Regardless, that's an interesting perspective. My follow up would be "do you think the government is well-informed and future-thinking enough to pick the jobs with 20+ year lifespans?"
Personally, I don't think anyone is. Ten years ago, Microsoft was nearly unbeatable, Google hadn't IPO'd, Facebook didn't exist, and Apple was still a bit player. Ten years from now, who knows what it will look like.
(Granted, that's one field.. but outside of tech, the home construction market was the other boom.)
And I'd like to point out that just because that hypothetical job being created pays $50,000 per year, doesn't mean it only contributes that amount of value to the economy.
But you certainly wouldn't want to bet on any job that is repetitive (or "concave" to use a michaelochurch-ism) being around longer than a few years, because that work will be distilled into a software program and handed off to a machine sooner or later. It only makes sense to invest in creating jobs that involve creative, "convex" work, which has a smaller chance at a bigger payoff in terms of value creation.
And at some point, when you are talking about investing public funds to create jobs in a period of high unemployment, you need to think of it in terms of subtracting the cost of that person's potential unemployment benefits, medicaid, and other costs that an unemployed person imposes on the public.
So, it's kind of complicated to do a full cost/benefit analysis of this.