Vernor v. Autodesk: Copyright, Software Upgrades and Secondary Markets

Earlier today, the Ninth Circuit decided an important case about the scope of the first sale doctrine in copyright as applied to computer software. In Vernor v. Autodesk, the court concluded that the documents in question created a license rather than a sale of the underlying software with the consequence that copyright’s first-sale doctrine did not apply. Early commentary—at least of the 140 character kind on Twitter—is largely critical. The case is one of statutory interpretation and, for the Ninth Circuit, figuring out what some of its earlier software cases had said. The case raises some broad issues regarding copyright and secondary markets, but there is a narrower issue regarding software upgrades embedded in the case and that is the one I want to focus on here.

The facts are detailed but to simplify Autodesk produces high-end computer-aided design software and turns over copies of that software for cash. Note that I didn’t say Autodesk sells the software as that is the relevant question for the first-sale doctrine. In the spirit of Fred Khan, I will refer to this transfer for now as a banana. Autodesk’s software has gone through many versions and two are relevant to this situation. Autodesk bananaed a number of copies of AutoCAD release 14 to one of its customers—CAT—who in turn eventually bananaed that software to Timothy Vernor. In turn, Vernor was bananaing the software on eBay. Autodesk believed that Vernor’s bananaing was a copyright infringement and Vernor defended under the first-sale doctrine of section 109 and what the Ninth Circuit calls the essential step defense under section 117.

We need a little more granularity. The software that Autodesk had bananaed to CTA included documents that imposed a number of limitations on what CTA could do with the software. CTA was barred from transferring the software without Autodesk’s consent; barred from reverse engineering it; and limited in other ways (including a term separating commercial markets and educational markets). The documentation also required CTA to destroy prior versions of the AutoCAD software if CTA upgraded from an old version.

Eventually, CTA wanted to upgrade from release 14 to release 15. New licenses from Autodesk were going for $3,750 a pop but the upgrade price for an existing user was only $495. Unsurprisingly, CTA chose to upgrade. More surprisingly—but I am a naïve soul—CTA then chose to sell its release 14 copies to Vernor. After a good chunk of analysis, the Ninth Circuit ultimately concluded that the documents in question created a license in favor of CTA rather than a sale and therefore CTA did not receive the benefits of either the first-sale doctrine or the essential-step doctrine. Parsing the statutes and the Ninth Circuit’s precedent is a task for another day. I have five hypotheticals.

Case 1 is our baseline. A firm wants software and gets it in a banana transaction and pays the relevant fee. After a while, the firm decides that it would rather have a competitor’s software and does another deal and pays another fee. What can the firm do with the software it would now like to discard? If we were talking about a novel rather than software—and now think a paper-based novel rather than an e-book—the answer would be clear. The first sale doctrine makes it possible for the owner of a physical copy of the book to sell that book in the used book market without the permission of the copyright holder.

Implementing this kind of secondary markets regime is tricky. The copyright holder of course sees the secondary market as a missed opportunity for sale. It has software or novels ready to sell and its original transaction has the potential for creating a competitor to it. Moreover, if the software seller is separating markets—selling at one price to the educational market and a second price to commercial markets—secondary sales put pressure on that separation. That segmentation can be socially useful. Not always, but sometimes. Moreover, unlike novels, software sellers and digital purveyors generally fear that the secondary market sale will actually just be copying, where the seller will retain an original and simply dump off a copy in the secondary market. We can have an extended discussion about the merits of the secondary markets but that really isn’t my issue here. Assume for now that we treated my baseline case as we would the novel and generally permitted sales of software in the secondary markets.

That gets us to case 2. The software company is the only show in town and new versions of the software compete with old versions. The software firm does not offer an upgrade path: if you want the new version, you have to pay list price even if you have the old version. As before, our bananaer get software in the initial transaction and pays. Eventually the software firm issues an update and our software user bites the bullet and pays full price for the new version. They then sell the old version in the secondary market. There seems to be no reason not to treat case 2 as we did case 1, so if we permit secondary market transactions in used software when switching to a competitor’s software, we should also permit them when the software user pays full price for a new version of the original software.

Now consider case 3. Our software firm issues software and the software user pays. Eventually, the software firm is prepared to release additional functionality but rather than selling that as a full new version it simply offers it as a freestanding addition. Of course, only pre-existing customers would buy that software as it would be worthless without the old version of the software. No secondary market issues at all are raised in this transaction.

Case 4 is case 3 except that at the point of buying the additional software, our software user decides to make a copy of the original software and sells that in the secondary market. That is a straightforward copyright violation and the software user has no plausible defense under the first sale doctrine because it is keeping the original software and not giving it up.

Finally, in case 5, at the point that the software vendor has additional functionality, it releases an updated version of the software. It sells that software to new customers at a high price but offers an upgrade path and a much lower price to pre-existing customers. Our software user pays the upgrade price to get the new software but then turns around and sells the old software in the secondary market.

Whatever we think about the statutory interpretation questions in Autodesk and whatever we think about the general question of whether we should have say a secondary market in iTunes songs, we should have a firm sense of how case five should come out. The low upgrade price is the equivalent of the sale of the freestanding incremental software in case 3. There may be good reasons of software management that it is more sensible to distribute the software as a whole rather than in piecemeal functionality but the economics of the transaction should be treated as equivalent. When the software user in turn pays the low upgrade price and then sells the old version of the software, we have now replicated case 4. You can’t both keep and sell the old version, unless you are paying full price for new version as in case 2.

There are still lingering legal niceties. For example, should we think of this sale as a copyright infringement or a contract violation? But we need to recognize that the upgrade case is a distinctive case in the secondary market debates over the first sale doctrine. There is no sense that the Ninth Circuit appreciated that and its analysis seems perfectly general as to the question of ownership versus license—hence the gnashing of teeth in some quarters—but the analysis here suggests that the court gets the upgrade case right, even if it didn’t fully appreciate that that was what was before it.

Comments

[James Grimmelmann]

Nicely put. The court reaches the right outcome, but for the wrong reasons. First sale is supposed to be a buy one-get one deal: buy a copy of The Castaway and you can resell it, regardless of whether Bobbs-Merrill slaps a sticker on it saying you can't. CTA here is economically trying to engage in a buy one-get two deal: it paid only the upgrade price but wants to be treated in law as owning two copies that it can use or sell as it chooses. (That's a deal that Autodesk would almost certainly not have offered if the law supported CTA and everyone did what it did.)

The danger of the holding is that the court's test effectively underrules Bobbs-Merrill in practice. Make the sticker on the book a "license" that doesn't just prohibit you from selling the book but also prohibits you from reading it aloud or taking it to the beach, and you have turned the transaction into a buy one-get none deal. While trying to protect the first sale compromise from being undermined in one direction, the court accidentally undermines it in the opposite direction.

I would have preferred that the court confront the economic substance of Autodesk's deal with CTA. Autodesk agreed to improve CTA's copy in various ways, in return for a fee. For first sale purposes, this should be regarded as CTA's having returned legal ownership of the version 14 copy to Autodesk, and received a copy of version 15 in return. CTA is left with bare possession of its copy version 14, and that only because Autodesk has not demanded return or proof of destruction of the physical medium (an inefficient step that it benefits neither party to require). My reading of the transaction would say, in other words, that Autodesk's license is effective to prevent the resale of version 14 but not the resale of version 15.

[Michael Risch]

Seems like this case was a no brainer, but I agree that the generalized rules may wind up being problematic. I've been wanting to write an article about how we determine whether or not a license restriction should be preempted by law, but I'm not sure we have a clue about how to make a general rule. The more I think about it, the more I think that it has to be an individualized inquiry: ProCD is right because market segmentation was good there. Davidson was wrong because battlenet was buggy and bnet.d added social value. Not a very palatable solution, though.

[Randy Picker]

James,

I think that you and I may disagree on the overall merits of first-sale. I tried to slide around the big picture question because--and here we seem to agree--this case isn't the general case but instead a very different case. We seem to agree on exactly what CTA was doing there and why that shouldn't be allowed. I don't think that the court gets that at all. I usually blame the parties--really, the lawyers--in that situation. A good example of why you have to understand the economics of the transaction to have a chance of getting the case right. That said, to be fair to the 9th Circuit, for them it was just a question of statutes and precedents.

Randy

[DJL]

Professor Picker -

Are your views the same if this were digital media, and Apple was offering non-DRM "upgrade" versions to my old DRM'd iTunes files (in which the non-DRM upgraded are perhaps even higher quality in terms of sampling, etc.) for 50 cents since I bought the DRMd version for $1? Can I resell my old DRMd iTunes files? I guess we should put aside the fact that my old DRMs iTunes file would only be useful to those brave enough to use the many useful tools out there that asssit in circumventing the DRM. I think your answer is no, I should not be able to resell my old DRM files, from an economics perspective. The question arises here because Ts&Cs for sale of digital media are obviously following the clever history of the software industry in distinguishing the term "owner of a copy" by specifically noting in the Ts&Cs that the transaction a license not a sale.

Returning from the silly hypothetical, the contractual restrictions (and additional express statements that the software is being licensed and not sold) discussed in the case are fairly standard and pervasive in all EULAs and SLAs today (curious whether that was the case in 1980 when Section 117 was adopted using the term "owner of a copy of a computer program" - since Section 117 seems simply to have been completely eviserated in terms of usefulness). It's not clear to me that end user agreements as discussed by the Court are the appropriate agreements to discuss in this case and the holding is a brave one for the Court to make here and perhaps it did not have to get that far.

Perhaps this case, intead, could have been more conservatively decided on the fact that the already-used license keys/codes were handwritten on the box based on a DMCA circumvention of a technological protection measure type argument.

While the software industry did quite a good job in structuring end user agreements to be construed as licenses, it seemed somehow a little less rigourous in its reseller and distribution agreements, which are a core distribution mechanism for many software companies.

Off-the-shelf Microsoft Office software sold in Best Buy is in some sense "purchased" by Best Buy pursuant to reseller or distributer agreements which have terms and condiions that are different from the EULAs, since Best Buy won't be "using" the actual software and will be just reselling the tangible medium on which that software is stored (i.e., box, CD-ROM, etc.). What is Best Buy? From an accounting perspective, it "owns" the software "inventory" after purchasing from MS - is it simply the owner of a "copy of the tangible medium" on which the software is stored, but not the owner of a "copy of the software" itself? Does it really need that "standard" license to distribute that Microsoft gives to it in the distribution agreement or can it rely on the first sale doctrine? Is Best Buy really getting just a license to re-license the software? That doesn't gel with its inventory concepts where it needs to have "title" to inventory from an accounting perspective. Verner is arguing that he is Best Buy - the only rub is that the license key was already used (this ties into who has the right to get the upgrade price). If CTA never used the license key (i.e,. would not be entitled to the upgrade) and had sold the software to Verner who then put it up on eBay, maybe this is a different case since neither CTA nor Verner would have agreed to the EULA/SLA and the license/sale issue would not come up.

Finally, as a commentary, standard software end user licenses seem to be very interesting legal documents from a historical perspective as they obviously borrowed language regarding warranties, liabilities, remedies and disclaimers from contract experts of a prior era trying to get around default liability rules in the UCC as the UCC relates to the "SALE" of goods (not licensing). Then the software guys realized, they shouldn't be thinking of their stuff like other UCC "sale of goods" contracts and started adding all these terms in to construe the agreement as a license not a sale - so software agreements in general are a confused mix of sale versus license concepts themselves.

This case also brings forth many other interesting issues less discussed in academic circles. For example - tax advisors who advise companies on putting in place inter-company licenses/agreements for transfer pricing type IP holding structures need to make use of the conception of "selling" (and not licensing) software via distribution channels to make their structures work - unfortunatey, tax advisors don't follow these sale versus license cases as closely as IP attorneys do which often makes these tax structures confusing from an IP perspective.

[James Grimmelmann]

Randy,

I tend to think of first sale as a negotiability regime; it makes sure that holders in due course of copies can give good title. That rule makes a lot of sense for analog goods on information-costs grounds. I'm open to arguments either that we need a corresponding rule for digital goods that's less tied to the MAI-style formality of the "copy" or that first sale is unnecessary for digital goods in light of their easy reproduction and the decreased costs of licensing. Regardless of where we each come down on this question, I think we share a concern that the court may have just weighed in on it without realizing the implications.

[Ariel Katz]

Randy and James,
You seem to share the view that CTA behaved opportunistically and that preventing it from behaving this way is the right outcome. But the case was not about CTA. I think CTA wasn't even a named party, neither did it intervened as amicus. The case was about Vernor who purchased the copies from CTA and sold them on eBay, and the court's holding also means that anyone who purchased a copy from Vernor and installs it infringes copyright.
It's still an open question whether the 1st sale doctrine renders any contract that prevents resale unenforceable, and this case may be an example of a case where there are good reasons to enforce the contract between Autodesk and CTA despite the 1st sale doctrine. But the distinction between copyright infringement and breach of contract is highly important. It's much more than a lingering legal nicety.

[Randy Picker]

Ariel,

I agree that the contract vs. copyright question is important, though one that I have done some on before, but maybe should do more again given this case. I hadn't focused on the particular issues raised by software upgrades until I read the decision on Friday.

As to opportunism, you are right to say that the post frames this as to CTA, but I suspect Vernor--the seller of more than 10,000 items on eBay--is the real agent of opportunism here, hence why Autodesk wanted to chase him. Controlling someone like Vernor--or not--is obviously part of the contract vs. copyright discussion.

Randy

[Ariel Katz]

I agree that whether Autodesk should be able to chase Vernor is part of the discussion; that's why I raised it. It seems to me that discussions about the 1st sale doctrine there is a tendency to move very quickly from identifying situations in which post-sale restrictions may be efficient, to concluding that the doctrine is misguided (or the opposite, supporting the doctrine and therefore concluding that any attempt to impose post-sale restrictions should be unenforcable). I think that taking the doctrine seriously, while enforcing contractual post-sale restrictions where there are compelling reasons to do so is the right approach.
I wonder why you see Vernor as the "real agent of opportunism". You wouldn't give this title to Alibris (www.alibris.com) or to all used-cars dealers, would you? What makes Vernor special?

[FvL]

Your hypothetical #5 could easily have been addressed by Autodesk if it had required return of the original physical media as a condition of the upgrade price (something that other software vendors have done in the past). That would neatly resolve any lingering first sale questions (as first sale attaches to the tangible media), without jeopardizing a host of other socially desirable outcomes that flow from a robust copyright exhaustion doctrine.

[Randy Picker]

Fred,

Sure, but this is a question of what kind of transactions costs do you want to run the system. Trade offs both ways: burdens on enforcing the upgrade rules vs. burdens on secondary markets in software.

But a physical disk return policy might be good practical advice to the Autodesks of the world.

Randy

[Brett]

From someone who works with AutoDesk, ArcGIS and related software...
Autodesk is enterprise software that is normally integrated into multiple 3rd party applications and intra- and inter-department workflows. I have departments that are six versions behind other departments on their upgrade paths.

While you can opt to receive new media with each full license, media and upgrade licenses are not issued one-to-one. We get one media copy of each upgrade, even though we possess over 60 licenses that we upgrade (all licenses possessed are upgraded simultaneously for one fee based on the number of licenses).
If we were required to return or destroy the original physical media with each upgrade, we have a situation where lagging departments were unable to upgrade or replace hardware due to unavailability of the older version upgrade media.
There paths around this. You could issue media for every upgrade license, and allow the individual upgrade of each license. But when discussing this path, you have to recognize that there are organizations (such as the USGS, FEMA, USDA) out there with tens of thousands of licenses who do silent installs without every touching physical media. Physical media installs of AutoDesk or ArcGIS easily take more than an hour. Such a requirement would be imposing a very large cost on these organizations as they physically return, physically install likely well over one hundred thousands DVDs per software package every year.

[Alex]

Nothing prevents Autodesk from implementing the DRM based scheme that would make copies unfunctional after upgrade activation. It would still be not a crime to sell deactivated copies under first sale. A researcher may well be interested to buy such a copy and attempt to break the DRM scheme to tell Autodesk that the scheme is not reliable. And it *would* be a crime to break the DRM scheme (and sell copies with broken DRM) allowing use of deactivated software after upgrade. That is precisely what DMCA is about.

But instead of gently pointing out to Autodesk the right solution to their problem, the imbeciles in the 9th Cir. has created utterly idiotic anti first sale "legal test." The people of the 9th Cir. shall impeach sillies in robes.

Sorry for the strong words (and feel free to XXXX it) but Vernor opinion is really unbelievably stupid ruling in my opinion.