We can imagine all day long, just like we can imagine terrorist attack scenarios. That doesn't mean it's a good idea to expend state resources to prevent our imaginary scenarios, even if
some of those scenarios will indeed come to pass.
This isn't a philosophical debate. This is an empirical debate. Non-competes have both pros and cons, and empirical studies suggest the pros are exaggerated and cons under appreciated. They're also not the only tool in the toolbox. For example, there are NDAs, trade secrets, and tortious interference, among others. Compensation and other incentives can be restructured.
Finally, your scenario only begs the question: so what? For the sake of argument, let's assume that it's somehow unfair to the original employer. What about the customers? Why is it fair that they've lost the option of working with a guy they like, probably for less money? There are two sides to the coin. That's why I said it's an empirical question: we can work out which option produces the better outcome overall, and the facts strongly suggest the best outcome is the world where non-competes are generally unenforceable.
One of the biggest and last remaining areas of productivity growth in modern service economies is putting people to their most productive task. There are huge structural inefficiencies that result in person doing task A when they'd be much more productive doing task B. This is much more of a problem in a service economy than a manufacturing economy. And one source of this inefficiency, and an increasingly common source at that, is non-compete agreements.