While I have always enjoyed Easterbrook's mockery of coach-protective strategies, using average gain per down to prove his point is oversimplified. One would have to analyze not only the probability of conversion by the offense and the subsequent increase in expected scoring, but the expected-points-allowed based on a turnover-on-downs in a given position on the field, and contrast that with expected-points-allowed of giving the ball away at some averaged distance down the field (however far your punter kicks). The coach in the article seems to recognize this, although he's using college stats and is guessing on his team's ability to convert. It can be assumed that expected-points-allowed increases on turnovers-on-downs as one is closer to one's own endzone (and probably sees a major jump inside some point because of average FG ranges). In addition, the increase in expected scoring of the offense should they convert on 4th down is less and less as you get closer to your own end zone (you have more field to go, and more potential 4th down situations). It would get particularly bad when your punter's dropback is shortened by being in the endzone. And this is all before taking into account the actual situation of the 4th down. (4th and 1? 4th and 10?) If you chose some arbitrary yardage, say 4th and 5 or less, you could in theory determine where on the field the threshold is for making it worth it to go on 4th down. This is pretty much what the HS coach in the article tries to establish. When comparing levels of competition, I would imagine that the threshold is closer to the end zone in HS and further away in the NFL mostly given quality of kickers and passing offenses. So, this at least nominally supports Easterbrook's using of the oversimplified theory at his lower level of competition. Anyway, the larger points are the ones that have already been made in this thread. Depends on the offense, depends on the defense, depends on the punter.