somethings' while you had only 1 'nothing'!
What I'm saying is that my theory is that 1 'nothing-particle' could, very rarily, be substituted for a particle and an anti-particle. Which could go different ways under the right circumstances. When there are no other particles around, perhaps.
Compare it with waves. You can have a horizontal line, very boring. Then you take a wave out of it. Suddenly you have two waves: the one you took out and the one that's left over in the line, which used to cancel out the wave you took out, which is why there was a horizontal 0-line to begin with!
Interesting experiment just entered my head, so ending this reply now.