Quantifying the cost of assholes on the subway

(Disclaimer: tongue-in-cheek and no claim of scientific accuracy)

Living in NYC I am frequently confronted by herds of ‘subway assholes’ — individuals who attempt to enter the subway car before anyone has exited.

On the surface it doesn’t seem to make any sense. It defies physics — no two objects can occupy the same space at the same time. And doesn’t it slow the whole process overall? One morning the rage reached acute levels which created my next personal project. I was going to study and attempt to quantify the cost of the ‘subway asshole’.

Asshole cost: the additional cost incurred in emptying then filling a subway car caused by the presence of assholes versus ‘normal’ people.

Selecting p5.play and p5.js I built a simulation. (Unfortunately medium is not embed-friendly so simulation is here: https://geod.github.io/subwayassholes/)

Simulation with those who are polite (note channel at bottom)
Simulation with assholes (note they are cramming in before people have remotely exited)

Asshole: individuals who try to force their way onto the subway car as soon as the doors open irrespective of people exiting

Normal Person: individuals who queue at the sides of the doors and wait until the exiting group are all out before entering

Results 1

The first experiment was to run a simulation for 10 people entering and 10 leaving with different proportion of the entering group as polite/normal people versus assholes. Data below: key result 10 assholes takes ~600 units of time versus 10 normal people ~500 units of time.

  1. 10 assholes take 20% more time than 10 normal people
  2. The marginal cost of each asshole is 10 units time

Results 2

Ok — so the above really got me going (I was marginally surprised that the cost of assholes was linear. My simulation rules are perhaps slightly more favorable than the real world in that the group can ‘jostle’. Whereas in the real world people get stuck and the process of un-working is much more time consuming/inefficient. However, this is as much stress relief as science so I continued)

The next question was what is impact of larger groups switching position? Are assholes more damaging when the subway is busy? To test this I re-ran the same simulation for different group sizes. Data below!

And presented as a 3D-plot just to look snazzy (made with plot.ly)


  1. Increasing the size of the group switching places increases the total time taken (not surprising)
  2. Perhaps slightly surprising is that the impact of adding more people is more expensive than adding assholes (note that increasing the group size by 2 adds 4 people — two entering and two exiting)
  3. A bit of (incredibly flimsy) statistical regression creates Thomas’s first law of subway assholes which is:
Thomas first law of subway assholes

So there you have it — an asshole is ‘scientifically’ proven to be ~40% more expensive than a polite person in terms of time impact on efficiency on the subway.


On the surface it appears irrational to be an asshole — yet it still happens? Let’s delve further into the mind of an asshole to understand the possible benefit/utility function. Is it:

  1. To Get on the subway first (just to stand around?)
  2. To Increase the probability of securing a seat? (unlikely in rush hour as seats are filled or filled by those already on the car)
  3. Driven by fear of not getting on a subway car when there is finite space? (likely in rush hour)

Therefore, let’s for one second not assume that assholes are entirely irrational. Simulation does confirm that the assholes are frequently on the subway car before the polite individuals. What if there is a utility of being an asshole (most likely centered around #3)? To assume they are rational introduces Thomas’s second law of subway assholes which is to say that if they are rationale their behavior must have benefits which exceed the cost of being an asshole

Thomas second law of subway assholes

I am sure many of the assholes do it because …. well … they are assholes. However, perhaps some assholes are actually smart and realize that although they damage the overall system they have an improved outcome. This establishes Thomas’s third law in which there are two types of assholes

Type 1 ‘Irrational’ Assholes: Those who are assholes even when it is apparent there is excess capacity on the subway car and there is negative personal value and negative overall value in being an asshole. (59th street B-D — I am looking at you)

Type 2 ‘Rational’ Assholes: Those who identify that there is scarce capacity on the subway car and choose to be an asshole in recognition that the personal payoff exceeds the negative impact to the system (2–3 train on rush hour at 72nd street — take a bow)


Subway assholes are probably here to stay. Their impact of an asshole has (pseudo)scientifically proven to be +40% versus a normal person. Perhaps, having increased understanding it will reduce my rage. Perhaps not.

As for constructive advice — my new favorite tactic is to be first at the door to exit. Then instead of forcing my way out — I instead hold my arms outstretched to a) indicate I have nowhere to go and b) prevent assholes from entering (this needs to be combined with strong eye contact). I have found that assholes when ‘out-assholed’ have no choice but to scurry backwards and open up a channel. Success.

(Team america world police really was right about this if you have seen it …)