John Carmack recently Tweeted, “I remain easily optimistic in the face of everything happening. Consider the most amazing person you personally know, by any quality metric you choose. Odds are that there are literally millions of their caliber in the world, which is plenty to build a bright future.” (Double spaces sic.) Let’s be those high-caliber people: to deal with the many unfolding crises we’ve been handed, how can we have agency as local beings in a global system? Local-global shows up in many phenomena, such as the Us/Them distinction in social group dynamics, the topology of social networks, geopolitical hierarchies, biology, and so on.Us/Them: “We” are local, “They” are a separate locality, and together the localities compose a whole with global properties.
Social networks: My Facebook friends are my locality, your Facebook friends are your locality. One globality on Facebook is the set of all users residing in the United States. Without doubt, there are more ways to draw both “local” and “global” boundaries.
Geopolitics: Cities are local to states, states are local to countries, and countries are local to international geopolitics. Global relations hold in reverse.
Biology: Organelles are local to cells, cells are local to organs, and organs are local to organisms. Global relations hold in reverse.
Local-global is a basic property of the systems we are made of and the systems we partake in. There is, however, a problem! Until recently in evolutionary history, we did not interact with many people other than our hunter-gatherer group, so humans evolved for local contexts. For millions of years and many iterations of evolution, this worked swimmingly (so to speak). Internalizing local-global pays dividends specifically to that end. In this essay, I want to introduce you to the idea well enough for it to stew. First, I’ll define local and global. Then, I’ll explore how humans have evolved for the local but not the global, and finally and most extensively I’ll demonstrate the utility of local-global by analyzing two interacting local-global systems of intense relevance right now: the pandemic and social media.The Idea
Definitions
As a flexible concept, local-global can be defined in more than one way. The definitions I’ve chosen for this discussion are a) inspired by networks/graphs, b) not formal, c) not going to cover every quirky edge case, of which there are many, and therefore d) meant to be semi-specific heuristics rather than definitive definitions, because e) a definitive definition of local-global doesn’t exist due to its flexibility. I would argue that is a strength! Local: Regarding a network, "local" pertains to those nodes or collections of nodes which are "close" together, where we get to choose a definition of close. There are at least two kinds of closeness, connection closeness and coloring (or property) closeness. Connection closeness defines closeness in terms of connections in a network, whereas coloring (or property) closeness defines closeness in terms of nodes or collections of nodes that all satisfy some property—pictorially, all colored in some way. (I’ll use the phrase “coloring closeness” in this essay because I think the visual component helps its understandability, even though “property” is substantially more general.) Connection closeness can be defined relative to a reference node, or more nuanced techniques from graph theory can be used that define localities based on connections, but don’t output a result relative to any one node in particular; an example is a “component” in the graph theoretical sense. In an example setup, we can look at my Facebook locality. The nodes are Facebook users, and the connection closeness is defined as “is Facebook friends with [me].” My locality, then, is the set of all people who I’m friends with on Facebook.
Here, I have two Facebook friends, and therefore the three of us combined are my locality (I’ve decided to include myself).
Importantly, we can generalize the idea of locality to apply to collections of nodes instead of individual nodes. We could talk about the locality of families who have children in 2nd grade at an elementary school (coloring closeness, via “has a 2nd grader at elementary school X”).
We could describe the locality of all employees of a company (coloring closeness, via “employed by company Y”)
Another example is the locality of all donors who donated to a given non-profit (coloring closeness, via “donates to nonprofit Z”), and so on.
The most reduced localities are individual nodes. We can think of “generalized localities” as collections of those nodes which can themselves be thought of as localities relative to other collections—this can be done either by having connections between collections of nodes (c.f. hypergraphs), or by having properties defined on collections of nodes rather than individual nodes. Each student in elementary school A is their own locality because they are an individual node, but the group of all students is a (color-connected) locality relative to elementary schools B and C.
The “closeness” measurement in this generalized locality is “goes to school with.”
Global: There are two kinds of global phenomena, descriptive ones and causative ones. A descriptive global phenomenon is a state that occurs across many localities, and arises when many nodes could be colored the same way (i.e., share a value assigned to a property, particularly values that can and/or will change) or when many nodes colored the same way are connected together.
A causative global phenomenon is something—anything—that causes state changes in many localities.
“Many” in “many localities” is loosely defined, and its instantiations in specific scenarios depend on what kind of locality you’re describing, and what closeness metric you’re using. If only four humans out of 7.8B are presently experiencing a state of enlightenment (“present” pun intended)—one wouldn’t say enlightenment is a descriptive global phenomenon. However, if 690M people in the world are undernourished, that’s a distressing descriptive global phenomenon. How many people would need to be enlightened for it to be a descriptive global phenomenon? There is no single principled way to say; in that sense, globality is vague (as a term of art in philosophy). Note that the set of localities from which we pick a subset big enough to be called “many” can be either the entire network (like all of Facebook) or a sub-network (like one Facebook group). Similarly, we can also choose to describe globality relative to fully-reduced node localities (like individual Facebook users), or generalized localities (like Facebook groups). Making a crucial insight explicit, something that is local in one setting can be global in another. Technically, given the above definitions, the coloring that describes one set of nodes or generalized localities as a locality can similarly be used to define a descriptive global phenomenon. The easy way to think about this insight is through an example; Facebook works well. A left-leaning Facebook group has a natural locality—its members are part of the locality—but it can also itself be a local member of the set of all left-leaning Facebook groups. The set of all left-leaning Facebook groups is a descriptive global phenomenon, based on the property of being left-leaning and being on Facebook.Evolution
It's clear that we evolved for the local and not for the global. As an insightful Aeon essay from February of this year puts it, "Global-scale interaction clashes with human biology. We evolved as members of small tribes, in hunter-gatherer societies of fewer than a thousand people. Our methods of communicating and self-organising into social groups evolved for interactions on this scale. We smile and respond to smiles, and we are experts at reading each other’s facial expressions. We follow social norms and, when others don’t follow those norms, we exert pressure on them to conform by being morally outraged at their transgressions. We are motivated to show our moral outrage because it signals our own virtue to the tribe, which raises our value in it. These mechanisms are beneficial for tribal social cohesion, and encourage cooperation in a small tribe." Robert Sapolsky's Behave talks at length about humans’ penchant for thinking about the world in terms of “Us” and “Them.” He writes, “the strength of Us/Them-ing is shown by: (a) the speed and minimal sensory stimuli required for the brain to process group differences; (b) the unconscious automaticity of such processes; (c) its presence in other primates and very young humans; and (d) the tendency to group according to arbitrary differences, and to then imbue those markers with power” (p.392, Kindle edition). The “Us” refers to my locality; the “Them” is out there in the global arena. Sapolsky provides evidence to support a basic intuition, namely that we perceive ourselves to be part of many such Us’s. “Crucially, which Us is most important to me constantly shifts—if some octopus moved in next door, I would feel hostile superiority because I have a spine and it didn’t, but that animosity might melt into a sense of kinship when I discovered that the octopus, like me, loved playing Twister as a kid” (p.405, Kindle edition). We have faculties for quickly assessing the localities to which we belong, but not for descriptive or causative global phenomena—states of large systems, and the causes that influence them.Applying The Idea
Let’s take a look at two interacting local-global systems, the pandemic and social media.Pandemic
The COVID-19 pandemic is a global phenomenon, a bona fide clusterfuck. It spread through globalized travel (the descriptive global phenomenon of having connections between many localities), and has caused chaos for economies and politics around the world (hence, a causative global phenomenon). The basic "local" units in this context are individual people, and people are part of families and organizations; organizations operate in accordance with laws established by the government; each city operates within the context of a state, each state within the nation, each nation within the international geopolitical-economic milieu. In other words, every element of the hierarchy is local with respect to some global, and every element (except the individual, in this analysis) is global with respect to some local. All the way up and all the way down, the pandemic is local-meets-global. To intervene with the biological disease itself we must intervene locally at the level of individual people, such as by wearing masks, maintaining physical distance, air ventilation, vaccination, participation in contact tracing, administering treatment, and generally responsible decision-making (if I had three wishes, I'd use all three on responsible decision-making...). Of course, to design, produce, and distribute vaccines, we need organizations with causative globality to coordinate with one another (organizations like pharmaceutical companies and Congress are generalized localities with causative globality over lower localities; the state of coordination between pharma, government, and all of healthcare is a descriptive globality).
For the economy, we have causative global interventions. Take interest rates, for example: a causative global intervention is necessary because of the structure of the system—only the Federal Reserve can set interest rates. Another causative global intervention is the passing of a stimulus package by Congress (well, hypothetically). Trillions of dollars distributed into the American economy impacts many localities, and therefore it is a global causative phenomenon. A notable difference between the disease and the economy, considered in local-global terms, is that certain disease interventions like mask-wearing are ultimately up to the base locality, individual people, whereas economic interventions (and the state of the economy in general) are out of any individual’s control. Unfortunately, people cannot simply will money into existence; a global causative intervention is required. All of these interventions rest on people, who as we established above, do not count global decision-making as a strength because we evolved for small groups. In a situation not unlike the Tragedy of the Commons, our local decision-making leads to local choices—”let’s have a small dinner party, we haven’t seen our friends in months”—that may be optimal for this locality in the present, but may turn out suboptimal for this and other localities in the future when one of the dinner attendees is unwittingly infectious. Compounded across a country, this promotes the descriptive global phenomena of more disease and a slower economic recovery. That so many people make local decisions like this is just a fact of our evolution: we care deeply about the people immediately around us, and less about people far away. We make decisions weighted more by present reward than future reward. We simply struggle to apprehend how small local choices unfold in complex systems with many interacting localities and globalities: a small decision to attend a dinner party might ultimately spiral into 1000s of infections and likely some deaths.Social Media
Social media is the theatre of our times: on these platforms we watch our comedies and tragedies unfold. Structurally, social media platforms are networks ("social networks"), so they lend themselves well to local-global analysis. In social networks, local users see information shared by other local users, either about their own lives (local updates) or about something bigger (descriptive global updates). A "global" media organization like the New York Times disseminates global information down to the local users (an act of causative globality), who then share this information to other local users. Above, we covered the nested structure of social networks; as with the pandemic, all the way up and all the way down, social media is local-meets-global. An interesting property of systems whose nodes are cognitive—humans, non-human primates, birds, and so on—is that the nodes can directly perceive descriptive global variables of the system as expressed in a condensed way. Social media is of course one such system. For example, a descriptive globality in America right now is the presence of intense and conflicting political belief. This descriptive globality can be summarized by the title of a New York Times article, such as this one from the halcyon days of 2014. The title, “Polarization is Dividing American Society, Not Just Politics,” can be displayed on a Facebook news feed. The users of Facebook, nodes of the system, see the article’s titles, and therefore receive information about a (summarized) descriptive global state of the system. In this manner, the descriptive global state becomes a causative global state and influences the system, since individual people’s behavior might change due to their ability to observe this descriptive global characteristic. This leads to the formation of tight local-global interaction cycles in social media because most social media now aggregates global descriptions and feeds them to users. In essence, participating in social media is the process of 1) creating content that often contains your belief about a global description, 2) sharing that content with other people and thereby being a local influence on them, and 3) the lion’s share of social media interactions, looking at the content other people have created, which often captures global descriptions. Of course, this is muddied by the fact that content rarely, if ever, gives “accurate” or “good faith” summaries of descriptive global phenomena, and all sorts of cognitive biases will lead the recipient to acquire that summarized information imperfectly and mix it in with their own views, thus creating localities everywhere with distorted perceptions of the global phenomena. No wonder social media is messy. And then there was a pandemic!Social Pandemedia
The pandemic and social media, two local-global systems, interact with one another. Social media provides the information infrastructure upon which knowledge about the pandemic is shared (knowledge about a descriptive global phenomenon like case counts, or knowledge about a local phenomenon like an individual's battle with illness). Social media is also the infrastructure for spreading beliefs and behaviors. Beliefs can represent local perception of a descriptive and/or causative global phenomenon, and behaviors can represent how the localities react to descriptive and/or causative global phenomena, given their beliefs. (Not to over-apply, one can certainly believe things and exercise behaviors that have nothing to do with globalities.) It’s interesting to note how large a portion of people’s beliefs and behaviors these days, as observed and enacted on social media, do pertain to the global. Though I have no evidence per se, the truism that focusing on the present leads to calm probably holds for the social media scenario as well: the more the social network’s activity pertains to local beliefs/behaviors about the global, the more tumultuous the social network activity will be. If we regard accurate summaries of descriptive and/or causative global phenomena as (compressed) truth, then ideally truth should inform belief which in turn informs behavior. Because belief influences behavior, problems arise when truth and belief diverge. For a complex system to function well, its components (localities) have to be constrained to some patterns of interacting. We see it with cells: when cells become runaway proliferators because their interaction patterns are no longer constrained, a cancer forms, spreads to become a descriptive global presence, and the organism dies at the hand of the cancer’s global causation. Likewise with the complex system of civilization: when people's behaviors are no longer constrained, bad things happen. Whether by definition or by empiricism—whichever you prefer—truth (as compressed information about the environment, including global phenomena) has provided the necessary constraints to flourish, vis-à-vis determining the beliefs that constrain behavior. When truth doesn't constrain belief, belief doesn't constrain behavior. Why is this relevant to flourishing? Look at the pandemic. Flourishing means the maintenance of health and economic productivity (health is definitional flourishing, and economic productivity is a contingent fact of capitalism). The impact of the pandemic on biological health is tautological, and economic health has clearly been impacted by the pandemic as well. All of epidemiology, all of biology, all of medicine, all of economics have been developed in the pursuit of compressed truth (with major missteps and some screwy incentives, but on average tending towards truthful understandings of global phenomena and other things). Our flourishing to date is therefore the result of a pursuit of truth. If truth, a characterization of descriptive global phenomena, is either thrown out the window or manipulated in pursuit of nefarious goals, it spells trouble. The pandemic provides examples, like mask-wearing. It’s clear for those folks focused on truth that mask-wearing is a necessary route to flourishing, i.e. widespread mask-wearing is a causative global phenomenon that will yield a desirable descriptive global phenomenon. The truth informs beliefs held by localities, which constrain behaviors of the localities—e.g., I wear a mask every time I walk out of my house. But with the pandemic, bad information (non-truths) about mask-wearing means that beliefs are being formed based on non-truths, which therefore induce local behaviors that do not produce global flourishing. How does this truth, belief, behavior chain happen right now? Largely on social media. Social media spreads bad information (dis-information and mis-information) about global phenomena. Because people can forward expressions of their beliefs/behaviors to so many people simultaneously either via direct messages or via posting to a feed, the "belief points" can build up rapidly within localities of people (c.f. complex contagion theory). Of course, the echo chamber phenomenon then prevents the penetration of information (truth) inwards through those walls, which leaves the within-locality super-synchronization unchecked (a widespread descriptive global phenomenon occurs without causative intervention to dissipate it). Lastly comes the addictiveness. Humans have short attention spans—especially nowadays. If we weren't hooked on social media, we would spend less time observing the information and belief/behavior expression, which would not only slow the spread of bad information-belief-behaviors, but simply put—we would care about it less.The Human in the Pandemedia Machine
What are the specific failure modes where motivations/incentives/cognitive architectures designed for small groups fail us when our information/beliefs/behaviors come from participation in large groups? The aforementioned Aeon piece argues that social media creates uncapped outrage expression because we don't see the pain of the accused, which would act as a dampener on our rage expression and moral signaling, and subsequently prevent runaway rage. Beyond that, "cognitive social locality" comes into play: as per the often-cited Dunbar's number, we have bounds on the capacity to hold social graphs in memory. Speculating, I suspect this creates a new category of "just-in-time" social entities which, speaking in terms of how our brains respond, are sort of like people but don't elicit the same type of response as real bona fide people. The differences are that we don't have typical signals to assess power and trustworthiness. We can assess power and trustworthiness for local people in our local social graphs, but not people out there in the non-local other. I'd also wager we have weaker theory of mind for someone whose existence to us is restricted to our Twitter feed, for example. Not only do we not have information about people on social media like we would if we knew them locally (either in person or closely connected through in-person relationships with other members of an in-group), but even if we did we couldn't hold all of our 100s-1000s of social media follows in memory as true participants of our social graph—because we didn't evolve for that kind of global memory. Because it's so easy to share information and therefore signal a belief, people do so with minimal filtering (and again, as per the above, without the inhibition coming from real human interaction)—and since people's beliefs are influenced by multiple other in-group members expressing a belief, beliefs can catch on rapidly. Other people signal thoughtlessly, and you thoughtlessly relay the signal once you see enough signals firing. Local regions ignite rapidly and eject their pillars of flame off into the non-localities without any inhibitory feedback, and all the sudden one piece of local mis- or dis-information becomes a causative global phenomenon. Because adjacent localities are closely connected through social media, it doesn't take much flame to ignite part of the adjacent localities, which are susceptible for the same reasons, and so everything continues without inhibition and with local actors participating in causative global dynamics well beyond their cognitive capacity to truly apprehend. This is the story of undamped virality.Your Turn
It turns out that the local-global concept can yield results: in a September 3rd Facebook post, Mark Zuckerberg wrote that “We're reducing the risk of misinformation and harmful content going viral by limiting forwarding on Messenger. You'll still be able to share information about the election, but we'll limit the number of chats you can forward a message to at one time. We've already implemented this in WhatsApp during sensitive periods and have found it to be an effective method of preventing misinformation from spreading in many countries.” What can you do with this idea?