> Choosing a handful of successful winners in comparison to "most firms only last about 10 years" is a selection bias.

"Most firms only last about 10 years" is selection bias because most firms don't even last 10 years. Infant mortality is very high. The firms that do last 10 years usually last 20 or more.

Meanwhile you're ignoring my point -- it isn't a question of how many companies are destroyed, it's a question of destroying even more of them. Some evidence that maintaining a stable company is hard is not an argument for making it even harder.