Selection Pressure: Evolution in Roster Construction and Player Outcomes (Part 1)
General ideas and concepts in this article
Viewing the development of rosters and player niches through the lens of evolutionary processes offers us many insights.
It’s the environment that players exist within that determines their fitness and subsequent success or failure.
A positional prototype’s shift over time towards new standards is much like population shifts in nature.
Understanding the various forms of selection can help us to broadly understand the driving forces behind why teams make the decisions they do.
A larger population to draw players from and long time frames can be used to explain positional optimization and league trends over the years.
Ideas like adaptive radiation and the species-area relationship can help explain positional and player type diversity relative to roster size and the period in league time.
Historical contingency/path dependence matters when it comes to the way rules and players evolve.
Artificial selection and natural selection can be applied in terms of how we think about the player selection process and outcomes on the field.
Selection processes at lower levels, cumulative advantage, and randomness raise interesting questions about what prospects reach the NFL landscape.
Fullbacks are an excellent case study of a position going nearly extinct as a result of environmental pressures outside of their control.
The league as a whole is a complex adaptive system. Everything is interdependent and tinkering in one area leads to changes in others.
Table of Contents
1. A Few Thoughts
2. The Most Boring Game
Using a hypothetical league’s field dimensions and player archetypes to illustrate how the environment shapes those within a system.
3. Less Boring Games
Real life examples of different leagues and the subsequent changes in the game that arise from their varying rules.
4. Changing Chess Boards
Illustrating the importance of field dimensions by showing changes in a chess game if the board were to change.
5. What is Evolution?
A brief definition of evolution and its application to the rest of the article.
6. Fitness and Environmental Pressures
Defining fitness in the evolutionary sense through beetles.
7. Beaks and Population Shifts
Showcasing types of selection through the evolution of a finch population in the Galapagos.
8. The NFL’s Archean Eon
Reflections on the league's history, both in its rules and players.
9. The Small Supply of Big People
On the implications of a scarce supply of players with certain physical attributes.
10. Rat Islands and the Species-Area Relationship
What island biogeography tells us about diversity and niches.
11. 53 Forms Most Beautiful
On adaptive radiation and the development of new species.
12. By Generation Not Time
Viewing evolution through the lens of generations and mutations rather than by days, months, or years.
13. Replaying the Tape of Time
How small differences in the past can create vastly different outcomes in the present.
14. Potters and Pigeon Fanciers
Artificial selection and its comparability to team selection processes.
15. Marble Racing, the Matthew Effect, and Genetic Drift
How cumulative advantage and randomness influence the population of players.
16. Correlated Traits and Spandrels
On measurements that relate to each other and their value in success.
17. Fitness Landscapes
Using the metaphor of fitness landscapes to map player populations.
18. Going the Way of the Dodo
On the extinction of certain types of players.
19. Peacocks, Oxpeckers, and the El Farol Bar
The league as a complex adaptive system and the difficulty of understanding the whole through its parts.
20. Application and Closing Thoughts
Putting these ideas into practice and reflection.
A Few Thoughts
“All models are wrong, but some are useful.”
– Statistician George E. P. Box
“I suppose it is tempting, if the only tool you have is a hammer, to treat everything as if it were a nail.”
– Psychologist Abraham Maslow
I did not initially intend to venture deeper into the natural sciences when I bought an audiobook copy of Rob Dunn’s A Natural History of the Future in November 2022, the first of the many books used in the composition of this article. But the more time I spent learning and listening to those in these fields, the more I saw patterns between the systems of the NFL and the systems of nature.
The harsh circumstances that drive an animal’s adaptation to be faster, to be built a certain way, or to fill a particular role in the ecosystem more and more began to resemble the need for long arms in an offensive tackle or quick twitch in a wide receiver. The limitations and values of selection mapped surprisingly well from one domain to the next. I hope this article represents both those concepts in their native form (as studied by experts in the natural sciences) and in an applied form in a way that is cohesive yet respectful of the material and the differences between domains.
I don’t come from a background in the field and there are many concepts within it where I lack the intuitive grasp that a true expert in the field would have. Perhaps that is a good thing in some ways, as I can transfer concepts in a way that an expert would be unwilling to because of the inherent deficiencies involved in it. There is a price paid as some of the value of the information embedded in the work of true experts gets lost in translation.
That said, I have found great usefulness in viewing these concepts as applied to roster construction and the outcomes of players. It is important to recognize the differences and the limitations of these concepts when transferred from the natural world to the study of a strictly defined game space. But when used carefully and in the right places, I believe they can shed new light on old problems.
These ideas don’t provide some grand theory, a skeleton key, or perfect answer to every problem with roster development and scouting, but I believe they can be valuable in avoiding the treatment of such a complex task with the bluntness of Maslow’s Hammer.
I’ve structured this article in the form of chapters in hopes that the length won’t be off putting and that someone uninterested in a particular idea can skip ahead. The first 8 chapters are more introductory oriented and shorter, whereas the last 12 chapters get more into the weeds. While the topics do build on each other throughout the article, you could pick any particular chapter that seems interesting and jump to it without reading everything.
Thank you, and I hope you enjoy.
The Most Boring Game
“As many more individuals of each species are born than can possibly survive; and as, consequently, there is a frequently recurring struggle for existence, it follows that any being, if it vary however slightly in any manner profitable to itself, under the complex and sometimes varying conditions of life, will have a better chance of surviving, and thus be naturally selected.”
― Charles Darwin, The Origin of Species
The roar of the crowd is nearly deafening, which is unsurprising given the size of the stadium. It’s second down and eight yards away from a first down. Seven offensive linemen get into their stance. A 280 pound tight end is at the end of the line of scrimmage, and he is lined up across from a 270 pound linebacker.
The defense is not much different from the supersized offense, with multiple 330 pound defensive tackles eating up the middle of the field. There are no cornerbacks or wide receivers on the field.
The center snaps the ball and an immediate scrum begins. The 250 pound running back gets a lead from his fullback who collides with another enormous linebacker. The condensed space makes it hard to gain any yardage, and the play is blown dead after the offensive line generates enough push to make it third down with five yards to go.
The offense has driven around 16 yards so far and has another 12 yards to go before they can punch it in for a touchdown. It’s a tedious, grinding, and physically imposing approach, and it will repeat the other way when the teams change possession.
Is this what happens if you put two NFL teams on a field that is 30 yards long and 13 yards wide?
This is a fictitious scenario, of course. No professional football league plays on a field with these dimensions, and those who do play on smaller fields have much different rules than the 11-on-11 4-down clash you would find in the NFL or college football.
Perhaps this story plays out differently in your mind, but I believe the general sentiment of a compact run-game based approach would tend to permeate across most scenarios where the field shrinks this much while other rules remain fixed.
The receiver with 4.3 speed would not have room to run on this field, and defending his speed would be an easier task for defensive backs regardless of their athletic attributes. There isn’t enough space to function in a way that provides value to the offense. Smaller backs who make people miss would face a similar struggle, as their strengths would be mitigated in the traffic that makes up the new line of scrimmage. The quarterback with rare arm talent or deep passing acumen would be easier to replace with a big, physical runner who can function within the constraints of a 30 yard field.
These aspects of how the offense would change would also directly impact what defenses are bringing to the table. Without 4.3 speed receivers there is no premium on speed, and therefore defensive backs would be more suitable as large and imposing types who can function in coverage but must play the run first and foremost. The emphasis on controlling the line of scrimmage would lead to a focus on big, long, and strong run defenders who can control their opposition and dominate up front. There would be major restructuring to how we view the scheme, the responsibilities of linemen, and how we build a competitive roster.
Less Boring Games
We already see a form of this on NFL and college football fields when teams line up near the goal line. While the width of the field still allows for a competent passing attack, the shift in personnel and condensed vertical element has a similar overall influence on how teams approach the game. The lack of roster construction purely for this aspect of the game means the personnel is less equipped for it than if you were only to play in this manner, but the end result is close enough to compare here.
Let’s take a look at some other examples where we can see these elements of space influencing how teams and players function.
The Oklahoma drill is the gold standard of “playing in a phone booth” and is an example of what it looks like if we condense things as much as we can while still fitting bodies on the field. If you grew up playing football at any organized level, this narrow corridor is one you’re probably familiar with.
To state the obvious, there is no passing, room for maneuverability is limited, and generally hard collisions are the only thing possible. Success here is determined by size and physicality more than anything else. If we were to determine a team’s success or failure by this method I imagine we’d be looking at mostly very large individuals participating in this gridiron equivalent of a duel.
Let’s continue on our trip towards expanding the field and look at the Arena Football League.
This condensed field’s dimensions aren’t too far away from our peculiar starting field, as it is 50 yards long and 28 yards wide with 8 yard end zones. Most of the comparisons stop there. This new field has 4 foot tall rubber barrier sidelines rather than traditional sidelines. The goal posts are half as wide as an NFL goal post and on each side is a rebound net. It’s a live ball if a pass ricochets off the net.
This is 8-man football. There is no stunting and twisting from the three defensive linemen. Only one linebacker can blitz. There are only 20 players on the roster and prior to 2007 teams used a “one-platoon” system where players saw action on both the offensive and defensive side of the ball. There is no punting.
A look at the league’s statistics would tell you a lot about how it functions. In 2012, Utah’s Tommy Grady led the league with 142 passing touchdowns, and over half the league’s quarterbacks threw for over 100 touchdowns. Iowa’s Jesse Schmidt led the league in receiving production as he collected 172 receptions for 2218 yards and 58 touchdowns. The league’s best back was Philadelphia’s Derrick Ross, as he ran 146 times for 645 yards and 32 touchdowns. No other back rushed for more than 300 yards.
Scoring was the name of the game for the league in that period. In that same 2012 season, the Philadelphia Soul led the league in points scored over their 18-game regular season. They averaged 68.22 points per game. They ultimately lost in ArenaBowl XXV to the Arizona Rattlers 54-72.
We’ve got one more field to talk about before wrapping things up, and this time we’re going to go much wider and longer than what we are used to. Let’s look north and see what the Canadian Football League is playing on.
Instead of the 100 yards long and 53 yards wide NFL field, the CFL field is 110 yards long and 65 yards wide along with longer endzones. There are 12 players on each side of the ball rather than 11. The focus of the game is on speed and passing. They play three down football, allow multiple receivers in motion pre-snap, and enforce a one yard neutral zone among many other notable differences.
Winnipeg Senior Assistant General Manager Ted Goveia’s explanation of the way they play football is more illustrative than anything I can write, so here is Goveia at the 7th Annual Inside The League Combine Seminar talking to agents about the league.
So we've looked at what different types of leagues, both real and hypothetical, look like given various field dimensions and rules. Here's the key question we have to ask about all of these.
How do these differences influence what types of players are selected?
This question is the foundation of this article and in my opinion it is integral for understanding why players make a football team, regardless of what level or league they are in.
In our condensed hypothetical field, there is a place for multiple 330 pound nose tackle types who eat up space and try to shut down the run game. On a CFL field with 3-down football, a one yard neutral zone, and a short play clock there is no room for any of them.
Nothing about this nose tackle’s inherent physical and athletic characteristics change regardless of which league you put them in. But in one environment you see great success and value and in the other you see the equivalent of extinction.
It is therefore the environment, not the individual, that determines any chance at success or failure.
Viewed in this light, an NFL field shifts from being something simple and familiar into an evolving ecosystem of different types of players excelling or failing based on the field’s dimensions, the rules, and the interactions among the entire population of players.
Changing Chess Boards
One way to visualize the influence of the environment would be in the reshaping of a chess board. If we were to take a standard 8x8 board and restructure it as a 4x8, it would change the way almost every piece could possibly move. Any pre-existing plans of certain opening moves or strategies would be lost. There are over a thousand openings in chess and this new board would basically toss them out the window.
Every game would begin as a pawn on pawn slugfest until there was enough room to maneuver a rook or knight. The queen could be more devalued because of its starting position and the spatial limitations of the board. The king’s limited range would be less important on this condensed board, so strategies using it offensively could become more prominent. Although perhaps the more limited space would increase the risk of checkmate for any error in king movement.
Pieces take on new values in this new environment. It would not only be debilitating to overarching strategy, but it would change how any one piece would be used in pursuit of exchanging material, taking advantageous positions, or leveraging against the positions of other pieces.
Games would play out faster. New strategies would emerge and develop over time as there becomes familiarity in the patterns of how pieces get positioned and exchanged on this board. But it would take plenty of time and practice for players to reach an equilibrium point where pieces are valued properly and a consistently good model of gameplay is formed.
One thing to keep in mind in this 4x8 version of chess and the arc of gameplay development is that our past experiences and mental models are what guide us in decision-making. These are not necessarily what is most efficient or optimal within the system, but rather a result of how we process information. We would intuitively try to map our experience from 8x8 chess onto this new board and as a result we would struggle tremendously.
Studies in 1946 (translated 1965) by Adriaan de Groot and in 1973 by Herbert Simon and William Chase on chess expertise illustrate this pattern recognition aspect to our thinking. These lessons in how we perceive information matter because we have to understand their limitations when thinking about how the environment impacts its inhabitants. It takes time and effort to learn those patterns within new environments.
de Groot studied the ability of amateurs and chess masters when it came to reconstructing boards they had seen for only a brief period of time. Amateurs struggled while experts had little trouble. Initially de Groot believed the difference to be photographic memory on the part of experts. However, he tried the experiment with pieces scattered randomly across the board rather than in places that made sense. If it were truly photographic memory differentiating amateurs and experts, then it would not matter.
But it did. Experts struggled just as much in recreating boards that made no sense. They were not simply inputting and outputting information perfectly. They had developed specific pattern recognition abilities due to their experience, and those abilities allowed them to “chunk” the board in a way that allowed for easier processing ability.
This concept of “chunking” is outside of the scope of this article, but the same pattern recognition tools are what we use when we think about what should succeed or fail within different systems, even when those systems are unfamiliar to us.
To quote Simon (as quoted by Farnam Street).
“The streetwise slum resident has good intuition about how to react to the situations that are often encountered in a slum environment. The manager has good intuition about how to react to the situations that are often encountered in organizations. Both skills have the same basis in knowledge and recognition capability.”
Put into a football context, when the agent with a nose tackle, tight end, or fullback calls Winnipeg’s Goveia telling him, “my guy will kill it up there, he was in an NFL camp,” it’s because he is in an unfamiliar environment making incorrect assumptions. He is struggling in the same way that we would struggle in porting our understanding of chess on an 8x8 board to a 4x8 board.
What is Evolution?
Let me begin by saying that the remainder of this article is not intended to conflict with beliefs about evolution on a philosophical, religious, or moral level, but rather to adapt the frameworks and ideas from it in a way that I believe applies to the topic of interest. By necessity, I’m going to cite a lot of examples documented by experts in the field to help explain these concepts. I hope this does not push away any readers, and I want to be very clear in stating the goal I have in mind here.
Before diving further into some of the topics that relate to environmental pressures, population shifts, or selection processes it is necessary to define what evolution means, both in the sense of evolutionary biology and in our application to football.
The primary focus of this article so far has been on the environment any player exists within and the impact it has on the success or failure of that individual. Evolutionary processes in the wild are much less neat than the ones we will be poking at in the world of football. There is no collective bargaining agreement or 53-man roster in nature. The constant shifting of resources and interactions inside incomprehensibly complex ecosystems certainly creates different pressures than those I will discuss.
Doug Emlen’s introduction in Animal Weapons is much less environmentally focused than how I’ve approached the topic so far, but he has a succinct explanation for evolution as a whole. It explains variation, transmission of characteristics, adaptation, and selection in a brief package. Bolded for emphasis on these aspects of this quote.
"At its most basic, this evolutionary process is a simple one. Entities exist in multiple forms. They vary from individual to individual, and their individual characteristics are transmitted from one generation to the next. The transmission of this information is efficient but it's not perfect and mistakes get made along the way. Errors in the transmission of information lead to new characteristics, novel variations that crop up in the population from time to time. New variants now exist side by side with earlier forms, competing with them for resources and reproduction, and only some of them will persist.
Whenever individuals differ in how successful they are at propagating their kind, evolution occurs. This can be due to chance, random deviations in the reproductive success of types, or it can result from selection. Individuals with some characteristics perform better than individuals with other characteristics, and as a result they leave behind more copies of their kind. Such a process iterated over generations results in the eventual replacement of inefficient with more efficient forms.
The population evolves as the least effective types are culled and replaced by more effective types. As transmission errors yield new types of individuals they inject fresh forms into the mix. If the new individuals do worse than the rest they gradually disappear. If they do better then the new forms can spread, replacing earlier forms in the process. This turnover is evolution."
While there are key differences between evolution in nature and something like the player pool, such as the propagation of certain types and characteristics through imperfect reproduction in offspring, there is still plenty to chew on here.
There is variation from individual to individual, and certain individual characteristics are selected for based on their suitability to the given environment. There is competition between new variants and earlier forms for limited resources (success on the field, playing time, salary). There must be winners and losers, so only some persist.
When individuals have success they are, in a way, lighting the path for those with similar characteristics coming after them, and as such a form of detached evolution occurs. A smaller quarterback succeeds and suddenly many more of their type join the fray in a way that was not selected for previously. A defensive tackle with subpar measurements becomes an all-pro and the threshold for measurements collectively nudges downwards.
Generation after generation, there is a slow and steady replacement of old forms with new forms, much like the replacement of the large run stuffing linebacker in favor of light and agile movers who must cover space to succeed. Evolution occurs not on the level of any one individual, but on the level of the entire population shifting towards new standards in pursuit of those limited resources.
This is the lens I want to put on the evolutionary processes in roster construction and player selection. It is not a perfect match for evolution in the wild, as we will explore later, but many of the same mechanisms that drive evolution fit neatly within the framework of population changes or outcomes in player types.
Fitness and Environmental Pressures
“But a plant on the edge of a desert is said to struggle for life against the drought, though more properly it should be said to be dependent upon the moisture.”
― Charles Darwin, The Origin of Species
The term “survival of the fittest” has been popularized outside of evolutionary biology, but in the Darwinian sense it does not necessarily mean fitness in the sense of physical prowess like raw strength or speed. Instead, its usage is more akin to “the type which leaves the largest number of copies of itself in future generations” or “the type which is most suited to survive (and reproduce) within a given environment.”
This definition of fitness matters, because while there is no direct genetic lineage from one class of players to another, there are general trends for what makes certain types successful or unsuccessful as they perpetuate forward.
The term phenotype (set of observed characteristics or traits) that comes from this domain strikes some similarities with the “prototype” terminology that is often used in scouting and roster construction. While no player has the exact same phenotype, there are certainly similarities such as two defensive ends who are 6’2”, 250 pounds, with 33” arms and 4.6 speed.
A key point here is to understand that these traits and characteristics do not change significantly for individuals when thrust into different environments, but that instead they find success or failure based on what environment they land in. An animal cannot force its teeth to grow longer or sharper, nor can a man stretch hard enough to grow several inches. If you run with 4.8 speed, you likely cannot run with 4.3 speed regardless of your training regimen. It’s not part of your phenotype.
In viewing fitness as a matter of domain or environment, we can take very common examples in sports to showcase where certain traits that can be beneficial in one area would be harmful in another. A world class boxer would put on a masterclass against a sumo wrestler inside of the boxing ring, but that same boxer would be treated like a child inside the sumo’s dohyō ring. A typical offensive tackle would be useless trying to make a touch throw over the head of a linebacker, just like a quarterback would be steamrolled if pushed out to left tackle.
Emlen has a great example of the environmental pressures that drive fitness for individuals in a population with his study of dung beetles, and the key question of “why do some beetles form weapons while others don’t?” was perhaps the largest driving influence in the aforementioned Animal Weapons. Here is an excerpt from an article in the Montanan about how these beetles function.
"But until recently, Emlen was puzzled as to why some beetles developed weapons and others did not. Even among dung beetles, Emlen’s particular specialty, species with horns and species without horns can be found feeding on the same pile of poop. A massive swarm of beetles descending on a fresh pile of elephant dung in Tanzania one evening revealed the answer. Emlen and his students watched some species of beetles pack the dung into balls, which they rolled away to bury. These beetles fight above ground, and their confrontations are chaotic scrambles out in the open. Beetles without horns would be just as likely to win in these brawls as those with horns, so the males don’t waste the resources to develop weapons.
But the beetles that dig tunnels underneath the dung do develop weapons. Fights inside burrows unfold as orderly duels, and only the best-armed males mate with the females."
These are otherwise nearly identical species of dung beetles evolving in different directions to fill different niches and survive in different environments. The ball rollers have no need for long horns or the equivalent of beetle antlers because they are in a mad dash to push their pile of dung far away from competitors. They are often fighting with multiple other beetles at the same time and would not improve their fitness with the addition of cumbersome weapons on their head.
However, very similar types of dung beetles evolving for duels in underground tunnels adapt for the condensed spaces in a way where bigger weapons provide major advantages in the pursuit of reproduction. When a male braces his tunnel and blocks a challenging male from getting past him to mate with the female, he needs this advantage to succeed. As such, bigger weapons usually lead to more breeding and higher fitness inside the tunnels. This same beetle would not see that success trying to roll dung balls while trying to shed tackles from competitors.
One beetle evolved for the equivalent of the Oklahoma drill while the other evolved for the equivalent of playing keep-away on a field with no sidelines or endzones.
Beaks and Population Shifts
Imagine if, over the course of about a year, the league shifted so the average height of offensive tackles went from 6’6” to 6’10”. Or maybe imagine their average weight shifts from 315 pounds to 331 pounds. Perhaps they get much longer instead, and the average arm length for an offensive tackle goes from 34 ¼” to 36”. It would drive the path of almost every conversation about the position from people inside the league, to media pundits, to fans who follow the happenings of offensive tackles.
On a hot rocky little island in the Galapagos called Daphne Major, this is essentially what happened with the beaks of finches.
Peter and Rosemary Grant, two trailblazers in the field of evolutionary biology, studied a number of these “Darwin’s Finches” on Daphne Major over the course of decades. The isolated island is home to a variety of species of finches of varying beak sizes. One particular finch, the Medium Ground Finch, specializes in eating seeds. Of these finches, those with smaller beaks are more proficient at quickly cracking and eating smaller seeds, while the ones with bigger beaks are the only ones capable of eating bigger seeds. In many years, there is enough diversity between the seed sizes for finches across the spectrum of beak depth to find success in foraging.
In 1977, a drought hit Daphne Major that significantly changed the types of seeds available to the birds. Larger seeds were more available than small seeds, and the scarcity of the small seeds made it hard for the small-beaked birds to find the necessary food to survive. The result was a dramatic shift in the average beak depth of the finches, as the birds with smaller beaks had a much lower fitness for that environment than those with big beaks.
The opposite occurred a few years later, as a season of non-stop rain covered the island in vines and made smaller seeds plentiful. The eventual result of this was a shift towards smaller seeds during the next drought. As a result, fitness increased for the small-beaked birds and led to a swing in the population towards favoring those who could proficiently crack and eat small seeds.
The shift in the population of finches is what is known as directional selection, as the population was constrained by one side (small beaks less effective in first drought, big beaks less effective in second drought) and as a result during each drought there was a shift in the population going in one direction.
This type of selection is common across evolution, and some of the most prominent examples like the darkening of peppered moths in Great Britain or the rock pocket mice in New Mexico give us great insight into how this process works to shape a population.
One way to visualize this type of selection is to imagine a funnel with marbles being dropped into it. The marbles are all of various sizes, but the hole at the bottom of the funnel is too small to take in all of those marbles. Let’s assume the marbles that don’t fit in the hole get removed rather than stuck. We may start with 100 marbles but the process of dropping them in the funnel selects for marbles below a certain size (the size of the hole), and as a result the population that exits the funnel shifts to be collectively smaller. The average shifts downwards and the variation decreases as the upper end of marbles is removed.
We can work this in reverse as well. Imagine we take this same population of marbles and place them above a net that will only catch some of the marbles, as the holes in the net are too big to catch the smallest of marbles. The smallest of these marbles fall through the net and get removed from the population while the big marbles catch on the net and remain in our initial group.
The net and funnel act as simplistic models for what a selection pressure would look like to directionally change the population. This would be like teams selecting for players who are a certain speed. If you’re a wide receiver or cornerback with 4.8 speed, you fall through the net.
Let’s once again go back to the well for an Emlen quote on directional selection. As he states, directional selection doesn’t necessarily mean permanent stabilization in one direction. The beaks of the finches don’t grow bigger or smaller in perpetuity. Rather, directional selection is the process of moving from one optimal state to another as the environment shifts.
"Such is the nature of most natural selection. Populations adapt to surrounding conditions until they reach a local optimum. Environments may change, and when they do, selection kicks in once again favoring new trait sizes or colors that work better in the new surroundings. But these shifts are also expected to stabilize. Ocean sticklebacks once sported three long spines and 52 armored plates for hundreds of thousands of years. When some of these fish found themselves dumped into freshwater habitats, natural selection in their new environment led to a rapid shift from three spines to one, and from many armored plates to few (14 plates). But once these populations reached their new optimum, the change in armor stopped.
Even when natural selection is directional, its effects often cancel so that the net effect is still stasis. When physical environments change they tend to fluctuate back and forth. Winter gives way to summer, but reverts back again each year to winter. Wet years are offset later by dry years and vice versa. Even glacial ice sheets advance and retreat in cycles, and ocean levels rise and fall. When animal populations evolve in response to these changes, they do so in an oscillatory fashion. Mice become lighter, then darker, and then lighter again. Many populations adapt continuously to changing patterns of natural selection, but the directions of these changes cancel so that the long term trend is stasis."
Of course, that aspect of oscillating populations is dependent on the environment. The mouse population becoming lighter, then darker, then lighter again depends on the conditions around them. On the black rock of that New Mexico landscape home to the rock pocket mouse, there isn’t enough change in the environment to create that cycle of light to dark to light again. They’ve reached the local optimum of dark fur to camouflage themselves and hide from predators on the dark rock.
At 6’1” and 270 pounds, the offensive guard with 30” arms lines up across from a typical NFL defensive tackle. What happens next?
At 6’10” and 370 pounds, the offensive guard with 37” arms lines up across from a typical NFL defensive tackle. What happens next?
This is an example of where stabilizing selection plays a role in the size of the player relative to their position. If you are too small, you lack the necessary strength or length needed to effectively engage with your opponents and win the battle. If you are too big, you lack the necessary body control or leverage needed to stay low and take proper angles or use functional technique. Selection for size and strength keeps you from pushing too far to one end of the distribution while selection for control and adjustability keeps you from pushing too far to the other end of the distribution.
While it almost certainly never hurts to be a tenth of a second faster, there is a general range of acceptable heights, weights, and lengths that players must fall within relative to their position to effectively do their job, and selection pressures on both ends provide a squeeze.
Every aspect of the phenotype comes with benefits and costs that will be exposed within the environment, whether it be a tall running back who struggles to drop his pads, a short-armed linemen struggling to win an engagement, or a high-cut cornerback who trades off fluidity for top end striding ability.
One commonly used example of stabilizing selection in nature is the size of human babies at birth. The improvement in care for premature babies and regularity of c-section has relaxed some of the selection in this regard, but prior to those developments there was a large cost for babies delivered on the far ends of the distribution.
In the past, babies that were too small dealt with severe health problems or were often too weak to survive. Those that were too big had trouble passing through the birth canal and would do life-threatening damage to either themselves or their mother in the process.
There are some other forms of selection that could be combed through to try to find applications to the populations of players, but for the sake of both time and relevance these are the two primary selection types that are worth knowing and trying to apply.
The NFL’s Archean Eon
We don’t have to look at other leagues or hypothetical scenarios to find examples of these evolutionary processes and “environmental changes” in action. We just have to roll back the clock to look at some of the early days of the NFL.
The New York Giants, one of three NFL teams who still retain their original name from the 1930 season, were a peculiar group by today’s standards. In their 17 game season, the Giants averaged 18.12 points scored per game while ceding 5.76 points per game. The league as a whole was incredibly different at the time both in the rules utilized and the players who took the field.
Teams attempting more than one forward pass during a set of downs would be penalized. The rules stated that passes must come from at least five yards behind the line of scrimmage, a rule that would not change until 1933. Rosters were incredibly small and rules against substitutions led to “one-platoon” systems that saw players on both offense and defense much like the aforementioned Arena League.
I collected the heights and weights of the players on New York’s roster in 1930, 1980, and 2022 to compare the gradual change in positional standards and the league-wide trend towards bigger individuals. These comparisons give us a great example of the way evolution can occur for certain types as the environment changes and populations are expanded in a more diverse manner. Numbers courtesy of Pro Football Reference (Sheet w/ numbers)
1930 (23 players rostered)
Average height and weight of all players: 5’11 ⅝” and 203 pounds
Average height and weight of linemen: 6’0 ⅛” and 216 pounds
Average height and weight of non-linemen: 5’10 ¾” and 186 pounds
1980 (67 players rostered)
Average height and weight of all players: 6’2” and 224 pounds
Average height and weight of linemen: 6’4” and 261 pounds
Average height and weight of non-linemen: 6’1 ⅛” and 208 pounds
2022 (78 players rostered)
Average height and weight of all players: 6’2 ½” and 247 pounds
Average height and weight of linemen: 6’4 ⅝” and 303 pounds
Average height and weight of non-linemen: 6’1 ¼” and 213 pounds
To understand this growth, we have to look from a few different angles. We have to ask about why the absolute skill (and size) of players has grown over the years. We also have to ask why players are trending towards certain physical standards that are best suited for the playing environment.
The first professional football player was the son of a Union veteran who was wounded during the Battle of Gettysburg. Pudge Heffelfinger, an all-American guard for Yale from 1888 to 1891, was a talented athlete in many regards. At 6’3” and 200 pounds, he was physically imposing relative to many of the other players at the time. He was a standout in many sports, including rowing, baseball, track, and boxing.
But like many who were stars in college at the time, Heffelfinger did not have an NFL draft to go to or a professional team to sign with. Instead, he went to work and played amateur football on the side. It wasn’t until a rivalry between two athletic clubs promised to pay him a handsome amount to play that he broke ground as the first player to truly garner the professional label, according to NFL Operations.
A Nov. 12, 1892, contest between Pittsburgh-area rivals the Allegheny Athletic Association and the Pittsburgh Athletic Club changed everything. As the game approached, both teams were determined to beef up their squads. Heffelfinger — a low-salaried railroad office employee in Omaha, Neb. — became one of their targets.
Heffelfinger’s post-college status, however, was typical for football’s rag-tag days. He went to work at the railroad, but continued playing for independent teams. In the weeks before the Allegheny-Pittsburgh contest, he took a leave of absence from the job to take part in a six-game tour of the East with the Chicago Athletic Association, which used the expense-money method to attract players.
The Pittsburgh Athletic Club scouted one of those games and, according to a Pittsburgh Press story at the time, offered Heffelfinger $250 to play for them on Nov. 12. Allegheny offered him $500. The star played for Allegheny and earned his keep by forcing a fumble, scooping it up and racing for a touchdown — the game’s only score.
Allegheny did not admit to paying Heffelfinger at the time, but the Hall of Fame discovered a club expense sheet that shows the $500 transaction. It calls the document “pro football’s birth certificate.”
Pay for play would continue throughout the early 1900s, but the limited financial firepower of clubs would prevent them from providing enough to keep players full time. Players would often work throughout the week and play games on the weekends.
The emergence of college stars turning pro like Red Grange and Jim Thorpe created the pipeline from college to pro, and the jump in talent from sandlot players to athletic phenoms was officially underway. But the financial stability of the league was still questionable, and it was hard to draw players in when other jobs were more profitable.
In 1936, the first overall selection in the first NFL Draft was Jay Berwanger, a 1935 Heisman winner. Berwanger declined the low wages of the league in favor of becoming a foam rubber salesman. Only 24 of the 81 players drafted that year would sign contracts.
Even throughout the 1960s, players were working other jobs as a means of providing for themselves and their families. Specialization slowly became a regularity after player pay rose to sufficient levels and the rules against substitutions were removed. The days of one-platoon football were done, replaced by increasingly specific types of players who fit niches that only they were tuned to fit.
The pressure of pay (or rather, a lack of pay) is an example of selection on the players who were joining the league throughout the early days. As pay rose, it allowed for more talented players to pass up the benefits of a typical job like being a salesman or doing factory work in favor of the benefits of playing football professionally.
The pressure of ironman football with no substitutions selected for players who could “wear multiple hats” and work on both offense and defense, which prevented players from getting too big or too small in pursuit of speed or strength that would only be serviceable in one role but not others.
The pressure of racism limited the talent pool by selecting against African American players, which led to fielding less talented players in favor of their skin color.
As rules and resources changed to eliminate these issues, the population of available players increased and that would play a key role in breathing life into the league’s growth as the most dominant professional football league.
The Small Supply of Big People
“In the post–Big Bang of body types era, whether with height or reach, almost no player makes the NBA without a functional size that is typical for his position and often on the fringe of humanity.”
– David Epstein, The Sports Gene
Michael Mauboussin’s book The Success Equation touches on the paradox of skill, which is the idea that as skill gaps narrow there is more luck involved in any outcome. In this explanation of how skill functions league to league, he covers the NBA’s skew very far into the skill side of the luck-skill continuum rather than the luck side of the continuum. One thought experiment he uses to explain this gap is particularly pertinent when thinking about the quality of available talent (and subsequent ability to specialize) in the NFL.
Mauboussin’s thought experiment goes like this. We want to build a league of 25 teams, each containing 20 players per team. To do this, we are drawing from a town with a thousand players with a skill level ranging from one to six in a normal distribution. I’ll let Mauboussin explain how this goes as we draw from this pool, then what happens as we add more towns.
Source: Mauboussin’s The Success Equation
"To populate our league, we start from the right distribution and select the five hundred most skillful players. So we will take all 20 of the 6s, all 140 of the 5s, and all 340 of the 4s. We can then calculate that our players will have an average skill level of 4.36 with a standard deviation of 0.6. In this league, the 6s will perform head and shoulders above the rest of the players…
If we were to continue to add towns of a similar size and distribution of talent, we would fill our league with a rising percentage of 6s until we reached the limit at twenty-five towns… Baseball provides a concrete example of this process. One century ago, professional baseball players were white and came primarily from the northeastern United States. Baseball now draws players from all racial groups, and from not just the entire United States but the entire world. Just under 30 percent of players in Major League Baseball today were born outside of the United States, up from about 10 percent in the 1960s. Soccer and ice hockey have experienced similar trends. In chapter 3, we observed the paradox of skill. Now we have a mechanism to explain it: if the population of players to choose from expands and the size of the league remains constant, the average level of skill will tend to increase."
Mauboussin continues on the discussion of a paper by David Berri, which discusses a peculiar imbalance in the NBA. While the NBA has seen many of the same growths as other leagues as far as a larger talent pool and fewer factors like race impacting the selection process, the variation of skill within the NBA is still significantly affected by one key aspect.
"But there is a relatively straightforward explanation for the strong and consistent contribution of skill in the NBA: the height of the players. In most sports, the most skillful players within a wide range of heights can make it to the pros. But a relatively small percentage of men are tall enough to play in the NBA, and the average height in the league has risen over the years. Only about 2 percent of the male population is 6’4” or taller, but more than 80 percent of NBA players meet that threshold. Average stature in the NBA is 6’7”, and 27 percent of NBA players are at least 6’10”, which is four standard deviations above normal. For some context, four or more standard deviations means you’d expect about thirty-two hundred men of that height in a population of 100 million.
This mental exercise suggests that a small population of big men should lead to a wider variation of skill than a large population of smaller players. A pair of economists, David Berri and Stacey Brook, along with some colleagues tested this prediction. They compared players who were 6’4” or shorter with players who were 6’10” or taller, using points scored and productivity each minute to gauge the variation. The two groups had roughly the same number of players. They found that short players had lower standard deviations in performance than the tall players did, which means that the range of scoring and productivity was smaller than it was for the big men.
The high standard deviation of the tall players appears to be an important factor in explaining why skill is the primary determinant of the results for NBA teams. Even though the NBA is more international than ever, skill levels show relatively high variance because of what Berri et al. call “the short supply of tall people.” Think back to our mental exercise. The NBA’s need for tall players means there’s a small pool of talent from which to select, so it’s like drawing players from one instead of many towns. As we saw, there is a wide range of talent in one town, and the range of talent narrows as more towns are included."
We have two things to account for here, but they both tell us a lot.
Firstly, we have a representation of the way skill increases as the population available increases. As selection pressures like low wages, racism, or one-platoon systems that stamp out player variation get removed we get a change in who can enter our league. If we are paying a large amount of money, Jay Berwanger is taking the field rather than selling foam rubber. Our league’s skill increases. If we aren’t preventing black players from entering the league, star running back and color barrier breaker Kenny Washington spends the prime of his career in the NFL rather than playing for the Hollywood Bears, a Pacific Coast Professional Football League team. Our league’s skill increases again.
Secondly, we have another way of expressing directional selection on a population as we change the parameters of leagues. In the NBA, directional selection favored taller players. In the NFL, we can assume that it favored bigger players as the offensive linemen went from the bulkier guys in the sandlot to gigantic individuals that would probably be part of Greek mythology if they fought in the Trojan War. We can assume this based on the growth in average New York Giant lineman size from 6’0 ⅛” and 216 pounds in 1930 to 6’4 ⅝” and 303 pounds in 2022, a total growth of four and a half inches and 87 pounds.
While the individual skills of offensive linemen are judged mostly subjectively, there is a persistent narrative out there about the limited supply of talented big men, which particularly batters the bottom of NFL rosters and secondary leagues like the XFL and USFL. This idea of a “smaller town” to draw from provides a good explanation for that, as the population to pull from who meet baseline measurables is much smaller than the population for other positions.