Rather, imagine her shouting it out at a mall on a day when the mall is relatively empty. (So that she had reason to believe the audience tiny, but still broadcast it) Unfortunately for analogy-world Justine, a news crew was downstairs shooting B roll for background and she's there clearly visible and audible in the corner of the screen, and the clip got leaked. That's the analogous real-life situation.
> but may be overheard accidentally by others
There is nothing accidental about tweeting in public. If she wanted to limit the exposure, she could have used the "protected tweets" feature.
It's a sad state of affairs if our only option for having semi-open communications on the internet - things that friends of friends can see without having to sign up to a walled garden - is if we only say things that are acceptable to everyone possible, for fear of mobs semi-randomly forming, or worse, forming at the behest of a for-profit company driving the mob in a frenzy to get ad revenue (which is the case for times companies like Gawker have stirred the mob up).
That sort of absolutism of policing thought just doesn't end well, ever.
To translate it away from Twitter: imagine you're at a public space, such as a mall, and you happen to be walking by a table of people leaning together just as one makes a racist joke.
Is your response a) give them a sort of disgusted look and go on about your day or b) call everyone you know and repeat the joke to them, along with the location of the table in hopes you can whip up a mob of people to shout obscenities at them, continuing to follow them around for the day, and eventually form a mob outside of their office building, continuing to shout at them until their employer fires them?
Most of us would go with 'a' and not 'b', because we realize that what happened doesn't really fit the extreme response. It's interesting to me that people feel that more extreme response is okay, as long as they don't feel personally too involved with it (eg, they can shout remotely).