Big Tech’s Latest Infringement Loophole: Mass Filings of NOIs to Avoid Paying Statutory Royalties (Part 2)

“1 NOI” Means “1 Excel file for the NOIs Filed That Day, each Excel file contains tens of thousands of songs

As noted in Part 1 of this post, Google, Amazon and others are filing what are reportedly “millions” of “address unknown” NOIs with the U.S. Copyright Office.  I fully expect that Pandora will eventually do the same for its on-demand service and Spotify is likely to do the same.  Note–this type of carpet bombing of NOIs would not have helped Spotify in the David Lowery litigation because David Lowery registered his copyrights that are the subject of that litigation.

If you click here, you will find the most recent iteration of these massive NOIs, which apparently are being posted on a regular basis.  The screenshot above is the first page of these filings on the Copyright Office site, most of which came this month (September 2016).

Each Excel file can be downloaded–a word of warning, even the zipped files are large and may take a while to open on an average home computer.

Remember what you are looking at in these files–this is the list that results from comparing the list of sound recordings that the services are using to the data dump that the service purchases from the Library of Congress.  Take a tip–you’ll never find the page on the LOC website unless you know where to look, which is right here.


Willful Blindness on Song Titles

This is an overwhelming amount of data, so in order to have any idea what is really going on, spot checking will be required.  And since it’s Google, you know there’s a scam afoot, your challenge is just to figure out which scam it is this time.  (Of course the entire exercise is a scam, but leave that to one side for now.)

Scam # 1 appears to be treating any song title that has any text in it other than the actual song title as a song for which the owner cannot be identified.  Here’s two examples from Sting in the Google 9/16/16 NOI file:


Of course, the song “Fragile” is registered, but Google’s filing claims that there is a different song “Fragile (Live)” that is not registered by that title.  Google has, no doubt, sent another NOI for the song “Fragile” (or has a direct license) and if so has actual knowledge of the song copyright owner.

And here’s the loophole–by claiming that “Fragile (Live)” is an “unknown” song, Google can try to get out of paying for the live version.  (Because how would you know that “Fragile” performed by Sting for which you know the copyright owner is the same as “Fragile (Live)” performed by Sting for which you now claim to be shocked that is the same song–unless, oh, maybe if you listened to the two?)  The government’s compulsory license says this:


If you search for live recordings in Google’s NOI filings, you will find many, many live recordings by artists such as Bob Dylan, Heart, Quincy Jones, Lynyrd Skynyrd and Chicago.  And then there’s the medleys like “Hotel California Dreaming” which lists the Eagles writers with John and Michelle Phillips of the Mamas and the Papas.

Not to mention a ton of foreign songwriters who are under no obligation to register their songs with the U.S. Copyright Office.

Let’s also set aside for the moment whether the recordings that Google has listed on their certified filing are all lawfully distributed–some certainly look like bootlegs to me.  Of course “bootlegs” these days have to include illegal live recordings posted on YouTube and then stream ripped into mp3 files to be distributed through Tunecore, CD Baby or someone else who doesn’t pay much attention to where the recordings come from and then subsequently distributed through Google–who invented the game.

So what appears to be happening is that Google and Amazon (which has hired MRI, I believe) are playing the willful blindness game.  What can be done about it?

That will be the subject of the next and final part of this post.


Bad Medicine, No Spoonful of Sugar: How Useful is Blockchain for Music?

A Guest Post By Alan Graham of OCL.  Let’s get real. There’s a lot of talk about data and transparency and blockchain these days when it comes to the music industry. This will solve everything! However, there are issues no one seems to want to discuss that also need immediate attention. So let’s open up and take a very bitter […]

via Guest Post: Bad Medicine, No Spoonful of Sugar — MUSIC • TECHNOLOGY • POLICY

In this guest post, UK technologist Alan Graham offers some penetrating insights into what the much ballyhooed blockchain will and won’t do, and some technologies that are not so far over the horizon that we should all be thinking about how to bring into a safe and sane environment.

Are Legacy Revenue Share Deals More Trouble Than They Are Worth?

By Chris Castle

As an important publisher panel observed at MIDEM this year, revenue share deals make it virtually impossible for publishers to tell songwriters what their royalty rate is.  That’s especially true of streaming royalties payable under direct licenses for either sound recordings or songs or the compulsory licenses available for songs.

There are some good reasons why streaming rates developed without a penny rate–or at least some reasons that are the product of sequential thought–but there are also good reasons for creators to be distrustful of the revenue share calculation.  This is particularly true of compulsory licenses for songs where songwriters and publishers don’t even have the right to examine the services books to check if the service complied with the terms of the compulsory license (known as an “audit” or “royalty compliance examination”).

If you thought record deals were complicated, you will probably have to find a new vocabulary to describe streaming royalties.  (Calling Dr. Freud.)  But even under direct licenses for songs or sound recording licenses where there usually is an audit right, the information that needs to be audited is so closely held, so over-consolidated and the calculations so complex that there may as well be no audit right.

The result is that smart people with resources at big publishing houses cannot determine the penny rate coming out of Spotify and others with the information that is on their accounting statements.  That is hard to explain to songwriters (or artists for that matter, as they have similar problems).

Why is the calculation so complex?  The artist revenue share calculation looks something like this in its generic configuration:

[Monthly Service Advertising Revenue or Monthly Subscription Revenue] x [Your Total Monthly Streams on the Service/All Monthly Streams on the Service] x [Revenue Share] = Royalty per stream

Both monthly revenue and monthly usage change each month–because they are monthly.  In order to get a nominal royalty rate, you have many calculations on both sides of the equation.  Because these calculations are made monthly, it is not possible to state in pennies the royalty rate for any one song or recording at any one time.  There’s actually an additional eye-crossing wrinkle on subscription deals of setting a negotiated minimum per subscriber which can vary by country, but we will leave that complexity aside for this post–YouTube’s “Exhibit D” lists 3 pages of one line entries for per subscriber minima around the world.

In a simple example, if both advertising revenue and subscription revenue were $100, your one recording was played 10 times in a month, all recordings were played 100 times in a month and the revenue share was 50% for the sound recording then you would get:

$100 x [10/100] x .50 = $0.50 for that month.  How you get to the multiplicand in the revenue pot is not so simple and has gotten more complex over the years.  In fact, the contract language for these calculations make the Single Bullet Theory seem more plausible.

Revenue share formulas produce a different product when the factors change–which for the most part changes every month.  The formula we’re using is for the sound recording side, but publishers have a version of this calculation for their songwriter’s royalties, too.  The statutory rates are a version of this formula (see the nearly unintelligible 37 CFR §385.12).

Most of this information is under the exclusive control of the service, and largely stays that way, even if you are one of the lucky few who has an audit right.  Bear in mind that the “Monthly Service Advertising Revenue” in our formula is a function of advertising rates charged by the service, and “Monthly Subscription Revenue” is a function of net subscription rates charged by the service.  These calculations take into account day passes, free trial periods, and other exceptions to the royalty obligation.  There is essentially no way to confirm the revenue pot when the royalty rates appear on the publisher or label statements.

The problem is that the entire concept of revenue share deals is out of step with how artists and songwriters are used to getting paid, even for other statutory mechanical rates such as that for downloads.  If a publisher or label can’t come up with a nice crisp answer for what the songwriter or artist royalty is based on, the assumption often is that the creator is being lied to.  And who’s to say that’s an unreasonable conclusion to jump to?  The question is–who is lying?  Here’s a tip–it’s probably not the publisher or label because they’re essentially in the same boat as the artist.

How Did We Ever Get Here?

Let me take you back to 1999.  Fish were jumpin’, the cotton was high, and limited partners showed up for capital calls.  Startups were starting up their engines–some to drive into a brick wall at scale, others to an IPO (and then into a brick wall at even greater scale).

On the Internet, you didn’t just do business with a company, they were your “partner.”  You didn’t just negotiate a commercial relationship with a behemoth Fortune 50 company that could crush you like a bug–in the utopian value system your little company “partnered” with AOL for example.  Or Intel.  Or later, Google.

What that meant for music licensing was that startups wanted rights owners to take the ride with them so if they made money, the rights owner made money.  Rights owners shared their revenue, you know, like a partner.  Except you only shared some of their revenue.  You weren’t really a partner and had no control over how they ran their business even if the only business they’d had previously run was a lemonade stand.

The revenue share deal was born.  To some people, it seemed like a good idea at the time.  And it might have been if there were relatively few participants in that revenue share.  But revenue share deals don’t scale very well.

Enter Professor Coase and His Pesky Theorem 

Here’s the basic flaw with revenue share deals:  Calculating the share of revenue for the entire catalog of licensed music on a global basis requires a large number of calculations.  For companies like Spotify, Apple or YouTube, calculating the share of revenue for millions of songs and recordings requires billions of calculations.

Free services like Spotify or YouTube involve billions of essentially unauditable calculations, all of which are based on a share of advertising revenue.  Advertising revenue which is itself essentially unauditable due to the nearly pathological level of secrecy that prevents any royalty participant from ever knowing what’s in the pie they are sharing.

That secrecy runs both upstream, downstream and across streams.  And as we all know, keeping secrets from your partner is the first step on the road to ruining a relationship.

But before you get too deep into nuances, let’s start with a basic problem with the entire revenue share approach.  In order to get to a per unit royalty, you have to multiply one dynamic number (the revenue) by another dynamic number (the usage).  Meaning that the thing being multiplied and the thing by which it is multiplied change from month to month.  The only constant in the formula is the actual percentage of the pie payable to the rights owner (50% in our example).

Remember–this all started with the digital service proposing that artists, songwriters, labels and publishers should take a share of what the service makes.  If you have a significant catalog, however, you do what you do with everyone who wants to license your catalog–you require the payment of a minimum guarantee as a prepayment of anticipated royalties (also called an “advance”).

So in our simple example, if the service is pitching that they will invest heavily in growth and make the catalog owner $50 over a two year contract, the catalog owner is justified in responding that however much confidence they have in the service, they’d like that $50 today and not a burger on Tuesday.  The service can apply the $50 minimum guarantee against the catalog’s earnings during the term of the contract, but if the minimum guarantee doesn’t earn out, the catalog owner keeps the change.  This shifts the credit or default risk from the catalog owner partner to the digital service partner (who actually controls the fate of the business).

But–given the complexity of the revenue share calculations, at least three questions arise:

Question: How will creators ever know if they are getting straight count from the service due to the complexity of the calculations?

Answer: The vast majority will never know.

Question:  How will anyone know if the advance ever recoups with any degree of certainty if they cannot verify the revenue pot they are to share?

Answer: The royalty receiver has to rely on statements based on effectively unverifiable information.

Question:  And most importantly, if streaming really is our future as industry leaders keep telling us, then which publisher wants to sign up for a lifetime of explaining the inexplicable to songwriters and artists who question their royalty statements?

Let’s Get Rid of Revenue Share Deals

There’s really no reason to keep this charade going any longer.  If the revenue share deal was converted to a penny rate, life would get so much easier and calculations would get so much simpler.  There would be arguments as always about what that penny rate ought to be.  Hostility levels might not go away entirely, but would probably lessen.

Transaction costs should go down substantially as there would be far fewer moving parts.  Realize that it’s entirely possible that the transaction costs of reporting royalties in revenue share deals (including  productivity loss and the cost of servicing songwriters and artists) likely exceeds the royalties paid.  My bet is that the costs vastly exceed the benefits.

And the people who really count the most in this business–the songwriters and artists–should have a lot more transparency.  Transparency that is essentially impossible with compulsory licenses.

Because when you take into account the total transaction costs, including all the correcting and noticing and calculating and explaining on the publisher and label side, and all the correcting and processing and calculating and messaging that has to be done on the service side, surely–surely–there has to be a simpler way.


Investors and Music Licensing

by Chris Castle

Music licensing for online music services is a pay me now or pay me later kind of thing.  From an investor’s point of view, the paying is going to be done with your money and if you pay later, you’ll probably–almost certainly–pay a lot more.  This illustrates the rule of thumb–if you don’t have a license, don’t use the song or recording.

To better understand music licensing, let’s take the example of a single song and recording.  That’s right–the song and the recording are separate property rights, or copyrights in this case.  The song and recording may be owned by the same person or by different people.  A bit of nomenclature: Songs are created by songwriters and are frequently owned or administered by music publishers.  The symbol for a song is often a “©”.  Recordings are created by artists and are frequently owned or distributed by record companies or “labels”.  The symbol for sound recordings is often a “℗”.  Labels ≠ publishers and © ≠ ℗.

In the easy case, let’s take a single song written and recorded by one person who acts as her own publisher and label.  In this case, your portfolio company can get one license from one person for all the rights (except the right to publicly perform that song, which almost invariably comes from either ASCAP, BMI or SESAC under a blanket license.  Those entities are are called performing rights organizations).

If the same artist records “Yesterday” written by John Lennon and Paul McCartney, then the company will license the artist’s recording directly.  Your portfolio company will also need to get a license (probably a “mechanical license”) from Lennon and McCartney’s music publisher Northern Songs, Ltd.

A mechanical license can either be “direct” or “compulsory.”  If the license is direct, then your portfolio company will contract directly with Northern Songs for the use.  If the license is compulsory, then your portfolio company may be able to take advantage of the government-mandated mechanical license depending on the intended use.

The government-mandated mechanical license covers certain types of functionality only.  This license, sometimes called a “compulsory license”, is found in Section 115 of the Copyright Act (17 USC 115) and the corresponding regulations in the Code of Federal Regulations (especially 37 CFR § 201.18, § 201.19 and §385.10-§385.17).  (The Copyright Office has an excellent plain English summary in its Circular 73.)  Taken together, Section 115 and the regulations are a kind of uniform mechanical license that applies to all “mechanical reproductions” of songs, including streams.  (Not dissimilar conceptually from the way the Uniform Partnership Act covers partnerships.)

The thing to remember about compulsory licenses is that if your portfolio company qualifies for one, it’s entirely likely that all of its competitors do, too.  If your company’s services are truly innovative, then you’re probably in the direct licensing bucket as the compulsory license is for more standard types of functionality.

Direct Statutory or
Blanket License
Song © Contract (interactive) Sec 115 Mechanical (permanent download, limited download, streaming, a few others)

Non interactive usually does not require a mechanical license

Sound Recording ℗ Contract (interactive) Sec 114(g) SoundExchange (noninteractive) N/A for interactive
SoundExchange for non interactive

In order to take advantage of the compulsory license, the Copyright Act requires that your company comply with certain formalities.  Assuming that the song is subject to a compulsory license (many but not all are), the threshold formality is that your company send a notice informing the owner that your company intends to rely on the compulsory license.  That notice is sent either to (1) the copyright owner of the song, or (2) the U.S. Copyright Office if the copyright owner cannot be found.  The notice has to be sent before or within 30 days of making, and before distrib­uting, any reproductions of the song.  The reference to “making” is more of an analog concept, but the the import of “before distributing” is pretty clear.

In addition to sending the notice, your portfolio company’s other threshold obligation is to account and pay royalties to the copyright owner or authorized agent of the owner on or before the 20th day of each month for every copy “made and distributed“.  Notice that phrase “made and distributed”–not sold.  That means that unless you have an agreement to the contrary with the copyright owner, the compulsory license requires a payment for every reproduction of the song whether or not your portfolio company was paid.  This is one reason to get a free license for the iTunes “download of the week” for example.

The compulsory license establishes a royalty rate and a method of calculation of that rate. The Copyright Office publishes a list of those rates.   The method of calculation is especially relevant for revenue share rates.  Remember–if you don’t have a direct license and you don’t qualify for the compulsory license, the royalty rate is entirely at the discretion of the copyright owner.

If your portfolio company tells you that they are “escrowing” royalties without a license, the alarm bell that should immediately go off is how can they know what the rate is if they don’t have a license.  And if they don’t know what the rate is, how can they “escrow” a royalty?  (Setting aside the fact that no such “escrow” is permitted for unlicensed songs.)  And if they don’t know who is entitled to the money, how can that copyright owner have ever agreed to the escrow.

The closest that the Copyright Act comes to establishing an “escrow” concept is for users who want a compulsory license but can’t find the copyright owner’s name or contact in the Copyright Office records.  (Most users look beyond the Copyright Office records.)  For unfindable copyright owners, the user files the notice with the Copyright Office licensing division, but accrues the royalties in anticipation of the copyright owner being found or coming forward.  (This raises a question of what happens to any accrued royalties if the user goes out of business for whatever reason.)

The important thing to remember about filing these notices and complying with the license is that doing so can stop an infringement claim, assuming no other problems.  While there is a filing fee for each song that averages about $25 per song, that’s pretty cheap insurance to avoid a copyright infringement lawsuit or the other statutory rate–statutory damages.

If your portfolio company plans to use a large number of songs, say in the millions, you can see that filing fee adds up.  The first question for investors is whether the company really does need to use that many songs.  If there is reliable consumer research that supports the idea that a successful consumer offering requires tens of millions of songs, I haven’t seen it.  If you don’t have a good reason to get into licensing tens of millions of songs in order to launch then you probably should take a more leisurely and prudent pace. Because you’ll either pay now or pay later.

As I will address in a future post, there is a way to get direct licenses and avoid the entire process of statutory licensing.