Intro to Innovation

The topic of innovation comes up often here, so let us review some basics:

  1. Systems are parts in a structure; innovations are better part or structure designs. An innovation embodies insights whose value depends on a context, and so changes with that context.
  2. Most net growth in the number or size of large systems has been due to collecting innovations.
  3. Wars, quakes, and diseases may be distributed so most impact comes from the few largest instances.  In contrast, in large systems most innovation value comes from many small innovations.  Even big innovations require many matching small innovations to be viable.
  4. Innovation rates increase when early innovations make it easier to pursue later innovations, and decrease when the most valuable easiest innovations tend to be pursued first.  Steady (exponential) growth suggests that these factors roughly cancel.  Since growth rates commonly increase then decrease, usually the second factor eventually wins.   
  5. Innovation in large systems comes mostly from part innovation, so system innovation is steadier than part innovation, and the largest systems grow steadiest.
  6. System structures vary in how well they encourage and test innovations locally and then distribute the best ones widely.  Better structures for this are meta-innovations.
    • Good modularity reduces the need to match innovations in differing parts. 
    • Good abstraction puts similar innovation problems within the same part.
  7. If a barrier isolates two systems, the faster growing one eventually dominates.  A system that better promotes innovation can lose to a system with a larger source of innovation. 
  8. In large innovation pools, similar innovations commonly arise from several semi-independent sources at nearly the same time.  No single source was essential.
  9. Current human society can give incentives to innovate too much, when innovation is used to signal, and to innovate too little, when innovators are not paid the full value gained by others.

I learned this stuff long ago so I have little idea how commonly known this all is. 

GD Star Rating
loading...
Tagged as:
Trackback URL:
  • Silas

    Considering how vague and ambiguous it is, it’s kind of hard for much of it to be wrong. How do I test e.g. the claim that “lots of small innovations” are “more important” than “a few big innovations”?

  • http://shagbark.livejournal.com Phil Goetz

    The dynamics of diffusion are also important. The genetic algorithm, while overrated as a search mechanism, is a good model for the creation and diffusion of innovation. A GA which uses global selection (choosing the best n organisms in each generation) typically reaches a lower optimum than one which uses local selection. This is because the best n organisms at any given time are largely related. Selecting the global best n thus severely constricts the gene pool.

    For example, broadcast media can find and disseminate good performers and musicians rapidly; but the rate of musical innovation would probably be higher in the long run if we had more diverse radio playlists, and more cultural isolation between regions of the US.

    What do you mean by “when innovation is used to signal”?

  • Luis Enrique

    Robin,

    Are there any books that you would recommend to help me start thinking about innovation in this way? The only one I know of is Mokyr’s Gifts of Athena, although I’m not sure if that fits with your thinking here.

    thanks

  • ad

    growth rates commonly increase then decrease, usually the second factor eventually wins.

    How can you tell that growth rates do not decrease then increase?

  • http://profile.typekey.com/tim_tyler/ Tim Tyler

    The idea that innovators should be paid “the full value gained by others” for their innovations seems likely to result in huge overpayments to me. I would rather see sponsorship of inventors based on their expected future productivity.

  • Douglas Knight

    This list seems easily produced from a point of view that I might call economics (anyone have a better name?). Moreover, I doubt that it is of much use without that point of view (eg, Tim Tyler can’t read it). I think it would be more useful as part of an explicit attempt to teach the point of view by example.

  • http://stconsultant.blogspot.com/ John Daly

    This seems like a pretty good list. Of course, not all innovations are “better part or structure designs”, some are worse and some are merely different. I think systems may have control algorithms as well as parts and structure, and innovations may be in those algorithms as well.

    A system is of course in the eye of the beholder. While an entrepreneur may think of his enterprise as a system, an economist might think of it as a subsystem in a larger system such as an industry. One of the imnteresting aspects of the information revolution is that innovations have made many systems smaller rather than larger. Thus firms have innovated by outsourcing functions and thus simplifying their systems and focusing on core competencies.

    Innovations in organizations can also occur in ways people seldom seem to recognize, such as by hiring a new person, or replacing one employee by another.

    I too have been thinking in this way for a long time, but can not judge how wide spread that thinking is.

  • Rob McClary

    As James Q. Wilson points out in Political Order in Changing Societies, innovation within an organization involves 2 acts; the proposal of innovative idea and the adoption and implementation of the innovative practice. The rate of proposal is directly proportional to the diversity of an organization. However, the rate of adoption of those innovative ideas proposed is inversely proportional to the diversity of the organization.
    Diversity in this case is a measure of the complexity of task structure, control mechanism, and incentive system.
    His views make complete sense to anyone who has tried to get bring about change to a business, political party, or military staff. The greater the number of entities in an organization, and thus the greater number of interactions and relationships, the greater chance that new combinations of ideas and methods will be thought of. However, the greater the number of entities, interactions & relationships within an organization, the greater the chance that there will be enough opposition to the change (perceived loss of status, resource, or power) to prevent adoption / implementation.

  • http://profile.typekey.com/tim_tyler/ Tim Tyler

    I can read. IMO, the problem is with the phrasing that “innovators are not paid the full value gained by others”. Should the tax office regurgitate its sales taxes for the benefit of innovators? What about the wages of manufacturing/sales/marketing men involved in realising and propagating innovations? If this is an ideal, it seems like a silly one. Also, it ignores the whole controversy over intellectual property – and assumes something like the current legal situation.

  • http://hanson.gmu.edu Robin Hanson

    Silas and ad, for many specific devices, such as the steam engine, researchers have done detailed inventories of the value added from each specific innovation to that device.

    Phil, many people like to affiliate with innovative people, and so many people try to appear innovative.

    John, yes of course systems are part of other systems, and a control system can be a part of a larger system.

  • http://brokensymmetry.typepad.com Michael F. Martin

    Looks like you’re articulating a lot modularity theory, which I know Henry Smith at Yale has been working on in the patent law field within legal academia.

    But I would say that most of this stuff isn’t accepted as obvious by practicing lawyers, economists, or business people. More specifically, I would say that almost all of the points you make are worthy of a separate blog post, if not a separate academic paper.

    The larger problem that I see is that the dominant theories of innovation (and patent law) are supplied by economists, and economists since the 1940s haven’t had a good set of tools for analyzing dynamics. Rather, they’ve been cobbling together models of very rich phenomena by using a bunch of coefficients that don’t have apparent physical meaning. Paul Romer’s work on spillover effects is an example of a success with this type of technique.

    …but as every good physicist knows, sooner or later you have to go back and grapple with the underlying dynamics.

    I would go so far as to say that even general observations about how and how much innovations decrease the period of time that it takes goods or services to be supplied would be new for a lot of people in the field.

    You don’t have to have a whole complex theory of modularity to make improvements in the existing theory of innovation.

  • http://hanson.gmu.edu Robin Hanson

    Michael, can you be more specific about what claim I made “isn’t accepted as obvious” and why?

  • Ryan

    It probably goes without saying, but read Schumpeter! He has had the most to say about the nature of innovation and entrepreneurship.

  • http://brokensymmetry.typepad.com Michael F. Martin

    So the first thing to say is that the terminology of modular systems design is completely unfamiliar to anyone in accounting, economics, or patent law. This is no doubt part of the reason for why what seems obvious to you hasn’t diffused into these other fields. I’m actually performing an act of translation here in reading your claims, which I’m not altogether confident I’m doing correctly. But here’s my understanding of the state of the art in theory in the economic theory of patents (if not endogenous growth theory more generally):

    “1. Systems are parts in a structure; innovations are better part or structure designs. An innovation embodies insights whose value depends on a context, and so changes with that context.”

    The first second is already obvious to everybody. The second is not so much. For example, there isn’t a good appreciation (much less quantitative theory) for how much cross-disciplinary collaborations contribute to the rate of new invention. Within legal academia, the best theory I’ve seen has been done by Paul Heald and Scott Kieff, both of whom have a good grasp of how the alternative of trade secrets to patents is costly both from an institutional perspective AND from a perspective of decreased spillover effects. Among academic economists, I’m still trying to figure out myself what state of the art is. But I know that people are saying that Paul Romer will win a Nobel for his work modeling “spillover effects.” So maybe you will consider this adequately covered. I don’t because I don’t think modeling back from a macroeconomic level is as useful in developing insights into institutional design as developing an understanding of what generates the macroeffects at a micro level.

    “2. Most growth in the number or size of large systems has been due to collecting innovations.”

    Does this mean human capital development has been the biggest driver of economic growth in economies? As with 1, this point is generally appreciated, but witness the fact that a “Journal of Human Capital” didn’t exist until last year and you’ll see what I mean about how more work could be done.

    “3. Wars, quakes, and diseases may be distributed so most impact comes from the few largest instances. In contrast, in large systems most innovation value comes from many small innovations. Even big innovations require many matching small innovations to be viable.”

    I’m not even sure I agree with this. I would say that the invention of the printing press and the internet are as catastrophic (in a positive way) as large wars, quakes, famines, or diseases in their consequences for human capital development.

    “4. Innovation rates increase when early innovations make it easier to pursue later innovations, and decrease when the most valuable easiest innovations tend to be pursued first. Steady (exponential) growth suggests that these factors roughly cancel. Since growth rates commonly increase then decrease, usually the second factor eventually wins.”

    You’re describing what I think Schumpeter would have called waves of creative destruction. Since he published “Business Cycles” in 1939 and Harold T. Davis published in 1941, I don’t know that many economists (and certainly no patent lawyers) have given much thought to how the phenomena you describe could be modeled with time-frequency distributions. In fact, I think we will have to convince academic economists to introduce a new fundamental hypothesis (of periodicity) into economic theory before we make any progress in arguing for time-frequency distribution models as an alternative to the static models of aggregate supply and demand or time-series analysis that have been in vogue for the last fifty years. As far as I can tell, the good academic economists have developed a number of ways of cobbling dynamics onto their mental models for supply and demand (e.g., by imagining how elasticity shifts in time). But nobody has gone back to first principles and connected time-varying patterns of consumption and production up to the supply and demand curves that are taught to be fundamental in intro to Econ. What a blind spot!

    “5. Innovation in large systems comes mostly from part innovation, so system innovation is steadier than part innovation, and the largest systems grow steadiest.”

    This is already obvious. You’re describing how economies of scale diminish as scale increases. We’ve already got that one down. Still, I would say that there is a kind of prisoners’ dilemma that startups and large corporations face in negotiating technology transfer that hasn’t been solved yet. For example, what if either the startup or company with economy of scale refuses to deal on reasonable terms for technology transfer? Now the startup will wastefully repeat the process of establishing manufacturing capacity, and channels of distribution and marketing that have already been established by the large company. I realize that this is not the point 5. But if more startup CEOs (and their investors) had point 5 vividly in mind, then there might be less wasteful competition. But they are also quite reasonably worried that the large company will not pay enough if they don’t threaten to compete head-to-head. Stronger patent rights would obviously help solve the problem. Paradoxically, many startups and VCs are AGAINST stronger patent rights now.

    “6. System structures vary in how well they encourage and test innovations locally and then distribute the best ones widely. Better structures for this are meta-innovations.

    * Good modularity reduces the need to match innovations in differing parts.
    * Good abstraction puts similar innovation problems within the same part.”

    The way I understand what you’re saying here is that inventions in institutional design are more important than other product or process inventions. That might seem obvious. And the first subpoint about modularity is obvious — that’s like saying that standards-making bodies are going to promote commercialization of inventions.

    The second subpoint (about “abstraction”) is not obvious. This seems to go back again to the point about how cross-disciplinary work is beneficial to promoting innovation.

    I would say that this point is even more important in the field of education theory than it is in patent law or economics. Most universities are balkanized into various departments. It’s not a trivial point to observe that every department would benefit from spending more time translating their work into terms that all other departments could understand. At least not for educators. Incidentally, I think this is why legal academics tend to pick up on new ideas in their literature quicker than other academics. It’s because they’re trained to “translate” whatever they read in the sense that they’re trained never to assume anything about the definitions of the words the authors of the texts that they read are using. No two judges ever have the same thing in mind when they talk about “fairness.”

    “7. If a barrier isolates two systems, the faster growing one eventually dominates. A system that better promotes innovation can lose to a system with a larger source of innovation.”

    This point in particular is underappreciated by politicians and economists considering the question of immigration. You could right a very important paper about how our immigration policy could end up putting us behind the rest of the world.

    “8. In large innovation pools, similar innovations commonly arise from several semi-independent sources at nearly the same time. No single source was essential.”

    This point, if anything, is overstated at the moment. A version of this appears in the recent Malcolm Gladwell piece about Intellectual Ventures. It’s somewhat true, as the history of Interference proceedings in the U.S. demonstrates quite vividly. But if you dig into the details of the history, you’ll find that inventions are as distinct and original as inventors.

    “9. Current human society can give incentives to innovate too much, when innovation is used to signal, and to innovate too little, when innovators are not paid the full value gained by others.”

    Although I don’t understand the subtleties of your theory of signals, I think I understand them well enough to say that that part of 9 is not obvious. Even the second part (substituting “full” for “part” as an earlier commenter pointed out is necessary) is not obvious — although it was to the Venetians in the 15th century, the Founding Fathers in the 18th, and Abraham Lincoln in the 19th century. The current modes of thought on incentives for innovation seem to be (a) let the government fund it; or (b) we don’t need incentives. (a) is wrong for reasons that people at GMU understand. (b) is wrong for reasons that most economists understand. And politicians now too since so many Ph.D.s are leaving the U.S. after earning their degrees.

    One of the things I thought was so odd about being in grad school was how so many large corporations funded grad students. I thought, this is a rather mercenary move on their part, oversupplying the market for Ph.D.s so they can then underpay them the rest of their careers.

    Well it turns out that some of those Ph.D.s don’t have to live in the U.S. and be underpaid.

  • http://brokensymmetry.typepad.com Michael F. Martin

    1-7 and 9. NOT 8.

  • http://youtube.com/leearnold Lee A. Arnold

    Robin Hanson, some of this is standard cybernetics, starting from Norbert Wiener and Ross Ashby. There is a huge literature by now, and I have some questions and comments for you:

    “Systems are parts in a structure” In this and what follows, your definitions appear to be contradictory: “systems are parts in a structure” (in #1); “innovations are better part or structure designs” (#1); there are “part innovations” and “system innovations” (#5); there are “system structures” (#6).

    Exactly how are you using the word “system” in all of these?

    Wouldn’t it be better to say instead that systems are certain structures that have separable but interlocking parts, there may be separable subsystems, and that innovations are changes among the parts?

    #1 “innovations are better part or structure designs” Why are innovations “better?” The answer is, usually because they REDUCE the space, time, or personal physical effort that is required to effect a transfer, transformation, transaction, or transportation between two parts of the structure. There is a reduction or condensation or concentration quality to it.

    At the next hierarchical level up, this condensation may be seen as a specialization or differentiation, to be integrated with others, again via a larger type of transfer, transformation, transaction, transportation, in that higher level.

    #2 The growth in systems is due to both (a) “collecting” innovations, and (b) the simple enlargement or increase of the system, including by accretion. A example of (b) is population growth, such as in societies or ecosystems, or the “opening” of new markets. It is hard to maintain that “most” growth is due to innovations, unless you restrict the discussion to certain systems along certain time scales: e.g. modern economic growth, or the eons of evolutionary time.

    #3 In the case of institutions, which are homologous to innovations but instantiated at a different (usually higher) hierarchical level, the value often proceeds almost entirely from the single BIG change. Private property, for example, enables the easier transformation of resources and easier transfer of possession among individuals — which Locke was on to. There was of course a host of “lesser” innovations to attend to private property, such as contracts and so on. But your point was about where most “innovation value” comes from, and institutions are a type system-wide innovation with an enormous importance and value; they are a sort of technology at the social level. (You may dissent from likening private property, or any institution, to an innovation, but then you cannot define innovation as “structure design” in #1 — at least, not for all structures.)

    #4 I wonder if you would please give an example, real or theoretical, from social systems or biology, where innovation rates “decrease when the most valuable easiest innovations tend to be pursued first.” I am unfamiliar with this. Are you limiting this to a single technological lineage, as opposed to a whole system of them? Is there an example from biological evolution?

    #5 Again, what characterizes “system innovation” in this list of basics? I.e., how and when do you say that the whole thing has changed?

  • http://hanson.gmu.edu Robin Hanson

    Lee, most long term population growth and opening of new markets is due to innovation. Private property is a mix of lots of small innovations, not just one big one. It is trivial to describe a theory where agents choose the easy innovations first.

  • Tracy W

    and to innovate too little, when innovators are not paid the full value gained by others.

    But if innovators gain the full value their innovation creates, then what’s the point of encouraging innovation? An innovator creates something that makes $100 million for the world economy as a whole, they get paid $100 million, so the rest of us are no better off than we were before. Furthermore, if any gains I get from an innovation simply go to the innovator, why bother adopting the innovation in the first place? Why buy a more efficient washing machine if the capital cost of the washing machine plus payment to the innovator is equal to the discounted value of the energy saved? Why buy a new medicine if I’m just going to be taxed enough that I’ll be as miserable as I was without the medicine? So, under your scenario where the innovator is paid the full value of their innovation, there is in fact too little innovation as no one else has an incentive to make use of the innovation in the first place.

    The only reason it makes sense to provide incentives for innovation is that the innovator does *not* capture the full value of their innovation.

  • http://youtube.com/leearnold Lee A. Arnold

    Robin, thank you. It didn’t occur to me until after posting my comment that you are restricting the use of “systems” to “economic systems,” and “growth” to “per capita growth.” Because surely, long-term growth in wildlife populations, or even bacteria, is due to reproductive rates and environmental conditions, which are not innovations — unless you insist upon the fact that these things have evolved to do what they do; i.e., you make “evolutionary emergence” (“innovation,” in biology) an obvious precondition of any statement; which would be trivial and redundant.

    Even so, there is an exception to your basic rule in economic systems: comparative advantage. Here, each side can concentrate in the pre-existing differentiation where it has the best efficiency. The theory does not require the next step of an additional innovation to increase per capita income. After that, of course, growth is a steady alternation between innovation and trade.

    Indeed I meant “opening of new markets” in the older usage of the term, i.e. of exploration or contact with other countries and trade liberalization — opening a foreign country to trade, having more people to sell to, the prelude to comparative advantage.

    On #4: with regard to a theory where agents choose the easy innovations first, can it be other than trivial? It is a tautology that, if innovation rates decrease, the “most valuable easiest” innovations were pursued first. But to write that the rates decrease “when” this is done, made me think this was more than mere definition.

    I also did not realize that by “innovation value” (#3) you meant immediate monetary economic value, and not (as any old systems theorist would naturally be thinking!) something like what economists would call “positive externalities” or “network effects.” But then how do you define “big” vs. “small” innovations? I think you are incorrect about the specific importance of private property in the web of institutions under its rubric — but let’s take another example. It seems to me that a big innovation was the telephone, and I would define it as big because of the extraordinary positive externalities, leading to an incalculable chain of growth and profits over a centruy. Yet you write that “most innovation value comes from many small innovations” (#3). So how do you draw the line between “big” and “little,” in technological innovations?

  • Danni M

    I have found the work of Charles Leadbetter interesting on this subject (there is a Youtube video but here is his personal website)

    http://www.charlesleadbeater.net/NotaBlog/blog-2008.aspx