What if in the middle of the night someone dug a huge pit, so that when I step out my door I fall down and die?
Well, says I, I'll be safe if no one does that.
The more I describe the situation in which I will be safe, the more accurate it becomes.
Example
Opening a door will cause me to step safely outside.
What if it's the door leading to nowhere on a 10-story building?
Opening my house's front door will cause me to step safely outside.
What if there's an ice storm and it's slippery?
Opening my house's front door when there is no ice will cause me to step safely outside.
And so on.
In other words, I know when it's safe when I'm about to open the door and the event closely resembles one in the past when I stepped out safely.
So all I have to do is infinitely describe a scenario in which I was safe that still resembles the current scenario I am in and infinitely open the door and step outside safely. Then, to quote Russell, the general law that that I will be safe when I open my door and step outside shall "approach certainty without limit."
I don't think I've really said anything new here, but it's interesting to think about how we justify day-to-day actions as safe.
