In my opinion, the set of players should be inclusive of all football players. By limiting it to the most successful outliers in a power law distribution, we limit ourselves to the players with the best possible outcome. Players don't really get to choose whether they belong in this set; they choose to play highly competitive football.

In short, I think we should define the group by factors that result from choice, not outcome.

You're one of the few people who seem to actually understand this.

By comparison, a 4-year degree is probably worth more in the long-term than those NFL earnings. So the players are actually being better compensated in college than the NFL.

Edit:

The median salary for an NFL player is about $770K and most players are out of the league in under 3 years.

https://www.si.com/nfl/2016/03/01/nfl-careers-shortened-two-...

http://blog.futureadvisor.com/nfl-players-must-go-long-on-re...

It looks like the average 4-year degree is worth less than the average NFL earnings - the two figures I found Googling were $2.1M for lifetime earnings from a 4-year degree, and $4M for career earnings for an NFL player.

On the other hand, engineering degrees tended to hover around $4M for average lifetime earnings, and some engineering fields (petroleum engineer, SV software engineer) averaged ~$6.5M in lifetime earnings.