My intention was poorly portrayed but to try again, in all my years I've never known anyone who has been sued over a non-compete, a good chunk of those people have been software developers and most software developers sign non-competes these days, usually in overly broad topics (and I do know a bunch of people who have continued to work within the same relatively narrow industry).
Non-competes are nearly never enforced but their presence has a chilling effect, those of us aware of their unenforceablity and lack of attempts at enforcement generally waive off the consequences of signing them. People who are less familiar take them more seriously and they can have a real chilling effect, so their mere existence creates an imbalance in seeking future work.
But, yes, the chilling effect is much broader.
No, a thousandth of a percent = 1/100,000, which is more likely than the meteor or murder scenarios presented.
Also, if "some company" turns out to be Amazon or EMC, then it's not only higher than that, but certainty in some cases.
Perhaps you're exaggerating for effect, or perhaps you really don't understand the probabilities involved. But this is many orders of magnitude less likely than you getting killed by a meteor strike. E.g. something like [0] estimates a 10km meteor strike every 10 million years. That has (optimistically) a 50% survival rate for humans and you will live 50 more years, so 1 in 400,000 chance it gets you.
If you don't like the meteor calculation, you can look at the odds of, say, a murder in your area and both the judge and all the members of the jury are convinced you did it despite your alibi to the contrary. Maybe a million to one. Or, you know, something a thousand times more likely, like you get diagnosed with a rare untreatable form of cancer and have a year to live. Maybe 1000 to 1.
Either way, if that's what you think of your non-compete it's not worth worrying about.
[0] http://www.tulane.edu/~sanelson/Natural_Disasters/impacts.ht...