Voting v. Counting, and Payouts in CLs

1.  Aaron Swartz suggests using anonymous, encrypted voting to distribute CL money, rather than allocating funds based on estimating actual usage. His proposal isn’t entirely unique.  Professor Fisher notes two similar ideas on pages 55-60 of his CL regime: the Blur/Banff proposal and Peter Eckersley’s “Virtual Market for Virtual Goods” paper. 


Fisher suggests that there would be deficiencies beyod the encouragement of purposely distorted voting (see Ernest), including increased hassle for users and higher likelihood of privacy violations. Frankly, I’m not quite sure how a website operator recording everything downloaded from his site and then reporting it to the Copyright Office is much different from my telling the Copyright Office directly what music I’m using using anonymous voting like Aaron describes. In both cases, the data would need to be recorded and reported anonymously, like Aaron explains; moreover, both cases cause concern inasmuch as we don’t trust others to record and report information anonymously.


2. In going over all these proposals, I began to think again about two different ways of doing the payout: one-payment-one-consumption using a fixed rate or payments according to relative consumption (that is, if you get 75% of the downloads, you get 75% of the tax revenue). The two are similar in that, if the fixed rate is based on 100 total payouts, and the total number of payouts is 100, and your work is downloaded 75 times, you’re going to get 75% of the money in either system.  However, I see two possible meaningful differences:



1. The latter method would not risk running the money well dry because every payout is related to the total tax revenue. If payout goes up with each consumption, you risk not having enough money to pay everyone.


2. The latter might deal with methods of gaming the system slightly better. Let’s say everyone gets 5 people to ballot stuff for him/her.  With a proportional payout system, this would not affect anyone’s payout. But, if you payout per consumption, then everyone gets five more payments than they ought to have received.  I noted this in the pho discussion, and Ernest voiced his doubts whether such “cancelling out” would occur a significant amount of the time.


Perhaps this is an unimportant aspect relative to the other complications of CLs, but I figured I’d throw it out there again. Any thoughts? Are these distinctions incorrect?


3.  Speaking of how you payout, I’m curious about how Aaron thinks that aspect will work.


If everyone has a different amount of certificate money, then does that alter how many downloads one can make? If so, wouldn’t that defeat the purpose of the CL?  If not, wouldn’t that be rather unfair to the artists? I pay in 2 dollars in taxes, download 1000 songs of Artist X (and nothing more), and give him my two dollars. Aaron pays in 20 dollars, downloads 5 songs of Artist Y (and nothing more), and gives him all 20 dollars.


On the other hand, if everyone would have the same amount of “certificate” money, then would this fairly pay out to artists?  Say I download 100 songs by Artist X and Aaron downloads 50. Because it’s all we download, we give all our certificate money to Artist X.  Now let’s say Ernest downloads 150 songs of Artist Y and gives all his money to him. In a one-payment-one-consumption model based on 300 consumptions, Artist X and Artist Y receive the same amount. In a proportion model, they receive the same amount. But in Aaron’s, Artist X receives twice as much as Artist Y.  Is that entirely fair? What if the respective downloads were 15, 5, and 300?  In the first two models, Artist X gets substantially less than Artist Y; in the latter, Artist X gets twice as much as Y. 


In sum, is it fair to totally ignore volume of consumption? (alternate question: is my math completely incorrect?)

One Response to “Voting v. Counting, and Payouts in CLs”

  1. Gordon Mohr
    September 18th, 2003 | 1:25 am

    I think you’re a bit too fixated on making payments match consumption events. A person could vote their share that way; popular software and hardware might offer that as a default. But it’s not a necessary step: for any reason whatsoever, a person may intentionally vote their share however they like. Perhaps they hear one song more often, but another is more meaningful to them. Perhaps they make a judgement that a certain creator “needs” the compensation more.

    By not letting any behind-the-scenes, game-able formula reign supreme, it becomes less likely (IMHO) that “fake plays” and backroom deals adjusting the sampling rules at the compulsory agency subvert the system. Petty fraud would still occur, but no system is perfect.

    An added benefit of a flexible, audience-driven voting system is that the same official clearinghouse for routing the compulsory fees could also be opened for voluntary overpayments: “I liked that work so much, I want to throw another $10 towards the creation of more like it.” Most compulsory systems, by hiding the payment and the allocation, encourage complacency — audiences can feel like everything is done for them, they never need directly connect with creators. Individually votable shares restore the primacy of direct, intentional audience-to-creator relationships.