Beware Cultural Drift

Everyone without exception believes his own native customs, and the religion he was brought up in, to be the best.
~Herodotus circa. 440 BCE

Who or what do you most trust? You might think it’s your parents, spouse, or a teacher. But in fact, you most trust your culture, as you are part of a species, humans, with the superpower of cultural evolution. So as a kid, you eagerly accepted most of what your prestigious associates said, especially if they had long agreed on it. You’re also inclined to accept changes to your culture that come from your culture, and to dismiss the contrary opinions of outsiders. You trust your culture not just on practical facts but on deep values and sacred feelings. Your culture’s instructions feel like a warm, loving embrace. You even find it hard to notice when you’re trusting your culture; its claims feel like obvious facts that no sane person could question.

Do cultures deserve this trust? They used to. For most of human history, the fact that your culture existed was evidence that it had done well for your ancestors and would likely do well for you. (This didn’t mean your culture’s claims were literally true, just that they were a reliable compass for living.) But today is no typical day in history. Until a few centuries ago, the world had hundreds of thousands of cultures. Each small peasant community was mostly self-sufficient, in effect a separate “macro” culture. Suspicious of neighbors, but mostly peacefully coexisting, each was relatively free to defy a neighbor’s disapproval if it didn’t defy an encompassing empire. 

But in the last few centuries, states merged these local cultures into nation-scale macro cultures. (For more on this, see Eugene Weber’s 1976 book, Peasants into Frenchmen: The Modernization of Rural France, 1870–1914.) And lately we’ve seen the rise of a global elite culture. Top university graduates from around the world mix at events like Davos, set convergent policy worldwide, and serve as interchangeable professionals in global organizations. The small peasant cultures of yesterday were poor, reluctant to change, and subject to frequent epidemics, famines, and invasions. They faced strong selection pressures—one false move could kill them. In contrast, today’s handful of huge, rich, fast-changing cultures face weak selection pressures and can survive many false moves. Paradoxically, that’s just why you should maybe distrust them.

Here, I want to persuade you that your macro culture is likely drifting off the rails and taking you with it. I will start by explaining a bit more about how culture works with a couple of simpler examples.

Example 1: Corporate Cultures

While it’s hard to get good data on macro cultures, we have much better data on corporate cultures. Today, there are 166,000 US employers with at least 100 employees; half of new firms die in five years, and firms last on the Fortune 500 list for 15 years on average. 

A firm’s tangible assets include buildings, machines, contracts, and inventory. But for firms in the S&P 500, only 10 percent of firm value, on average, is accounted for by such assets. A bit more firm value comes from strategy: tech adoption, org charts, product designs, pricing strategies, and so on. But as these are standard, easy to change in response to success or failure, and easy for competitors to copy, they can’t make up much more than another 10 percent of stable firm value. (Yes, firms gain from customer and partner respect, but these are mostly downstream of prior firm behaviors.)

But firms are complex social systems. They have many parts that are hard to see, understand, or control. Parts that add up to “the way we do things around here.” Some of those parts are stable, shared across the firm, and hard to vary within the firm, such as status markers and behavior norms. These add up to a firm’s “corporate culture,” which seems to account for most firm value; firms with good cultures are worth far more than those without.

Corporate cultures change in response to market conditions, macro cultures, and managerial strategies. But they change even more from complex, unpredictable processes within culture, such as firm factions fighting over the directions of local cultural change. And if firm culture accounts for most of firm value, its changes should account for most changes in firm value, such as when they go in or out of the Fortune 500, or when they go out of business altogether.

Aware of this fact, firm managers try to steer their cultures as best they can. This can take five years or more, and even then, two in three attempts to change firm culture fail. The usual life history of a firm’s culture is that, after some initial improvement, it will tend to drift, accumulating dysfunction, until the firm ceases to be worth saving and dies. Of course, there’s a lot of variation; some firms improve for a long time before decaying.

This process explains slower growth in older firms, and in industries where a few big firms rule. That is, firms that don’t start out with sufficiently functional cultures never get off the ground. And for the rest, an initially functional culture drifts fast in the space of possible cultures, sometimes to better places, but more often to places that are bad for profits or survival.

On average, firms today are more productive than in past centuries because they benefit from better technology and better cultures. But why is that so if each firm’s culture degrades over time? The key is selection: firms with bad cultures go out of business fast, while those with good ones grow, inspiring new firms to copy them. If the world had only a few firms, each of which lasted centuries, we’d see vastly worse corporate cultures and far less progress overall.

Example 2: Biology

Compared to plants and animals that live in the ocean, species that live inland (especially in rivers or tropical rainforests) have smaller spatial habitats and consequently fewer members. Between 85 and 95 percent of all macroscopic species now live inland. And since it’s typically harder for biological innovations to jump across species than to spread within them, you might think this would make inland innovation slower, compared to ocean innovation.

But land (and river and rainforest) species have been especially innovative, with their descendants disproportionately colonizing other habitats. While innovations within a species must stay consistent with the defining features of a species, innovation in the features that define a species requires selection between species. And the latter sort matters more.

Now let us generalize from our two cases: species and corporate cultures. For each type of unit, innovation happens at two levels: within units and between units. Having larger units promotes within-unit innovation but hurts between-unit innovation. As that latter sort usually matters more, innovation gains overall by having more smaller units instead of few larger ones. For example, in retail firms, more innovation happens via new better firms replacing old worse ones, instead of firms getting better.

How Macro Cultures Work

Macro cultures are parts of our larger social systems—like nations—that are hard to see, understand, or control. These parts, such as status markers and behavior norms, are also stable, shared across the larger unit, and hard to vary within them. As with firms, variations in the cultures of such units likely account for most variation in unit success. For example, nations are rich mainly because they have good cultures.

Yes, our cultures have a complex fractal structure. We have cultures associated not only with organizations, but also with professions, hobbies, ethnicities, religions, cities, and generations. Here I focus on the cultures of our largest units, like nations, as their features are especially hard to change, and we have less global variation in them.

Just as for other kinds of culture, macro-culture innovation happens at two levels: within and between. The recent big jump in the size of macro cultures has boosted within-culture innovation, powering peace, trade, and fast-growing wealth. As a result, our few huge cultures today suffer much less from famine, disease, or war. But because of these effects, we should expect to now get much less selection of cultures, and thus less long-run innovation.

It’s not just that we’re forgoing opportunities to improve our macro cultures. Selection may also be too weak—at least in the short run—to cancel the mistakes of cultural drift. Shouldn’t we expect that macro cultures, when selection is weak, will drift into dysfunction just as firm cultures do?

After all, most of humanity’s cultural experience has been with cultures much closer to the size of firms than nations, and so our cultural skills also work better at firm scales. And compared to firm leaders, macro-culture leaders have far weaker incentives and powers to control culture. So, even with all these advantages, if firm leaders still can’t manage to prevent drift into dysfunction, what hope could macro culture leaders have?

However, powerful social factions often fight hard to influence our macro-culture changes, and the winners of such fights have a strong interest in framing such changes as improvements, not dysfunction. They offer two main supporting stories.

Story 1: Context-Dependence

Powerful social factions like to claim that cultural changes are reasonable adjustments to changing contexts, like tech, density, wealth, ease of talk, or lifespans. We aren’t changing our deep values, they say, just our surface behaviors. After all, when plants and animals change their behaviors with context, like the time of day, they typically execute sensible pre-programmed adaptations. Might human cultural changes be like that?

First, this can’t explain a single culture splitting into different versions that behave differently in the same new context. For example, if our ancestors were reborn, and acted here more like they did in their era than like we do now, that would suggest that our deep values have changed.

Second, cultural evolution is a new thing in the world, pioneered only recently by humans. So cultural processes are far cruder than most DNA-built structures and processes. Culture is more like the first powered planes, with a few hundred miles of flight per fatality, than commercial planes today, with ten billion miles per fatality. We shouldn’t expect culture to have as many subtle adaptations for context-dependence as DNA-designed organs.

Third, consider a few of the macro-culture changes in my lifetime: a big rise in parental attention; a switch from cornerstone to capstone marriage norms (i.e., waiting for stable career and personality); longer career prep; a preference for cities; great declines in religion, patriotism, and militarism; far more acceptance of homo- and trans-sexuality; far stronger norms against sexist or racist language; and national cultures merging into a global culture, especially among elites. If these were simply our adapting constant deep values to new contexts, why do they look more like values changes? As humans are familiar with consciously adjusting behavior to context, why didn’t culture just tell us our deep values, and then let us consciously adjust behavior?

Fourth, wouldn’t pre-programmed changes to context be more predictable, at least once you knew of the new contexts? Yet many cultural changes are hard to predict. Yes, one can predict that, once elites accept a new value, masses will follow later. But those initial elite values seem hard to predict. Also, many culture changes, like Neolithic pottery styles or US baby names, seem to quite clearly reflect random drift.

Consider that Hitler taking over Germany, and then losing WWII, were among the most influential cultural events of the 20th century. Yet these processes were hard to predict. Similarly, whether or not world elites will take a pro- or anti-Israel stance in a few decades’ time looks important for future culture but hard to predict now.

AIs Will Be Our Mind Children

We must free our artificial descendants to adapt to their new worlds and choose what they will become.

QuilletteRobin Hanson

Story 2: Learning

The second main story that social factions offer to frame their culture changes as improvements, not dysfunction, is learning. They say that biological or cultural evolution endowed us with constant deep values with unclear implications. So our beliefs about the practical implications of those deep values change as we think more about them and test ways to achieve them.

Now this learning theory has a clear implication: changes of best estimates regarding practical implications due to learning should follow random walks, where it is hard to predict the next step of the walk from all prior steps, on average. As above, there could be predictable delays where elites learn of something first, which then percolates to the masses with time. But elite view changes should be unpredictable. Yet many cultural changes seem to be more predictable than this.

More important, as-yet-unnoticed implications of deep values couldn’t have been selected for by biological or cultural evolution. As selection acts only directly on behavior, any prior selection would have been for the previous estimates, not their deeper supporting values. From the point of view of evolution, changes to behavior estimates represent out-of-control drift, not information on the adaptive power of the deeper values.

Fertility Decline: Proof of Culture Drift

The clearest proof of biologically maladaptive culture drift is fertility. Children per woman per lifetime has been declining worldwide for centuries, and is now below replacement levels almost everywhere. Earth passed peak births in 2016, and in a few decades, we’ll pass peak population. Absent huge AI advances, innovation rates will then fall even faster than the population, causing a many-centuries-long innovation pause, and then less liberal governance, perhaps including even the return of slavery.

This fertility fall is driven by many strong and beloved cultural trends, including more gender equality, more intensive parenting, longer inflexible career paths, less religion, more urbanity, capstone replacing cornerstone marriage, and less grandparent involvement. On the whole, these look more like non-adaptive value drifts than adaptive learning or context-dependence. And having fertility fall below replacement during times of plenty seems clearly maladaptive. While policy solutions exist, like big payouts to parents, they seem unlikely to be adopted, as they need us to care enough, and to allow the reversal of beloved trends.

How exactly did culture drift to hurt fertility? Maybe many independent trends just added up to that. But another possibility is that high-status folks had wealth to invest in kids and widespread status markers that could be improved by wealth. Then our general cultural habit of copying high-status behavior could have combined with a selection effect: having fewer kids causes each to have higher status. This pattern was widely reported in history, at least among elites.

Just as our cultural drift story predicts, the main fertility exceptions we see are in fragmented, very insular cultures like Mennonites, Amish, and Haredim. By doubling every two decades, they look on track to replace our mainline civilization in a few centuries, just as Christians once took over Rome with similar growth rates over a similar timescale. And just as Christians discarded many things they didn’t like about Roman civilization, these new groups may discard many aspects of our liberal civilization that we now treasure.

In fact, many ancient civilizations like Greece and Rome plausibly fell due to low fertility. And humanity may repeat this pattern: innovation causing wealth causing fewer richer cultures, which drift, fall, and fragment. Culture selection then heals drifts, letting civilizations rise again.

Is There a Fix?

Eventually, when our descendants spread across the stars, long communication delays will ensure cultural fragmentation, and thus more selection. (Fast or easily copied minds might also work, as in my book Age of Em.) Before then, I see only three fixes: conservative, totalitarian, and multicultural. And none seems likely to work (though we should try).

The conservative fix is to revert culture back to a point when cultural selection was strong, and then stop it from changing. If these cultural values are shallow, this would forgo gains from adapting deep values to changing conditions since then. But agreeing on deeper underlying conservative values seems hard.

The totalitarian fix is for culture leaders to take strong control of culture, to stop maladaptive change. But such leaders would need to manage this much better than either CEOs do for corporate cultures or autocrats do for nations.

The third fix is deep multiculturalism, wherein cultures with deeply divergent norms and values still somehow maintain peaceful trade and tech exchange. This fixes only variety problems, and not weak selection pressures, and it requires much stronger global tolerance than today’s “boutique” multiculturalism, which tolerates varying clothes, foods, or language, but not deep values. Deep multiculturalism does engage many who are disturbed to see world capitalism steamroll small indigenous cultures, and also some libertarians like Robert Nozick. But not most people. For example, I conducted Twitter polls on 12 kinds of diversity, and “fundamental values” was the kind where respondents most wanted less diversity.

Note that all of these fixes require either changing world culture as a whole, or changing some big part that is strongly insulated against mainstream cultural influence. Our world is like a big ship heading for an iceberg; we must either turn it or rush for the lifeboats. So, alas, our likely fate before we reach the stars seems to be as follows: world civilizations repeatedly rise, suffer cultural drift, and fall. As they chanted in Battlestar Galactica, “All this has happened before; all this will happen again.”

NOTE: The opening paragraph has been slightly amended for clarity since publication.