No True Scotland
Rule Technological Civilizations In, Not Out
Part of a series on the history of technological contingency & the future of AI.
Scotland, a well-known True Scotland (Source: James Craig (1768), Plan of the New Streets and Squares intended for the city of Edinburgh)
When we discuss the future of technology, when is history a good guide? Which historical cases are valid or useful sources of evidence? These are not (just) epistemological questions, but a load-bearing consideration for how we reason around the future impacts of new tech; and for how we judge the latitude and levers we have in shaping those.1
If we assume that we can learn anything at all from history, we face the question of what historical evidence counts; how we weigh it; and how to make sure that we don’t stack the deck on that evidence in ways that make our theories functionally unfalsifiable.2
Yet my experience is that in such debates, people can be quick to dismiss or minimize historical cases out of hand. And, sure, at face value, when you are trying to understand the drivers and constraints on modern states’ industrial strategy, there’s only so much that you can learn from, say, the persistence of the unusual Khitan dual-government system, the evolution of Polynesian mnemonic stick-chart navigational tools, or the way that Cahokia managed to develop a complex and dominant hierarchical society without any writing system.
To be clear: I think it’s fair to argue that we can learn more from innovation patterns in the recent past, i.e. in societies more (culturally, geographically, materially, economically) similar to ours, and at closer junctures in the tech tree.3 Still, I don’t think we get to write out surprising cases just because they occurred in premodern or seeming obscure contexts.
More to the point, I claim that none of the standard criteria (and many nonstandard ones) for restricting which civilizations count as properly ‘technological’ really draw a defensible line. And that as such we should adopt a permissive inclusion rule for historical evidence on convergent and divergent tech paths.
Are You Living In A (Technological) Society?
Let’s return with my collection of cases of surprising technological divergence found in premodern societies. Is this just a list of trivia to do well on pub quizzes? I mean, yes, it’s that, too. But such countercases should also carry some weight in persuading us that the viable branches on the tech tree are, if not unlimited, still broader and more numerous than we’d naively assume. Even so, are these genuine signals of What Could Have Been? Or are they still just not that relevant to understanding the dynamics and scope of technology in the modern era?
I’ve already reviewed some potential objections over whether such cases ‘really count’, rebutting suggestions that (all) cases of technological divergence occured in ‘frivolous’ technologies; were fully explainable by variation in the available material resources or in the needs created by the local environment; or were obviously dead-ends.
A determinist could still reasonably retreat to the claim that these technological differences still shouldn’t count as strong evidence, as divergent innovation paths were not really the products of ‘true’ technological civilizations. Let’s call this the ‘no true Scotland‘ argument.
Can we avoid (unjustified) appeals to civilizational ‘authenticity’ when comparing the technological track records of different societies?4 I think this requires a more principled account of what actually ‘qualifies’ a civilization’s track record on the tech tree; and when its technological achievements do (and don’t) reveal important, transferable insights. Otherwise, I think we’re at risk of arbitrarily ruling out any contrary evidence, and claiming that ‘no true Scotland’ would adopt a technology path that diverges from the assumed optimum one.
There are a few different accounts that people often draw on in judging which are ‘real’ technological civilizations; I don’t think many of them are very satisfying or without problems. But it's interesting to review them, so let's.
Is this the Sign of a Technological Culture?
1. The folk account: ‘I know Civilization when I see it’
‘what did I learn in high school about history?’ The Pyramids. The Great Wall. Things like that. Those things have got to show up in the game [Civilizations], because when you see them, when you run into them, you go, “Oh, I know, I’ve heard of that, I’m a smart person, I know this stuff.” So we wanted to put in the game, but then the question was ‘What effect would it have?’ If it was going to be a Wonder of the World, it had better be pretty dramatic. That was another rule of the game: stuff had to really feel important...
– Sid Meier (by way of Gavin Leech)
So is a ‘true’ civilization one that... possesses the right kind of trappings? Monumental (stone) architecture, road systems; a standing army; etc.?
In that case, we risk defining relevant civilizations based on their overall vibes; ones informed more by videogame design incentives than by in-depth historical research.5 Moreover, the claim that ‘civilizations’ are ‘whichever societies developed technologies X, Y, Z’, just makes any claim that all civilizations converged on X, Y, Z circular. It also doesn’t really constrain the divergence in technological practices, with even the subset of stereotypically ‘high’ civilizations (Roman, Aztec, Egyptian, Chinese, etc.) showing significant (and socially relevant) variation.6
This account is also under-inclusive, given that many societies without prominent (and legible) hierarchies could e.g. carry out large coordinated construction projects, such as monumental Mesoamerican canal-and-dam systems and ‘cosmograms’ that rival or exceed the extent of later cities; sedentary hunter-gatherer’s ability to create major fish-trap wall structures in France; or subsistence-level neolithic Brits’ ability to construct monuments such as Silbury hill; a structure of comparable age and volume as the pyramids of Giza, that demanded a tremendous resource investment, proportionally on the scale of modern megaprojects.7 This is not to suggest that there was no functional differentiation at all; but it does show that societies can adopt a reasonable range of organizational forms and technologies while meeting their own interests and goals.
2. The Ozymandian account: Civilization as enduring monumental art
Is the criterion for a technological civilization... high and legible and enduring artistic achievement?8
That still means we are likely to undercount the achievements of many non-state societies that nonetheless held their corner of the world, producing artistic contributions that should rank high in the proverbial akashic records even if they aren’t as easy to recover for our museums. I was once struck reading this memorable argument by ‘Jane PSmith‘:
We associate human achievement, striving, and greatness with the archaeological remains that testify to them — things like written works and monumental architecture — because often that’s our only evidence that it ever happened. But sometimes, a little clever digging (literal or figurative) can uncover glories of a barbarian past. The most obvious example, of course, is that of the Iliad and the Odyssey, products of a non-state people’s oral culture in the Greek Dark Ages and only recorded with the reintroduction of writing centuries later. How many other texts would be considered classics of world literature if only they had ever become, you know, actual texts? But let’s go beyond art: if you want to talk world-bestriding greatness more broadly, look no further than the ferociously expansive Proto-Indo-Europeans, whose obsession with “imperishable fame” left their DNA all over Eurasia and their culture and even mythology so deeply embedded in their daughter cultures that it can be convincingly reconstructed today. Or the Polynesians, whose expansion is arguably even more impressive given how much harder it is to travel across ocean than steppe. Sure, it’s not the Lion Gate or the Mona Lisa — or even the cuckoo clock — but the remains we do have should remind us of the other cultural achievements that have doubtless been lost like tears in the rain.
In fact, the evidentiary problems and selection effects here are severe, given what Benjamin Breen has called the ‘wooden stupa effect of history‘, named after the gargantuan (400-700ft tall) wooden structures built in the second century CE by the Kushan in modern Peshawar. The towers ‘dwarfed almost anything standing in Rome or Constantinople at the same period’; but, being constructed of wood and topped with metal, they eventually repeatedly burned down as a result of lightning strikes; and when they did so, they burned down entirely.
A hypothetical reconstruction, since, er, we don’t really know what it looked like
Breen concludes that “When we look at what remains and call it “the historical record,” we are seeing the fire-resistant remnants of something immeasurably bigger than we can easily imagine.” The point here is that not all complex or sophisticated or valuable technologies even leave the same material traces in history.9 Yet that does not mean that these were marginal civilizations in their accomplishments or (as we'll discuss shortly) in what they achieved for their people.10
Illegible records do not mean that we should simply down-rate or exclude such societies from a survey of technological possibility. Just because past branches of the tech tree have been hard to recover, does not mean they did not represent (socially, religious, military) marginal technologies in their time.
3. The normative account: Civilization as the city on the hill
Maybe we should define ‘civilization’ instead normatively, as those societies that best manage to fulfill the values, welfare and needs of their people, at least relative to their contemporaries?
You can see echoes of this view of civilization in essays such as Bryan Caplan’s ‘What is Barbarism?‘ (”Maximally civilized societies don’t just scrupulously respect members’ rights to life and property; they afford exactly the same rights to all intelligent beings.”), or Scott Alexander’s ‘In Favour of Niceness, Community and Civilization‘.
I am sympathetic to this normative account; it reminds us to keep thinking about why we actually (should) care about technology: not because of the novelty or complexity of new artefacts per se (that way lies a ‘Disneyland without children‘); but rather because of the beneficial sociotechnical changes that technologies can drive. The ways that they can improve (or degrade) human welfare, and so on. They need to do something for someone.
However, I’m unsure that this approach makes sense when studying the historical dynamics of technology. For one, it’s not such good practice to pack a lot of normative weight into your analytical categories. Moreover, contra Caplan, who holds that there is a correlation between “degree of civilization” and “degree of technological advancement”,11 I don’t think it is the case that more technologically developed societies were reliably more moral even to their own people, let alone their enemies.12 In fact, (as Caplan would grant) that was not the case for pre-modern societies, as many ‘classic’ states imposed severe costs in freedom and quality of life on most of their subjects, as tracked even on identifiable biological dimensions (e.g. stature and height; lifespan). That is even before considering rates of military conquest, or cultural practices such as human sacrifice. That track record isn’t much better in the colonial era.
Yet presumably we don’t want to rule out the technologies of oppressive societies from the study of the historical dynamics of technology.
4. The citadel account: Civilization as the walled garden
Perhaps civilizations are defined by their persistence, resilience, and ability to weather external challenges or shocks better than their peers and (especially) better than ‘non-state’ societies?
However, while throughout early history state-based agrarian societies did tend to outcompete or ‘enclose’ their tribal pastoralist neighbours, there are numerous cases where ‘classic’ civilizations fell against rivals that lacked their tech package (e.g. the Sea Peoples during the Bronze Age Collapse), or where complex supply chains and tech packages in fact exposed states to overreach in ways that made them more exposed to ecological shocks.
Conversely, there are numerous cases where ‘barbarian’ societies without central bureaucracies prospered alongside empires, or as ‘inner barbarians’ during periods of crumbling imperial authority or outright collapse, establishing sheltered compounds or safe havens that may have left less legible traces behind, but which resisted (or outlasted) their imperial neighbours all the same.
To be sure, societal competition (and competitive fitness) can play a large role in the history of technology, as I’ll explore in future writing. But it’s hard to enlist this consistently without introducing a quite significant survivorship bias.13
5. The technocapital take-off account: Civilization as boot-loader for the industrial revolution
Perhaps the technological practices, choices and heritage of premodern civilizations were just the early embers of technology, which should be weighted by their proximity (or their contributions) to the eventual flame of the industrial revolution?
This is the worldview that I think informs a lot of modern thinking on AI; on modern technology more generally; and (by the by) grounds Western societies’ sense of identity and heritage. In this view, there is some deep, qualitative difference between pre-modern and modern societies (the hint is in the ‘pre-’ qualifier), such that the former only really count in discussions of modern technology insofar as they nourished early forms of those technologies.
This distinction is key to a lot of popular writing on history.14 You can hold this view from many different positions. The anti-humanist accelerationist Nick Land opens his Fanged Noumena with the line: ‘The story goes like this: Earth is captured by a technocapital singularity as renaissance rationalization and oceanic navigation lock into commoditization take-off’. More conventionally, the economic historian Robert Heilbroner, in a chapter on ‘Technological Determinism Revisited‘15, pins the distinction on the ‘economic force field’ of capitalist logic that ensures that:
“The impact of precapitalist technical change therefore appears more contingent, less open to systematic elucidation, than when an economic force field guides its applications and consequences. We can say many more things about the “path” of technical change in the United States in the nineteenth century than about its course in ancient China or the Roman empire.”
Ultimately, I think that this is the criterion many people implicitly have in mind. What’s interesting is that this view in some sense incorporates elements of the normative and competitive account: since the industrial revolution unleashed such significant forces for human welfare and societal competitiveness, the real question, on this account, is not over what divergent choices societies managed to make; nor whether or not they were locally functional; but simply whether or which of these technological choices could lead towards lighting the fuse on industrial take-off. Any other technological branches, on this view, may be interesting experiments, but ultimately were dead-end paths or rounding errors in the larger macrohistorical view.
Still, this account is worth a closer look.
Reflections on the take-off view
Of all these views, the technocapital take-off account is perhaps the most interesting. However, it has its limits.
The persistent continuity of technological history
Luke Kemp writes in Goliath’s Curse, in important ways “the modern world is simply an intensification of the past”.
That is, while there are undeniable quantitative breaks in the history of technology (both in particular metrics of technological performance, and in aggregrate societal statistics), we should not overestimate the qualitative differences across history. Premodern agrarian societies were hardly static or lacking in innovation (or ‘proto-industrialization‘). There isn’t a qualitative phase transition in technological history.
For instance, in her recent conversation with Dwarkesh Patel, Ada Palmer observes;
history has always moved fast. But when we teach it in high school, we’re trying to move over large chunks of time quickly, and so we pretend that it moved slowly. We have this lie that there were long periods of stagnation. But you can zoom in anywhere, and you’re going to find every decade feels different, and people in the 1320s are nostalgic for people in the 1300s.
[...] One thing that makes the pace pick up in modern day is simply the population grew and grew and grew and is now much, much larger. The majority of people who ever lived in the entire history, since humans have been humans and not hominids, have lived in the last 200 years because the population became massive.
Moreover, as Peter Perdue argued in his reply to Heilbroner, premodern agrarian societies were hardly static.16 And even when their pace of change was slower, this might actually make them better test-beds for understanding the impacts and dynamics of specific innovations, since it is easier to isolate the effects of particular technologies when their impacts aren’t cut short by the next technology’s arrival.
Which tech tree branches were load-bearing for the industrial revolution?
More to the point of this essay, the technocapital take-off view doesn’t really narrow down whose technological practices to give most weight to, given the amount of disagreement over what were the actual sources or enablers of the industrial revolution. It therefore doesn’t by itself provide clarity over what are the load-bearing (or ‘relevant’) branches of the tech tree.
After all, while, say, palaeolithic non-farming sedentary peoples never created the cultural and material conditions for an industrial revolution ... neither did the vast majority of (farming-based) state societies, including a great number of advanced, powerful civilizations that came remarkably close. As such, our view on which societies had a shot for the industrial revolution, and which set of preceding technology branches were actually necessary to achieve take-off, will depend on our assumptions about what were the actual bottlenecks to (or preconditions for) that industrial revolution.17
For instance, if you think the key factor was:
Cheap and accessible energy from easily accessible coal deposits (or from peat)? --> only mining tech choices matter, along with pre-defined geography;
a sufficiently large or dense population to enable a large enough rate of appearance of talented individuals to make improvements --> this implies that complex non-agricultural sedentary societies might never have had a shot, since aquatic resources alone could not secure a enough surplus;
a minimum package of underlying manufacturing technologies (e.g. cast iron, mechanical linkages, standardization) to facilitate industrialization? --> a broader range of tech branch constraints apply
information dissemination technologies, especially the printing press, to enable the exchange of ideas amongst innovators? --> the tech tree constraints only bite on the printing press and its pre-requisites;
A ‘culture of improvement‘? --> Preceding tech branch choices were less relevant than cultural factors (or they constrained rather than conditioned). However, innovation might be robust so long as the ‘culture of improvement’ remains widespread in the modern world.
various of the above, along with a sufficiently long period without domestic conflict or external military pressure or intervention? --> geographic and exogenous political factors are more determinative. Many societies might have run viable paths across the tech tree, including towards alternate industrial revolutions; except most just got unlucky.
etc.
An inclusion rule for Technological Civilizations
Why does this all matter? In short, I don’t think it is easy to draw a principled distinction that allows us to firewall historical evidence of divergent technological practices in premodern societies as irrelevant just because it came from premodern societies.
Going on vibes (folk account) just falls prey to our existing cultural biases, while looking to grand and enduring artistic agreement (Ozymandian account) smuggles in the idea that only some kinds of art mattered. Defining civilizations by their degree of human welfare or freedom they and their technologies actually enabled (normative account) is appealing but arbitrarily disregards the important technological practices of more oppressive societies, while focusing on societies that were especially resilient (citadel account) doesn’t seem to closely track premodern societies’ attitude to- or investment in technologies. Finally, focusing on ‘the most innovative’ historical societies that laid the foundation for the industrial revolution (technocapital take-off account) doesn’t provide that much clarity over how narrow a scope we should entertain.
So what do we do? Have a free-for-all in terms of the inferential leaps tech historians (or modern tech commentators) can draw from even historically remote cases? I don’t think that’s warranted either. But I do think there should be a presumption to look broadly at the (material, cultural, political, etc.) drivers, enablers and inhibitors of technological innovation across more than the usual suspects.
In an old essay quite unrelated to all this, Scott Alexander referred to a decision rule of ‘positive selection‘ for public intellectuals, where “a single good call rules you in – as opposed to negative selection, where a single bad call rules you out.” I think we could do with a similar broad inclusion rule for premodern societies--with both cases of convergent and divergent technologies weighted, on their own terms, even if they do not hit all of the conventional accounts of technological civilization. Rule technological civilizations in, not out.
Such a rule doesn’t prejudge the way we’ll interpret and weight that historical evidence (that has its own set of hurdles) or the answers we’ll get to. But it does help give us a better, unbiased empirical basis for studying the drivers of technological trajectories in the past and today.
And it is one way to help us better chart the roads to the True Scotland.
Acknowledgements: thanks to Luke Kemp, Jake Slosser, and Janna Tay for discussion.
I’m talking here (of course) about AI. Everyone loves a game of ‘reference class tennis‘: is AI technology like electricity? Social media? Nuclear weapons? i.e. is AI fundamentally a ‘normal’ technology that plays by some set of established historical rules (whatever they are)? Or is it an abnormal technology that doesn’t even have any meaningful historical precedent, and should be studied from first principles? Or is it a mix--a ‘social technology‘ that plays by some rules but also introduces some new problems? Side-sidenote: the whole debate over which-analogy and is-there-an-analogy framings around a technology is a well-known problem in technology law: see Crootof & Ard in one of my favorite papers.
Of course, even under the best circumstances, it can remain tricky to take our cue from history, if only because of how many misconceptions remain rampant in public debates and even scholarship.
I’m here bracketing the important debate over the ways that distinctions between ‘civilized’ and ‘uncivilized’ peoples have been levied around and abused historically. And indeed my point is that it doesn't make much sense to try and decide which societies ranked as ‘(technological) civilizations', and rather we should just explore the technological practices of a wide set of civilizations (qua complex societies), before evaluating which of these tell us more or less about the shape and texture of the full tech tree.
See again; Étienne Fortier-Dubois: “The tech tree can be traced to the 1980 board game Civilization, widely popularized by game designer Sid Meier’s video game version in 1991. The tech tree in Civilization functions as a branching pathway of scientific discovery that unlocks new units, buildings, and abilities — from primitive tools to space-age technologies. The story goes that Meier threw the tree together on only cursory research, with the intention to eventually come back and make the “real tech tree” later. But this initial version was so fun that it stuck.”
It also doesn’t really work for many other less-well-known civilizations, such as the Indus Valley/Harappan civilization, which combine many of the trimmings of civilization (cities, agriculture, script, trade, monuments) with a radically different, more egalitarian social structure than the earliest states they were concurrent with.
For instance, of Silbury Hill, Atkinson 1974 wrote that “in view of the small size of the neolithic population, this [structure] represents a fraction of the ‘gross national product’ at least as great as that currently devoted by the United States of America to the whole of its space programme.” (H/T Joseph Levine).
Speaking of enduring artistic achievement (OP unfortunately and appropriately lost to time):
This is similar to a point in Ursula Le Guin’s famous essay on the ‘Carrier bag theory of fiction‘: that when we trace the earliest inventions of humanity to flint sticks, we neglect the fact that contemporary or earlier inventions of containers (e.g. baskets) may have been as decisive, yet are far more fledgling in the archeological record.
For another case, take the Pre-Contact civilizations of central Amazonia; as Isaac Wilks memorably chronicles,
Only one group of men from the Old World ever saw the civilizations of central Amazonia in their flower. This was the expedition of the Spaniard Francisco de Orellana in 1540-1. […] Never before, and never again, will the recollection of an entire civilization fall to the pen of one man. Gaspar de Carvajal, a priest from those rugged marches of Extremadura which rendered up so many conquistadors, had found his way to Ecuador and thence to Orellana’s brigantine; to his account, Relación, we owe our eternal thanks. What he saw was so fantastic that it was immediately discounted, and thereafter forgotten: Cities of wood which stretched for over a hundred miles, flotillas of thousands of armed marines, floating musical orchestras, undulating flash mobs covering the hills, even glazed pottery and naturalistic painting equal to that of Europe—all gone a century later, when Europeans next sailed the mighty length of the river. Old World disease had claimed as many as 95 percent of the basin’s people, in an apocalypse that went all but unrecorded. […] Yet the past forty years of archaeological work have revealed that Carvajal was not hallucinating. Long considered to be a “counterfeit paradise”—a lush but sparsely populated waste—the massive region was, in fact, home to a series of complex, dynamic civilizations. Painstaking pedological and botanical surveys have revealed, for example, that vast swathes of the Amazon Rainforest—at least an area the size of France—are actually a project of ancient, intentional human cultivation: the world’s largest garden. Extensive road networks, some nearly 150 feet wide, once crisscrossed the basin. In some regions, people built wooden bridges that were at least a half-mile long.
On the theory that “”Societies that respect the rights of members and outsiders alike encourage work, investment, and trade, all of which foster technological progress. Societies that disrespect the rights of members and outsiders, in contrast, discourage work, investment, and trade, all of which foster technological stagnation.”
To be sure--I agree that technological supremacy can in some cases provide (theoretical) affordances for more ‘moral’ societies to continue to act in alignment with their moral codes, especially at war. However, it feels like these are downstream from other factors that determine and constitute their morality.
Obligatory:
For instance, history blogger WSCFriedman clarifies that: “*I will try to include the disclaimers in advance, and, instead of saying “all societies,” say “agricultural societies between the invention of history and the Industrial Revolution.” But when I say “all societies”, or “everywhere,” or “the default society,” I do, yes, mean agricultural societies with extant towns before the Industrial Revolution or else in areas the Industrial Revolution hasn’t reached.”
(pg 170) “I try to connect the analysis of agrarian society with the concepts used by analysts of industrial societies. If we apply the same types of argument to both, we undermine the all-too-common assumption that real” technological change began with the Industrial Revolution and that subsequent developments followed a logic uniquely dictated by industrialism. Such an assumption is cast into doubt by recent studies of the economic history of the Industrial Revolution which explore continuities between the pre-industrial and the post-industrial age. These studies underline the importance of “proto-industrialization” and “growth recurring” (Mokyr 1985; Jones 1988). “Take-offs” are not so obvious as we once thought.” (emphasis added)
In a future essay I may explore these debates over the causes, preconditions and limits of the industrial revolution in much more depth.