Power and Class

16 min. read

The world really seems to have fallen to pieces.

We’re deep into the throes of a global pandemic where hundreds of thousands of Americans have died and millions of people across the world have experienced discomfort, loss of work or wages or housing, and/or loss of loved ones. It also just happens to be a time in our country where there are enormous uprisings to demand that something be done to address the inequity in our country and stop our police officers from repeatedly taking black lives.

“Ugh, 2020,” you mutter under your breath, rolling your eyes at the crazy mess that this year has brought on us. But it’s weird it has all happened at the same time, right? Yes, stress from one problem can aggravate hurt from other but the connection seems deeper than that. It’s easy to blame Trump for everything that is happening to us, but what else could we possibly have expected from a man like that? All he is is a selfish idiot who was given the keys to the castle. Who gave him those keys? Who decided that the country would best be run by an egotistical millionaire whose largest contribution to society was a reality TV show that netted us precisely one mediocre meme? We did. America did.

Before you start to blame this all on our president or bad luck that these things all hit at once, consider the following:

Power structures

Warning: I am blending my thoughts an opinions with ones I have heard or read about. It is possible I will mischaracterize some of them at some point and I apologize for that in advance. It’s also possible that my thoughts on this will evolve as I write it or as I reflect on what I wrote. :)

I recently was talking to some people about class as it is interpreted in socialist circles. The traditional Marxist view of class is generally thought of a binary divide between the bourgeoisie (the ones that own the “means of production”–land, factories, capital, etc) and the proletariat (the people who work the land and and in the factories to earn a wage). A common refrain among Marxists and others is that in a capitalist society the upper class must extract value higher than the wages paid to the lower class in order to make a profit. In other words, the system is, by design, unfair to those in the lower class. And I think any capitalists that are reading this would probably agree that this is an accurate (although perhaps reductive) description of how capitalism is supposed to work.

Anyways, back to our discussion. We were asked to talk about race and sexuality and other issues of identity and to talk about how they fit into discussions of class. My thought was something along the following lines: what Marx was talking about was a power differential. People belonging to the upper class (whether you believe that it was given or earned or ordained from God) have an outsized power over the lower class which they can (and generally do) use to control and extort them. But it is not only money that creates power differentials. There have been similar power structures in place separating men and women, white from non-white, straight from gay, and so on for nearly every dimension you can consider.

The messy intersection between class and other identity variables is the reason this is a difficult discussion to nail down (and also why it is worth having). I think the only conclusion for an informed, empathetic, and reasonable person is that class is inextricable from race and gender and other signifiers of identity.

Power and individuals

Let’s start by discussing how individuals view differences between people and how they incorporate these differences into their worldview to inform how they treat others. In my opinion, one of our greatest strengths as a species is our ability to form abstractions and quickly generalize experiences (and even stories from others) to apply them to new situations. A necessary part of this skill is to quickly and reliably identify important similarities and differences in objects and situations and people so that one can make inferences about what to expect or how one should act in order to preserve one’s safety and way of life.

It truly is an amazing skill we have. Tech companies and universities have spent untold sums of money and dedicated huge amounts of brain power to solving the problem of how to teach machines to reason about new situations as “well” as humans do naturally. But one need look no further than a generative model trained to “depixelate” images to realize that the state of the art isn’t up to snuff. A question we should be asking ourselves: if our judgement-making machines distill situations and people down to a handful of discrete variables, are we ourselves subject to the same kind of bias as a machine learning algorithm?

I think the answer is yes. And I think there are a lot of people that agree with me. For years people have warned against the danger of stereotypes and of institutional racism and what are these other than misclassifications our wonderful meat machines are making in the moments leading up to a decision? I have experienced many times in my life where my gut instinct is wrong and based in troublesome suppositions. In many of these instances I have observed things over time that made me realize just how wrong my instincts were. We have all done this. This is just growing and learning. But if we accept that our impressions of things are malleable and subject to change, it seems wrong to rely on such instincts in scenarios where our life and safety is not in question, but that of others is.

Many people are interested in critically examining their own implicit bias and the bias in the systems around them, but some are not. Some are actively opposed to it. I think the two most common manifestations1 of this mindset are eradication and power assimilation. Eradication comes from a place of fear: we fear things that are different from us because they are not things that we know how to control. We worry that the other people don’t share our values and therefore pose a risk to us and our way of life. Eradication doesn’t necessarily mean extermination, but can also mean forcing the foreign group to assimilate (or at least become invisible), largely in hopes that once they become “like us,” they will no longer be a threat to us. People who seek eradication include people as violent and aggressive as the KKK and white nationalist groups as well as people who espouse the joys of being “colorblind” or think that everyone should be treated equally, regardless of cultural and/or historical information.

A more callous lot are those that wield labels and divisions in the interest of gaining power. In a way, these people are diametrically opposed to the people that seek eradication in that they rely on the existence of these categories to succeed. The most obvious lot here, and the ones that are the subjects of classical Marxism, are the ultra-wealthy who rely on the control over the means of production to continue to amass money (and thereby power) by exploiting the working class. I think that politicians, although not always purely driven by money, are another group often deserving of this title. One needs to look no further than the Republican leadership (or the Democratic leadership, but the former has more egregious recent violations of basic morality and humanity) to find examples of people who leverage differences in people to empower themselves. And it’s not just the fact that they are more wealthy than the average American (although they certainly are in general) that they leverage, but they can draw on other divides to ignite our lizard brains to act out of fear to give them more of what they want: more campaign donations, another four years, wider powers with less oversight, etc.

I think this last point is the one that some socialists make when they say that we shouldn’t allow ourselves to get caught up in issues of identity politics when what really matters are issues of class. And to some extent I get it. One of my pet examples in this arena is gay marriage–it was an enormous step for our country when we made it so that states had to officially recognize the love between same-sex partners. And while I am so happy that this occurred (it empowered my sister to get married to her wonderful partner just recently), I found myself at the time feeling a bit confused. We had huge debates about whether this should happen and people with strong opinions on both sides arguing about what was “right”, but the CEOs and the legislators, I realized, didn’t care one ounce about what was happening. Nabisco couldn’t care less whether you are allowed to marry the person you love. They only care about selling more Oreos. If public opinion has swayed in a way that them inserting rainbow frosting into their cookies will cause them to make more money than it costs to do it, they will do it. Look no further than the way that politicians’ stances on abortion and gay rights evolve over time as the opinions of their electorates change (let alone the many instances of Republican politicians with anti-gay stances have been exposed for their own secret homosexual relationships). So then are the identity politics games we engage in just distractions from the larger discussion of who gets to have (and keep) power?

Power and society

While the individual mindset helps us to understand how other people have different views as us, the difference between how the average American views gay marriage as opposed to an enormous conglomerate highlights that there are different levels on which we can view these issues. When we elevate our view to how these biases coalesce into systems around us, we no longer see a distinction between different aims but a superposition of all possible incentives for upholding an idea. It is by definition reductionist to say that racism is purely a tool of power assimilation, but it is equally reductionist to say that it is purely a byproduct of individual fears. Hate and fear and desire for control all come together to form a messy institution that it is difficult to nail down, let alone address.

Consider for a moment the following plot I created from US Census American Community Survey data for 2019 (some information about sources and some assumptions I made in visualizing this data can be found using this2 footnote):

A plot showing family income distributions for Blacks, Hispanics, Whites, and Asians

Money is the primary factor we are visualizing here, and while your eyes might be drawn more to the differences between Asians and other groups, it’s worth recognizing the (at first glance) subtle differences between White non-Hispanics and member of the Black and Hispanic/Latino groups. The general trends are the same and the overall range covered ($0 to $250k+) are the same for each group considered, but these distributions are not the same. A simple first metric to consider is the median:

Racial group Median family income (2019)
Black/African American $44,124
Asian $94,588
White (Not Hispanic/Latino) $72,045
Hispanic/Latino $56,117

One could argue that the median is too simple to capture the complexity of the issue and in some ways they are correct. But if you are right in the middle of White America in terms of income, that puts you above 71% of Black Americans. If we plot the (experimental) cumulative distribution functions for all races, we get the following:

A plot showing CDFs for family income distributions for Blacks, Hispanics, Whites, and Asians

I like to think about these CDFs in the following way: if you draw a vertical line at any point on the graph, you can consider the points where it intersects two different races’ CDFs. The difference in the y-values you find tell you something about the power differentials between the two groups. The fact above (that 50% of White Americans have more money than 71% (or more!) of their Black counterparts) is just a specific instance of this3.

There are a few things to take away from this data: first, classist policies are inherently racist. If a policy is enacted that, on average, benefits everyone who makes over the median4 US salary of $68,703, White people could look among themselves and realize that the policy would help 52% of them and therefore that the policy was in their best interest. The problem is that this would only help 32% of Black people, so they would see it as a bad and/or ineffective policy. Population statistics (Black people comprise about 13.4% of the US population, compared to 61% for non-Hispanic/Latino White people) further complicate the matter in that the 68% of Black people left in the dust by such a policy would only comprise about 9% of the electorate.

On the other hand, racist policies are inherently classist. Perhaps I am beating a dead horse, but the fact of the matter is that the correlation between these two variables is high enough that it is very difficult or impossible to create a policy which is fair by one measure but not by the other.

To elucidate the problem using numbers: in a preprint I recently wrote with some collaborators (shameless plug), we investigated whether two sets of labels on the same dataset were independent using the metric of mutual information. Here two variables are independent if and only if their mutual information is zero. We found that two strongly correlated label sets had a mutual information of 2.318 bits. In this dataset (where race is just Black/White and income is grouped by $5,000 brackets) race and income enjoy a robust 3.056 bits.

What this means

I would be willing to bet that most people that have made it this far are already (at least partially) sympathetic to the cause I a promoting and so I am, to some extent, preaching to the proverbial choir. For all of you, here’s something: I picked two variables that are often the subject of disagreement among people when discussing which policies we, as a society, should support. I would be willing to bet most lefty people have had discussions about this in the past. The argument is often goes something along the lines of the following

  • Alicia: Policy A, which disproportionately benefits billionaires, is racist.
  • Blake: In America, we have rules that laws can’t be explicitly racist! At worst, Policy A is classist and is decided without any concern for race.
  • Alicia: That’s exactly the problem.

at which point we leftists can walk way smugly and all the conservative tears will run in cigar lounges and country clubs across the world. But this same analysis tells us that choosing to rebel against the upper class is functionally identical to attacking the power structures that protect white, straight, cis, male, “Western”, and Christian people and subvert the wills of the other.

I think it is perfectly possible that a person could read through to this point in my rambling post and agree with most of what I said and come to the following conclusion:

Fighting a war against identity politics is futile, since (as is pointed out above) it is the same as the fight against the class divide. Resolving one is resolving the other, and thus we shouldn’t let issues like race and gender distract us from our quest of attaining a more equitable society.

I’ll admit, it’s tempting. But the problem I see with this take is that I don’t view identity politics as something that distracts from the issue of class. In fact, consider the following little nugget: until the fall of capitalism, the class divide will remain. Which, like…. duh. But if we are only interested in eliminating capitalism and the class struggle, these other fights will languish and none of us can be liberated until we all are. And while that makes for a romantic statement to make in solidarity with the subjugated, I would be very upset if all the complicated and venomous power structures that supported our broken society remained in place until the very last moment when everything changed. What a tragedy that we would not be able to alleviate the suffering even of the smallest part of society before we were completely done dismantling gargantuan system of oppression that hangs above us all.

Truly a shame.

My take

I choose to believe that the best way to attack the whole is to dismantle the pieces that comprise it. “Go for the eyes” in demanding that women be treated and paid fairly to men. “Sweep the leg” in demanding that companies actively counter biases seen in hiring against people with “black sounding names”. Attack the Achilles heel by demanding that trans people deserve human rights and create a culture where they feel less isolated and may be less likely to kill themselves for how they feel.

These are just right. It’s not hard to justify such actions to yourself (or if it is, you should really take a long look at who you are and what makes you feel otherwise). And the plus side is that all of these fights are aligned with the socialist struggle. You haven’t wasted your time helping your fellow humans become more free and accepted and happy, you have done a good thing. And if we can push the agenda for fairness and equality forward on all the other fronts except for class (which, again, is actually inexorable from these other issues), we will be left in a world founded on equality and mutual empowerment in which one class is unjustly seated in a position of power over the other. I can’t help but think that is actually the ideal situation for dismantling the last bastion of power differential in society and pushing us forward into the society we all dream of.


  1. I don’t think that these are the only two kinds of people who avoid engaging with their own implicit biases and I doubt that anyone is purely comprised of one or the other. In reality people are most likely a blend of these mindsets and more. 

  2. I got my data from here, here, here, and here. Since we are only given the number of people in each income range, I generated data points uniformly from the range in proportion to the number of people in each range. This likely causes the “waviness” in my charts but the picture is more clear than if I had given every family the same income. Furthermore the last block is for $200k+, which I sampled uniformly from the $200-250k range. Since there are many people who make more than $250k, this is likely what causes the bulges above 200,000 (which is especially pronounced in Asians and Whites). 

  3. It ends up that what I am doing is a form of the Kolmogorov–Smirnov test. The distributions are pretty clearly different, though, so interpreting it this way isn’t much help. 

  4. I realize that basically no one would draw the class divide at the median income, I decided to use this because the number was readily available and a reasonable number for illustration’s sake. In the interest of fairness, I should mention that the differences would be less stark if I chose a number closer to the upper end of the spectrum.