I read the Spolsky blog the post with title "Evidence based scheduling" (http://www.joelonsoftware.com/items/2007/10/26.html) but I don't understand the mechanism to predict and estimate interruptions.

If I understand, Splosky says to prepare 100 scenario for a task in ramdom way

At point 3 of the article, Spolsky says

1) "While calculating each possible future for a given developer, you’re going divide each task’s estimate by a randomly-selected velocity from that developer’s historical velocities, which we’ve been gathering"

2) "Do that 100 times; each total has 1% probability, and now you can figure out the probability that you will ship on any given date"

3) Spolsky says that, more the velocity of a developer has been regular in the past, more the differents values are convergent in the same value (as every Montecarlo simulation, I says too).

4) In next paragraph ("Obsessive-compulsive disorder not required") I tell us that there are a lot of different interruptions with differen amount of time (him example is about "Painful conversation about marlin"), but EBS don't receive any problem in precision of estimation.

Well, I don't understand why

• at the point 2, what total have 1% probability in every case? The total of example? But it has only 5 tasks.
• at the point 4, I don't understand the sentence "is exactly the same as the probability that John’s boss would interrupt him during any given feature"

I think to lose same details, but I feel a bit confused.

What is my error?

PS I don't know if this question is more correct here or in Project Management Exchange: if I got wrong I'm sorry.