July 19, 2024

Devil in the Details

There’s been a lot of discussion lately about compulsory license schemes for music. I’ve said before that I’m skeptical about their practicality. One reason for my skepticism is a concern about the measurement problem, and especially about the technical details of how measurement would be done.

To split up the revenue pool, compulsory license schemes all measure something – some proxy for consumer demand – and then give each copyright owner a share of the pie determined by the measured value. Most proposals require measuring how often a song is downloaded, or how often it is played.

Most compulsory license advocates tell us what they want to measure, but as far as I know, nobody has gone into any detail about how they would do the measurement. And based on the thinking I have done on the “how” question, there doesn’t seem to be an easy answer.

So here is my challenge to compulsory enthusiasts: tell us, in technical detail, how you propose to do the measurements. You don’t have to give us working code, but do tell us which programs you would write or modify, and what specifically they would look for. Tell us how you would cope with backward compatibility, and the diverse formats in which people download and store music. Tell us how you would deal with non-PC platforms such as Macs, Linux boxes, and iPods, as well as non-traditional network setups such as public WiFi access points.

The devil is in the details; so show us the details of your plan.


  1. The CPCC is a model of corruption more than anything else. It only “began” distributing money (4 years into its life!) after threats of lawsuits by artists for failure to distribute the money. Snort! 72 million bucks in, 23 million out. The sort of thing that makes one go “hmmmm”, eh?

    Furthermore, the CPCC has lately been taken to cranking up the “levy”, as well as adding a “levy” to essentially everything that can store and transmit bits. For the children, of course. It takes no rocket science to figure out what is happening here…

  2. Apparently Canada has had a system in place since 1999, and which began distributing royalties this year. Why not begin by examining their mechanism?

    See http://www.infoworld.com/article/03/12/15/HNcanada_1.html

    I quote:

    “The board also imposed a levy on manufacturers and importers of MP3 music players with non-removable memory, depending on their storage capacity. Players with memory capacity of less than 1G byte will be subject to a levy of C$2 (US$1.52), those with between 1G byte and 10G bytes will be levied C$10, and any players with more than 10G bytes will face a levy of C$25.

    “The levies are paid to the Canadian Private Copying Collective (CPCC), which distributes the money to authors, performers and producers of music in Canada.”

    See the website for the Canadian Private Copying Collective at http://www.cpcc.ca/english

  3. The music industry had a perfect system in place for recording downloads, but they destroyed it.
    Napster search requests would probably have correlated almost perfectly with downloads, and even if it didn’t, it would have been trivial to modify the protocols to relay the download request through the server.
    The RIAA decided to destroy Napster, and any other centralized music distribution system. This directly lead to the development of the decentralized P2P systems in place now.
    Now that there’s no central server for file sharing, it’s going to be extremely difficult, if not impossible to accurately measure music file sharing.

    One possible soulution might be to convince people to go back to a napster type system. In order for that to work though, the new system is going to have to beat all existing P2P systems in a number of areas.
    File quality.
    Broad selection of music
    FREE downloads of both the music and the sharing program.
    No DRM of any sort on the downloaded files.
    No spyware included with the sharing program.

    The first one should be easy for the RIAA to arrange, but unless they can get together and beat existing P2P systems on all of these features, centralized file sharing is dead, along with any sort chance of accurately recording download statistics

  4. Cypherpunk (Lochner?), using old-media profits for dividing is an attractive idea but has two serious flaws:

    1. If the system takes away a large portion of sales (especially if it does so disproportionately) the numbers quickly become inaccurate. This seems quite likely to me.

    2. It eliminates one of the major benefits (for me, at least): freeing artists from those who control the distribution channels. That is, if we somehow counted actual playings, artists could try to publish songs on their own and see recoup the profits if they were successful. But if old-media sales are required to make money, they will likely be forced to hand over their copyrights or sign onerous contracts with RIAA members.

    All the other concerns raised here have been amply addressed, I think. (Although we should probably do a better job of publishing these results to people other than those who went to the conference.)

  5. There are problems, but I don’t think this is one of them.

    Tax the monthly cable modem bill, when the bill is paid, email a blinded token to the user. Have the token automatically opened by the MP3 player (e.g. iTunes). iTunes keeps track of what songs are played and coordinates with the iPod to make sure that the counts are up to date across both devices. At the end of the month, it anonymously sends in the token and the playcount records.

    Simply generalize this to all MP3 players and portable devices (the devices have to get the MP3s off the Internet somehow!), with some sort of either market- or government-based encouragement to get providers to release and promote updates. (Get Your Vote Counted! Install the free upgrade today.)

    What am I missing?

  6. Of these, I like David’s solution the best: divide up the compulsory licensing revenues exactly in proportion to physical and online music sales. This keeps the market in operation and sends the most money to the artists for whom people are willing to vote with their pocketbooks.

    I don’t think any system will work that relies on monitoring P2P activity. It’s too easy to fake and hide, and there are no incentives for honesty. Maybe you can catch webcasters and make them pay licensing fees, but if shutting down unlicensed P2P were that easy, we wouldn’t be in this situation in the first place.

    You still have the problem of determining the size of the compensation pool, and funding it. Music becomes a public good. Ultimately this probably has to be a political decision made by the government through its usual processes.

  7. And don’t forget the other side of the problem: creating that wonderful pool of money that is going to be handed out to needy and deserving artists. Where is that going to come from? How do we decide how much of society’s resources go to music vs education vs defense?

    And is this going to mean that people get taxed on their P2P downloads? What if they try to evade the tax by encrypting their communications?

    You’re pretty tough with your skepticism when it comes to DRM. You flat out say that it won’t work. I’d like to see you judge these schemes on an equal footing with DRM. Don’t let your ideological preferences get in the way.

  8. fuzzy gorilla says

    Considering the fact that the current webcasting rate rules already require a webcaster to keep track of the artist, the song title, the album title, the label, the ISRC, the copyright owner, the aggregate tuning hours (#listeners * #minutes), the start date and time of each transmission, I figure this is already a solved problem. If the webcasters can do it already for royalty payments, then they and peer-to-peer networks can do it for compulsory licenses. Next case please, this is a non-issue.

  9. Actually, the Central Limit Theorem does not apply to this kind of sampling as it is not normally distributed. Think – what are the average sales of a song?

    Naive application of statistics will give radically wrong results, favouring those few who already do well out of recorded music at the expense of the many orders of magnitudes higher number of creators poorly served by the royalty/advance publishing model.

  10. You quite simply can’t track how many times a song is played, but you easely track how many times a song gets purchased or downloaded.

    It seems logical to me that most shops that are offering songs for download will keep statistics of the sales numbers per song (it’s what i would do anyway). Since keeping these statistics happen on the server-side, there are no worries about different file-types or systems.

    Gathering all these statistics would be a severe pain in the ass though, unless some kind of standard system is developed that online music-shops could use. Maybe the easiest solution would be to just determine a standard way of formatting the statistical data (be it CSV, XML, whatever).

    If the number of online music-shops participating would be high enough, the analises of the statistics will be a fairly accurate representation of reality.

  11. The Linux box immediately to my left looks a lot like
    a PC to me! Perhaps you should just say “diverse

  12. > …tell us, in technical detail, how you propose to do the measurements.

    Is it a technical question or a challenge? 🙂

    See, I’ve seen and participated in many discussions on that subject. Sometimes these discussions are educational, sometimes they are amusing, sometimes they are boring, but I have a feeling that ultimately they are – most of all – irrelevant.

    I suspect that the real method of statistics-gathering will be determined by a whole bunch of factors, and the technical details and the precision of the chosen method won’t be anywhere close to the top of the list.

    If the question is: “Is there a bulletproof technical method of gathering the statistics and dividing the money?” – the answer is most definitely: “no”.

    But if the question is: “What method should be used for this purpose?” – then my answer is: “For all practical purposes, almost any one will do”.

    (It is probably entirely possible to hack and subvert, say, Nielsen TV Ratings. But who cares?)

    Best wishes –
    Serguei Osokine.
    10 Dec 2003.