The Technium

Increasing Specialization

Evolution moves from the general to the specific. The first version of the cell was a general purpose survival machine. Over time evolution honed that one generality into multiple specialties. In the beginning the domain of life was restricted to warm ponds. But most of the planet was far more extreme; volcanoes and glaciers. Evolution devised cells that specialized in living in boiling hot water, or within freezing ice, or special cells that could eat oil, or trap heavy metals, or glow in the dark. Specialization enabled life to colonize these major, but varied, extreme habitats, and also to fill millions of niche environments – such as the insides of other organisms, or on the dimples of dust particles in the air.  Very soon every possible environment on the planet sprouted a specialized variety of life making a living there. Presently there are no sterilized places anywhere on the planet that we’ve inspected, except in a very few temporary spots within a hospital setting. The cells of life keep specializing.

The trend toward specialization holds for multicellular organisms as well. Cells within an organism specialize. The human body has 210 different types of cells, including the specialized cells in your liver, and a different type for your kidneys. You have special heart muscle cells, different from ordinary skeletal muscle cells. The original omnipotent egg cell that initiates every animal divides into cells with greater specificity, until after less than 50 mitotic cell divisions you and I wind up with a unified assemblage of 10^15 bone cells, skins cells and brain cells.

Over evolutionary time, there is a significant rise in the number of cell types in the most complex organism. In fact, these organisms are more complex in part because they contain more specialized parts. So specialization follows the arc of complexity. 


The number of cell types found in organisms over evolutionary time.

The organism itself also tends toward great specialization. Over the course of time, for one example, barnacles (comprised of 50 specialty cell types) evolve into specialty barnacles: the Six-plated Barnacle specializes in extreme high-tide locations that are flooded (with food to eat) only several times a month. Or, the Sacculina Barnacle that grows only inside the egg sack of a living crab. Birds focus into specialized type of seed eaters with specialized beaks: fine ones for small seeds, but fat beaks for hard seeds. A few plants are opportunistic (we call them weeds) occupying any disturbed soil, but most plants dedicate their survival skills toward a particular niche: dark tropical swamps, or dry, windy alpine mountain peaks. Koala bears are famously specialized on eucalyptus trees, and pandas on bamboo. Parasiticism is particularly specific and yet so common in life that some experts estimate up to 50% of living species are parasites. Most parasites, such as lice, will feed only on one species, and often only on one bodily part of one species. There are parasites that prey only on other special parasites.

The trend toward specialization in life is propelled by an arms-race. More specialized organisms (such as a clam feeding in sulfuric emissions in lightless deep sea vents) present more specialized environments for competitors and prey (crabs that feed on them), which breed more specialized strategies (parasites on the crabs), and solutions, and in the end yet more specialized organisms.


This urge to specialize extends into the technium. The original tool of the hominids, a roundish rock with a broken edge, was a general purpose tool used for scrapping, cutting, and hammering. Once taken up by Sapiens, the general tool morphed into specialty tools: a separate scraper, cutter, or hammer. The variety of tool species increased over time as specialty tasks increased. Sewing required needles; sewing hide required special needles, sewing woven fabric another. When simple tools were recombined into composite tools (string + stick = bow) specialization increased further. The astounding diversity of manufactured items today is primarily driven by the need for specialized parts of complicated devices.


At the same time, just as in organic life, tools tend to start out useful for many things and evolve toward specific tasks. The first camera with photographic film was invented in 1885. Once incarnated, the idea of the camera started to specialize. Within years of its birth inventors devised tiny spy camera, extra large panoramic cameras, compound lens cameras, high-speed flash cameras. Today there are hundreds of specialty cameras, including those for use deep underwater, to cameras designed for the vacuum of space, or those to capture the infrared, or the ultraviolet. There are telephoto cameras for the  very far and microscopic cameras for the very close. While one can still purchase (or make) the original general purpose camera, they count for an increasing smaller proportion of cameradom.

This sequence from general to specific holds true for most technologies. Automobiles start off broadly appealing and over time, they evolve to specific models, while the general purpose variety fades. In cars you choose among compacts, SUVs, vans, sporty models, sedans, pickups, hybrids, and so on. Scissors are specified for hair, paper, carpet, mesh, or flowers.

As we look into the future, specialization will continue to increase. The first gene sequencer sequenced any gene. The next step is a specialized human DNA sequencer, which only does humans or another species, say the mouse for researchers. Then we’ll see sequencers that specialize in racial genomes (for African-Americans or Chinese), or extremely portable ones, or ones that are extremely fast and sequence in real time, letting a person know whether pollutions are damaging their genes right now. The first commercial virtual reality consoles will serve up virtual realities for all purposes, but over time, VR consoles will evolve special versions with special gear for games, or military practice, or movie rehearsals, or shopping.

At the moment computers seem to be headed in the opposite direction. They seem to becoming evermore general purpose machines, as they swallow more and more functions. Entire occupations and their worker’s tool have been subsumed by the contraptions of computation and networks. Computers have already absorbed calculators, spreadsheets, typewriters, film, telegrams, telephones, walkie talkies, compasses and sextants, television, radio, turntables, draft tables, mixing boards, war games, music studios, type foundries, flight simulators, and many other vocational instruments. You can longer tell what a person does by looking at their workplace because they all look the same: a personal computer. Ninety percent of employees are using the same tool. Is that the desk of the CEO, the accountant, the designer, or the receptionist? This convergence is amplified by cloud computing, where the actual work is done on the net as a whole, and the tool at hand merely becomes a portal to the work. All portals become the simplest possible window: a flat screen of some size.

This convergence is temporary. We are still in the early stages of computerization – or rather intelligenation. Everywhere we currently apply our own personal intelligence (in other words, everywhere we work and play) we are rapidly applying artificial and collective intelligence as well, and rapidly overhauling our tools and expectations. We’ve intelligenized bookkeeping, photography, financial trading, metal machining, airplane piloting, among thousands of other tasks. We are about to computerize automobile driving, medical diagnosis, and speech understanding. In our rush toward large-scale intelligenation, we first installed the general purpose PC with its mass-produced small brain, mid-size screen, and conduit to the net. So all chores get the same tool. To complete the dispersion of intelligenation into all occupations will probably require another decade. Silly as it now sounds, we will put artificial intelligence into hammers, dental picks, fork lifts, stethoscopes, and frying pans. All these tools will gain new powers by sharing the universal intelligence of the network. But as their newly augmented roles become clear, the tools will specialize.  We can see the first glimmers in the iPhone, Kindle, Wii, and netbooks. As display and battery technology catches up to chips, the interface to ubiquitous intelligenation will diverge and specialized.  Soldiers and other athletes who use their full body want large-scale enveloping screens, while mobile road warriors want small ones. Gamers want minimal latency, readers want maximum legibility, hikers want waterproofing, kids want indestructibility. The portals into computation, or the net, will specialize to a remarkable degree. The keyboard, for one, will loose its monopoly. Speech and gesture input will gain a major role. Spectacle and eyeball screens will supplement walls and flexible surfaces.

When we pick up a smart hammer, the handle will contain its smartness. A very smart air hammer pounds in nails at the right speed, of the correct type, for that immediate material and job. Computational intelligence will mostly disappear into the tool itself, so we only deal with the interface. The interface for tools will specialize as the jobs specialize. The ultimate specific technology or technique is customized to a particular use by a specific user for a specific time. Very niche-y functions may summon devices that are assembled for only that task and then unassembled. Ultra-specialized artifacts may live for only a day like a mayfly. Specialization may be reckoned in half-lives.

All the things we make with our mind, and the creations of artificial minds, will all tend over time to become more niche based. The “long-tail” is not merely a characteristic of media, but of technological evolution itself: the tail of niches gets longer and longer. We can imagine the future of almost any invention working today by imagining it evolving into dozens of narrow uses. Technology is born in generality and grows to specificity. Technology wants specialization.

  • Andrew

    @kevin more than the study it’s the way they think about things that’s inventive. Reading Steigler and having him come to the conclusion that Humanity is a denaturalizing process or Gourhan’s peculair way of shifting through the ideas of early tool making and coming up with new ways of conceptualizing the role of technology in our society and the way it affects us is the interesting part. Reading your prose it seems remarkably inline with evolutionist ideas, when it’s the anthropologists and sociologists who seem to be better at conceptualizing the life of technology and how it plays a role as a social actor and shapes culture. BTW Gourhan felt that alphabetic writing is on the way out:–CorporealWriting.pdf


  • Alex Tolley

    “Over evolutionary time, there is a significant rise in the number of cell types in the most complex organism. In fact, these organisms are more complex in part because they contain more specialized parts. So specialization follows the arc of complexity.”

    Stuart Kauffman suggests that cell types are attractors and is a function of the number of genes an organism has. (See “The Origins of Order”)

    Regarding specialization in biology. I wonder if metabolism is a possible counter example? Core parts of our metabolisms are extremely conserved across all organisms. While we may think that this is due to selection of optimal pathways, what if it reflects selection of a more diverse metabolism in the pre-biotic or early biotic earth? Likewise, the genetic code is almost completely universal, but why should it be? Diversification would suggest diverging codes, not something we see.

    I think what we have is a tension between selection for optimum fitness and the opportunities to find new niches. Selection reduces the diversity, whilst mutations and sex increase diversity.

  • Alex Tolley

    “I think what we have is a tension between selection for optimum fitness and the opportunities to find new niches. Selection reduces the diversity, whilst mutations and sex increase diversity.”

    In the technium, the tension is between selection of the lowest cost, mass produced, commodity item, and the creation of new species of niche items to fill profitable marketing opportunities. The product life cycle initially expands into the new niches with diversity, but then commoditization reduces it. In computers, we had many different chips, bus architectures and OSs back in the explosive growth period of the early 1980′s. Within a few years the choices standardized around the Wintel architecture plus a few niche players using the PowerPC chip and Mac OS.

  • Gino Tocchetti

    btw specialization brings to a short intensive application then the obsolescence: as human being, evolve avoiding specialization, and composing specialized tools in a new machine, in a new organism

  • Patrick McNeely, PT, CEAS

    Hmmm. I really enjoy reading your writings. I find my thinking stimulated by both your articles and the responces of your readers. In this one I am reminded of something I have observed in medicine regarding specialization. In therapy, we have: Physical Therapists who tend to be generalists; studying the whole body; Occupational Therapists specializing in upper body therapy and functional activity; Speech Therapists specializing in cognition, speech, and oral motor function; and Respiratory Therapists focused on pulmonary function. Unfortunately I have found the more specialized therapies lose out on the core education on anatomy and physiology. This has caused some disparity in the professions such that PTs tend towards more independent work while the other therapies have become more dependent on the medical system itself to function…. I am afraid that this does a disservice to our patients and allows more errors.

  • Clem Weidenbenner

    On the issue of computers seeming to buck the diversity trend… “ever more general purpose machines”… I think this is in some part due to what we consider a computer. Is the electronic box the computer? Or is the code that is processed the actual computer (or isn’t it the combination of the two?). And while the “box” may look similar on the outside to its predicessors of some thrity years, there are some pretty significant changes under the hood. But why limit the definition of computer to the box on the desk? Surely a setback thermostat is a computer, and there’s a computer in my car taking stock of many different sensors to govern aspects of the engine to improve performance. There are computers all around us, so the diversity model really doesn’t seem violated to me.

    Software does seem to better follow the diversity pathway – with code progressing into a long tail model quite nicely. While you may not be able to tell what a worker does by looking at the box by the desk – if you have a peek at the software used you’ll often have enough to go on. And those specialized computers also give us clues. The automechanic has a special tool to plug into my car’s computer to read the codes that help diagnose issues. Few Fortune 500 CEOs will have one of these on their desks. The bar-code reading hand held that a retail worker uses to inventory shelves fits this category as well. The ‘box’ side of computing is diversifying, maybe just not at the same rate as the software.

    But lets turn this diversifying phenomenon on its head for a second. In a previous post a lot was made of the English alphabet (however many letters it contains). And one might argue that there are many different alphabets in existance, and there are many that once existed only to morph over time (become more diverse… or evolve) into a different form. But the rate of change in something so technically competant as an alphabet seems slow compared to other artifacts of human creation(hats, crackers or spark catchers for instance).

    When evolution finds a winner it doesn’t stop trying to improve it, but its success rate surely suffers.

  • Andrew


    As always your thoughts are very interesting. Have you looked into the work of anthropologists studying the origins of tools? The French anthropologist André Leroi-Gourhan for instance came up multiple concepts to deal with the emergence of technology in human societies:

    The teleological aspect of Leroi-Gourhan’s ideas is clearly expressed in his use of the concepts of tendance and fait in the study of technology. Tendance is defined as a an evolutionary phenomenon which underlies specific manifestations of technology. To quote from L’Homme et la Matière: “[La tendance…pousse le silex tenu à la main à acquérir un manche” (LHM 27).

    Another example of tendance is decoration of the self. For Leroi-Gourhan it is because of the underlying power of the tendance that there are
    parallels in the ways societies around the world decorate themselves.

    The fait is the inverse of tendance. The fait is the unpredictable local manifestation of the tendance. In Leroi-Gourhan’s words the faite “c’est un compromis instable qui s’établit entre les tendance et le milieu” (LHM 27). A forge is the compromise between fire, metal, combustion, fusion, commerce, fashion, religion that is the concrete manifestation of the tendance of metallurgy.

    End qoute

    Tendance in other words seems to posit that technology has a social presence that in turn influences its creators: tribes learn how to smelt iron in turn it occurs to them, why not make swords etc?

    These ideas are summarized here:

    Recently I’ve been reading Bernard Steigler’s Technics and Time in which he takes ideas from Leroi-Gourhan and concepts from Derrida and uses them to locate the origins of the idea of human, which he thinks was a double invention when man first invented tools, the tools in turn created human. Exteriorization, the ability of humans to take things from their mind and perserve them in technology is a form of memory, but also an essential part of what makes human. Steigler eventually posits that human isn’t a zoological or biological category, but rather one produced by life pursueing evolution by a means other than life. He uses the term epiphilogenesis (based on the evolutionary idea of an epigenesis) to represent the evolution of tool use and the memory they contain. At heart, according to Steigler, Human is a denaturalizing process. Intelligenation strike me as continuation of this process in which human is being redfined not by their externalized tools, but rather by the deepening of the communication between human and technology.

    Anyway, you’re on different pages from the two thinkers above, but the ideas do connect and the interplay interests me.

    Steilger’s book:,_1


    • Kevin Kelly

      @Andrew: I have not read Gourhan but will check him out. The French are amoung the few who have studied “material culture” (technology) seriously.

  • Tim S

    My personal intuition (which is all that it is) is that “computers” will actually become more general. This is in the sense that an element of computation will be incorporated into more and more devices e.g Clem’s engine management system.

    The software front end will become ever more specialized to each device, but the computational hardware will be standardized for ease of installation, repair and upgrade.

  • Vasu Srinivasan

    Fascinating article Kevin.

    Generalization in a System’s functionality, is a function of diversity in specialization of its components.

    -Vasu Srinivasan