What this covers
Kevin Kelly sits down with host Russ Roberts to explore how complex systems—from biology to economics to the internet—operate through decentralized self-organization rather than top-down control. The conversation ranges across technology, philosophy, and economics, anchored in Kelly's observation that the natural world and human-made systems increasingly follow the same logic: order emerges from seemingly chaotic, "out of control" networks. Kelly argues that networked computers, Wikipedia, open-source software, and ant colonies all demonstrate that distributed intelligence produces results superior to what any single designer could mandate.
The discussion covers several distinct territories. Kelly traces how the web's true power arrived only when isolated computers connected into networks—a transformation still in its infancy. He sketches the evolution toward machine-readable systems and a future where software agents conduct most communication silently in the background. He examines why people adopt technology and migrate to cities not for net benefit but for marginal increases in choices and options, a principle he sees as fundamental to human progress. Kelly contends that biology and computation are converging from both directions—machines becoming biological, organisms being engineered—and that applying computational thinking to DNA and immune systems reveals deep structural similarities.
The conversation turns contentious on several fronts: whether the "hive mind" can escape mob stupidity and conspiracy, whether economic models should abandon physics-like precision for messier biological frameworks, how intellectual property law went wrong, and why innovation worth 51 percent good still drives adoption despite being nearly half bad. Kelly also discusses his involvement in shaping The Matrix and Minority Report, the economics of reputation and accountability, and why culture has been slow to accept that copies themselves lack value.
Kelly argues that complex systems—biological, economic, and digital—share a common 'out of control' logic where decentralized, emergent self-organization produces order and value superior to top-down design, and that technology's true gift is expanding human choices.
- The realm of the born (biology) and the made (machines) are merging, so biological and economic models illuminate each other bidirectionally
- Networked 'hive mind' systems (web, open source, Wikipedia) produce emergent intelligence greater than their parts despite being out of control
- People migrate toward cities and technology not for net benefit but for marginally more choices, which is the deep driver of progress
This asset isn't compiled yet
You're seeing its claims, ranked. Compile it to build the argument threads, weight them, and check each claim against your library — the full view.
Computers did not change the world until they were networked together; isolated PCs only accelerated a few jobs like word processing and Photoshop, but networking dumb PCs together unleashed enormous cultural and technological force—and we are still only at 'the beginning of the beginning' of that power.
“computers really did not change the world until they were all hooked up together... computers accelerated a few jobs like word processing and Photoshop and Excel but but they didn't have anywhere near the cultural force... until they're all networked”
Human language frustrates our ability to describe emergent, uncontrolled phenomena because it deals so often in intention and design; people are so accustomed to cause-and-effect that 'out of control' or 'unmanaged' sounds negative, like 'unmanaged trade,' when it is merely a neutral or positive description.
“Hayek talked about the challenge of language how our language itself frustrates the ability to describe these phenomena partly because they so ought the language so often deals in intention design as opposed to emergent uncontrolled things”
If Kelly had stood up ten years prior and described everything the web would deliver for free—street maps, Google Earth in 3D, real-time stock quotes, weather radar, encyclopedias—he would have been laughed off the stage, the main objection being that there was not enough money to build it, and economics still lacks a good explanation for how this free abundance happened.
“if I were to stand up and lift all the things that I believed would come in ten years... most of it would be for free I would be laughed off the stage”
Although new technology often brings more problems than solutions, innovation remains worth accepting because most problems are merely opportunities not yet realized, and even when a technology is roughly 49% bad and 51% good, the small positive delta—especially the increase in choices—is what drives adoption and progress.
“most problems are just kind of opportunities that we haven't realized yet and secondly that in those albums as well as it benefits our choices”
People build the web and contribute things like Wikipedia not just for money but for love, and sometimes act irrationally—these 'miracle agents' driven by love rather than rational self-interest must be incorporated into our economic thinking, contradicting the straw-man view that people are only motivated by money.
“people do things for love and that people do things not just that agreed and maybe even do they do things irrationally sometimes too and so that's not been tarred economy economic formula”
The Semantic Web (web 3.0) is the next evolutionary stage where the web shifts from pages to databases and becomes machine-readable, so that software agents can read and extract meaning—producing a vastly more powerful hive mind in which links and tags add structure, letting the web 'sense' meaning and answer natural-language questions intelligently.
“the next stage in the evolution of the web which is a move away from seeing the web as pages and Mort seeing it as a databases... the web is becoming will become more what they would call machine readable”
Once you view DNA as a computer that calculates solutions, you can apply the same lens to see cells computing their metabolism and the immune system performing computation, because abstracting their functions into logic flows reveals identical patterns—evidence that biology and computation are fundamentally one.
“imagine that cells are actually calculating or computing their metabolism that their responses or the immune system is another great example”
An economic marketplace can emerge among non-conscious, non-biological, even erratic and irrational mechanical agents, provided there is an exchange of resources in sufficient numbers, with sufficient degrees of freedom, over persistent time—agents need not be rational to produce market phenomena.
“there can be marketplaces of mechanical agents and that that when you have an exchange of resources in sufficient numbers with sufficient degrees of freedom over persistent time even if those agents aren't conscious or even biological you can still have the phenomenon of an economic marketplace”
Most communication over the next 25-30 years will not be between humans, but between our software agents and other machines operating in the background, shifting our relationship with the web from navigation toward meaning or experience—we will be served and supported without having to consciously navigate.
“most of the communication that will happen in the next 25 30 years and beyond is going to be not between humans in the web or between human and human but between our agents and machines behind them and other machines”
We are in a Renaissance of television programming via the long form (TV on DVD), with 20-hour 'movies' like The Sopranos and Lost featuring storylines far more complex and richer than anything previously written—the medium was simply looking for the right form.
“I think we're actually in the Renaissance of TV programming with the long form of some of these shows you know basically 20-hour movies you know Sopranos or lost or others very deeply rich storylines far more complex than anything that's been written till now”
The hive mind has a serious downside—the stupidity of the mob—manifesting as herd behavior, conventional wisdom that isn't wisdom, conspiracy theories, rumor, gossip, and mass spam, so networked collective intelligence is not a utopia.
“there's a downside to the hive mind which is you know the wisdom of them the stupidity of the mob... conspiracy theories and rumor and gossip are all part and parcel of this hive mind”
To gain the full power of an economic system, a biological system, or a complex computer system, you must surrender a certain line of control—being 'out of control' is a feature and a price worth paying, not a bug.
“to really gain the full power of an economic system of a biological system and eventually of a very complex computer system that you actually have to surrender a certain line of control to maximize it”
We are still in a cultural period of discovering how far out-of-control decentralized systems can go in solving problems; the wise recognize such systems cannot get you even halfway to your goal, yet they remain the best way to start, and they consistently go further than expected before requiring complementary approaches.
“we are seeing how far you can go with an out-of-control decentralized system and solving things I think the wise realize that you can't go even halfway to where you want to go with just that model but at the same time it seems really the best way to start”
We migrate forward in time—one million people move from countryside to cities daily not for net improvement but for roughly 1% more choices, freedoms, and options; this pursuit of marginally more choice, not net benefit, is the fundamental human driver of progress and technology adoption.
“every day 1 million people moved from the countryside into cities... they're coming as we all do for the slightly increased number of freedoms and options... to have 1% more choices”
Long Bets was created to make people accountable for their predictions by introducing a penalty for being wrong, since currently there is no hurt for confidently making forecasts that fail; the web's search function increasingly restores reputational accountability that was easy to escape a decade ago.
“Stewart Brand and I started something called long Betts which was a website to encourage people to be responsible and accountable for their predictions”
DNA computing solves complex problems by exploiting massive parallelism: each possible solution is encoded as a DNA sequence, millions of variations are generated simultaneously, and the physical molecule that best fits the 'lock' (the puzzle) is extracted as the answer—useful for problems like the Traveling Salesman problem that don't scale on linear von Neumann machines.
“rather than just do one at a time one after another it's actually doing all the solutions or all the possibilities at once”
The convergence of the born (nature) and the made (human-constructed)—machines becoming biological and biology becoming engineered—has not only continued since 1994 but accelerated, shifting from poetic metaphor to conventional, almost cliched, background assumption supported by far more concrete examples.
“I would say the trend is definitely still happening if not accelerating the primary thing that has changed is that that insight has become conventional almost cliched”
The holistic, fuzzy way of understanding complex systems should supplement rather than replace the reductive, precise approach; for centuries there was an intuition that this holistic view was productive but no tools to pursue it with rigor until personal computers arrived, whose ability to synthesize and simulate complexity finally provided that tool.
“what we have not been able to do before is to tackle this other way this holistic way in any kind with any kind of rigor until computers came along”
The intellectual property and copyright regime was mishandled because the political process is structured so that established business interests with a vested stake in the status quo speak loudly and outweigh diffuse, uncoordinated citizens who would benefit from change—producing laws that fail to coincide with what the technology wants.
“I actually think that we messed up in our IP regime in copyright law”
Economics has been dominated for at least 60 years by a 'physics model' of precise equations modeling simplistic aspects of the economy, but economists are increasingly dissatisfied; a biological, messier, fuzzier, more holistic model is a more fruitful way to understand how the world actually works.
“economics has been dominated for the last Oh probably sixty years... by what we might call the physics model a set of very precise equations modeling very simplistic aspects of the economy”
Predictions of national economic doom often turn out ghastly wrong yet their authors suffer surprisingly little reputational damage—people who claimed Japan would destroy America and cost all our good jobs disappeared without lasting penalty, and today's China-fear predictions will likely take an equal hit.
“all the people who said Japan was going to destroy us and we're going to lose all our good jobs and that just disappeared... a lot of people today who are worried about China I think will take an equal hit”
Copies themselves don't have value; rather, value lies in the things you do with copies, and recognizing this 'things want to copy' principle should drive a more favorable philosophy of fair use and public domain—an insight our culture was slow to recognize over the past 20-30 years.
“things want to copy and the copies don't have values themselves that Ziller re things you do with copies that have values”
The 'hive mind' (the wisdom of the crowd) holds that nobody is as smart as everybody: connecting many dumb individual agents—like ants in a colony—with communications produces an aggregate intelligence that exceeds any individual, an emergent 'upcreation' where the whole is less error-prone than its parts despite intuition suggesting errors should compound.
“the hive mind is what some other people called the wisdom of the crowd... the idea that nobody is as smart as everybody”
A large cultural shift toward accepting decentralized, out-of-control systems was driven by three factors: the demise of alternative models like communism, the Boomer generation's personal failed experiments with communal living, and the ongoing demonstration by the computational metaphor and the internet (e.g., the miracle of Wikipedia) that decentralized systems work.
“the demise of our alternative model like communism made uh we focused it... the boomer generation of which I am who had their experiments with communal living... I do believe the computational metaphor and things like the internet seeing how it works”
It is not impossible to be happy and to a certain extent spiritually content with very little, as Kelly learned spending his formative post-high-school years living in 15th-century-like Asian villages that often lacked electricity and even metal.
“instead of going to college I spent my formative years after high school in Asia living in the 15th century living in villages that not only back to electricity but often lacked metal... it was not impossible to be happy and to a certain extent spiritually content with very little”
The critique that current technology isolates us ('bowling alone' at our computers) is a misreading; the connections people make with strangers via blogs, shared music, and podcasts are extraordinary and deeply delightful, fulfilling the fundamentally human desire to connect with others.
“people decry our current world of technology because it's isolating... Bowling Alone and yet it's a misreading of what's really going on... the connections that we make with with strangers now via the web... are just they're extraordinary”
There is no technology that is completely bad—every technology, like Darth Vader, has some good in it; the task is to find a good home or niche for each technology where its virtues blossom and its negatives are diminished, as is still being worked out for nuclear energy.
“bad technology this is bad parents... there's no technology that is completely bad... it's trying to steer that technology to the right place where it's genius Canon can blossom”
The mainstream media has no sense of the long view—nothing more than six months to a year old gets written about—whereas a futurist worth their salt must be a good historian, and cultivating a long-now perspective (thinking in terms of generations and centuries, increasing options over time) is needed to solve problems that short attention spans handicap.
“no one writes about things that are more than 6 months old or year old”
The Traveling Salesman problem is easy to state but very hard to solve because as the number of cities increases, the number of possible routes balloons much faster, making it computationally intensive with no clean algorithm that scales.
“if you have a... very large number of cities that you want to travel within any country and you want to devise the shortest possible route... as the number of cities increases the number of possible solution balloons much faster”
The Wachowski brothers were influenced by Out of Control (along with Baudrillard's Simulations) when making The Matrix, requiring the actors to read the book; its influence appears in the swarm super-organism imagery and the simulated-reality/hyper-reality concept where a simulation becomes so rich it attains its own level of reality.
“the brothers... who made the film were very influenced by reading out of control and two other books apparently and they felt so strongly... that they required all the actors on the film to read that book”
Out of Control was reviewed in only two places when first published in 1994 and was largely ignored, yet now sells better and receives far more attention ten years later—suggesting timeless ideas can be vindicated long after release.
“when it was first issued in 94 the auto control was reviewed only in two places Wow it was completely ignored and it has far more attention now ten years down the road”
The principles in New Rules for the New Economy (1998) remain valid and would require only a global search-and-replace of 'new economy' with 'networked economy'; the examples are dated but anticipated developments like Google before they existed.
“I don't think I would change anything I said or anything principal's... do a global search and replace a substitute the word networked economy with new economy”
Spam control has become an arms race in which self-organizing filtering tools (e.g., SpamSieve on a Mac) can reduce inbox spam to about one per day, roughly the rate from ten years prior, and this equilibrium is expected to persist going forward.
“this is are like an arms race yep and that about one spam a day or so it's kind of what we had 10 years ago and that III expect that'll kind of be about the the rate going forward”