I have written a decent amount about the concept of nonrival consumption. It is the economic characteristic that distinguishes public goods from private goods. Ideas are nonrivalrously consumed because the marginal cost of allowing another person to access and consume the idea is zero. Sure, the costs of distributing ideas and explaining them may be positive, but it costs nothing to allow another person to consume an idea because ideas do not deplete – – that is, if you consume the idea, there is no less of it for me. (Jefferson wrote a famous paragraph on this that people like to quote.) Apples, on the other hand, are another story.

Now some disagree with the notion that ideas are nonrival, and I have always been puzzled by this. Typically, they conflate consumption and distribution or something like that. But here is a somewhat different approach; it occurred to me that maybe what they are doing is focusing on the value generated by the idea. If we focus on the value or surplus generated by the production of an idea, that value is finite (even if potentially dispersed across many people and well into the future) and thus rivalrously consumed. If you get some chunk of the value, then I cannot get that same chunk. Then, if we allow everyone to access that value without restriction – – e.g., as direct consumers or as competitors seeking to appropriate some value by using the idea to sell products or services to consumers, viola, a tragedy of the commons?

Well, no. Because the value itself is depletable no matter what; it is not a sustainable, renewable resource (unlike the idea, which is sustainable and nondepletable). Overconsumption is not the risk with which we are concerned.

I have more to say on this, and will be doing so in some essays and perhaps on the blog, but let me see if this sparks some comments…

11 thoughts on “Ideas and (non)rivalry

  1. Mark McKenna

    Brett –

    I’m with you in sentiment, but I don’t think it’s really true that the value of ideas necessarily is depletable (assuming you mean “ideas” broadly). Even if we don’t think that the ability to maintain value (artificially) is not a legitimate goal, is it not true that Coca-Cola has been able to maintain, and build the value of its mark? And what about the value of Disney’s stories?

  2. Mark McKenna

    oops, that should say *is* a legitimate goal

  3. Brett Frischmann

    Well, I am not so sure. I agree Disney can continue to generate and appropriate more value from its stories as more people continue to read them and watch them and buy things based on them and so on. There is plenty that has been appropriated and tons left to get in the future. The point is simply that if I appropriate the value you get from reading a story (say I sell you an infringing copy of a story), Disney cannot get that same chunk. In this sense, ideas are nonrival but value ($) is rival.

    My point is really that the argument that ideas are somehow rival is wrong and that the argument tends to inappropriately conflate the idea with other things that may be rival (distribution devices or perhaps simply $).

  4. Kathy Strandburg

    I’m kind of late responding here, but, as Brett knows because we’ve been talking about this, I’m one of those people who think there may be some hidden rivalrousness in ideas. Here’s why: I think that rivalrousness is not a property of things per se, but of uses of things. Thus, the proverbial apple is certainly rivalrous with respect to eating. But suppose you want to paint the apple? Or just enjoy looking at its beauty? (I don’t like apples, so I am definitely in this camp.) Those uses of the apple are non-rivalrous.

    Now, as to ideas. Ideas can be used in many ways. Is it possible that some of these uses are rivalrous? For example, if I want to use an idea for my PhD thesis, no one else can do so. But if I want to just think about it, that is clearly non-rivalrous.

    Now how are ideas different from apples so that we tend to think of them as inherently non-rivalrous? Ideas generally have many more and more varied uses than apples. And more of those uses are non-rivalrous.

    But maybe not all.

    By the way, I agree with Mark that the value of ideas is often not depletable. But I think this question of depletable/non-depletable is bound up with the question of how the idea is being used.

  5. Brett Frischmann

    Well, my post was intended to get Kathy commenting, so I am glad. We have been chatting about this recently and there is more to tease out I think. I agree that it is the use of something that matters and, not surprisingly, that the variety of uses matters as well. Further, it matters whether we are talking about consumption or use as an intermediate good (as an input). Apples may be consumed and they may be used to produce views or paintings. (Ok, I admit this seems a bit tortured. But when you view or paint an apple, what is the consumption good, “the thing” being consumed, the apple or the view?)

    The rivalrousness question turns on whether a person’s use diminishes the consumption opportunities for others (or, put another way, on whether the marginal cost of allowing another to access and use the good is positive [rival] or zero [nonrival]). I don’t see how ideas are rivalrous in the same way as apples, though. Whether consumed for internal gratification or enlightenment or used to produce something else, ideas remain fully available for anyone else to consume or use in the exact same way – – absent some socially constructed resriction. I understand that doctoral programs would not give credit / value for the use of an idea also in another person’s PhD thesis; in a sense, you could not appropriate the value in the same way because of the socially constructed system of credit / reward. But I am not sure I see that as an example of rivalrousness (anymore than if we say that it was rival because someone obtained a patent on it). Anyone could use the idea to write a doctoral thesis, but not likely for credit / value. So the appropriable value may dissipate and the marginal benefits may decrease, but the marginal cost of allowing another to access and use the good remains zero.

    This was a quick ramble – hopefully it makes some sense

    thanks Kathy!

  6. Kathy Strandburg

    Brett’s response raises the question as to whether “using” and “consuming” something are different from “appropriating value” from it. If “use” and “consumption” are really just means of appropriating value, then the rivalrousness question is whether another person’s use diminishes my ability to appropriate value. In the apple example, my eating certainly does that — my viewing does not.

    Does use mean something more than appropriating value? To say that more than one person can use an idea to write a doctoral thesis isn’t really right. The second person can write something, but it won’t be a doctoral thesis. The “doctoral thesis” use is gone even if other uses are still available.

    It’s an interesting question the extent to which it matters that restrictions on re-use are socially constructed. Certainly it matters for social policy, since socially constructed restrictions potentially could be changed or removed. But does it matter for determining whether the use is rivalrous? (After all, marginal costs are to some extent socially constructed, aren’t they?)

  7. Brett Frischmann

    well yes, here we go. we need to seperate costs and benefits of access and use. rivalrous consumption turns on the marginal costs of allowing another person to access and use the good; it does not turn on the benefits of doing so. Appropriating the value of an idea relates to the marginal benefits of allowing another to access and use – while the marginal benefits may decrease as value is appropriated, that does not mean that the marginal costs are positive or increasing.

  8. Kathy Strandburg

    Of course, Brett, you’re quite right to point out that marginal costs and marginal benefits are different things and that we should be careful to distinguish them. However, I think trying to do this helps to point out the conceptual difficulty here. In order to talk about either marginal cost or marginal benefit, there are important questions of units to be resolved. (Ah, units, physicists love units . . .)

    So, if I talk about marginal cost, I mean the cost of producing one additional unit of something. If I am to compare marginal benefit to marginal cost, I have to talk about the benefit of one additional unit of the same thing. But once I bring in these units, I think I must define the use (i.e. means of appropriating value) that I am talking about.

    So if I talk about apples in the usual sense, marginal cost is dc/dq, where c is the cost of producing q apples and q is the number of apples I produce. Marginal benefit is db/dq, where b is the benefit received from consuming q apples. OK, now what do I mean if I say that the marginal cost of looking at the apples (rather than eating them) is zero? What I mean is that I am now thinking of a different “q”. So now, q is not the number of apples, but some measure of “amount of viewing”. So dc/dq is zero because there is no additional cost to “more viewing”. But db/dq is positive because there is additional benefit to “more viewing.”

    So I think we can say that non-rivalry means that, for some particular “q”, dc/dq is zero, but db/dq is not. To answer this question, we must specify what “q” we are talking about. Thus we must be talking about a specific means of appropriating value.

    OK, so now if we think about ideas, the issue is “What is q?” Take the Grokster-type situation. There q is something like “number of people who have a copy of a particular song”. dc/dq approaches zero because copying and distributing costs are so low. db/dq is positive. So the “song” is approximately non-rivalrous for the use of having a copy.

    OK, now what about the dissertation idea? If q is “number of people who know about this idea”, the dc/dq is approximately zero, db/dq is positive and the idea is non-rivalrous. But if q is “number of PhD dissertations based on this idea”, then the cost of producing another PhD dissertation is not zero (in fact, I guess it’s infinite, since duplicate PhD dissertations are forbidden?).

    So, while it’s true that marginal cost and marginal benefit are distinct, they are both dependent on a particular means of producing/consuming/appropriating value.

  9. Brett Frischmann

    I understand what you are saying, but am slightly confused. q is the quantity of some good. It can be apples, a view, an idea, etc. I agree with you that the difficulty is identifying the good (or better yet, the “thing”) that we are producing and consuming.

    You lost me on the “What is the q?” analysis for Grokster-type situations and a dissertation idea. Perhaps we are just focusing on the marginal costs and benefits of different things. I do not think rivalrousness of consumption turns on the marginal costs of distribution, except I suppose if the “good” is defined as distribution itself, as you seem to have done with the song example.

    I do think that the marginal costs of distribution matter, don’t get me wrong, but I think that we should be careful not to conflate different types of supply-side costs. I think doing so obscures the meaning of nonrivalry, as I explained in my article on infrastructure.

    Perhaps I have been tricky in avoiding this in the past, but I have tended to say someting like this (which I just grabbed from a draft):

    Nonrivalry describes the sharable, nondepletable nature of a resource: sharable in the sense that the resources can be accessed and used by multiple users for multiple uses at the same time; nondepletable because consumption by one user does not reduce the quantity of the resource available for other users to consume.

    Nonrivalry is a function of the marginal cost of allowing an additional person to consume a good. When *this* marginal cost is zero, consumption is nonrivalrous. Information resources are nonrivalrously consumed and thus, once produced, are infinitely sharable. Of course, even though additional costs do not have to be sunk to produce information for each additional consumer, there generally are other supply costs involved, such as the cost of distribution[1] or the cost of absorption[2].

    [1] The marginal cost of distribution depends on the medium of communication. The combination of digital technology and the Internet has drastically reduced distribution costs for many types of information resources. Nonetheless, distribution costs remain relevant in most cases.

    [2] The marginal cost of absorption depends on whether a user or consumer has the capacity to appreciate the benefits of an information resource. For example, while the marginal costs of allowing another person to access and use Einstein’s Theory of Relativity may be zero, the cost of educating someone to the point that access and use yields appreciable benefits may be substantial.


    All this being said, I completely agree with you that we need to pay more careful attention to the different ways in which resources can be put to use to generate value. This gets into the demand side analysis that we have been chatting about and working on. But that is another post.

  10. Kathy Strandburg

    Yes, q is the quantity of some good, but I think that may be the whole crux of the issue. If dc/dq = 0 is to be the definition of non-rivalry, then c(q) has to be the cost of producing one additional unit of consumption. Thus, some “things” have more than one “good” associated with them — more than one way of being consumed and thus more than one c(q) function. For viewing the apple, the “good” (the “q”) is not the number of apples. Perhaps it is “people who can view.” Producing more apples costs something. But producing more views does not. So, there are different q’s (and different c(q)’s) for different types of consumption.

    In the song case, what is the q? Maybe it’s number of people who can listen to the song. Or maybe it’s number of people who can “own” the song. (Those are two somewhat different ways of consuming a song.) But we have to define a q if we want to talk about non-rivalry.

    In the idea case, I would argue that there may be different “q”‘s, different consumption goods, for different ways of “consuming” an idea. “Thinking about the idea” is one way of consuming it and if “q” is “number of people thinking about the idea” it is certainly non-rivalrous. But I’m not yet convinced that is necessarily true about all ways of consuming the idea (e.g., the dissertation).

  11. Brett Frischmann

    When we are talking about viewing an apple, the “good” is the view and the apple is merely an input into producing the good. The nonrival character of the view would be the same regardless of what you are viewing – i.e., replace apple with sunset, toothbrush, house or whatever else and the fact that consuming the view leaves as much of the good for everyone else to consume remains intact. So yes, a different good and q and dc/dq.

    If we define the good to include distribution (or actual access), then I agree with you that nonrivalry seems to disappear.

    If someone paints a particular scene – say of a house at dawn, it may be the case that that painting will effectively appropriate the commercial value [of that particular use of the house/scene] because it fully satisfies the demand for paintings of that house at dawn. Still, the view and the house remain available for all others to “consume” in the same way (painting). The benefits of doing so may have decreased significantly, perhaps to zero, because the first comer preempted the second, but the costs don’t change.

    Put another way, the costs of the inputs don’t change although the value of the output does.

    I am still trying to see how your approach fits in with mine because I do believe the multiple uses of ideas and other nonrival resources drives the analysis. I just do not see how the rivalrousness of the [downstream] uses render the input itself rivalrous.

    Ok, I need some more coffee…

Comments are closed.