The problem is how do you find them. You have a thousand people in front of you. Now please make a prediction that actually works reliably for big numbers of people.
Your argument suffers from the same problem as any of the others that always come up after something happened, "why didn't they...". You look backwards. You make a "prediction"about something in the past and of course you get 100% accuracy and think "that's easy". Okay, now please make tat prediction for the future. You have people in front of you whom you know zilch about. Now make - en mass - statements about their future work that when looking back from the future will still look good.
About your
> Was Heaviside a great engineer?
Now please make that backward-looking prediction about his greatness for the same person before it turned out that he was good. And not just for one person either, do it for thousands of not-famous people with some degree of consistency.
And then report about all your predictions, not just the few that worked, and don't forget to mention if you did some extreme filtering beforehand in order to ensure you get a lot of "hits", leaving a large part of potential candidates without any chance to begin with. That kind of thing works for some (like so many "recipes for success") but not for society.
The simplest answer is to predict that 0 of the 1000 are great developers. That will be reliable for most cohorts of 1000 people and most active industry applicants for developer positions.
You can only reliably find greatness by filtering through many college applicants, training and mentoring them, keeping, motivating, and incenting the best, or by using networking and references to find the standout performers. It’s so rare for these industry experienced people to need to apply for a job cold that that is a tremendously adverse pool from which to select in my experience If greatness is the goal.
If good enough is the goal, there are plenty more of those devs available.
But yeah, you have a light filter and then let them grow and evolve. You can’t possibly learn in 4 hours anywhere near what you can in 2 years. In college recruiting, you’re looking for “could be solid” and you get the raw material for that a lot. Some happen to turn out great.
Finding greatness in industry is driven mostly by asking some of your best employees to search their networks for their best ex-colleagues and otherwise by dumb luck (someone tried a startup that went bust and no one in their network had an opening, someone is moving to a new city with no network, etc)