COIN-OPERATED CAPITALISM

Introduction

If you believe what you read on social media, the world of venture finance is undergoing a sea change. Old institutions like banks and venture capital firms are finding themselves supplanted by masses of individuals coordinating through new financial platforms.

Excessively compensated elites are on the outs. They are being replaced—so say the believers—by equity crowdfunding, peer-to-peer lending, and the wisdom of the crowd.

The rise of the Initial Coin Offering (ICO) is a chapter in this story, and this Article’s subject.

Obviously, the ICO was named after the IPO, or “Initial Public Offering.” But though the IPO has been familiar for almost a century, the ICO is exotic. Unlike its namesake, an ICO does not typically involve the sale of equity in (or governance rights pertaining to) a corporation.

Instead, ICO participants buy an asset—a “token”—that enables its holder to use or govern a network that the promoters plan to develop with the funds raised through the sale.

It would be as if Coca-Cola had funded its initial deployment of vending machines through the sale of tokens its machines might one day require. The token holders’ interests would have been imperfectly aligned with the interests of investors who owned shares in Coca-Cola, Inc. Rather than caring about share value, they would have cared about token value, which would relate to the supply of the tokens and demand for vended Coke.

For this hypothetical Coca-Cola, it’s easy to imagine physical tokens and real vending machines. But for ICOs, the tokens and the “machines” they operate are digital. They exist on the internet, embodied in software code. The key forms of software are known as “smart contracts”—automated, “if-this-then-that” rules that coders can design to govern the functionality of the digital “crypto” assets sold in ICOs.

Smart contracts may be digital and automated, but they help structure real-world relationships. At present, relationships between ICO promoters and token buyers are quite nebulous.

Imagine that those Coca-Cola token investors lacked established legal means to enforce any promises made by Coca-Cola, Inc., cap the supply of tokens, require the use of those tokens to buy Coca-Cola from vending machines, limit sales of Coca-Cola through non-vending-machine channels, or even deploy machines at all. That scenario roughly captures the state of ICO legal contracting and governance today. This is a financial form ripe for fraud, and it has allegedly been used to that precise end.

But fraud also went hand-in-hand with early financial markets;

its presence settles little about the fate of the ICO form. According to some, the ICO is an innovative, low-cost method to raise capital and enables a widened range of potential investors to support the development of new, software-based enterprises.

In 2017—the year that ICOs entered popular consciousness

—453 ICOs raised an estimated $6.58 billion.

By July 1, 2018, an additional 684 ICOs had raised an estimated $17.47 billion.

Yet only a few months later, ICO project valuations were at fractions of previous years’ highs, causing some analysts to proclaim a “crypto winter.”

Twenty-four billion dollars raised over eighteen months is not chump change, but Facebook raised sixteen billion dollars in one day with its 2012 IPO.

Though one might not jump to read an entire law review article about Facebook’s IPO, an article about the strange world of public coin offerings may present a more compelling proposition. Indeed, an inquiry into ICOs could be fascinating even if (perhaps especially if) the entire ICO market were to dry up tomorrow.

As we aim to show, ICOs have much to teach us about the uneasy relationships between law and technology in our present moment.

To students of capital markets, the interest should be obvious. One basic question about our new financial contracting world is simple: How are investors protected from exploitation?

For regulators, scholars, and investors this issue is an increasingly pressing one. As of early 2019, government agencies at both the federal and state levels have launched ICO investigations, and multiple firms have been charged as fraudulent or criminal enterprises.

Even blockchain technologists admit that ICOs as a form of fundraising suffer credibility problems, as many projects have still not delivered functional products.

Less obviously, an understanding of the ICO experience can also inform debates about the digital future of capitalism.

ICOs represent the increasing financialization of internet-based peer production, and they also reflect the informational ecosystem the internet has wrought. The legal system’s interactions with these trends are on display in what follows.

This Article is built around a survey of the 50 ICOs that raised the most capital in 2017 and the role that computer code plays in structuring them. The presence of a cryptoasset at the heart of an offering enables entrepreneurs to deliver investor protections through computer code, rather than through legalistic means. This technological capacity was central to the ideological and practical case advanced by the entrepreneurs who engaged in ICOs. In the 2017 market, founders spoke of automated, “[d]ynamic [c]eiling[s]” for cryptoasset supply;

of placing founders’ cryptoasset allocations in “time-locked smart contracts” to align incentives for productivity;

and of replacing trusted parties with decentralized and verifiable computation.

We take an initial look at examples of smart contract design to establish that code does have the potential to become either a substitute for or a complement to old-fashioned legal governance in financial contracting.

But potential is not “reality,” and this study shows just how far code falls short of expectations for the top 50 ICOs of 2017. We analyze the relationship between the “paper” promises made by ICO promoters in their offering documents and the actual functionality of the digital assets they deliver. This Article establishes actual functionality by examining the smart contracts associated with each ICO, along with the broader software environments through which those smart contracts function.(These are known as “distributed ledgers” or “blockchains,” which we discuss further below.) Through careful auditing of the gap between what ICOs promise and what their code delivers, we aim to present coin offerings at a deeper level of institutional detail than is currently present in the literature. Indeed, though legal scholars have begun writing about smart contracts in theory, we are the first to take smart contracts seriously as real-world objects of study.

We evaluate our sample on three aspects of governance that ICO proponents have claimed can be delivered through code and which economic theory suggests should be salient to ICO investors. First, did ICO promoters make any promises (and encode those assurances) to restrict the supply of their cryptoassets? Second, did ICO promoters pledge (and build their promises into smart contracts) to restrict the transfer of any cryptoassets allocated to insiders according to a vesting or lock-up plan? Third, did ICO promoters use code to retain the power to modify the smart contracts governing the tokens they sold, and if so, did they disclose (in natural language) that they had allocated themselves that power? Credible commitments regarding these salient cryptoasset qualities should matter to an investor interested in the economic fundamentals of an ICO.

Our basic finding is that ICO code and ICO disclosures often do not match. In a financial ecosystem built around the proposition that regulation is unnecessary because code is the final guarantee of performance, the absence of coded governance protections is troubling. We also show that at least some popular ICOs have retained the power to modify their tokens’ rights but have failed to disclose that ability in plain English.

One takeaway is that no one reads smart contracts,

making them a rickety wheel on the ICO investment vehicle. Why might this be, and how significant is it? In evaluating our findings, we consider a few potential explanations for the mismatches between code and disclosure that we observe. We ultimately conclude that while the disjunct is troubling, the normative implications of our project will turn on learning more about who buys ICOs and why.

We proceed as follows. Part I provides clear and precise definitions of various aspects of ICO machinery. It also presents the history of various components: cryptocurrencies, blockchain-based networks, smart contracts, and ICO technology. Part II describes the three ways that we evaluate the quality of an ICO’s paper–code match and offers an introduction to the mechanisms by which tokens can vouch for quality. Part III presents the methods of our empirical study. It describes our sources, collections, coding, and smart contract audit procedures. Part IV offers evidence that the ICO market does not vet smart contract code for the qualities we have identified and offers theories as to why. It also suggests how researchers could help regulators and lawmakers in better understanding and overseeing this new business form.

I. An Introduction to Tokens

To set the stage for our analysis of ICO quality—and our premortem on the current market’s pathologies—this Part presents an operational account of ICO components and mechanics.

A. From Debt and Equity to Native Coin

Consider a group of entrepreneurs who want to create a soda company. Though they have an amazing recipe, they lack sufficient seed capital to quit their day jobs and market their soda to the world. To access the traditional capital markets, they might form a corporation and seek a business loan, or perhaps a few rounds of private venture capital funding. If successful, they might then choose to issue shares on the New York Stock Exchange (NYSE). In exchange for payment of a price (in dollars) set by investment bankers through careful underwriting, the team would part with shares of its company. The purchasers of those shares would then possess a bundle of rights to govern the corporation, along with residual claims on its assets in proportion to the number of shares they own. Once built, the corporation could charge its customers in dollars, pay its employees and suppliers in the same, and then distribute the leftovers to its shareholders.

The new world of coin-based finance looks different from this traditional model. Instead of issuing contractual claims on the assets of a legal entity (in the form of debt or equity), the team might now issue a token—call it Colacoin—that it promises will be the only way to buy sodas from its (yet to be deployed) vending machines.

The team could also pledge that possession of Colacoins would enable their holders to vote on proposed alterations to the vending machine’s prices. Further, they could even commit to paying suppliers—bottling companies, truckers, lawyers who work for them—in Colacoin. If, and as long as, the dehydrated people of the world want access to machine-vended cola, then Colacoin will hold value. And if Colacoin is easily exchangeable for dollars, then the nascent company’s truckers and lawyers will not mind receiving their initial payments in a strange currency. Replace Coca-Cola with a software-based venture (like a file-sharing service or a platform for streaming video), and Colacoin with a cryptoasset, and you have an ICO.

Obviously, the scenarios differ in a few ways. First, they diverge in terms of how they allocate claims on the entrepreneurs’ business. Traditional capital markets require business owners to contractually divest themselves of various rights over their corporation’s assets.

In contrast, the ICO method can leave economic ownership and legal control unencumbered.

Second, they vary in their source of value. While stock prices should reflect the net present value of the legal rights to the company’s expected future cash flows,

cryptoasset pricing should reflect an equilibrium between token demand, which is driven by the present value of expected future use and exchange options within the token’s native ecosystem, and token supply, which is driven by the token’s monetary policy.

Third, the infrastructure of capital markets enables vetting, trading, and liquidity in established ways. A mighty edifice of regulation and institutional capital stands behind each issuance: Investors know, or at least have the tools to inform themselves about, what they are getting. By contrast, cryptomarkets are new, their players mere years or months old.

No Wall Street investment bank has backed an ICO.

Indeed, the absence of ICO-specific regulation and intermediaries is seen to be a feature, not a bug, by many enthusiasts.

Finally, and perhaps most significantly to our lawyer-readers, ICOs expand the role played by computer code in governing transactional relationships. Traditional capital-market transactions are heavily mediated by laws, regulations, contracts, and social norms.

ICO transactions augment, and perhaps replace, those mediators by embedding controls within the smart contracts through which rules function.

At the same time, they also create new roles for lawyers and other legal-adjacent personnel.

The Colacoin clearly would be a far more experimental way to raise capital for the underlying soda company than through the sale of debt or equity.

Yet despite their differences, the scenarios share something at a particular level of abstraction: The value of debt, equity, and Colacoin tokens all depend heavily on the success of the entrepreneurial team in building and attracting customers to the product.

B. Understanding Cryptoassets

A working conception of ICOs begins with the cryptoassets—the digital coins and tokens—at the center of the operation. Like a physical coin, a cryptoasset is scarce and control over it is transferable. But while physical coins are transmitted hand-to-hand (or hand-to-machine), changes in control of cryptoassets occur through the networks that host them (via the transfer of a digital key).

Indeed, a cryptoasset is nothing more than an entry in a ledger that specifies that a particular user, identified by a certain “private key” (essentially, a fancy password) is the sole party able to exercise a discrete set of powers associated with the ledger entry. While their private keys might travel hand-to-hand in the physical world, the actual cryptoasset is destined to remain a mere ledger entry, forever locked inside its “native” protocol.

Cryptoasset history begins with Bitcoin currency and the Bitcoin ledger (also known as a “blockchain”).

Prior to their advent, money was either held in physical form (for example, coins or paper notes) or on the ledger of a centralized intermediary (for example, bank deposits or PayPal balances).

Bitcoin is the first significant digital currency system that needs no centralized intermediary to maintain proper books.

The key to the ledger’s design—and that of the public blockchain-based systems in its wake—is how it maintains a trustworthy record of ownership rights. Rather than being centralized within a single firm, the Bitcoin ledger is replicated and distributed across a network of computers that communicate with each other via the internet.

These computers are called “nodes.”

When a holder of bitcoins distributes a message to the network’s nodes asking to transmit some bitcoins to another user, the transactors need not rely on the trustworthiness of any actor in the system to revise their copy of the ledger appropriately.

Rather, they rely on economic incentives and code-based controls that govern the nodes’ behavior to ensure that all copies of the ledger are updated identically.

The shift toward a broad range of blockchain-based business plans was realized in another network: Ethereum. The designers of Ethereum produced a general-purpose computational system that operates through a public blockchain.

To perform computations on this decentralized “world computer,” users must pay a per-function fee of “ether”—a “gas” charge—which functions as Ethereum’s currency.

As a result, the value of ether depends significantly on the supply of, and demand for, computational power active on the Ethereum system. One of the key reasons for Ethereum’s popularity is its support for snippets of computer code that interact with the ledger known as smart contracts.

One can think of smart contracts as a prewritten set of system-performance rules. Just as legal contracts govern the allocation of paper money among transactors, smart contract code governs the transmission of ether, or other stored assets, among transactors on the Ethereum system.

To understand how Ethereum works, imagine that you drop a quarter into a vending machine slot and down falls a can of Coca-Cola. This “humble” mechanism serves as the inspiration for wide-ranging creativity on Ethereum, where smart contract engineers write scripts about how the system will behave in response to various inputs.

These inputs might include basic information about where to send ether, and also more complex information, like data from a weather vane.

Ether plays the role of both the vending machine’s quarters and its most important payload—the Coca-Cola of the system. Indeed, because ether acts as a decent (if volatile) currency, one can engage in smart contracting that attempts to mimic paper-age agreements for insurance,

escrow,

or even something akin to corporate formation.

To build increasingly complex and interoperating mechanisms within Ethereum, its community has begun developing standards—“fill in the blank” templates that perform agreed-upon functions. One of those—standard “ERC-20”

—plays a large role in our story. It establishes a simple template to create (or “mint,” in crypto-lingo) and operate entirely new cryptoassets within the Ethereum system. This is what the description of the standard looks like in code:

Figure 1: The ERC-20 Interface

Creating a new cryptoasset typically requires a minimum of approximately fifty lines of code and three decision components: the asset’s name, its ticker symbol, and the number of units—or “tokens”—to mint.

C. ICOs Hit the Bigtime

In 2014, Ethereum raised real money by selling ether to the public.

The next major ICO was Augur, which concluded in October 2015.

The market grew slowly until 2017, when it hit the gas.

Figure 2: Number of ICOs by Month

Figure 3: Total Raised in the ICO Market by Month

As the ICO market exploded so too did regulatory interest in its activities.

Such scrutiny is no surprise: ICOs, like many internet-based phenomena before them, intentionally take place at the regulatory perimeter.

They exploit a basic tension between the cross-jurisdictional and pseudonymous aspects of cryptocurrency transactions on the one hand and the objectives of regulators on the other.

The question of just how significant the demand is for cryptoassets among money launderers and tax evaders is not one we answer here, but it sits as a backdrop to the inquiry that follows.

In the traditional IPO context, the Securities and Exchange Commission (SEC) and state securities regulators oversee issuer activity from soup to nuts.

They mandate registration of securities issuances, require pages and pages of disclosures over the life cycle of a security, restrict the trading activities of various parties, and possess myriad investigation and enforcement powers to effectuate their portfolio of laws and regulations.

As of 2018, no similarly clear regime was in place for ICOs.

In lieu of the heavily lawyered products of IPO documentation, the ICO market agreed upon a less formal document known as a “whitepaper.”

Like governmental and nonprofit whitepapers that seek to exemplify authoritative subject mastery while gesturing toward collaborative openness, cryptoasset whitepapers are public documents that describe promoters’ plans for development and solicit community involvement.

Authoritative copies are typically available in PDF form on promoters’ websites and are provided through listing services like coinschedule.com.

This makes whitepapers a transparent form of investor information but obviates the need for outside vetting before they go live.

The legal status of whitepapers (and accompanying tweets, Medium posts, Reddit comments, and social media buzz) is unclear at best. Sometimes, whitepapers refer to—and embed—contractual terms and conditions of sale.

In such cases, they provide information about product attributes, which would function as contractual warranties. In other cases, they resolutely speak in future tenses, offering difficult-to-parse details about what is promised and what is merely aspirational.

Absent clearly communicated and defined offers, it is unlikely that buying a token in reliance on such documents constitutes a traditional contract, though other regimes of consumer protection law (state consumer Unlawful Trade Practices statutes, false advertising, securities laws) might fill the regulatory gap.

Beyond the informational environment, ICO issuances also differ from IPO issuances in terms of where they are traded. While public equities trade on established secondary markets like the NYSE or NASDAQ, cryptoassets trade on hundreds of upstart markets, sometimes under light-to-nonexistent regulation.

They are located in diverse jurisdictions and have been embroiled in a range of legal controversies.

Despite these significant divergences between IPOs and ICOs, the near-identical nomenclature is no mistake. Both entail the issuance of assets whose value depends on the success of a business venture, and both are offered to so-called “retail” investors. These essential similarities in economic function have not been lost on federal securities regulators in the United States, who lately have begun to apply the wonderfully medium-agnostic securities laws to regulate ICOs.

A number of state regulators are also actively policing bad actors in the ICO market.

Assuming the ICO market matures, these outlier-policing activities will likely be augmented with broader regulatory schemes aimed at standardizing disclosures for the mine run of ICOs.

For that effort to be successful, it is imperative for policymakers to understand the contours of ICO transactions, and the institutional environment in which they take place, in detail. We turn to offering such detail now.

II. Smart Contracts in the Wild

This Part seeks to better understand some of the basic economics of cryptoassets, and the roles that code—specifically, smart contracts—might be playing. The central relationship we investigate is that between “paper” and “code.”

Ever since the cryptogra­pher (and law graduate) Nick Szabo first introduced the concept of smart contracts, their artisans have sought to use code to replace and augment traditional institutions for ensuring performance within transactional relationships. The utopian ideal is a “grand merger of law and computer security,”

which might render the protections offered by the former to be at best superfluous.

That hope is emphatically present in some of the offering and promotional materials that crypto investors receive. These materials speak of sales where smart contracts will “stop accepting commitments at 888,888ETH hard cap,”

of automated destruction of excess cryptoasset supply,

and of “Reserve Tokens . . . locked in a smart contract” according to predetermined specifications.

They promise with precision that “new founders’ tokens [are] distributed pursuant to the launch of an EOSIO Platform in a smart contract and [that the default EOSIO Software configuration] releases 100,000,000 of such tokens . . . linearly to Block.one every second over a period of 10 years.”

While markets of unsophisticated investors typically require investor protection laws and intermediaries to protect against market manipulation,

the “crypto industry” has “greater transparency, fewer middle men . . . [and] programmatically enforceable contracts.”

That is, this community tries to make concrete the ideological project of using code to replace the rules of entity governance that law currently creates.

Practical realities also motivate a turn to code in this space. Even if the paper surrounding ICOs created legally binding obligations—which it sometimes will not

—legal rights are only as valuable as their practical enforceability.

Because cryptoassets can move freely and pseudonymously through the internet, it can be difficult to pin them down to particular jurisdictions.

And the promoters of many ICOs have set up shop in ways that make it challenging for U.S. courts and regulators to reach their assets.

Thus, promises that are made in marketing documents and terms and conditions of sale, even if legally binding, might lack an easy and practical form of legal remedy.

Given this background, an ICO that promises particular, encodable governance terms but does not encode them is not delivering on an archetypal feature of this financial form. According to those who argue the form is novel—so novel as to deny the need for wise intermediaries, venture capitalist (VC) vetters, and regulators with teeth—it is the immutable, transparent code that enables (and creates) a trustless but trusted market.

With that foundational, code-centered principle in mind, we ask the classic question that motivates so much of the law of finance and corporate governance: How can investors turn over productive control of their money to entrepreneurs while also protecting themselves against exploitation?

This is a timeworn problem. In the old-growth public markets, investors can rely on disclosure regimes (imperfectly backed by public agency enforcement) and fiduciary rules (imperfectly backed by court enforcement) to manage risk. In private firms—ranging from family-owned businesses to VC-backed startups—contracts must generally suffice. What is new here (if anything) is that the cryptoasset community proposes a technological solution—the token’s coded rules—to manage some crucial sources of agency cost.

One type of bargained-for protection is a constraint on the supply of the investment asset for sale. In the traditional corporate context, each share sold to investors provides a legal right to a piece of an enterprise’s residual assets. In an efficient market, changes to the number of outstanding shares would affect share price but not firm value.

Put another way, the enterprise’s assets are like a pie, and every newly issued share makes each slice smaller. Because they want big pieces, early shareholders seek protection against late-breaking stock issuance.

Traditional corporations act through human agents; those humans are only able to issue as many shares as the corporation’s (amendable) Articles of Incorporation allow.

Exploitative issuances are deterred by the common law of fiduciary duty.

Supply constraints matter to cryptoasset investors, as well. Remember, tokens are not typically claims on the enterprise’s residual assets.

Rather, they normally provide investors the right to use or govern the actual system whose hypothesized construction is funded by their money.

Shareholders in Coca-Cola care about the value of their residual claims on Coca-Cola, Inc.’s assets. But the holders of Colacoin care about the demand for, and supply of, use-rights to the future system. The number of use-rights available—in other words, the “money supply” of circulating tokens—is thus a central determinant of individual token price.

The value of a token, like the value of a stock, can be diluted through new issuance. Just as our Colacoin owners hope that legions of thirsty people demand vending-machine cola, they also pray that Coca-Cola will not engage in rampant inflation of the token supply. Similarly if Coca-Cola promises to remove tokens from circulation (so-called ‘burning’), Colacoin owners would expect the value of their investment to rise.

ICOs, unlike corporations, are not birthed through the filing of Articles of Incorporation that limit stock issuance. There is no analog to the fiduciary rules, or the Delaware Chancery Court, that govern when dilution can occur. Cryptoassets are instead created, limited, and used up according to code controlling the contents of a blockchain.

Thus, a purchaser’s protection against wanton inflation of supply comes directly from the cryptoasset code.

That is not to say that ICO promoters might not also make soft-law promises about supply—in fact, they often do, and such promises likely bear on value.

But when such promises are not manifest in the code, investors’ ability to enforce constraints will be limited to their very uncertain ability to sue and recover founders’ assets. Because ICO project founders can do business entirely over the internet, they may be hard to find and sue. Further, it remains to be seen which causes of action might be successfully pursued in the ICO context.

A second bargained-for protection has to do with the threat that key members of the entrepreneurial team will walk away from the project. Investors generally protect against desertion (and motivate exertion) through a set of carrots and sticks offered to managers. They incentivize them with equity options—rights that enable managers to share in the firms’ future profits—but condition those options’ exercise on contractual conditions, in other words, vesting.

Option, lock-up, and vesting rules attempt to align managers’ incentives with those of the firm and are endemic in the early-stage VC financing world.

In ICOs, classic options are quite rare, but token-vesting promises are common.

As one project (marketing its vesting promises) wrote, it “is a governance practice designed to ensure long-term alignment of interests and is standard for any serious project.”

Another wrote that “[v]esting is a must. There are no excuses not to do it. It aligns everyone’s incentives and ensures that no founder dumps happen.”

As with promises regarding supply, vesting promises that are coded are enforced automatically.

Those merely present in marketing materials or paper contracts are less likely to be enforceable.

Uncoded vesting promises might (or might not) be present in governing documents of the underlying formal organizations. They likely would be located in the employment contracts of the various managers and founders, but such contracts probably will not be publicly verifiable.

Perhaps to allay this very concern, ICOs often make claims about their smart contract vesting. For instance, one promises that:

20% of the BMCs will be allocated to the founding Blackmoon Crypto team and advisors, locked in a smart contract with a 24-month vesting period, and six-month cliff. These BMCs won’t be immediately tradable and will secure the core team members by ensuring their motivation after the Distribution Period.

Because promoters focus on it so much, examining how and whether vesting promises are coded sheds light on how strongly investors should buy the claim that a project’s key people will not exit with their newly raised capital.

That is not to say that failing to code vesting means that founders are about to abscond: Coded vesting rules are only one way to protect against looting. However, it is a way that is technically feasible and consonant with the industry’s ideological claim that law is a poor substitute for code.

A third and final protection against exploitation in ICOland is the supposition that the initial rights investors receive are not modifiable. Part of the appeal of cryptoassets and smart contracts that operate on blockchains hinges on their “immutable” nature. Legal contracts contain ambiguity and permit formal and informal modifications, but smart contracts are purportedly drafted in exhaustive, precise code that seems to set the parties’ obligations permanently.

Because cryptoassets are defined by smart contracts, whether those smart contracts are modifiable should profoundly impact price and receive intense investor scrutiny.

A fully disclosed regime that permitted a token to be modifiable should have uncertain effects on value. On the one hand, no social enterprise existing over any medium-length time period can have functioning governance rules immutably fixed at its inception. Human relations, including financial ones, evolve. Imagine a constitution that could never be amended, or a similar corporate charter.

Thus, investors told that every rule of a token ecosystem had been irrevocably fixed at their creation should (we think) recoil at the coders’ hubris.

On the other hand, when one party holds the power to modify formal relations, other parties bear risk. To the extent that a smart contract defining investors’ rights is mutable at the will of the issuer, investors ought to expect that the limits of that process would be explained in detail. Consider a fully modifiable Colacoin, for instance. One day the issuer might say that your coin, which you thought bought you a right to delicious fizzy soda, could only be used to purchase noncarbonated beverages or could be used to purchase cola only when you inserted additional fiat currency.

The “rights” you bought would be notional.

Surprisingly, until July 2018, the crypto industry rarely discussed modification.

That month, in response to a hack of a popular token, a handful of prominent cryptocurrency voices sounded the alarm that several circulating tokens were modifiable at will.

They were, to summarize a long Twitter thread, angry. This is not conclusive evidence that modifiability is seen as a negative characteristic of tokens, but it does suggest that the coded ability to modify a token is not an anodyne fact. In short: We would expect that if token code is explicitly modifiable, that fact would be disclosed. Similarly, if the token code’s governance provisions are not modifiable, we would expect that the marketing documents would explain how, and why, the project can evolve with the times.

With  these  three  investor-protection  ideas  in  hand,   we  now  will  provide  examples  of  how  they  are  actually  accomplished  in  the  real  world.  We focus  our  discussion  on  Ethereum  code.   Ethereum nodes  operate  a  simulated  computer  called  the  “Ethereum  Virtual  Machine”  (EVM).

This simulation runs by using both data and code (smart contracts) stored on the Ethereum ledger.

The smart contracts exist on the Ethereum ledger in a complex, hard-to-read machine language known as bytecode.

But they are most commonly written in an intuitive program­ming language called Solidity.

Solidity hides the internal details of the EVM and the complex machine language that it processes.

Before being uploaded to the blockchain, a program called a compiler is used to translate the Solidity source code into Ethereum bytecode.

This Article presents examples in Solidity.

Solidity code contains four major types of entities: variables, functions, events, and modifiers.

• Variables are the data-storage components of any smart contract and, in the case of a token’s smart contract, store balances for each user-address, along with other data required for the smart contract to operate.

• Functions describe the rules by which the smart contract operates, storing discrete chunks of code that perform specific

  Functions are executed (or “called”) by sending a specially formatted transaction to the Ethereum network.

  Functions are identified by a name and a set of parameters (or “arguments”) that are the inputs to the function.

• Events are signals that a smart contract sends to other applications or smart contracts programmed to receive them.

  They act as a form of logging.

• Modifiers allow a developer to easily restrict the execution of a function under certain conditions.

  For example, a developer may restrict the ability to mint new tokens to the smart contract owner alone.

Figure 4: An Example Code Snippet

To audit a given cryptoasset, we obtain a copy of the Solidity code (illustrated above), either from etherscan.io, where developers commonly upload their smart contract’s Solidity code, or from GitHub, a source code repository often used as part of the development process. Etherscan.io replicates the bytecode present on the blockchain but requires developers to upload Solidity source code for display.

The site additionally provides a verification feature, which allows users to check that the Solidity code matches the bytecode.

After obtaining source code, we then examine each function of a smart contract and manually track the role each line plays. We use code comments—explanatory lines of human-language text inserted by developers, which have no computational function—as guides to assist in identifying developers’ intentions.

A typical smart contract in our sample contains between five hundred and one thousand lines of code. We inspect that code, looking for the presence of our three investor-protection attributes.

A. Supply Promises

1. Minting. — Cryptoassets issued via ICOs are created through a process known as minting.

Recall that the Ethereum blockchain provides an extremely simple way to mint new cryptoassets through the ERC-20 standard.

But even if they do not conform to the ERC-20 standard, minted assets are typically created by executing relatively simple code on a blockchain.

In other words, a minted cryptoasset is created through an act of founder fiat. Billions or trillions of cryptoasset tokens are generated at a nominal cost reflecting fees paid to interact with the respective blockchain.

Then the team will typically commence an ICO, transferring the tokens to investors in private sales or to members of the general public in mass offerings. The sales are accomplished using smart contracts, automatically routing the project’s tokens to investors in exchange for other cryptoassets or, more rarely, for fiat currency.

Minting is an essential part of the ICO story. It creates the opportunity for early-stage blockchain projects to rapidly raise capital without the formalities required by corporate law and regulation. But it also opens the door to fraudsters, who can mint and sell tokens based on the expectation of a given supply schedule, only to mint more than expected—or to mint a special stash for themselves.

To understand minting, let’s examine an ICO for a cryptoasset called Kin (ticker symbol: KIN), orchestrated by a company called Kik Interactive (“Kik”). Kik runs a global messaging platform with approximately 300 million registered users.

Like other digital communications companies, it has sought to broaden its business model by turning to blockchain.

Ultimately the company would like to build a “decentralized ecosystem of digital services for daily life.”

If all goes according to plan, Kin will be the currency enabling and constituting this utopian ecosystem.

Building on Kik’s previous efforts to develop in-app loyalty points, Kin is meant to serve as a “transaction currency” that Kik users can exchange for premium features, like membership in “VIP” chat groups with celebrities.

It will also incentivize developers to work alongside the project.

According to its whitepaper, Kik planned to mint ten trillion Kin tokens, of which one trillion would be put up for sale.

A blog post from Kik’s founder and CEO states that 488 billion were sold for $50 million in a presale arranged with specific investors and venture capital funds active in the industry.

The remaining 512 billion tokens were offered to the public during the project’s ICO, which ran from September 12–26, 2017.

Ultimately, the ICO raised $98.8 million for the project, bringing the total amount raised to almost $150 million when including the private presale.

We audited the smart contract code to understand how these supply promises were accomplished. The cap on the number of tokens available is indeed coded in the smart contract. In addition, the smart contract mandates two discrete sale phases, and there are coded limits on how many tokens could be sold during each. One of these phases is the project’s ICO, and the other is presumably the private presale.

Figure 5 illustrates the code’s function.

Figure 5: Kin Project Code

That is minting. But there are other processes that can alter supply.

2. Increasing Supply.  —  The  full  supply  of  a  minted  cryptoasset  can  be  set  at  the  outset  of  a  project, or can fluctuate depending on how much investment the project receives.

The  circulating  supply  of  the  asset  can  also  fluctuate.  For  instance,   a  founding  team  could  retain  some  of  an  initially  minted  asset supply and use it to inflate the circulating amount in the future.

Similarly,  a  team  might  alter  rules  governing  the  ICO  process  to  achieve  various  supply  effects. For example, the Kin ICO smart contract contains code to enforce volume restrictions for individual purchasers.

Each address permitted to participate in the sales may only send a limited amount of ether to the smart contract that disburses KIN tokens.

However, these limits could be manually modified by the smart contract owner at any time.

The important point here is that maximum supply of a minted cryptoasset can be specified and enforced (or not) via the code comprising the cryptoasset itself. Projects can also contain an absolute cap. But some cryptoassets lack this feature. For example, there is no absolute cap on the amount of ether that can be created.

Indeed, there is heated debate about whether this is a desirable feature or not.

Supply caps are a typical part of an ICO’s marketing materials.

As one promoter said, “Even if on the last day of distribution Richard Branson shows up on a resplendent white yacht packed stern to bow with cash, we wouldn’t be able to sell him any more.”

3. Decreasing Supply (or “Burning”). — In prototypical blockchains, cryptoassets circulate like money. Think of Colacoin: If you drop a Colacoin in a vending machine for a pop, the coin will get picked up by a Coca-Cola employee, head to the corporate vault, be used in payment for the vault guard’s salary, and then—maybe after the vault guard goes for a jog—get dropped back into another vending machine in the system. To take one example, circulation is the default rule for ether.

When someone pays ether to complete a transaction on the Ethereum blockchain, its recipient can spend that ether right away.

But perpetual circulation is not always the fate of a cryptoasset. Cryptoassets also can be used up, or “burned”—that is, destroyed.

Burning can play important roles depending on the business model envisioned by project founders. Some might advertise that the token could be exchanged for the right to access the completed project. Then, the exchanged asset would be permanently “burned” upon use. Some projects described plans to actively buy tokens from holders and then burn them, creating token price appreciation similar to a stock buyback.

In others, only those tokens exchanged for certain features in the product—for example, tokens paid as fees—are burned. Finally, burning is used as a mechanism in ICOs, as a way to destroy unsold supply.

Burning on the Ethereum blockchain takes two forms. The first is a simple transfer of tokens (or ether) to the address of Ethereum’s “genesis” block,

consisting of all zeros. As this address has no owner, the tokens cannot be spent and as such are “burned.” The second is to use an Ethereum smart contract’s function programmed with the logic to either delete the ownership record and decrement the total supply accordingly, or that which destroys the entire smart contract, rendering any tokens or ether sent to that address inaccessible. The below snippet shows a characteristic burning function.

Figure 6: Burning Code

A smart contract with appropriate code can keep track of burned tokens, enabling investors to easily audit the current supply.

Not all burning promises are executed so cleanly. Consider, for instance, Paragon, an ICO that aims to “revolutioniz[e] all things cannabis with blockchain.”

Lest you think it’s all a smoky haze (and we promise that’s the first and last joke), the project does have a dedicated cryptoasset: an ERC-20 token called PRG. The whitepaper specifies that PRG holders will be able to interact with all of the project’s many initiatives; holders will be able to vote on real estate investments,

guide project governance decisions,

purchase access to coworking services,

and exchange tokens for local currency in cannabis-unfriendly jurisdictions.

In addition to these promises about governance, Paragon promised that any unsold tokens from the private or public sale would be burned.

And it describes a transaction fee system whereby “[a]ll fees on the Paragon ecosystem” incur a $0.000000005 charge (that’s five billionths of a dollar), half of which is burned and half of which replenishes the project’s PRG reserve.

We can perceive only a small part of this complex set of rules in the code.

PRG’s smart contract code does limit issuance to 200 trillion tokens. This is captured in Figure 7 below.

Figure 7: Paragon Supply Code

We also verified that Paragon contains code allowing users to burn a portion of their tokens. This is captured in Figure 8 below.

Figure 8: Paragon Coin Burning Code

However, we modeled the transaction fee system described in the paper and discovered troubling implications for supply. Following the creation of the smart contract, each transfer of a PRG token consumes approximately one-six-billionth of the total supply in transfer fees, half of which is paid to the owner of the PRG smart contract and half of which is burned. After a sufficient number of transactions the fee approaches the number of tokens remaining in the supply, causing the eventual demise of the network. This is captured in Figure 9 below.

Figure 9: Paragon Fee Code

B. Vesting Promises

If supply controls protect against the threat of dilution, vesting mechanisms protect against the threat of desertion.

They work either by delaying when the founder is granted assets or deferring the moment of their liquidity.

A smart contract usually provides for vesting by allocating a portion of minted tokens to insiders but then locking them up until some condition is satisfied.

The code prohibits the transfer, sale, or use of the tokens until the condition’s trip-wire is hit.

Most ICO-coded vesting is time-based, with few of the contractual conditions that come with stock vesting offline.

Let’s return to Kik and examine its vesting promises. In its marketing documents, Kik made fairly specific, detailed promises about token vesting. Of the ten trillion total Kin created, Kik’s whitepaper claimed that thirty percent would be allocated to Kik in exchange for its “startup resources, technology, and a covenant to integrate with the Kin cryptocurrency and brand.”

This stake would be subject to a vesting schedule that released ten percent every quarter, for ten quarters.

Further, sixty percent of the initial Kin was allocated to the Kin Foundation, the entity that is meant to gradually take control of the project.

This stake vests according to its own schedule.

A total of 0.061% of this stake will be released into circulation daily, or roughly twenty percent per year.

Kik even released a separate whitepaper detailing the vesting dynamics for the Foundation stake, specifying, for example, that the unvested portion of this stake will be around 4.6 trillion Kin on March 12, 2019.

Figure 10: Kin Allocation Code

The project implemented some of these promises in the code. The Kin smart contract creates vesting by maintaining a database of grants with a start date, end date, cliff, and installment length.

Grants are both creatable and revocable by the smart contract owner.

No more than 100 grants may ever be created and no address may receive a grant twice.

Every grant we have seen so far has a hardcoded cliff of one year, with two installments, one of which must be executed by the owner of the smart contract and on which is executed by the vesting trustee.

When the Kin ICO commenced, the developers created two new grants. One corresponds to Kik’s thirty percent stake and faithfully implements the ten percent per quarter vesting schedule described in the whitepaper.

Interestingly, the development team manually added a comment to the code showing that the address owning the stake belongs to Kik.

This suggests that Kik may have believed there would be at least some investor scrutiny over the technical governance features of its project.

The second grant corresponds to the sixty percent Foundation stake. We were unable to locate code for any of the highly detailed vesting mechanisms described in the whitepaper. We did observe that this grant is wholly controlled by the owner of a vesting trustee smart contract. Of course, offline ownership of that smart contract—the legal person within the Kin or Kik organization that actually receives the unlocked tokens—is not hardcoded into the Kin token code itself. It’s simply bestowed on whoever has the private keys for that smart contract. In other words, there’s nothing about the token code that enforces separate ownership of Kik’s stake and the Foundation’s. Instead, it depends entirely on the offline governance features of the project, enforced using traditional tools like corporate charters and bylaws (or not at all).

C. Modifiability

Beyond the specific protections against inflation of supply and desertion by key people, the promise of cryptoassets has also rested on the idea that investors are protected by the immutability of blockchain code. As we noted above, lawyers might well think of this as a wacky idea. And sure enough, immutability has indeed gone by the wayside for a number of ICO projects. Disclosure of what we refer to as “modifiability” is another matter. Though some token teams do advertise that tokens may provide new rights in the future, they do not explain that modification is a way to change any aspect of the token, not just activate valuable new features. And yet, as we will see, modification is built into the design of some ICO systems. How does this work?

In the simplest setting, a developer can simply copy the contents of the data stored in a smart contract, and create a new smart contract, prepopulated with the data from the former. While those who owned tokens in the context of the original contract also own tokens in the new smart contract, the developer is free to create new code controlling the behavior of the latter. More concretely, an issuer may refuse to honor the original token when they finally complete development of the product the ICO was designed to fund.

This can be accomplished using two sets of rules: a primary smart contract with which users interact and a series of secondary smart contracts whose code is incorporated by reference.

Our lawyerly audience can think of the typical relationship between a website’s Terms of Service and its Privacy Policy: The former usually contains a link to the latter, and purports to bind visitors to both.

Or, think of a public law that points to a private standard, like a city code that adopts LEED green-building standards.

The standard can be updated privately, thereby modifying the effect of public law.

A similar “pointing” mechanism enables the modification of cryptoasset smart contracts. All tokens using this method share identical code. The primary smart contract stores for each user the address of a secondary smart contract, containing the most recent set of accepted modifications.

The  owner  of  the  primary  smart  contract  can  modify  the  code  by  proposing  a  new  secondary  address,   defining  the  smart  contract  whose terms  will  be  incorporated. In one example we found (Monaco), the code gave users three days to opt in or out before the modification spread. When a user opts out, her current secondary smart contract address is frozen until the next time they explicitly opt in. The default state of all users is opt in, as illustrated below.

Figure 11: Monaco Modification Code

The Polybius project provides another example. It is a proposed “fully digital bank accessible everywhere at any time . . . with a very efficient cost/revenue ratio.”

Eventually, Polybius plans to “grow into your daily servicer and companion ecosystem . . . enabl[ing] secure and seamless connections between life and the things we love and use every day.”

Investors contributing to the project can supposedly expect “higher returns” than those investing in traditional banks.

The development team did make some limited claims about smart contract modification. The token purchase agreement explicitly states that “Polybius shall procure that the Smart Contract is modified and/or amended via an additional smart contract” to activate tokenholder voting.

It further specifies that the voting mechanism will enable the development team to propose changes to project smart contracts and to implement the changes if they receive two-thirds of tokenholder votes.

There are no further details.

However, we found modifiability functions in the smart contract code that extended well beyond changes to tokenholder voting rules, as Figure 12 details.

Figure 12: Polybius Modification Code

Through this code, Polybius can propose modifications by deploying an entirely new secondary smart contract and linking it to the primary smart contract via the commitUpgrade function.

The primary smart contract does not allow the owner to make modifications directly—the owner must first propose the upgrade, which only takes effect after three days unless the user opts out.

In terms that legal readers will be familiar with, it is a “sticky default.”

Using these mechanisms, a development team can unilaterally change the tokens purchased by investors—or sometimes, propose changes that will be adopted if a certain percentage of users do not object.

Unless investors scrutinize both the potential for their tokens to be unilaterally modified, and the substantive terms of the modifications actually proposed, they are unlikely to discipline hasty or abusive changes. As we describe in Part IV, investors hardly pay attention to even simple nontechnical markers of quality. It is thus incredibly unlikely that they have the technical skills to monitor a development team’s use of modification.

III. A Survey of ICOs

Having identified three salient attributes of ICO governance, we now attempt to step back to look at a larger set of issuances to see how (and if) they dealt with governance issues. We reviewed the fifty largest 2017 ICOs by amount raised in dollars.

For each listed promotion, we scrutinized the whitepapers, token sale agreements, and computer code posted by the promoters. Appendix C pulls quotes about supply, burning, vesting, and modification (if they are available) from the issuers’ public statements.

We compared those promises, read by investors, with what we discerned from close examination of software code. Our approach is empirical, but obviously neither comprehensive nor representative of all 2017 ICOs.

A. The Scene from 50,000 Feet

The fifty firms we studied were reported to have raised a total of approximately $2.6 billion at their ICOs and the notional initial market cap was $7.0 billion.

In the sample, nineteen were headquartered in the United States, six in Singapore, and the remaining in a variety of countries, including Switzerland (five), England (two), China (two), Estonia (two), and Thailand (two).

By January 2019, eleven of the projects had not released any kind of alpha version or demo of their project.

Our approach to auditing is limited: We try to take the position of a sophisticated, but time-constrained, investor. Consider, again, Polybius. Its whitepaper makes several claims that would lead us to expect certain features directly coded into tokens or other smart contracts. The most striking example is the team’s promise that “according to the conditions of the ICO, payouts to tokenholders are directly connected to the earnings of the Polybius project.”

The team goes on to specify a range of offline activities that will support payment of the dividend, like preparation of audited financial statements, and tells readers to expect dividend payments in Ethereum.

Beyond ERC-20 compliance and the presence of a modification feature, we did not verify that any of these features are present, largely because Polybius’s coded governance exists in bytecode (which, as you will recall, is the Ethereum machine language). Without spending a large sum of money purchasing the time and know-how of a very motivated and talented reverse engineer, an investor would have to rely on vernacular promises.

Below is an excerpt of what the public-facing code (incorporated by reference) looks like.

Figure 13: Polybius Bytecode 

This  shows  that  it  is  not  merely  the  case  that  the  investment  depends  on  the  development  team’s  decision  to  actually  build  the  products  it  hypes  in  its  whitepaper.

Investors must also have faith—either that ordinary contract law litigation will back up old-fashioned terms of use, or that the bytecode, which essentially no one can or will parse, renders those promises operable.

Putting unauditable smart contracts to one side, here are the results of our analysis, which compares the software to promises made in whitepapers, blog posts, and websites marketed to investors.

B. Supply Promises: Scarcity and Burning

We begin with promises regarding supply. Of the fifty tokens, we audited the code of forty-five (four remain in bytecode or in proxy contracts, and one, FileCoin, has not released any code or token). We dropped projects without auditable code from our analysis. Figure 14 illustrates how such firms approached supply scarcity commitments.

Figure 14: Scarcity Audit Results 

Almost all issuers promise a supply restriction in their marketing documents (40 of 45, or approximately 90%). And most of those that promise a restriction deliver it (29 of 40, or approximately 75%). Overall, though, only about two in three (29 of 45, or approximately 64%) firms that we audited encoded a supply restriction. To be clear, this is not to say that the firms that did not deliver coded scarcity limits actually promised to do so—their marketing promises either did not mention scarcity or may not have discussed how it was to be effected.

The second sort of supply promise—burning—displays a different pattern. Figure 15 details our burning audit.

Figure 15: Burning Audit Results 

Here, fewer firms promised to burn tokens than promised to cap supply in the initial mint (19 as compared to 40). Of those that promised to burn supply, approximately 36% (7 of 19) did not fix that claim with code.

C. Vesting Promises

Of the forty-five auditable issuers, thirty-six promised vesting in their marketing documents or whitepapers, while nine did not. Figure 16 illustrates our findings.

Figure 16: Vesting Audit Results 

Figure 16 illustrates first that almost 20% of the sample did not promise to vest at all, which is a surprising result given the amounts raised. Second, of the 80% that promised to vest, the vast majority apparently did not use smart contracts to encode those rights.

D. Modification Promises

Finally, we describe the modification rules in the sample. Modification is rarely discussed in marketing materials: Only seven of the fifty firms discussed the token’s modifiability in their marketing materials or soft contracts. But overall, twelve of the fifty firms permit modification through their code. While most (4 of 7) of the firms that disclosed modification had code that backed up their promises, eight firms that did not discuss modification permitted it.

Figure 17: Modification Audit Results 

E. Summary

To  sum  up:   There  are  significant  differences  between  code  and  contract  in  our  sample.

Projects are making governance claims that look to be modeled off of offline VC or traditional equity-based rules intended to reduce agency costs, but they are not encoding those promises into the sort of trustless, decentralized systems which undergird their networks’ purported sky-high values.

IV. Coin-Operated Capitalism?

So far, our inquiry has been motivated by two goals. First, we have tried to capture the reality of the ICO form as it existed in 2017–2018—a snapshot of a supposedly revolutionary innovation just after its birth. Second,  we  have  attempted  to  understand  smart  contracts  at  a  deep  level  of  contextual  detail.  They  are  at  the  heart  of  the  innovation  story  told  by  ICO  proponents,  some  of  whom  claim  that  code will increasingly be able to replace traditional law.

We have traced their early history,

explained  how  they  were  expected  to  function  in  the  ICO  market  of  2017,

and  taken  stock  of  the reality.

In this Part, we evaluate the distance between expectations and reality.

As we established in Part III (and in detail in Appendix B), for over 20% of ICOs in our sample where promoters promised cryptoasset supply restrictions, and 35% of promised token burning, we could not observe restrictions hard-coded into smart contracts. More starkly, we could not find hard-coded vesting restrictions in twenty-five of the thirty-six ICOs where promoters promised to adhere to such restrictions. Finally, of twelve ICOs for which our audit revealed that a central party could modify the functionality of the cryptoasset’s smart contract code, only four disclosed that ability in their promotional materials.

Our results raise serious questions about the role of code in ICOs. Do investors punish ICOs that fail to build key protections into code or fail to disclose the power of modification? If not, is that because code does not matter as much as its proponents claim it does? Or is it because the ICO market is broken? We examine those questions in the sections that follow.

A. Paper, Code, and Market Response

For a minute, let’s look at our results from the perspective of an ICO advocate who believes that code has the potential to be a cheaper and better way of delivering investor protections than traditional venture financing routes. Should this person be troubled by our results? At one level, the answer has to be yes. The fact that a majority of the leading ICOs—each of which raised over $20 million—fail to write their own vesting promises into code is inconsistent with a story about code replacing law. It also raises serious questions about whether investors are adequately protected from founder desertion.

But our ICO advocate might push back. Perhaps we are wrong about the absence of hard-coded rules (and if we are, we hope to be corrected). Or, maybe, investors do take the problems we observed in Part III into account when investing. That is, maybe problems with coded investor protection are reflected in market prices.

Though the ICO market is young, we are skeptical of this “investor-protection code is priced” thesis. As a first cut, the sheer number of problems in our sample suggests otherwise. Our results show that the majority of the top-grossing ICOs of 2017 had major problems with how code bore out their antiexploitation disclosures.

To  quantify  the  idea  of  paper–code  distance,  we  refer  to  any  uncoded  investor  protection  for  supply,  burning,  vesting,  or  incongruence  between  code and disclosures regarding modifiability as “distance.” Using these data, we score each ICO from zero to four.

Of the fifty ICOs, we give forty-nine a score because we can evaluate either the token or the associated smart contracts. Fourteen have no distance, ninteen have one marker, twelve have two, three have three, and one (Monaco) has four. If investors know about the problems we’ve identified, then the makeup of the top fifty suggests that they don’t much care.

Nor do the postsale market metrics we are able to observe enable us to say a great deal about the “code is priced” thesis. We do not see significant changes in code congruence over time, and we lack a natural experiment on initial code pricing. What we can observe is whether (over time) firms that encode their disclosures have different returns and trading volumes. An approach suggested to us by a commentator on an earlier draft of this paper

was to develop a rolling weighted portfolio of the prices (and trading volume) of our fifty projects, controlling for their disclosed and coded governance rules. Using this approach, we find that—consistent with earlier work—disclosed governance rights do seem to promote better returns.

Figure 18: Vesting Disclosures and Rolling Average Cumulative Returns

The next figure repeats the first, but now breaking out projects that coded vesting and those that promised but did not code it.

Figure 19: Vesting (Coded vs. Disclosed) and Cumulative Returns

Here, we can see that firms that coded vesting had returns that were indistinguishable from those that did not code it. Professor Robert Bartlett reported similar results on scarcity, as well as trading volume. In a series of regressions, he found while disclosure of vesting and scarcity were correlated with higher returns, coding of those attributes had no consistent and significant effects.

Trading volume and price were, however, closely tied to Bitcoin’s price and trading volume, a result that fits with other recent research.

Finally, we are skeptical of the “investor-protection code is priced” thesis because buy-side literature in 2016–2018 rarely treated the guts of code as something worth considering. Like stocks, ICOs have developed a wide range of secondary information sources, including “ratings” websites. But most of these raters do not vet smart contract code. Of the top five English-language rating sites by Alexa ranking, which measures how popular a website is,

only one posts information about code quality, though not of significant detail.

Similarly, code takes a backseat to other investment drivers in the retail valuation literature.

In the period before 2017, advisory publications focused on a project’s ability to deliver anonymity and decentralized governance, which in turn was thought to help hedge against regulation.

In the period after 2017, guides focused on the potential for widespread functional use within the startup’s system,

the reputation and involvement of the founders and creative team,

and the avoidance of obvious scams.

Eventually, some investors gave up on ICOs completely.

But there’s never been an emphasis on checking that coded governance actually happens.

For instance, while the bestselling Cryptoassets: The Innovative Investor’s Guide to Bitcoin and Beyond does exhort investors to scrutinize developer activity,

it does not view the actual product of developer activity—the code—on the same plane. Indeed, the book does not include a project’s codebase in the materials that it suggests a fundamental-analysis investor would want to consider.

To the authors of most buy-side advice, cryptocurrency investment is an exercise in reading whitepapers, blog posts, and commentary—and watching the social-media trade winds—but rarely involves inquiry into code.

Taken together with analysis of our sample, these impressionistic sources of evidence lead us to believe that investor-protection code is not a significant driver of market pricing.

ICO advocates might reasonably respond to this absence of evidence for the importance of code in a number of ways. First, it might be the case that investor-protection code will manifest itself as a driver of market returns in the future. Perhaps future researchers will develop measures that capture price tremors in response to phenomena like the one we identified in Part III. It is also possible that the ICO market’s “crypto winter” was driven by investors who scrutinized the code of circulating tokens and found it lacking.

Some commentators do advise investors to pay attention to the underlying code of cryptocurrency projects, and their approach may be gaining adherents.

Further, some ICO promoters take to Reddit message boards to offer bounties to independent parties interested in auditing smart contract code—an indication that attention to code (or at least the perception of attention to code) is valuable from the promoter perspective.

These audits focus on the antihacking aspects of cybersecurity, not specific instantiation of economically relevant promises.

But perhaps the recent “modifiability crisis” after the Bancor hack will bring our investor-protection concepts to the fore. In other words, the market will reflect investor protections found in code sooner or later.

A second potential response from our ICO advocate might take a different tack. Instead of defending the importance of code in delivering investor protections, the advocate might retreat and take up a holistic defense. Specifically, even if code is failing to protect investors, there still remain legal and reputational checks on exploitation and desertion by ICO teams. That is, there will be substitutes for coded governance rules. Instead of the law of the blockchain, the law of the Swiss stiftung, the California Business Practices Code, and the Securities Act of 1933 will ensure that bad actors are punished, and the market will do the rest.

As we argued above, the legal safeguards against ICO investor exploitation are, at present, significantly weaker than in other investment markets.

It is easy for an issuer to set up shop in a low-regulation jurisdiction,

and  the  architecture  of  the  cryptoeconomy  enables  far  more  user  and  promoter  anonymity  than  typical  markets.

And even for transparent issuances conducted in the shadow of U.S. law, our background legal regime presents untested forms of investor protection. While a number of class-action suits, largely premised on state law violations, have been filed against some prominent ICO teams, the viability of any of their claims remains unclear.

The deterrent threat of legal ramifications is not nearly as strong as in typical markets—and, of course, is far weaker than the automated enforcement of code.

At a deeper level, arguments about the power of traditional legal deterrence are dangerous for ICO advocates. They show that advocates have already abandoned the high ground of “lex cryptographica.”

Smart contract code was, after all, supposed to render traditional interme­diaries useless, obviate the need for regulation, and reduce transaction costs for participants.

Without those justifications, it becomes harder to see what benefits ICOs provide, other than regulatory arbitrage.

To be explicit, if the value of blockchain-based financial products turns on the reputations of their creators or the vitality of legally enforceable wrap contracts, we see no good reason why traditional regulatory tools—securities law, know-your-customer regulations, and fiduciary suits—should not heavily police a space that currently is rife with the opportunity to bilk investors. The analogy to the failures of the pre-1933 securities regime would be unavoidable.

However, we are not ready to make that sort of strong claim about the missing role of intermediaries. Some projects encode all of their governance protections, and others appear to fall short largely only on vesting.

We simply do not know enough at the moment about what incentives encouraged particular turns to coded governance. Nor have we investigated the more mature 2018 market. Today, several sites are working to develop informally rich certification systems.

Perhaps such systems will evolve and further depress the need for old-fashioned intermediation in the absence of regulation.

But perhaps not. If problems with investor protection code are not priced into the market, and traditional law presently has trouble deterring abuses, where does that leave us?

B. Whose Market Is This?

The absence of evidence suggesting that investors are well-protected in the ICO market raises a natural question for legally-minded readers: Should we regulate this thing? Some see evidence of fraud and call for the whole market to be shut down.

Others would like the state to keep out.

Each approach has costs and benefits, of course—a conundrum where good things like innovation, investor protection, and regulatory clarity sit uneasily alongside each other.

There are tradeoffs galore.

For the pragmatists out there, a lot depends on who is being protected, and who benefits from innovative change. Are the investors actually grandparents risking their retirement savings?

Or are they day-traders enjoying a virtual casino?

We  might  want — really,  we  do  want — to  protect  mistaken  elders  more  than  thrill-seekers.

We also must be aware that regulations often will protect first-movers against competition by setting up new barriers to entry. And any serious regulatory strategy needs to help combat cryptoassets’ role in supporting illicit markets.

To inform the best approach to regulation, we need to know a lot more about the ICO buy side.

We see four archetypal participants on the buy side in the ICO market. Each has different implications for how to interpret the sell-side picture we have painted in this Article. Gaining a better read on the precise ratios and combinations of each will be a key next step for scholars and policymakers who deal with ICOs.

1. Irrational Exuberance. — The conventional wisdom about ICOs—the meme that drives most headlines—is that explosive valuations were the result of a massive financial bubble. As one leading analyst put it in the New York Times, “It’s not going to last forever, but it’s fun in the interim. The space is giddy right now.”

A massive financial bubble would certainly help explain why the market didn’t seem to care about the investor protections in smart contract code.

The possibility of a bubble accords well with the existing literature on what drives cryptoasset performance. While we are the first to study investor-protection measures found in code, numerous researchers have investigated the relationship between market performance and a host of potential predictors, including founder profiles, business plan charac­teristics, social media factors, known cybersecurity incidents, and more.

A consistent theme in this emerging literature is that reputation is the key to understanding the ICO market. Unfortunately, reputation is hackable. For instance, one paper finds that management team quality, as rated on a website called ICObench.com, predicts market performance.

ICObench, however, has been accused of operating as a “pay-to-play” operation.

Indeed, many rating platforms at the heart of the ICO informational ecosystem

operate on a “pay-to-be-rated” model.

Project owners place a high value on their project’s rating and are willing to pay as much as $20,000 for a rating on the most influential sites.

Such paid systems have well-known pathologies, as reflected in the credit-ratings experience during the financial crisis.

As a result, when academic papers find that some proxy for social “hype” or “buzz” correlate with higher returns,

we are not heartened. Instead, they only make us worry about targeted ads

and “pump-and-dump” cartels that coordinate massive social media pushes to temporarily inflate prices before selling their tokens to their marks.

These sources of noise and misdirection have contributed to many bubbles in the past.

Of course, reputation-driven markets are not necessarily all bad; it is the particular characteristics of this one that cause concern. We are not alone in this worried hand-wringing. Even researchers who hold out hope that “the wisdom of crowds” might one day triumph still characterize the ICO market as a series of “information cascades” susceptible to insanity.

Cooler heads suggest that taking market returns seriously during the 2017–2018 highs would have been seriously misleading, given the market’s immaturity and “speculative frenzy.”

As of early 2019, there is compelling evidence that valuation highs were more bubble than accurate assessments of promising projects. The  market  capitalization  of  all  cryptocurrencies  fell  over  eighty  percent  in  2018,

and  trading  of  certain  coins  has  essentially  stopped  completely.

Research that identifies the particular sources of air for the bubble will be valuable going forward.

In a sense, a bubble would be the least surprising and most manageable explanation of the ICO market’s rapid price swings. Regulators would need to focus on the time-honored, if difficult task of popping future bubbles with better informational requirements. But the “animal spirits” of irrational exuberance are not the only plausible drivers of ICO demand.

2. Illicit Demand. — As a complement to the bubble theory of cryptoasset success, many signs suggest that a material portion of cryptoasset demand is driven by money launderers, tax evaders, and other holders of illicit cash.

Some of these illicit holders might be inspired by the original, anarcho-capitalist vision for Bitcoin: to “win a major battle in the arms race and gain a new territory of freedom” from centralized governments.

Others might not have politics on their mind.

This  second  piece  of  “conventional  wisdom”  about  the  cryptoasset  market  was  initially  suggested  by  accounts  of  how  Bitcoin’s  growth  was  fueled  by  the  drug  trade.

Recent allegations that Russian hacking of the Democratic National Committee in 2016 was bought and paid for using Bitcoin have made this concern more salient.

Indeed, one recent paper found that approximately half of all bitcoin transactions were associated with some form of illegal activity.

Another found that the imposition of “Know Your Customer” policies designed to enforce tax and anti-money laundering laws shrank ICO returns.

This source of demand would have entirely different implications for ICO regulation than the “bubble” story. Obviously, it would seriously weaken the case for ensuring an “innovation-friendly” environment through well-tailored regulation. It would also counsel in favor of greatly increasing scrutiny on the major players in an ICO ecosystem who are benefiting from their dalliance with criminal underworlds.

Along with the “bubble demand” hypothesis, the “illicit demand” hypothesis also comports with some of our results. For instance, if criminal payments facilitation is indeed a major driver of demand for ICOs, then it is unsurprising that investors do not seem to care about whether founder vesting promises are delivered via smart contract code. Instead, they might simply be treating all ICOs like new printings of black-market money. If this is the case, then the high-flying business plans found in ICO whitepapers are merely window dressing, or an initial spark to help create a network effect for a new cryptocurrency. This form of demand could dovetail with the speculators driving the bubble described above. And it seems fair to say that gamblers, bubble speculators, and criminal cartels alike will not be inordinately attentive to smart contract code.

3. Crypto Winnings. — A third possible source of ICO demand might be coming from investors who raked in gains on investments in Bitcoin and Ethereum. These two cryptocurrencies have appreciated enormously since the beginning of 2015. This has led to massive wealth-creation for a cohort of so-called “Bitcoin millionaires,”

and their decisions about what to do with their winnings might be driving a fair bit of ICO success.

This hypothesis might play out in two ways. First, ICOs might serve as a decent place to park winnings that are trapped in crypto purgatory. To the extent that the “crypto winners” have been the illicit actors described above, they will have trouble converting their cryptocurrency holdings to fiat money through traditional channels. To be explicit, even if they could easily turn ether or Bitcoin directly into cash, they might not want to—they might be worried that governments would investigate the owners of fiat cash hoards.

Instead, they might attempt to wait until cryptocurrency affords them more access to consumption in the real world. In doing so, ICOs would provide a reasonably good vehicle through which to diversify their holdings and to attempt to invest their winnings in potentially lucrative ventures.

Second, to the extent that some investors treat cryptoasset markets like casinos, they might be simply gambling with the house’s money.

That is, it is easier to imagine investing in speculative assets, without caring too much about the details, when the stake one uses to invest with is itself the product of recent, sharp, gains. This is why people sometimes (foolishly) play the roulette wheel after winning at blackjack at the casino.

The “crypto winnings” hypothesis is the least-explored in literature about ICO demand and market performance. Nevertheless, there is preliminary evidence supporting it. Specifically, one time-series analysis suggests that blockbuster ICOs have negative effects on Bitcoin and ether prices.

This suggests that investors are trading between ether and Bitcoin on the one hand, and ICOs, on the other. Other analysts observe that ICO teams who amassed huge Ethereum war chests from the proceeds of their token sales were eventually forced to liquidate them as the price of ether dropped. This intensified price declines in not only ether but tokens as well.

If research continues to bear out this effect, it would only further support the kinds of regulatory responses that are appropriate in light of the “bubble” and “illicit demand” scenarios described above.

4. Smart Money. — Finally, it is possible that some ICO demand is driven by legitimately smart money. Anecdotal reports indicate that a wide range of old-growth VC firms, hedge funds, and family offices are, in fact, investing in ICOs.

Sometimes, they invest directly, as with the participation of Sequoia Capital, Andreessen Horowitz, and Union Square Ventures in the Filecoin ICO.

In other cases, they invest through intermediaries, whether due to regulatory restrictions on their holdings, or simply to work with other investors who are experts in the cryptoasset class.

In either case, these investors are the most likely to be engaging in fundamental analysis of ICOs, and thus the most likely to be scrutinizing smart contract code.

The presence of these investors in the market raises numerous questions for researchers and regulators alike. First, recall the colloquy with the ICO advocate in Part IV.A above. In a world where the code of “lex cryptographica” is not performing crucial investor-protection roles, we must look to traditional sources of protection. One of those is public regulation, but another is private gatekeeping. In the IPO world, for instance, the involvement of initial underwriters and primary market-makers channels pricing toward a fundamental valuation.

So, too, does the participation of institutional investors on the long and short sides of the market.

These investors do the heavy analytical lifting that helps protect retail investors from succumbing to irrationality. And (most of the time) these investors read the investment contracts.

Are “smart money” investors playing similar channeling roles in the ICO market? It is hard to say. Maybe investors like Sequoia Capital are entering into side letters with ICO teams to contractually ensure that supply and vesting promises are upheld.

Maybe the Andreessen Horwitzes of the world are scrutinizing modifiability and holding private corporate-governance fiduciary powers to rein in its use. They might also be embedding important information into market prices—for instance, information about ICO project activity, founding team reputation, and the quality of an ICO’s informational disclosures.

On the other hand, maybe they’re not. There is nothing stopping the “smart money” from riding cryptoasset volatility for all it’s worth. Bubbles are profitable for smart money, too, so long as they can cash out before the music stops. It would be valuable for future research to suss out the strategies and tactics that old-growth investors have been employing in this market.

From a regulatory perspective, the presence of smart money presents both a reason to care about preserving ICOs as a potentially valuable innovation and a potential lever to use. Indeed, one happy story that might be told a decade hence is that the ICO market of 2017 merely represented a period of growing pains, where reliable information sources and reputable gatekeepers were taking formation.

C. Whose Market Might It Become?

Based on the strong evidence that smart money is not leading this market, it can be tempting to cast doubt on all aspects of ICOs, including smart contracts. Though it will take future research to prove it, the ICO buy side today looks to us like a mixture of a bubble and an illicit market, with some smart money in the mix. And yet, this doesn’t mean that smart contracts are meaningless.

As John Maynard Keynes (didn’t) say, “The market can stay irrational longer than you can stay solvent.”

But over a long enough time horizon, every bubble must pop. This leaves open the possibility that fundamental aspects of smart contract quality will, eventually, sway the outcomes of the market, with smart money at the helm.

In many ways, the ICO market of the past couple of years resembles the dot-com boom that took place at the end of the last century. That boom featured massive reallocations of investment capital toward nearly any company that proposed a business strategy that incorporated what was then called the “world wide web.”

The same has been observed in relation to “blockchain” and “token”-based business plans in today’s climate.

In the dot-com boom, investors also broke from fidelity to traditional investment metrics like price-to-earnings ratios, instead relying on new valuation drivers like the sheer number of “eyeballs” viewing a website or the “stickiness” of the website experience.

Short-term performance on these metrics turned out to have little relation to a company’s long-term success.

It is hard not to see the rise of crypto-investment metrics like GitHub reputational stars, Twitter followers, and Instagram likes as representing a similarly problematic set of proxies for the possibility of network success. Financially, between the years of 1997 and 2000, internet stocks zoomed up and up, suggesting a new paradigm for corporate finance. The cryptoasset investor subcultures devoted to rejecting “fear, uncertainty, and doubt” may be in for a similarly painful fall to earth. Almost without question, both the dot-com market and the ICO market would have benefited from clearer and more reliable information environments to curb their excesses.

And yet, from a distance of twenty years, the economic follies of the late 1990s look less like utter madness, and more like a kind of overeager prescience. The clothing retailer boo.com may have gone belly-up, but e-commerce represents 40% of sales for even classic footprint companies like J.Crew,

and leading apparel startups like Everlane and Rent the Runway are decidedly “online-native.”

And though the grocery deliverer Webvan.com was widely derided as one of the biggest flops of the dot-com bust,

Amazon is pushing in that direction.

The  rush  for  eyeballs  has  become  a  rush  for  data,  and  online  shopping  continues  its  remarkably  paced  growth.

Will we look back on the cryptoasset craze initiated in 2017 with similar curiosity twenty years from now? What will fall away as the ephemera of the moment, and what will work itself deeply into our economic institutions? Given the froth of the market, it can be tempting to focus on the gut-level question of whether the ICO market is a financial bubble, and if so, how regulators should address it.

But our view is that legal policymakers might do well to look beyond the bubble (and its certain fate). Bubbles misallocate capital to unproductive uses and divert the energy of those who respond to the capital’s call. They also harm unsavvy investors who fall prey to the salesmen who are selling a bull market.

These animal spirits cause huge amounts of mischief. It ought to be—and indeed is—the province of lawmakers and regulators to temper them.

And yet, we are convinced there is something useful to be learned from this first experiment in blockchain governance. Some firms are encoding their promises, though it’s not obviously rewarding to do so. Others are working to create intermediaries and certification regimes despite the contrary incentives present in a sharply rising market. Rewarding such good actors should be as important to regulators as punishing fraudsters.

Conclusion

The computer code at the heart of ICOs enables a new way of founding and governing enterprises. It allows entrepreneurs to adopt the ICO method, whether for good or ill. But while smart contract technology may be a driver—indeed, a definitional component—of the ICO phenomenon, we believe our study demonstrates in detail that smart contracts are also embedded in the social world. Just like Coca-Cola’s vending machines, ICOs are products of their time and place. They are built atop innovative “technical systems” that only recently came into being, and they are conducted within particular “communities of discourse” that happen to exist here and now.

To make sense of the technology’s role, scholars and regulators alike should study the unique forms that this embeddedness takes.

Our study demonstrates that the current structures—markets, formal organizations, and professional communities—where ICOs take place are producing a disconnect. Far from replacing (or seamlessly extending) law and norms, code is often falling short of expectations. It sometimes fails to deliver key investor protections, and can provide founders with significant, undisclosed authority to alter the terms of investor engagement. While ICOs are promoted by an industrial community that espouses technolibertarian beliefs in the power of “trustless trust” and carefully designed code, actual ICO practices do not uphold that ideology.

The disconnect we observe reflects the informality of the ICO world. Paper contracts and IPOs are joint products of law firms, investment banks, regulators, and a panoply of buy-side institutional intermediaries. Smart contracts and ICOs, at least at the moment, largely result from coders and entrepreneurs working at greater distance from risk-averse gatekeepers. Befitting their relatively informal production setting, smart contracts have been plagued with quality control problems. They suffer vast amounts of hacking,

and, as we show, standards as to how code is produced and made legible are wanting.

Unlike the traditional legal world, there are currently no guilds or expert institutions governing smart contract coders’ practices to encourage quality. To withstand market ups and downs, the ICO community should invest in developing reliable institutions and promulgating best practices for the long term.

The informality of smart contract production leads to risks, to be sure, but it also breeds creativity. Lawyers tend to recycle language from agreement to agreement without much thought, but the smart contract community is full of “makers,” excitement, and avocational energy. This distinction suggests that the rate of innovation within smart contracting is driven by social factors, as well as technological ones.

It also suggests that whether or not the ICO market is a bubble, professionals and hobbyists working on ICOs will be able to port smart contract governance into new settings over the years to come. As their ranks increase, the “no-reading” problem for smart contracts might also be tempered. Right now, one aspect of the disconnect we’ve identified is that so few people can read smart contracts. The community of people who are able to vet and audit smart contracts has much room to grow. As it does grow, and as existing institutions develop vetting capacity, we would expect to see quality improve.

We think that optimal regulation depends heavily on a better understanding of the buy side of the market. But whatever the fraction of investors who deserve protecting, our results show that computer code is not presently a reliable part of the ICO form. Our results strongly suggest that an increased presence of gatekeepers and regulators might help that process along. The SEC, with its newly developed “Cyber Unit,”

is increasingly active in patrolling the scene. Other regulators, along with courts, will also contribute to increasing formalization of ICO code standards. So, too, will private standard-setting organizations within the industry itself. The rise of trusted intermediaries will be the next necessary step in any maturation of this novel financial form.

Appendix A

Summary of Top 50 2017 ICOs 

ICO Name	Country	Announced Raise ($M)	ICO Date	Initial Market Value ($M)	Market Value 12/31/18 ($M)
Filecoin	USA	$257.0	9/10/17	N/A	N/A
Tezos	USA	$232.0	7/13/17	$1,138.6	$281.0
EOS Stage	USA	$185.0	6/11/17	$654.9	$2,326.3
Paragon	RUS	$50.0	10/15/17	$18.7	$10.9
Bancor	ISR	$153.0	6/12/17	$98.8	$38.6
Kin	CAN	$98.0	9/26/17	$88.0	$28.7
Status	CHE	$100.0	6/20/17	$194.9	$59.8
Tron	CHN	$70.0	9/2/17	$29.3	$1,254.5
TenX	SGP	$80.0	6/24/17	$115.4	$28.3
MobileGo	USA	$53.0	5/25/17	$139.2	$51.3
KyberNetwork	SGP	$49.0	9/15/17	$254.8	$20.6
MCAP	USA	$44.3	5/7/17	$29.9	$0.1
Loopring	CHN	$45.0	8/16/17	$42.4	$31.5
Enigma	USA	$45.0	9/11/17	$51.9	$21.6
ICON	JPN	$45.0	9/22/17	$1,976.4	$112.0
PeerBanks	USA	$42.6	9/22/17	N/A	N/A
Electroneum	GBR	$40.0	10/20/17	$349.8	$64.0
Aeternity	BGR	$24.4	6/19/17	$100.6	$90.6
Monetha	LTU	$37.0	8/31/17	$61.1	$4.2
Basic Attention	USA	$15.0	5/31/17	$171.0	$156.9
Stox	GIB	$33.0	8/4/17	$36.0	$0.6
Civic	USA	$33.0	6/22/17	$60.0	$17.5
Request Network	DEU	$32.9	10/17/17	$38.7	$15.4
Grid+	USA	$32.2	11/12/17	$35.8	$1.8
ChainLink	USA	$32.0	9/19/17	$66.2	$101.3
Polybius	EST	$31.0	6/30/17	$12.8	$3.9
Unikoin Gold	USA	$28.6	10/23/17	$15.8	$3.0
DomRaider	FRA	$45.0	10/9/17	$16.8	$1.2
Blackmoon	CYP	$30.0	10/12/17	$25.3	$3.8
Bankera	LTU	$150.9	2/27/18	N/A	N/A
Agrello	EST	$35.0	8/17/17	$34.5	$3.3
Storj	USA	$30.0	5/25/17	$23.5	$19.3
Eidoo	CHE	$28.0	10/16/17	$25.5	$22.3
Monaco	CHE	$26.7	6/18/17	$23.9	$35.3
Power Ledger	AUS	$13.2	10/6/17	$76.6	$32.0
Everex	SGP	$26.7	8/31/17	$31.5	$4.4
Decentraland	ARG	$24.0	8/8/17	$18.7	$47.7
FunFair	GBR	$20.0	6/23/17	$348.3	$22.9
BitClave	USA	$25.5	11/29/17	$81.7	$0.3
Tierion	USA	$25.0	7/28/17	$81.6	$5.0
OmiseGo	SGP	$25.0	7/23/17	$42.6	$189.5
Aragon	ESP	$25.0	5/17/17	$36.9	$11.2
0x	USA	$24.0	9/15/17	$134.6	$163.9
Enjin Coin	SGP	$23.0	10/31/17	$18.7	$28.6
BLOCKv	USA	$21.5	10/25/17	$31.6	$7.2
FinShi Capital	SGP	$21.4	10/6/17	N/A	N/A
UTRUST	CHE	$20.0	11/20/17	$68.1	$10.0
Target Coin	IND	$20.7	8/31/17	$28.8	$0.8
ATB Coin	USA	$20.4	9/1/17	$35.2	$0.4
Giga Watt	USA	$15.0	7/31/17	$4.3	$1.6
Total	N/A	$2,584.0	N/A	$6,969.7	$5,335.1

Appendix B

Summary of Code/Contract Audit

ICOName	Scarcity Claimed (Y/N) Scarcity Coded (Y/N)	Burning Claimed (Y/N) Burning Coded (Y/N)	Vesting Claimed (Y/N) Vesting Coded (Y/N)	Modification Disclosed (Y/N) Modification Coded (Y/N)
Filecoin	N/A	N/A	N/A	N/A
Tezos	ScarcityNN	BurningYN	VestingYN	ModificationNN
EOS Stage 1	ScarcityNN	BurningNN	VestingYN	ModificationNN
Paragon	ScarcityYY	BurningYY	VestingNN	ModificationNN
Bancor	ScarcityYN	BurningYY	VestingYN	ModificationNN
Kin	ScarcityYY	BurningNN	VestingYY	ModificationNN
Status	ScarcityYN	BurningYY	VestingYY	ModificationYY
Tron	ScarcityYY	BurningNN	VestingYN	ModificationNN
TenX	ScarcityYN	BurningNN	VestingYN	ModificationNN
MobileGo	ScarcityYY	BurningYY	VestingNN	ModificationNN
KyberNetwork	ScarcityYY	BurningYY	VestingYN	ModificationNN
MCAP	ScarcityYY	BurningNN	VestingNN	ModificationNN
Loopring	ScarcityYY	BurningNN	VestingNN	ModificationNN
Enigma Catalyst	ScarcityYY	BurningYN	VestingYN	ModificationNN
ICON	ScarcityNY	BurningNY	VestingYN	ModificationNN
PeerBanks	ScarcityYY	BurningNN	VestingNN	ModificationNN
Electroneum	ScarcityYN	BurningNN	VestingYN	ModificationNN
Aeternity	ScarcityYY	BurningYN	VestingYN	ModificationNN
Monetha	ScarcityYY	BurningYY	VestingYY	ModificationNN
Basic Attention Token	ScarcityYY	BurningNN	VestingYY	ModificationNN
Stox	ScarcityYY	BurningNY	VestingYY	ModificationNY
Civic	ScarcityYY	BurningNN	VestingYN	ModificationNY
Request Network	ScarcityYY	BurningYY	VestingNY	ModificationNN
Grid+	ScarcityYY	BurningYN	VestingYY	ModificationNN
ChainLink	ScarcityYY	BurningNN	VestingNN	ModificationNN
Polybius	Unauditable	Unauditable	Unauditable	ModificationYY
Unikoin Gold	ScarcityYY	BurningNN	VestingNN	ModificationNN
DomRaider	ScarcityYY	BurningNN	VestingYY	ModificationNN
Blackmoon Crypto	Unauditable	Unauditable	Unauditable	ModificationNY
Bankera	ScarcityYN	BurningNN	VestingYN	ModificationNN
Agrello	ScarcityYY	BurningNN	VestingYN	ModificationNN
Storj	ScarcityYY	BurningYY	VestingYN	ModificationNY
Eidoo	ScarcityYN	BurningYY	VestingYY	ModificationNN
Monaco	ScarcityYN	BurningYN	VestingYN	ModificationNY
Power Ledger	ScarcityYY	BurningNN	VestingYN	ModificationNN
Everex	ScarcityYY	BurningNN	VestingYN	ModificationNN
Decentraland	ScarcityNN	BurningYY	VestingYN	ModificationNN
FunFair	ScarcityYN	BurningYY	VestingYN	ModificationYY
Bitclave	ScarcityYY	BurningYN	VestingYN	ModificationNY
Tierion	ScarcityYY	BurningNN	VestingYN	ModificationNN
OmiseGo	ScarcityYN	BurningYN	VestingYN	ModificationYN
Aragon	ScarcityNN	BurningNY	VestingYY	ModificationYY
0x	ScarcityYY	BurningNN	VestingYN	ModificationNN
Enjin	ScarcityYY	BurningNN	VestingYY	ModificationNN
BlockV	ScarcityYY	BurningNN	VestingYY	ModificationNY
FinShi Capital	Unauditable	Unauditable	Unauditable	ModificationNY
UTRUST	ScarcityYY	BurningYY	VestingYN	ModificationYN
Target Coin	ScarcityYN	BurningNN	VestingNN	ModificationNN
ATB Coin	Unauditable	Unauditable	Unauditable	Unauditable
Giga Watt	ScarcityYN	BurningNN	VestingYN	ModificationYN