Archives For Deception

What does it mean to “own” something? A simple question (with a complicated answer, of course) that, astonishingly, goes unasked in a recent article in the Pennsylvania Law Review entitled, What We Buy When We “Buy Now,” by Aaron Perzanowski and Chris Hoofnagle (hereafter “P&H”). But how can we reasonably answer the question they pose without first trying to understand the nature of property interests?

P&H set forth a simplistic thesis for their piece: when an e-commerce site uses the term “buy” to indicate the purchase of digital media (instead of the term “license”), it deceives consumers. This is so, the authors assert, because the common usage of the term “buy” indicates that there will be some conveyance of property that necessarily includes absolute rights such as alienability, descendibility, and excludability, and digital content doesn’t generally come with these attributes. The authors seek to establish this deception through a poorly constructed survey regarding consumers’ understanding of the parameters of their property interests in digitally acquired copies. (The survey’s considerable limitations is a topic for another day….)

The issue is more than merely academic: NTIA and the USPTO have just announced that they will hold a public meeting

to discuss how best to communicate to consumers regarding license terms and restrictions in connection with online transactions involving copyrighted works… [as a precursor to] the creation of a multistakeholder process to establish best practices to improve consumers’ understanding of license terms and restrictions in connection with online transactions involving creative works.

Whatever the results of that process, it should not begin, or end, with P&H’s problematic approach.

Getting to their conclusion that platforms are engaged in deceptive practices requires two leaps of faith: First, that property interests are absolute and that any restraint on the use of “property” is inconsistent with the notion of ownership; and second, that consumers’ stated expectations (even assuming that they were measured correctly) alone determine the appropriate contours of legal (and economic) property interests. Both leaps are meritless.

Property and ownership are not absolute concepts

P&H are in such a rush to condemn downstream restrictions on the alienability of digital copies that they fail to recognize that “property” and “ownership” are not absolute terms, and are capable of being properly understood only contextually. Our very notions of what objects may be capable of ownership change over time, along with the scope of authority over owned objects. For P&H, the fact that there are restrictions on the use of an object means that it is not properly “owned.” But that overlooks our everyday understanding of the nature of property.

Ownership is far more complex than P&H allow, and ownership limited by certain constraints is still ownership. As Armen Alchian and Harold Demsetz note in The Property Right Paradigm (1973):

In common speech, we frequently speak of someone owning this land, that house, or these bonds. This conversational style undoubtedly is economical from the viewpoint of quick communication, but it masks the variety and complexity of the ownership relationship. What is owned are rights to use resources, including one’s body and mind, and these rights are always circumscribed, often by the prohibition of certain actions. To “own land” usually means to have the right to till (or not to till) the soil, to mine the soil, to offer those rights for sale, etc., but not to have the right to throw soil at a passerby, to use it to change the course of a stream, or to force someone to buy it. What are owned are socially recognized rights of action. (Emphasis added).

Literally, everything we own comes with a range of limitations on our use rights. Literally. Everything. So starting from a position that limitations on use mean something is not, in fact, owned, is absurd.

Moreover, in defining what we buy when we buy digital goods by reference to analog goods, P&H are comparing apples and oranges, without acknowledging that both apples and oranges are bought.

There has been a fair amount of discussion about the nature of digital content transactions (including by the USPTO and NTIA), and whether they are analogous to traditional sales of objects or more properly characterized as licenses. But this is largely a distinction without a difference, and the nature of the transaction is unnecessary in understanding that P&H’s assertion of deception is unwarranted.

Quite simply, we are accustomed to buying licenses as well as products. Whenever we buy a ticket — e.g., an airline ticket or a ticket to the movies — we are buying the right to use something or gain some temporary privilege. These transactions are governed by the terms of the license. But we certainly buy tickets, no? Alchian and Demsetz again:

The domain of demarcated uses of a resource can be partitioned among several people. More than one party can claim some ownership interest in the same resource. One party may own the right to till the land, while another, perhaps the state, may own an easement to traverse or otherwise use the land for specific purposes. It is not the resource itself which is owned; it is a bundle, or a portion, of rights to use a resource that is owned. In its original meaning, property referred solely to a right, title, or interest, and resources could not be identified as property any more than they could be identified as right, title, or interest. (Emphasis added).

P&H essentially assert that restrictions on the use of property are so inconsistent with the notion of property that it would be deceptive to describe the acquisition transaction as a purchase. But such a claim completely overlooks the fact that there are restrictions on any use of property in general, and on ownership of copies of copyright-protected materials in particular.

Take analog copies of copyright-protected works. While the lawful owner of a copy is able to lend that copy to a friend, sell it, or even use it as a hammer or paperweight, he or she can not offer it for rental (for certain kinds of works), cannot reproduce it, may not publicly perform or broadcast it, and may not use it to bludgeon a neighbor. In short, there are all kinds of restrictions on the use of said object — yet P&H have little problem with defining the relationship of person to object as “ownership.”

Consumers’ understanding of all the terms of exchange is a poor metric for determining the nature of property interests

P&H make much of the assertion that most users don’t “know” the precise terms that govern the allocation of rights in digital copies; this is the source of the “deception” they assert. But there is a cost to marking out the precise terms of use with perfect specificity (no contract specifies every eventuality), a cost to knowing the terms perfectly, and a cost to caring about them.

When we buy digital goods, we probably care a great deal about a few terms. For a digital music file, for example, we care first and foremost about whether it will play on our device(s). Other terms are of diminishing importance. Users certainly care whether they can play a song when offline, for example, but whether their children will be able to play it after they die? Not so much. That eventuality may, in fact, be specified in the license, but the nature of this particular ownership relationship includes a degree of rational ignorance on the users’ part: The typical consumer simply doesn’t care. In other words, she is, in Nobel-winning economist Herbert Simon’s term, “boundedly rational.” That isn’t deception; it’s a feature of life without which we would be overwhelmed by “information overload” and unable to operate. We have every incentive and ability to know the terms we care most about, and to ignore the ones about which we care little.

Relatedly, P&H also fail to understand the relationship between price and ownership. A digital song that is purchased from Amazon for $.99 comes with a set of potentially valuable attributes. For example:

  • It may be purchased on its own, without the other contents of an album;
  • It never degrades in quality, and it’s extremely difficult to misplace;
  • It may be purchased from one’s living room and be instantaneously available;
  • It can be easily copied or transferred onto multiple devices; and
  • It can be stored in Amazon’s cloud without taking up any of the consumer’s physical memory resources.

In many ways that matter to consumers, digital copies are superior to analog or physical ones. And yet, compared to physical media, on a per-song basis (assuming one could even purchase a physical copy of a single song without purchasing an entire album), $.99 may represent a considerable discount. Moreover, in 1982 when CDs were first released, they cost an average of $15. In 2017 dollars, that would be $38. Yet today most digital album downloads can be found for $10 or less.

Of course, songs purchased on CD or vinyl offer other benefits that a digital copy can’t provide. But the main thing — the ability to listen to the music — is approximately equal, and yet the digital copy offers greater convenience at (often) lower price. It is impossible to conclude that a consumer is duped by such a purchase, even if it doesn’t come with the ability to resell the song.

In fact, given the price-to-value ratio, it is perhaps reasonable to think that consumers know full well (or at least suspect) that there might be some corresponding limitations on use — the inability to resell, for example — that would explain the discount. For some people, those limitations might matter, and those people, presumably, figure out whether such limitations are present before buying a digital album or song For everyone else, however, the ability to buy a digital song for $.99 — including all of the benefits of digital ownership, but minus the ability to resell — is a good deal, just as it is worth it to a home buyer to purchase a house, regardless of whether it is subject to various easements.

Consumers are, in fact, familiar with “buying” property with all sorts of restrictions

The inability to resell digital goods looms inordinately large for P&H: According to them, by virtue of the fact that digital copies may not be resold, “ownership” is no longer an appropriate characterization of the relationship between the consumer and her digital copy. P&H believe that digital copies of works are sufficiently similar to analog versions, that traditional doctrines of exhaustion (which would permit a lawful owner of a copy of a work to dispose of that copy as he or she deems appropriate) should apply equally to digital copies, and thus that the inability to alienate the copy as the consumer wants means that there is no ownership interest per se.

But, as discussed above, even ownership of a physical copy doesn’t convey to the purchaser the right to make or allow any use of that copy. So why should we treat the ability to alienate a copy as the determining factor in whether it is appropriate to refer to the acquisition as a purchase? P&H arrive at this conclusion only through the illogical assertion that

Consumers operate in the marketplace based on their prior experience. We suggest that consumers’ “default” behavior is based on the experiences of buying physical media, and the assumptions from that context have carried over into the digital domain.

P&H want us to believe that consumers can’t distinguish between the physical and virtual worlds, and that their ability to use media doesn’t differentiate between these realms. But consumers do understand (to the extent that they care) that they are buying a different product, with different attributes. Does anyone try to play a vinyl record on his or her phone? There are perceived advantages and disadvantages to different kinds of media purchases. The ability to resell is only one of these — and for many (most?) consumers not likely the most important.

And, furthermore, the notion that consumers better understood their rights — and the limitations on ownership — in the physical world and that they carried these well-informed expectations into the digital realm is fantasy. Are we to believe that the consumers of yore understood that when they bought a physical record they could sell it, but not rent it out? That if they played that record in a public place they would need to pay performance royalties to the songwriter and publisher? Not likely.

Simply put, there is a wide variety of goods and services that we clearly buy, but that have all kinds of attributes that do not fit P&H’s crabbed definition of ownership. For example:

  • We buy tickets to events and membership in clubs (which, depending upon club rules, may not be alienated, and which always lapse for non-payment).
  • We buy houses notwithstanding the fact that in most cases all we own is the right to inhabit the premises for as long as we pay the bank (which actually retains more of the incidents of “ownership”).
  • In fact, we buy real property encumbered by a series of restrictive covenants: Depending upon where we live, we may not be able to build above a certain height, we may not paint the house certain colors, we may not be able to leave certain objects in the driveway, and we may not be able to resell without approval of a board.

We may or may not know (or care) about all of the restrictions on our use of such property. But surely we may accurately say that we bought the property and that we “own” it, nonetheless.

The reality is that we are comfortable with the notion of buying any number of limited property interests — including the purchasing of a license — regardless of the contours of the purchase agreement. The fact that some ownership interests may properly be understood as licenses rather than as some form of exclusive and permanent dominion doesn’t suggest that a consumer is not involved in a transaction properly characterized as a sale, or that a consumer is somehow deceived when the transaction is characterized as a sale — and P&H are surely aware of this.

Conclusion: The real issue for P&H is “digital first sale,” not deception

At root, P&H are not truly concerned about consumer deception; they are concerned about what they view as unreasonable constraints on the “rights” of consumers imposed by copyright law in the digital realm. Resale looms so large in their analysis not because consumers care about it (or are deceived about it), but because the real object of their enmity is the lack of a “digital first sale doctrine” that exactly mirrors the law regarding physical goods.

But Congress has already determined that there are sufficient distinctions between ownership of digital copies and ownership of analog ones to justify treating them differently, notwithstanding ownership of the particular copy. And for good reason: Trade in “used” digital copies is not a secondary market. Such copies are identical to those traded in the primary market and would compete directly with “pristine” digital copies. It makes perfect sense to treat ownership differently in these cases — and still to say that both digital and analog copies are “bought” and “owned.”

P&H’s deep-seated opposition to current law colors and infects their analysis — and, arguably, their failure to be upfront about it is the real deception. When one starts an analysis with an already-identified conclusion, the path from hypothesis to result is unlikely to withstand scrutiny, and that is certainly the case here.

by Berin Szoka, President, TechFreedom

Josh Wright will doubtless be remembered for transforming how FTC polices competition. Between finally defining Unfair Methods of Competition (UMC), and his twelve dissents and multiple speeches about competition matters, he re-grounded competition policy in the error-cost framework: weighing not only costs against benefits, but also the likelihood of getting it wrong against the likelihood of getting it right.

Yet Wright may be remembered as much for what he started as what he finished: reforming the Commission’s Unfair and Deceptive Acts and Practices (UDAP) work. His consumer protection work is relatively slender: four dissents on high tech matters plus four relatively brief concurrences and one dissent on more traditional advertising substantiation cases. But together, these offer all the building blocks of an economic, error-cost-based approach to consumer protection. All that remains is for another FTC Commissioner to pick up where Wright left off.

Apple: Unfairness & Cost-Benefit Analysis

In January 2014, Wright issued a blistering, 17 page dissent from the Commission’s decision to bring, and settle, an enforcement action against Apple regarding the design of its app store. Wright dissented, not from the conclusion necessarily, but from the methodology by which the Commission arrived there. In essence, he argued for an error-cost approach to unfairness:

The Commission, under the rubric of “unfair acts and practices,” substitutes its own judgment for a private firm’s decisions as to how to design its product to satisfy as many users as possible, and requires a company to revamp an otherwise indisputably legitimate business practice. Given the apparent benefits to some consumers and to competition from Apple’s allegedly unfair practices, I believe the Commission should have conducted a much more robust analysis to determine whether the injury to this small group of consumers justifies the finding of unfairness and the imposition of a remedy.

…. although Apple’s allegedly unfair act or practice has harmed some consumers, I do not believe the Commission has demonstrated the injury is substantial. More importantly, any injury to consumers flowing from Apple’s choice of disclosure and billing practices is outweighed considerably by the benefits to competition and to consumers that flow from the same practice.

The majority insisted that the burden on consumers or Apple from its remedy “is de minimis,” and therefore “it was unnecessary for the Commission to undertake a study of how consumers react to different disclosures before issuing its complaint against Apple, as Commissioner Wright suggests.”

Wright responded: “Apple has apparently determined that most consumers do not want to experience excessive disclosures or to be inconvenienced by having to enter their passwords every time they make a purchase.” In essence, he argued, that the FTC should not presume to know better than Apple how to manage the subtle trade-offs between convenience and usability.

Wright was channeling Hayek’s famous quip: “The curious task of economics is to demonstrate to men how little they really know about what they imagine they can design.” The last thing the FTC should be doing is designing digital products — even by hovering over Apple’s shoulder.

The Data Broker Report

Wright next took the Commission to task for the lack of economic analysis in its May 2013 report, “Data Brokers: A Call for Transparency and Accountability.” In just four footnotes, Wright extended his analysis of Apple. For example:

Footnote 85: Commissioner Wright agrees that Congress should consider legislation that would provide for consumer access to the information collected by data brokers. However, he does not believe that at this time there is enough evidence that the benefits to consumers of requiring data brokers to provide them with the ability to opt out of the sharing of all consumer information for marketing purposes outweighs the costs of imposing such a restriction. Finally… he believes that the Commission should engage in a rigorous study of consumer preferences sufficient to establish that consumers would likely benefit from such a portal prior to making such a recommendation.

Footnote 88: Commissioner Wright believes that in enacting statutes such as the Fair Credit Reporting Act, Congress undertook efforts to balance [costs and benefits]. In the instant case, Commissioner Wright is wary of extending FCRA-like coverage to other uses and categories of information without first performing a more robust balancing of the benefits and costs associated with imposing these requirements

The Internet of Things Report

This January, in a 4-page dissent from the FTC’s staff report on “The Internet of Things: Privacy and Security in a Connected World,” Wright lamented that the report neither represented serious economic analysis of the issues discussed nor synthesized the FTC’s workshop on the topic:

A record that consists of a one-day workshop, its accompanying public comments, and the staff’s impressions of those proceedings, however well-intended, is neither likely to result in a representative sample of viewpoints nor to generate information sufficient to support legislative or policy recommendations.

His attack on the report’s methodology was blistering:

The Workshop Report does not perform any actual analysis whatsoever to ensure that, or even to give a rough sense of the likelihood that the benefits of the staff’s various proposals exceed their attendant costs. Instead, the Workshop Report merely relies upon its own assertions and various surveys that are not necessarily representative and, in any event, do not shed much light on actual consumer preferences as revealed by conduct in the marketplace…. I support the well-established Commission view that companies must maintain reasonable and appropriate security measures; that inquiry necessitates a cost-benefit analysis. The most significant drawback of the concepts of “security by design” and other privacy-related catchphrases is that they do not appear to contain any meaningful analytical content.

Ouch.

Nomi: Deception & Materiality Analysis

In April, Wright turned his analytical artillery from unfairness to deception, long the more uncontroversial half of UDAP. In a five-page dissent, Wright accused the Commission of essentially dispensing with the core limiting principle of the 1983 Deception Policy Statement: materiality. As Wright explained:

The materiality inquiry is critical because the Commission’s construct of “deception” uses materiality as an evidentiary proxy for consumer injury…. Deception causes consumer harm because it influences consumer behavior — that is, the deceptive statement is one that is not merely misleading in the abstract but one that causes consumers to make choices to their detriment that they would not have otherwise made. This essential link between materiality and consumer injury ensures the Commission’s deception authority is employed to deter only conduct that is likely to harm consumers and does not chill business conduct that makes consumers better off.

As in Apple, Wright did not argue that there might not be a role for the FTC; merely that the FTC had failed to justify bringing, let alone settling, an enforcement action without establishing that the key promise at issue — to provide in-store opt-out — was material.

The Chamber Speech: A Call for Economic Analysis

In May, Wright gave a speech to the Chamber of Commerce on “How to Regulate the Internet of Things Without Harming its Future: Some Do’s and Don’ts”:

Perhaps it is because I am an economist who likes to deal with hard data, but when it comes to data and privacy regulation, the tendency to rely upon anecdote to motivate policy is a serious problem. Instead of developing a proper factual record that documents cognizable and actual harms, regulators can sometimes be tempted merely to explore anecdotal and other hypothetical examples and end up just offering speculations about the possibility of harm.

And on privacy in particular:

What I have seen instead is what appears to be a generalized apprehension about the collection and use of data — whether or not the data is actually personally identifiable or sensitive — along with a corresponding, and arguably crippling, fear about the possible misuse of such data.  …. Any sensible approach to regulating the collection and use of data will take into account the risk of abuses that will harm consumers. But those risks must be weighed with as much precision as possible, as is the case with potential consumer benefits, in order to guide sensible policy for data collection and use. The appropriate calibration, of course, turns on our best estimates of how policy changes will actually impact consumers on the margin….

Wright concedes that the “vast majority of work that the Consumer Protection Bureau performs simply does not require significant economic analysis because they involve business practices that create substantial risk of consumer harm but little or nothing in the way of consumer benefits.” Yet he notes that the Internet has made the need for cost-benefit analysis far more acute, at least where conduct is ambiguous as its effects on consumers, as in Apple, to avoid “squelching innovation and depriving consumers of these benefits.”

The Wrightian Reform Agenda for UDAP Enforcement

Wright left all the building blocks his successor will need to bring “Wrightian” reform to how the Bureau of Consumer Protection works:

  1. Wright’s successor should work to require economic analysis for consent decrees, as Wright proposed in his last major address as a Commissioner. BE might not to issue a statement at all in run-of-the-mill deception cases, but it should certainly have to say something about unfairness cases.
  2. The FTC needs to systematically assess its enforcement process to understand the incentives causing companies to settle UDAP cases nearly every time — resulting in what Chairman Ramirez and Commissioner Brill frequently call the FTC’s “common law of consent decrees.”
  3. As Wright says in his Nomi dissent “While the Act does not set forth a separate standard for accepting a consent decree, I believe that threshold should be at least as high as for bringing the initial complaint.” This point should be uncontroversial, yet the Commission has never addressed it. Wright’s successor (and the FTC) should, at a minimum, propose a standard for settling cases.
  4. Just as Josh succeeded in getting the FTC to issue a UMC policy statement, his successor should re-assess the FTC’s two UDAP policy statements. Wright’s successor needs to make the case for finally codifying the DPS — and ensuring that the FTC stops bypassing materiality, as in Nomi.
  5. The Commission should develop a rigorous methodology for each of the required elements of unfairness and deception to justify bringing cases (or making report recommendations). This will be a great deal harder than merely attacking the lack of such methodology in dissents.
  6. The FTC has, in recent years, increasingly used reports to make de facto policy — by inventing what Wright calls, in his Chamber speech, “slogans and catchphrases” like “privacy by design,” and then using them as boilerplate requirements for consent decrees; by pressuring companies into adopting the FTC’s best practices; by calling for legislation; and so on. At a minimum, these reports must be grounded in careful economic analysis.
  7. The Commission should apply far greater rigor in setting standards for substantiating claims about health benefits. In two dissents, Genelink et al and HCG Platinum, Wright demolished arguments for a clear, bright line requiring two randomized clinical trials, and made the case for “a more flexible substantiation requirement” instead.

Conclusion: Big Shoes to Fill

It’s a testament to Wright’s analytical clarity that he managed to say so much about consumer protection in so few words. That his UDAP work has received so little attention, relative to his competition work, says just as much about the far greater need for someone to do for consumer protection what Wright did for competition enforcement and policy at the FTC.

Wright’s successor, if she’s going to finish what Wright started, will need something approaching Wright’s sheer intellect, his deep internalization of the error-costs approach, and his knack for brokering bipartisan compromise around major issues — plus the kind of passion for UDAP matters Wright had for competition matters. And, of course, that person needs to be able to continue his legacy on competition matters…

Compared to the difficulty of finding that person, actually implementing these reforms may be the easy part.

Last week, the FTC announced its complaint and consent decree with Nomi Technologies for failing to allow consumers to opt-out of cell phone tracking while shopping in retail stores. Whatever one thinks about Nomi itself, the FTC’s enforcement action represents another step in the dubious application of its enforcement authority against deceptive statements.

In response, Geoffrey Manne, Ben Sperry, and Berin Szoka have written a new ICLE White Paper, titled, In the Matter of Nomi, Technologies, Inc.: The Dark Side of the FTC’s Latest Feel-Good Case.

Nomi Technologies offers retailers an innovative way to observe how customers move through their stores, how often they return, what products they browse and for how long (among other things) by tracking the Wi-Fi addresses broadcast by customers’ mobile phones. This allows stores to do what websites do all the time: tweak their configuration, pricing, purchasing and the like in response to real-time analytics — instead of just eyeballing what works. Nomi anonymized the data it collected so that retailers couldn’t track specific individuals. Recognizing that some customers might still object, even to “anonymized” tracking, Nomi allowed anyone to opt-out of all Nomi tracking on its website.

The FTC, though, seized upon a promise made within Nomi’s privacy policy to provide an additional, in-store opt out and argued that Nomi’s failure to make good on this promise — and/or notify customers of which stores used the technology — made its privacy policy deceptive. Commissioner Wright dissented, noting that the majority failed to consider evidence that showed the promise was not material, arguing that the inaccurate statement was not important enough to actually affect consumers’ behavior because they could opt-out on the website anyway. Both Commissioners Wright’s and Commissioner Ohlhausen’s dissents argued that the FTC majority’s enforcement decision in Nomi amounted to prosecutorial overreach, imposing an overly stringent standard of review without any actual indication of consumer harm.

The FTC’s deception authority is supposed to provide the agency with the authority to remedy consumer harms not effectively handled by common law torts and contracts — but it’s not a blank check. The 1983 Deception Policy Statement requires the FTC to demonstrate:

  1. There is a representation, omission or practice that is likely to mislead the consumer;
  2. A consumer’s interpretation of the representation, omission, or practice is considered reasonable under the circumstances; and
  3. The misleading representation, omission, or practice is material (meaning the inaccurate statement was important enough to actually affect consumers’ behavior).

Under the DPS, certain types of claims are treated as presumptively material, although the FTC is always supposed to “consider relevant and competent evidence offered to rebut presumptions of materiality.” The Nomi majority failed to do exactly that in its analysis of the company’s claims, as Commissioner Wright noted in his dissent:

the Commission failed to discharge its commitment to duly consider relevant and competent evidence that squarely rebuts the presumption that Nomi’s failure to implement an additional, retail-level opt out was material to consumers. In other words, the Commission neglects to take into account evidence demonstrating consumers would not “have chosen differently” but for the allegedly deceptive representation.

As we discuss in detail in the white paper, we believe that the Commission committed several additional legal errors in its application of the Deception Policy Statement in Nomi, over and above its failure to adequately weigh exculpatory evidence. Exceeding the legal constraints of the DPS isn’t just a legal problem: in this case, it’s led the FTC to bring an enforcement action that will likely have the very opposite of its intended result, discouraging rather than encouraging further disclosure.

Moreover, as we write in the white paper:

Nomi is the latest in a long string of recent cases in which the FTC has pushed back against both legislative and self-imposed constraints on its discretion. By small increments (unadjudicated consent decrees), but consistently and with apparent purpose, the FTC seems to be reverting to the sweeping conception of its power to police deception and unfairness that led the FTC to a titanic clash with Congress back in 1980.

The Nomi case presents yet another example of the need for FTC process reforms. Those reforms could ensure the FTC focuses on cases that actually make consumers better off. But given the FTC majority’s unwavering dedication to maximizing its discretion, such reforms will likely have to come from Congress.

Find the full white paper here.