Back to page

📄 posts/
#lang pollen

◊(define-meta title "The Medieval-esque Internet")
◊(define-meta published "2021-09-07")
◊(define-meta topics "Society, Tech, Themelio")

◊section{What is a good decentralized society?}
In the previous blogpost, I argued that developing decentralized technology only makes sense under the aim to decentralize society as a whole. But what does a good decentralized society look like? To answer this question, let us look at a time and place with remarkable parallels to a decentralized Internet --- Western Europe during the Middle Ages. 

The phrase "Middle Ages", describing the period of European history stretching from approximately A.D. 500 to 1500, is strangely undescriptive. It tell us absolutely nothing about what that thousand-year period of time was like --- only that it's sandwiched between the fall of Rome and the rise of the "early modern" period. This in part reflects a largely unwarranted anti-medieval streak in European historiography. The Middle Ages were considered at best an uninteresting period of stagnation between the loss of Roman civilization to "barbarian" invaders and its revival in the Renaissance◊sidenote["JF"]{Which the same historiographical current often paints in an ◊link[""]{unreasonably positive light}!}. At worst, they were seen as the "Dark Ages" of misery, backwardness,  poverty, and superstition. 

In the scholarly world, no one takes the "Dark Ages" narrative seriously anymore, but many outlandish (and easily debunked) popular myths◊sidenote["TM"]{A great book that demolishes the pop-culture slander of Medieval Europe is ◊em{
Those Terrible Middle Ages: Debunking the Myths} by Régine Pernoud} still linger. We hear of peasants leading disgusting lives wearing rags and never bathing, knights clad in armor so heavy they needed cranes to mount their horses, and nobles using expensive, exotic spices to cover up rotten food. None of this, of course, ever happened.

◊figure["knight.jpg"]{Mobility in full plate armor (◊link[""]{Source})}

Yet in a way, it's not surprising that the Middle Ages are a widely misunderstood "gap" in European history. It was indeed a time where Europe ran on principles fundamentally alien to those of the preceding Roman and following (early-)modern period, which in many aspects were more similar to each other than to the medieval period. In particular, medieval Europe exhibited a rare combination of ◊em{decentralization and interconnection} that powered its success. This combination, I will argue, has been and will be crucial in the health of the Internet.

◊section{Medieval Europe: an ordered anarchy}

Unlike the preceding age of the centralized, international Roman Empire and the following era of centralized nation-states, medieval Europe was radically decentralized. The Western Roman Empire did not find itself replaced by another continent-spanning empire, but by a lack of empire altogether. Despite a few ultimately unsuccessful attempts at Roman-style empire-building, most famously by Charlemagne, Western Europe fractured into innumerable tiny statelets. According to Eric Jones' well-known ◊em{The European Miracle}, fourteenth-century Europe had a thousand independent polities. Even within these small polities, authority was highly decentralized and informal. Governance was reduced to essentially personal, non-transitive vassal-lord relationships --- the common phrase "the vassal of my vassal is not my vassal" referred to the fact that even a powerful king only had direct authority over those who personally swore fealty to him (say, dukes or counts), but not anyone else (say, commoners working for some subordinate duke). No apparatus for centrally governing large territories existed.

◊figure["Mitteleuropa_zur_Zeit_der_Staufer.png"]{Holy Roman Empire, circa 1200 (◊link[""]{Wikimedia Foundation})}

Somewhat ironically, the "Holy Roman Empire", pictured above, is now virtually a poster-child for medieval decentralization. As the legacy of Charlemagne's conquests, the HRE started out as a nominal successor to the Western Roman Empire. However, the "empire" quickly became a loose alliance of hundreds of ◊em{de facto} independent polities, and the Emperor turned into a weak figurehead elected by the most powerful princes of the realm. Due mostly to the detailed records kept of its internal political organization, we now get to draw a kaleidoscopic map of constituent states that strikingly visualizes the extraordinarily fragmented medieval political order.

Maps of medieval Europe, like that of the HRE above well illustrates, are shaped like Swiss cheese. Such geopolitical landscapes immediately call to mind times and places of destructive chaos, such as Warring States China or the Syrian Civil War. Lacking central authority, such places are usually plagued with violent fragmentation and disruption of cultural and economic interconnectivity. Conversely, periods of high interconnectivity and international peace, from the ◊em{pax Romana} to the ◊em{pax Americana}, are generally associated with the dominance of a centralized power.

But medieval Europe was neither of these two usual cases. Even during the early part of the Middle Ages, when Europe was still recovering from the fallout of the Germanic migrations and the end of the Roman Empire, complex, decentralized trade networks exceeded the previous Roman trade in volume. Contrary to a widespread misconception, medieval trade was not limited to exotic luxuries that most people could never afford. Richard Hodges in his ◊em{Dark Age Economics} documents a vibrant international economy of basic commodities and products during the early medieval period: for example, common pottery from Anglo-Saxon England was sold all across the European continent.

◊figure["english-trade.png"]{Carolingian pottery trade (Richard Hodges, ◊em{Dark Age Economics, 93)}}

International travel, though certainly lacking the convenience of modern vehicles, was quite common and fairly accessible, at least to the well-to-do. Notably, the multitude of jurisdictional borders posed little impediment to free movement --- modern borders with passports, visas, and immigration controls were centuries away! The famous thirteenth-century theologian and philosopher Thomas Aquinas is said to have frequently traveled between "Italy", "Germany", and "France" in today's books. In reality, he would have been traversing dozens of principalities, duchies, republics, and free cities on every journey.

This unique combination of decentralization and interconnection produced an environment that fostered an exceptional amount of cultural, economic, and technological advances. These advances eventually propelled Europe far ahead of the rest of the world and set the stage for momentous later developments, such as the Industrial Revolution. Free competition between independent jurisdictions and property owners --- a situation only possible in a world with both decentralized power and large-scale interconnectivity ---  spurred great increases in economic efficiency. The 2017 article ◊link[""]{"Mills, cranes, and the great divergence"} in the ◊em{Economic History Review} examines how independent feudal lords competing with each other for peasants and revenue heavily invested in labor-saving devices like water mills, greatly advancing agricultural technology and boosting economic growth. In the same time period, these inventions were available but neglected in the Islamic world, which was under the quasi-feudal, yet much more centralized ◊link[""]{iqṭāʿ} system. In the Islamic system, the closest equivalent to a Western manorial lord was a state-appointed farm manager (muqṭaʿ) with no title to the land, meaning no incentive to improve long-term efficiency. 

In the area of finance, secure property rights generated by pervasive jurisdictional competition powered sophisticated international financial arrangements. Banking and insurance contracts, for instance, were commonplace by the 15th century. Banks from the northern Italian merchant republics opened branches all across Western Europe. The Middle Ages sowed the most important seeds of Western capitalism: free markets, formalized finance, and international trade. During the same period, in the centralized Ottoman Empire, no such commercial revolution took place. Instead, property rights were highly insecure, and investment markets were close to nonexistent. Even in the late 16th century, Ottoman markets were so undeveloped that even the ruling elite distrusted investments to such a degree that they customarily lugged around large hoards of cash. For example, after the Battle of Lepanto in 1571, the defeated Turkish admiral Ali Pasha's massive stash of gold coins was found hidden in his flagship. "Without a system of banking, fearful of confiscation should he displease the sultan, and always careful to keep his assets hidden from the tax collectors, Ali Pasha toted his huge personal fortune to Lepanto"◊sidenote{Victor Davis Hanson, ◊em{Carnage and Culture: Landmark Battles in the Rise to Western Power} (Anchor, 2007)}.

◊; examples from the arts and humanities? how would that fit in?

◊section{The medieval-esque Internet}

Perhaps surprisingly, the same combination of decentralization and interconnection found in 12th-century Europe is also found to a great extent in 21st-century cyberspace. In the highly centralized, state-based modern social order, the Internet is a polycentric anomaly. Largely voluntary governing bodies like ICANN notwithstanding, no global authority centrally plans Internet infrastructure or standards. These nearly all organically grow from innovations by participants. The movers and shakers of the Internet world are generally private, non-state actors --- content providers like Google, Facebook, and countless other website owners, global backbone ISPs like NTT and Telia, submarine cable companies like Subcom, etc.

◊figure["bgpmap.jpg"]{Visualization of BGP autonomous systems on the Internet (◊link[""]{OPTE Project})}

Decentralization on the Internet extends even to governance and its enforcement --- Internet "law" operates on an informal system largely independent of state-proclaimed laws. The rules of the Internet are in fact often quite different from those of state-based legislation. For instance, legitimate multinational ISPs routinely carry traffic for networks harboring notorious, illegal piracy websites like The Pirate Bay, yet a network that tolerates email-spammer customers will quickly be blacklisted by most ISPs and ◊em{de facto} kicked off the internet, despite engaging in something that's in most jurisdictions no more illegal than sending physical junk mail. 

This high degree of decentralization is matched by equally fluid interconnection. Despite the fragmentation of governance, crossing jurisdiction boundaries is nearly completely frictionless on the Internet, allowing us to speak of a coherent online "cyberspace". An average netizen living in Canada takes it for granted that they can follow local news hosted on a Canadian server, post a Facebook status to a server in the US, check their ProtonMail inbox hosted in Switzerland, and finish their day by downloading some pirated songs from Russia. 

The Internet's decentralization and interconnection are made even more remarkable by the fact that essentially all other modern communications networks are heavily controlled by the centralized state. Telephone networks, for example, are nearly everywhere◊sidenote["som"]{An interesting counterexample is found in Somalia, where the lack of an effective central government has resulted in an ◊link[""]{unregulated and uniquely vibrant telecom market.}} regulated as public utilities, with the most important components (such as phone-number allocation) under direct state control. Development of new protocols is also highly centralized whenever non-Internet telecom is involved: witness the highly political, state-driven, and top-down process of 5G standardization and implementation. Compare it to how new Internet protocols like HTTP/3 evolve. On top of all of this centralization, most other networks also suffer barriers to interconnection, including border controls, customs forms, and international dialing codes that are simply absent on the Internet.

And just like that of medieval Europe, the Internet's crucial combination of decentralization and interconnection spurs astonishing innovation and growth. Without a centralized single point of failure that could force stifling regulations, Internet "law" quickly evolved rules that promoted growth, such as punishments focused narrowly on activities that impose large, widely-distributed negative externalities (like spamming and distributed denial-of-service attacks). Free experimentation by network participants of different protocols, applications, and business models led to an exuberantly rapid expansion of novel use cases. Meanwhile, innovation in networks still using the usual centralized model continued to stagnate. Even ideas like LTE video calls that merely play catch-up with Internet-based solutions never really got anywhere.

Again, competition between different cyberspace jurisdictions (including both traditional states and authorities that enforce "Internet law") is key. Operators can "jurisdiction-shop" for the most favorable place to host an Internet service while fully retaining the ability to reach a global audience. This competition creates pressure to formulate and enforce good rules that protect liberty, even in highly restrictive jurisdictions. For instance, despite the overwhelming amount of censorship and surveillance, restrictions on Internet use in China is still much less oppressive than those surrounding traditional print publications◊sidenote["lel"]{For example, international traffic is by default allowed, though with a large and expanding fraction of traffic filtered by the "Great Firewall". Print publications, on the other hand, are subject to a strict whitelist where only publications explicitly approved are allowed.}.

Furthermore, interconnection allows victims of perceived injustice to easily leave a jurisdiction, the same way that medieval cities swore fealty to foreign monarchs when under threat by local lords. Dissident Chinese websites like ◊link[""]{Pincong} find refuge in foreign cyberspace, while privacy-conscious American netizens wary of local spying run encrypted VPN tunnels to Europe. The controversial American social media platform Parler is a fine example of escaping from a ◊em{private} cyberspace jurisdiction. After being kicked off of mainstream American providers like Amazon Web Services due to allegations of fanning political violence, Parler quite seamlessly switched to Russian infrastructure and continues to serve its mostly American user base.

Despite its decentralization and interconnection, however, the current social order of the Internet does not seem poised to last a millennium and define a historical era. After mere decades of decentralization, the Internet is under attack on multiple fronts by powerful centralizing forces. Conventional state-based regulation is on the move: behold the proliferation of both heavy-handed censorship schemes like the Great Firewall and seemingly benign but no less centralized regulatory frameworks like the European Union's ◊link[""]{GDPR}. New centralized centers of social power in "cyberspace", such as social media giants, are also taking over the governance of our online society --- often in concert with the coercive state. If this trend continues, it appears inevitable that the Internet will soon be integrated into the centralized social order of our day, just like the other telecom networks.

It is clear, then, that the oasis of decentralization and interconnection found in the Internet is fragile and does not come for free. In order to defend it, we must understand not only why decentralization and interconnection are good, but also how this happy combination can be robustly upheld. In the next blogpost, we will again return to the Middle Ages to discover and learn from what stabilized the decentralized medieval world.

◊; ◊section{How do we get there?}

◊; To understand the prerequisites of a stable decentralized and interconnected society, we again return to medieval Western Europe, looking for clues as to what held that world together. There is a large body of research on the factors that upheld the medieval polycentric order, but in this blogpost I will focus on two factors that both important and especially relevant for the Internet --- one social, and one technical.

◊; ◊subsection{Means-restricting norms}

◊; An important property of medieval Western Europe was the existence of very strong social norms about the legitimate use of political power. Despite a lack of a central rule-enforcing authority, a nearly universal consensus on the permissible ◊em{means} that can be utilized to achieve political ends emerged based on a synthesis of Germanic tribal customs, Christian philosophy, and vestiges of Roman law. The interconnectedness of Europe at that time was no doubt also crucial to the formation of this body of what we would now call "international customary law".

◊; Just from looking at a map, we can see one of these means-restricting norms in action --- the inviolability of contracts, especially contracts pertaining to property and governance rights in land◊sidenote["feud"]{It's perhaps somewhat imprecise to speak of the oaths of reciprocal fealty that governed feudal relationships as contracts in the modern legal sense. But they were certainly mutual, voluntary exchanges of goods and services like agricultural produce and defense.}. Since these contracts were the basis of what we'd now call "political boundaries", this meant that jurisdictional borders within Western Europe were unusually stable. This was the case even with Swiss-cheese-shaped borders that are completely indefensible, which centuries of sales, purchases, marriages, and inheritances produced in great quantity. The strong social norms against arbitrary aggression and seizure of property generally safeguarded a medieval prince's cosmopolitan holdings even when traveling to them would require crossing the territory of many rival powers.

◊; Even in the case of France, whose ambitious Capetian dynasty gradually centralized governance to establish an absolutist monarchy by the end of the Middle Ages, this process took centuries and was conducted mostly through peaceful means (inheritances, purchases, etc). Vast, Napoleon-style conquests, even for some noble end (like unifying the kingdom), were out of the question. A king that attempted such an endeavor would quickly lose legitimacy and find himself deposed by either his subjects, or some foreign claimant to the throne.

◊; This brings up another, much more central means-restricting norm: the ◊em{lack of sovereign legislative power}. Unlike in modern times, "legislation" was not considered a core role of government. Political leaders, even the most powerful kings and emperors, could not simply define through fiat "the law of the land". Instead, their role was simply to lead the realm in times of crises and adjudicate legal cases, all in accordance with the customary law formed by tradition and legal precedent. In his seminal work ◊link[""]{◊em{Kingship and Law in the Middle Ages}}, medievalist Fritz Kern summarizes these legal order as the supremacy of "old and good" law:

◊; ◊blockquote{◊p{For us law needs only one attribute in order to give it validity; it must, directly or indirectly, be sanctioned by the State. But in the Middle Ages, different attributes altogether were essential; mediaeval law must be "old" law and must be "good" law. Mediaeval law could dispense with the sanction of the State, but not with the two qualities of Age and Goodness, which, as we shall see, were considered to be one and the same thing. If law were not old and good law, it was not law at all, even though it were formally enacted by the State.
◊; }
◊; ◊footer{Fritz Kern, ◊em{Kingship and Law in the Middle Ages} (Oxford, 1948), 149}
◊; }

◊; According to Kern, even legal innovation was framed in terms of restoring customary law to an older and better standard:

◊; ◊blockquote{◊p{For even if the Middle Ages actually made new law every day, mediaeval ideas forced them to assert that the reasonable, equitable law is also the old law. [...] Right and law are restored as they had been in the good old days of King Eric (in Sweden), of Edward the Confessor (in Anglo-Norman England), of Charles the Great (among the French and Germans) or of some mythical law-giver.
◊; }
◊; ◊footer{Fritz Kern, ◊em{Kingship and Law in the Middle Ages} (Oxford, 1948), 165-166}
◊; }

◊; Legislation --- the centralized, discretionary creation of new laws --- was thus out of the question as a legitimate means to achieve political ends. Even taxation, often taken for granted in modern times as an unquestioned right of the state, was governed strictly by customary law: "mediaeval taxation is a 'gracious aid', and the State or the monarch has an unquestionable right to a tax only when it has become traditional" (186). So strong was this limitation on the means of financing government that taxes were very difficult to raise even in times of great crisis, such as during wars. Even in late 15th century England --- at the very end of the Middle Ages in one of the most centralized kingdoms of Western Europe --- attempts to raise taxes to fund a war with Scotland sparked the massive Cornish tax rebellion of 1497.

◊; ◊; insert famous example of English king begging local communities to contribute more taxes for a major war
◊; ◊; insert HRE vs papacy example as illustration of how an attempt to co-opt a source of law led to a huge loss in social legitimacy

◊; ◊subsection{Defense/offense asymmetry}

◊; An important technological factor that stabilized this complex, decentralized world and its social norms is the ◊em{dominance of defensive over offensive military technology}. Especially during the "High" Middle Ages of 1000-1250, defensive systems such as castles, city fortifications, armor, and anti-cavalry infantry formations improved at a tremendous rate that greatly outpaced innovation in offensive weapons, so that "by 1300 the defensive obtained an almost complete mastery over the offensive" (CITE). (MORE EXAMPLES)

◊; Dominance of defense over offense has a twofold decentralizing effect. Firstly, it simply makes it easier to defend the existing social order, while increasing the costs to ambitious aggressors wishing to impose their own order. A king who attempts to forcibly centralize power at home or conquer vast empires abroad will quickly discover that it's much easier for his enemies to defend themselves than for his army to defeat them. The pervasive idea in medieval customary law that vassals had the right to defend themselves against a tyrannical lord, even by force, was no mere social custom, but backed by actual military capabilities.

◊; More importantly, a technological environment favoring defense localizes security. When offense is stronger than defense, the easiest way to achieve peace is through a centralized empire --- a monopoly on aggression that conquers all its challengers to achieve a Pax Romana or Pax Mongolica. In contrast, small local communities in the Middle Ages can achieve a reasonably peaceful existence by defending against rather than conquering their foes. Instead of relying on a "sovereign" state to ensure security, everyone took responsibility for their own security, now in economic terms a private rather than public good. In fact, throughout the Middle Ages there was quite a vibrant market for military defense. Communities often had many choices of mercenary bands to hire, liege lords to swear fealty to, etc, and this competitive market in large part drove continual innovation in defensive technology. 

◊; This is in stark contrast to the sudden reversal in the defense/offense balance brought on by cannons etc. 

◊; ◊subsection{Weak pillars in cyberspace}

◊; Unfortunately, Internet cyberspace has worrying deficiencies in both means-restricting social norms and defense/offense asymmetry. As we will see, this is directly connected to the precarious position of the decentralization and interconnection of our online social order.

◊; In the first area, "open Internet" advocacy has a tendency, like much social activism, to focus on immediate socially desirable ◊em{ends} like privacy, free expression, and open access. There is little attention directed towards limits on legitimate means to achieve these ends. Indeed, open-Internet advocates have occasionally championed means inimical to the decentralization and interconnection so crucial to the social ends they pursue. ◊sidenote["GDPR"]{A typical example would be "privacy" legislation like GDPR that enforces jurisdictional boundaries and empowers centralized regulators much more than it improves privacy}

◊; ◊; A little vague to talk about means-restricting norms, so here's a real-life example: "communication security should be the exclusive concern of endpoints". This led secure communication on the Internet in a radically different direction than on legacy telecom networks, focusing on encryption, authentication, etc rather than regulation of transit infrastructure. Importantly, it socially delegitimizes restrictive network regulation as a means to the end of securing communication, just like property rights delegitimizes conquest and expropriation as a means to the end of national defense.

◊; ◊; Find source on security in SS7 and similar legacy protocols

◊; ◊; Better 

◊; ◊; Find source on distribution of software vulnerabilities

◊; ◊; Map of France in 1300. Map of block-level swiss cheese borders.