141 | Zeynep Tufekci on Information and Attention in a Networked World
What this covers
Sociologist Zeynep Tufekci joins Sean Carroll to map how digital networks reshape information flows and collective attention. Drawing on Herbert Simon's foundational insight that abundance inverts scarcity—when information overflows, attention becomes the precious resource—Tufekci argues that modern power operates through glut rather than suppression. Authoritarian regimes no longer simply block facts; they drown crucial information in noise and competing claims. Meanwhile, attention economy platforms like Facebook achieve unprecedented scale by not needing ideological coherence; they can show different content to billions of people simultaneously, allowing them to profit by amplifying in-group/out-group dynamics that fracture shared reality. The conversation spans how technology reshapes human behavior without making humans fundamentally different, the mechanics of gatekeeping in a post-traditional-media world, and why the bonding benefit of finding like-minded communities online is inseparable from hostility toward outsiders.
The episode moves from theory into cases. Tufekci examines the Surgisphere fraudulent study on hydroxychloroquine—implausible on its face, yet passed by peer review during polarized times—as evidence that even institutional guardrails can fail when politics invades epistemic judgment. She traces communication history through writing, print, and telegraph, showing how each technology reshuffled the relationship between speaker, message, time, and space. On COVID-19 specifically, she details how Western public health anchored itself to flu playbooks and wrong assumptions about droplet transmission, illustrating anchoring bias at scale, and how countries with prior SARS experience adapted faster. She argues that paradigm shifts may proceed alongside collective amnesia—aerosol transmission shifted from heresy to obvious fact—and discusses why skepticism, though foundational to science, alone cannot empower people to navigate an epistemic fracture where different populations inhabit incompatible versions of reality.
Tufekci argues that in an information-saturated networked world, attention rather than information is the scarce and contested resource, so modern censorship works by flooding the zone rather than blocking facts, and platforms profit by exploiting in-group/out-group dynamics that polarize society.
- When information is abundant, attention becomes scarce, so the filter—not the data—is what matters (Herbert Simon).
- Authoritarian censorship now drowns crucial information in glut rather than blocking it.
- Social platforms monetize attention by amplifying tribal in-group/out-group dynamics, driving polarization and an epistemic fracture.
Platform design amplifies human tribalism by connecting in-group members while inherently defining out-groups, creating polarization through business models that sell attention.
- The bonding benefit of finding like-minded people online is inseparable from out-group hostility, because the demarcation process that lets you find your group necessarily defines an out-group—you cannot have one without the other, just as a sports fan needs an opposing team to cheer against.
“that's the flip side of finding people you don't like, because that demarcation process, that in-group/out-group. Like you can't have one without the other.”
- Because humans are social animals strongly motivated by group belonging, status, and out-group identification, tribalization is the best strategy for capturing attention; platforms financed by selling attention to advertisers therefore have design and business models that amplify in-group/out-group dynamics, feeding polarization even though they did not invent these human dynamics.
“if you really wanna get attention in a media environment where you can kind of play with these things, tribalisation is your best bet for getting attention”
- The same connection mechanisms that let isolated people—a lonely gay kid in a rural area, or a hobbyist with no local peers—find each other also let White Supremacists find each other; the error is treating the technology like a pony that does only what you want, leading to oscillation between utopian and dystopian disappointment instead of recognizing identical underlying dynamics.
“if people who like something certain geeky or cute and something I like can find each other, that doesn't mean the same mechanism doesn't operate for the White Supremacists”
- Facebook and similar platforms should be understood as 'people companies' rather than tech companies, because the sociological and psychological dynamics of human behavior are as central to how they work as the underlying technology.
“Facebook is a people business, it's not a tech company, it's a people company. It's very much focused on people stuff.”
- Unlike a physical newspaper that cannot morph depending on who picks it up and therefore had to ideologically center itself to reach advertiser-friendly scale, digital platforms like Facebook can achieve scale of billions while showing different content screen by screen, so they need not bring people together ideologically and can serve people across all political views simultaneously.
“you can have two billion people still be on Facebook, you have the scale, but they're not reading the same thing, that's a very different way to get scale”
Attention, not information, is the scarce resource in abundant information societies, making filters and gatekeeping mechanisms more consequential than ever.
- Following Herbert Simon's 1973 insight, in a society with abundant information the scarce and precious resource is attention—the thing that information consumes—so the filter becomes what matters rather than the information itself, meaning attention matters more than information.
“in a society where you have a lot of information, the thing that is precious is that which information consumes, which is the attention”
- Social media did not eliminate gatekeeping but replaced traditional media gatekeepers with new algorithmic ones; you can now get attention without approval of traditional gatekeepers, yet algorithms amplify content that is friendly to them and platforms can de-platform even a US President, so gatekeeping persists in a new form.
“it's not like the gatekeeping has disappeared, that it's all flat, it's just not the old kinda gatekeeping”
- At platform scale, transparency measures like publishing targeted ad data are insufficient on their own because they only matter if someone is actually examining them, analogous to publishing ingredient lists that go unchecked—and society has not resolved who should look or even who should be agreed upon to look.
“transparency itself isn't getting you everything you need if nobody's looking. It's kind of like saying, We should have ingredients, well fine, if the ingredients are published when nobody's checking, what's in them?”
Science works through community scrutiny and institutional guardrails that distinguish genuine skepticism from cranks, not through checklists or individual critical thinking.
- Skepticism is how science works and 'don't be a skeptic' is nonsense, but skepticism alone is not empowering; thinking you are a skeptic does not make you Galileo, and many self-styled skeptics are cranks or grifters, with the hard problem being how to tell genuine skeptics from cranks without a workable 'ministry of truth.'
“that doesn't make everybody who thinks they're a skeptic Galileo. There's a lot of people who are... They all think are, but there's a lot of people who think they are the new Galileo, and they're just cranks”
- Even when a fraudulent paper passes peer review at the highest-rated journal, the broader scientific community's willingness to publicly examine and reject it—as happened with the Surgisphere retraction within weeks—is what makes science work, though this process is hard to capture in a reproducible formula.
“the good part of science is then, there's a lot of us who kinda look up and say, Wait, I don't care that it got through, I don't care it got published in the highest rated journal”
- Most conspiracy theories are driven not by ignorance but by curiosity and a desire to figure things out without guardrails; conspiracy sites cite articles and connect dots in a pseudo-scientific way, but lack the institutional guardrails that keep the genuine scientific process from getting lost.
“most conspiracy theories are not driven by ignorance, but by a sense of curiosity and wanting to figure things out without guard rails”
- There is no critical-thinking checklist or simple hypothesis-data wheel that produces science; genuine science requires substantive domain knowledge embedded in a cultural institution with guardrails, debates, and acculturation into a community—yet that community is not the final authority because the human process still goes awry.
“there's no critical thinking checklist, there's no way to be a critical thinker, they think without a lot of substantive knowledge in the area and acculturation into a group of people and the discussions and debates”
- The fraudulent Surgisphere HCQ study—claiming standardized hospital records from 150 countries showing severe harm—should have been desk-rejected because standardizing medical coding across hundreds of hospitals worldwide is implausible, but political polarization (its undermining of a Trump claim) likely helped it pass peer review at a top journal like The Lancet before scientists detected the fabricated data and forced a retraction within weeks.
“How did you get data from 150 countries or hospitals and managed to sort of standardize the medical coding? We can't even standardize the medical coding between two hospitals in this country easily”
COVID's characteristics—aerosol transmission, super-spreading, age-targeting—required pathogen-specific responses that differed from historical flu-pandemic assumptions.
- Pandemic non-pharmaceutical interventions were set on the assumption that respiratory pathogens spread through ballistic droplets, but the WHO's micron cutoff for what floats versus falls is wrong by a large factor due to an old transcription error; the epidemiological record can only be explained by aerosol transmission being dominant, suggesting most respiratory pathogens transmit by floaty aerosols rather than large droplets.
“The World Health Organization to this day has the micron cutoff for what floats and what falls wrong by a large factor.”
- Western public health failed early in COVID partly because it was anchored on waiting for a flu pandemic—'we think in flu'—but COVID is a different pathogen: it is over-dispersed and driven by super-spreading events, kids are not its super-spreaders, so the flu playbook of closing schools and waiting for a vaccine was the wrong reflex, while countries with SARS experience moved quickly to correct non-pharmaceutical interventions.
“the Western public health, I think, was fairly anchored on waiting for a flu pandemic. We think in flu, and flu is a different pathogen. It's not over-dispersed the same way”
- By the end of March 2020 the epidemiology made clear that COVID super-spreading transmission occurred primarily indoors in poorly ventilated places, with outdoor transmission possible but rare, so authorities shaming people on beaches or outdoors are behind the scientific knowledge; outdoor activity is largely fine.
“it was pretty clear that transmission, the super spreading transmission was primarily occurring indoors in poorly ventilated places, that outdoor transmission wasn't impossible, but it was a pretty high bar”
- Despite half a million US deaths, COVID is a relatively mild pathogen in the space of possible pathogens because its fatality is strongly age-targeted rather than killing children, and it is highly susceptible to non-pharmaceutical interventions—controlling indoor super-spreading in crowded places yields a big effect per unit of effort.
“In the potential pathogen-scape, let's say, this is a mild one. It kills relatively low compared to many others. It is very targeting... It's targeted by age.”
- The key lesson for the next pandemic is to apply the precautionary principle and act fast under uncertainty—do everything when you don't know what you've got—but then remain highly responsive to early data and adjust interventions to the specific pathogen's characteristics, since there are no guarantees about what the next one will be.
“Act fast, but be very responsive to data early on, because it really depends on what do we have? And no guarantees what we're gonna get.”
Modern information environments fragment shared reality into epistemic fractures where different populations hold incompatible mental models of events.
- The pandemic demonstrated that there has never been a better time to be informed and never a worse time to be informed simultaneously, because the same abundance produces both rapid generous science and a flood of noise and crappy papers.
“there's never been a better time to be informed, but there's also never been a worse time to be informed. It's both.”
- Because shutting down the internet is incompatible with a modern economy, the modern form of censorship is information glut: regimes drown people in information, claims of fakery, and distractions, breaking the chain between information and action—not by removing the crucial information but by surrounding it in cacophony that makes it useless to the hearer.
“the way a lot of places have figured out how to do censorship is by information glut. You drown people in information and claims, questioning the credibility of whatever you don't like”
- Whereas mid-20th-century America had centralized broadcast media producing shared common knowledge, the fractured modern media environment produces an 'epistemic fracture' in which different portions of society hold completely different mental models of what is going on, changing how societies experience pandemics and political crises.
“if you have an epistemic fracture in your society, you can have a very different mental model of what's going on in one portion of the society, and the other portion has a completely different view of what's going on”
Communication technologies reshape the game board of human status-seeking and social dynamics without changing fundamental human nature or creating new paradigm shifts easily.
- The major breaks in communication history—writing, the printing press, telegraph, telephone—each reshuffled how speaking, time, space, and the body relate, with writing being the foundational break that first separated the speaker's words from the speaker, producing large social consequences such as changes in warfare and immigration.
“the big break is writing, where, for the first time, you can separate the words of the speaker from the speaker”
- New communication technologies are a case of 'more is different': they change how we live, acquire status, and get information without making us fundamentally different kinds of people—someone from the Pleistocene could adjust, because we remain social status-seeking creatures while the 'game board' on which we play changes.
“this is, I think, a case of more is different in that it affects what happens to us and how we live, but it doesn't make us different kinds of people”
- Kuhnian paradigm shifts may be more common in modern science than expected, and once a shift occurs people forget they ever held the opposite view—exemplified by aerosol transmission going from 'misinformation' a year ago to 'of course it was aerosols' today—so paradigm shifts proceed alongside a sociological process of amnesia.
“these paradigm shifts happen along with a very sociological process of amnesia, where the people who were saying the opposite, they just forget they were saying the opposite”
- Sociology as a discipline was born of an era of rapid technological transformation in the 19th century—urbanization, mass manufacturing, new communication technologies—as its founders tried to grapple with a world changing rapidly, including rural-to-urban migration producing cities of strangers.
“It was born off an era of technological transformation... trying to grapple with a world that was changing fairly rapidly from urbanization, mass manufacturing, just the communication technologies.”
Complex systems concepts like non-linearity and convergent evolution appear across disciplines yet remain siloed, limiting interdisciplinary understanding.
- COVID showed surprising convergent evolution, with the same few mutations popping up independently in many places and out-competing other variants, indicating strong selective pressure toward a local maximum; this convergence suggests that if left alone the virus will predictably reach similar adaptations everywhere, which matters for preparation.
“we have a lot of convergent evolution in the mutations, which has surprised the virologists. We weren't expecting this.”
- Exponential and non-linear dynamics are everywhere yet non-intuitive because people don't encounter them in ordinary macro life, but social media growth and epidemics finally provide intuitive examples; with an exponential threat you must act first because the steep part of the curve cannot be gently surfed—get in early or get overwhelmed.
“with an exponential process, you get in early or you get overwhelmed, right? The steep part of the exponential is not a place you're gonna very gently surf, you're just like need to go first”
- Complex-systems concepts—non-linear dynamics, phase transitions, convergent evolution, fitness landscapes—show up across physics, ecology, virology, and sociology because the world isn't siloed into fields, yet these languages are typically not spoken across disciplines.
“usually these are languages that are not spoken across fields... sociologists don't talk about convergent evolution... But I find them all... They show up, they come up in the world. The world isn't siloed into these fields.”