The Token Handbook

There won’t be millions of tokens. There will be millions of kinds of tokens. — Richard Olsen

In this essay, I set up the question: how should we allocate value to startups and new projects? This year, we’re doing it more with tokens than with stock, and there’s a general token explosion. So I want to explore tokens and their design space. I’ll start with a few definitions, then I’ll present the major design elements, and in the final section I present a token taxonomy that breaks out six “species” of tokens, how they are issued, and how they are used. In the next essay, I’ll discuss how to use tokens to fund start-ups.

Join me at the #PillarUnconference this July and learn everything about tokens and ICOs! Click to register now.

Part I: Definitions

**What is a Coin?**To help communicate, I use the term coin to mean a cryptocurrency. All blockchains have their own native cryptocurrency, which represents a store of value, a medium of exchange, and a unit of account within the blockchain. These are coins. There are only two things you can do with a coin:

Use it to pay the transaction fees in a system.
Trade it to someone else.

That’s all. If it does something more, it’s a token.

**What’s a Token?**Fundamentally, a token is an IOU. It is a contract. It represents rights and obligations. The only token that isn’t a contract is an unissued token that is “on the shelf.” Once a token is issued, it represents value. Examples of tokens: a poker chip, concert ticket, stock certificate, bond, coat-check token, dinner reservation, driver’s license, passport, airline ticket, etc. A dollar bill is a token, but a dollar is not. This is what a digital token looks like:

1HieAFgpQdrVLN8GPFMfG8yMcDxDsrXiLN

This token lets me read a book, because it combines with a token on my computer: https://read.amazon.com/?asin=B004J4WKUQ

I used this token to travel on a train in Slovakia:

Here’s one that tracked a package along its journey to my house:

LH018942129US

Tokens always have:

an issuer
a substrate (carrier)
a system it is meaningful in
a value to someone
some way to use them

Some tokens are identical (tickets to a party), and some tokens are unique (airplane tickets). They can be transferable (=tradable) or not. Tokens that are unique and transferable are said to be of the same species (=from the same smart contract).

A password isn’t a token, because it doesn’t have a substrate, but a password written on a piece of paper becomes a token. Most tokens have some kind of anti-counterfeiting measures.

Tokens can potentially be: used as a store of value, used as a unit of payment, used to pay fees, have identifying artwork and anti-piracy features, eventually become a unit of account, be traded, be identical, be unique, carry a payload, grant rights or access, be tied to other tokens, be used as inputs to smart contracts, and more.

Examples of early tokens:

Tokens today:

Generally, we store tokens in our pockets or wallets. But not always …

**A Tale of Two Tokens**See these garbage bags? One is just a garbage bag, the other is a token.

The garbage bag on the left costs around 5 cents and is an actual garbage bag. It doesn’t stand for or represent anything or give access. The garbage bag on the right, however, costs about a dollar. This bag entitles the bearer to throw away trash in its town in Switzerland, and when you pay for the bag you pay for the service. In most countries, you need an account to have garbage removed, and you need to pay your garbage bill monthly. In Switzerland, you buy a roll of bags and your use of the bags exercises your right to dispose of the contents. The bags come in different sizes and can be stored, exchanged, and used on demand. There is no billing. There are no accounts. The bag has a substrate, a printed logo/identifier, and carries a payload. When used, it pays the fees for the system anonymously. Anyone from anywhere can come to Switzerland, buy some bags, and throw away trash, without setting up an account. This is a pull system. This is the kind of system we’re going to see flourish in the digital realm.

Tokens represent things or do things; they are not things themselves. A birth certificate is a record. A photograph is a document. But a passport is a token, because the passport gets you places. A key is a token because it lets you in the door. A key is not a thing, it’s an enabler. A twenty-dollar bill is not a thing, it’s an enabler. A TV remote control is not an enabler, it’s a thing! The enabler (token) is whatever account you have that pays for the signal to come to the TV.

Token Standards are in their InfancyToday, we really have one token standard on the Etherum blockchain and another, completely different, one using bitcoin. In both cases, you can only really issue a fixed number of tokens and keep track of them using an official list of tokens, to prevent counterfeiting. As we’ll discover, this is good for fundraising but extremely limiting for at least 90 percent of real-world token use cases.

**The Best Projects Make the Best Tokens**The way to design a good token is to design a good project first and see if it lends itself to issuing a token that can legally be sold to your target market. Today, many tokens are seen as funding or investment vehicles without thinking hard enough about what the token really does. I say “does” because a token sold to the public must be active, not passive. Many “ICO” tokens play the pseudo role of equity and are passive.

It helps to study Kickstarter. As they say on Kickstarter: the platform makes it easy to collect money, but it doesn’t magically bring you customers. The same is true of token sales. Selling a token to finance the building of a couple of condominiums isn’t going to work, because the market for those condos isn’t large enough to warrant a tradable token sale (and they would be securities).

The first thing you need is a group of early adopters who want your product enough to buy at the stage when there is the most risk. You also need a team who can reach out to those early adopters and connect with them. I started the process with 7,000 connections on LinkedIn and 3,000 followers each on Medium and Twitter, all built over many years. Other members of our advisory team had similarly deep networks. Once you have your target market in mind, it’s time to start exploring token space.

**When Not to Use the Blockchain**Blockchains can eliminate middlemen and provide trust in an untrusted environment. Since many blockchain tokens are “bearer” instruments, the blockchain preserves safety and manages transactions securely, even when people are trying to cheat each other.

Canada Goose recently announced that each of its products would come with a blockchain token QR code, so you could look them up in a blockchain explorer and see that the clothing items are legitimate. This could probably be done easier and cheaper using a database and a lookup function on their website — similar to tracking a package. On the other hand, diamonds will eventually all be represented by blockchain tokens, and whoever owns the token will own the diamond.

We use tokens every day. Your browser gives tokens to web sites as you surf the web, and your phone gives tokens to the phone system every time you use it. From this point on, I will use the word token to refer to a particular type of cryptographic address sitting on a blockchain, accessible (“owned”) by the person who has the private key for that address. Soon, we’ll use many of these tokens without even knowing it, as we do today with digital tokens.

**Regulatory Uncertainty**Regulatory issues loom over tokenspace like a dark cloud. At the moment, cryptocurrencies are defined as commodities and subject to regulation by monetary or commodity authorities. Tokens are undefined. Even though many projects are open-source, most people are buying tokens to make money on the capital gain. As soon as a group of such people are harmed, there could be litigation. The SEC is watching this space carefully, considering what to do and when. At the moment, they are looking at each case individually. After the recent SEC ruling on The DAO, which I believe was a security, we’ll see more scrutiny on tokens in general. The Chinese authorities recently killed many good projects (and many more scammers) by shutting down all ICOs and making them return their tokens. Some exchanges have already closed.

We need a reasonable solution that allows innovation without cramming tokens into a set of paper-based laws created just after the Great Depression. I hope the next two sections will help guide regulators as well as issuers and buyers.

Book now for the #PillarUnconference: a week full of masterclasses, workshops, talks, panel discussions and much more — with stage open for you too!

Part II: Token Characteristics

William Mougayar talks about a token’s role, features, and purpose. In this essay, I propose a business-model canvas for tokens. First, I’ll explore the basic characteristics of tokens and their issuance:

Physical aspects

Substrate
System
Role: Active/Passive
Supply: Limited/Unlimited
Source: Open/Closed
Fungibility
Transferability
Token Flow Models
Tokens, Passes, and Coupons
Payloads
Divisibility

Policy Aspects

Rails
Forks
Monetary Policy
Fiscal Policy
Token Sale Structures

Think of these as the decisions — the dials and buttons you would need to set on the token machine to have your tokens come out. Let’s understand them first.

**Substrate**This is the blockchain (or other software) on which you’ll issue your tokens. Certainly, Ethereum has the lion’s share of this business, but some blockchains are specifically designed for token issuance and sale. See Wavesplatform.com. A token isn’t just a data format — it has to exist on a substrate that ensures the security and validity of the token.

**System**The token may exist on a particular blockchain, but it’s valid only in a certain system. In most cases (we’ll see some exceptions), tokens don’t cross systems. In many cases, the system maintains a database of what each token is allowed to do. In some cases, the system is made up of decentralized smart contracts running on a blockchain. For the purposes of what’s to come, these are roughly equivalent. The systems will change. The tokens will change, too.

**Role: Active vs Passive**This axis is a gradient, from purely passive to various degrees of active. Let me try to break that down.

A passive token simply plays the role of money in a system, holds value, or collects rent. Passive tokens include frequent-flier miles, poker chips, game credits, virtual currency, awards, points, etc. A passive token could represent the ownership of a diamond or piece of land, or it could be like a share of stock or any collateralized debt instrument. We’ve had digital tokens for decades without blockchain technology — most systems have some internal units of credit. A poker chip is a physical passive token. We do just fine with most of these tokens from centralized issuers — they don’t have to become blockchain tokens.

Yet putting them on the blockchain makes them much more tradable, and that can change the game for several systems. With a passive token, you should always ask the question: Would another common unit of value be just as good? Couldn’t you just use dollars, or ether, or bitcoins? And the answer in many cases is no, for the same reason we use poker chips — they aren’t spendable anywhere else, so they reduce the risk of theft. Why would you buy a $10 Bellagio poker chip on the street, when you can get one for $10 at the cashier? If someone wants to sell you a $10 Bellagio poker chip for $9, you know the story immediately. Holding a passive token is safer than holding something hackers want. In fact, a huge number of passive tokens for many different systems isn’t such a bad idea, because it reduces the attack surface to hackers. This also applies to local currencies — the more you atomize value, the less concentrated, the lower the global risk.

A fee-paying token is the next step. Not only is bitcoin a store of value, it’s also used to pay the fees that secure and operate the blockchain. This is a separate role — similar to the difference between paying for your concert ticket and then paying the fees and surcharges for issuing the ticket. It’s a judgment call, but I will say that if it only pays the fees for its own blockchain (like most cryptocurrencies), then it’s still a passive token. That said, I would label ether a more active passive token than bitcoin, because ether can also be used to pay the fees for smart contracts. I’ll still draw the line there — if it functions as currency and pays the fees for the operation of its own blockchain, it’s passive.

An active token can accomplish things. People are starting to call these “tradeable API keys.” An active token could allow you to start your car, buy a bottle of wine (transfering your qualification to buy to the seller), board an airplane, enter your home, vote, cross a border, collect a tax refund, get a discount, etc.

An active token doesn’t have business logic built in; it is a signed order in a system that recognizes the token and applies business logic to accomplish things. The software looks up the token and sees what properties, rights, and obligations it has in the system. An active token has to be more than just access to transact in its native system. It has to enable its holder to do things. As Josh Stark says, “… the question should be: is the token part of a necessary cryptoeconomic mechanism in the application?”

Smart tokens are coming. Today, the smart contract manages the logic and maintains a list of all tokens, but in the future, we’ll build more features into the tokens themselves. Eventually, I hope, the tokens will have business logic in them (as opposed to just a passive payload or a look-up in a database). This is an important area of research.

**Supply: Unlimited vs Fixed**Although most tokens today are issued in a limited supply, there will be plenty of unlimited-supply tokens.

Unlimited tokens: at TokenFactory.io, we’re working on a passive token that represents the value of a portfolio of cryptocurrencies. This is similar to an ETF — we make each token on demand, and we can make as many as the market wants. I use the example of poker chips — poker chips have a fixed price and unlimited supply. They can always make more. There are plenty of use cases for such tokens. In fact, I think many of today’s ICO tokens are limited artificially, simply because no investor wants to buy into an unlimited supply.

Tokens that are limited in number are usually very divisible. This is what investors want, and in many cases it makes sense. In the case of passive tokens, these could be currencies, asset, equity, or debt-issue tokens. For active tokens, it means that the price of those tokens is likely to rise as the system’s network effect kicks in. A limited-supply token becomes a de-facto store of value, but it also leads to speculation and volatility. It generally makes a terrible unit of account. Goods and services must be priced dynamically.

On the one hand, you can say that token competition lets the market sort out which system it wants. On the other hand, these things are usually beauty/marketing contests with winner-take-all network economics and hidden relationships driving the results. An unfair initial advantage often leaves competing tokens in the dust. A way to launch an unlimited token would prevent many of today’s problems — I explain how later.

**Source: Closed vs Open**This is a remarkable time for ambitious open-source projects. The token-sale model gives open-source projects a chance for funding they never had. Imagine meeting with a venture capitalist — you sit down and explain that your entire ecosystem will be open-source. That’s a short meeting! The token revolution has taken most VCs by surprise.

Open-source projects have no equity, but they do have community. The fact that we can fund them now means that world-class talent is attracted to them. As Albert Wenger has noted, we are witnessing the creation of a new protocol stack, a new infrastructure that carries a remarkable new business model with it. Any open-source project should set up as a non-profit foundation before the sale, to take advantage of the tax treatment when there is no viable income model.

Closed-source tokens are also viable — there are so many in real life! Examples include airline miles, virtual currencies, game points, insurance products, prescriptions, passports, and many more. Today, we have a handful of closed-source tokens. One is Numeraire, a closed-source token/system to bet on economic outcomes. Eventually, there will be far more closed-source than open-source tokens, but they may not have spectacular generation events.

**Fungibility**Poker chips are all identical, so are shares of stock. They are fungible. Today’s tokens aren’t identical (they have their own blockchain addresses), but, like cryptocurrencies, they are fungible. It doesn’t matter which one you own. The DIGIX Gold tokens all represent the same amount of the same commodity. So we can use the term fungible, rather than identical.

A unique token would have a unique payload or identifier. The card you carry to let you into your office or gym identifies you and is associated with your profile. A token that represents the ownership of a boat would be a unique token. Unique is the opposite end of the spectrum from fungible.

There are a few kinds of tokens in between the two poles — they identify subsets that are fungible but different from other subsets. Family tokens would be an example — you can transfer some tokens among family members but not with other families. Another example is movie tickets — all tickets are good for the same showing, but you could exchange your ticket for another showing if you wanted to. Or you can imagine tokens that represent all the bottles in a given batch of a vaccine or wine. These have limited fungibility.

**Transferability**Some unique tokens can be transferred, and some fungible tokens may be nontransferable. Now we can see a matrix with uniqueness and transferability:

A seat on a bus may be tradable to anyone who buys it, but a seat on an airplane can’t, because regulations require that the airline identify each passenger and sales must be approved by the airline. These are not binary categories. A token that lets you pick up your kids from day care could potentially have some limited or temporary transferability. Same with a token that lets someone rent a car or hotel room for a period of time.

Some tokens have a market price, some don’t. A golf reservation at a popular course might have a market price, but an appointment to see a psychologist may not.

What makes tokens unique is their ability to trade cheaply and securely on the blockchain, without an intermediary or escrow service. This is the fundamental difference that will not just enable a niche of blockchain-based tokens but will replace most of our existing digital tokens with blockchain-based versions that are more secure and more efficient. You can call them tokens today, but in a few years we’ll just call them dollars, tickets, reservations, stock shares, loans, etc. In twenty years, almost all of the physical or digital tokens we have now will simply exist on a shared ledger with far fewer intermediaries than we have today.

Learn from me during the world’s first and only certificated course on tokenomics — come with us to Vilnius this July 15–22 and book your seat now!

**Token Flow**There are only a few ways a token can flow:

Straight line, from source to sink. Tokens are created for a single purpose and destroyed when used.
The poker-chip model, where they are used as currency in the system, then the system providers send them back to the cashier to exchange them for cash.
Circular, where they exchange back and forth, as with currencies.

Tokens, Tickets, Passes, and CouponsA token is a unit of value in a system. It carries its value for its entire lifetime, the same as a dollar or a poker chip. When you spend the token, someone else gets it and can then spend it again or take it to an exchange to trade for something else.

A ticket pays for access to a system and then expires worthless. Tickets are unlimited in supply and generally fixed in price, though there are times when a ticket can be sold at a premium or dumped at a loss. A ticket flows from source to sink — whoever receives the ticket in payment must destroy the token to complete the cycle. Here is a good paper on crypto-tickets.

A pass gives the holder unlimited (or possibly restricted) use of the system. A pass can give direct access (let you on the bus) or it can enable something (like a half-fare card), but it can be used over and over again with the only limitation being the rules set by the system. For example, you could have a discount pass that gives you a 20 percent discount, but only on weekends. Another kind of pass would entitle you to multiple discounts, and once you have taken advantage of that discount, your token would be marked as “used” for that offer. Creating a token like this on the blockchain is currently impossible, but I believe a pass standard will emerge within the next few years. The technical details are tricky, because you have to avoid counterfeits and double spending.

A coupon entitles its owner to a rebate or a discount on the price of a product or service. It’s used in conjunction with another form of payment. You can imagine a coupon that entitles the owner to one free cupcake when you buy a full dinner, or a 20% discount to a wine-buying club, but both the coupon and the method of payment would have to be in the same system for the smart contract to execute the deal. It’s not possible to create a re-usable discount coupon using today’s smart contracts.

**Payloads**Tokens can carry value. The DIGIX Gold token carries the ownership of a gram of gold stored in a vault. The Bilur coin is worth one ton of brent crude oil. Bitland tokens carry land ownership. The NXT platform and others are designed specifically to tokenize and trade assets, and more will follow.

This works in two different ways. In the first way, the token itself carries the payload in the “comment” field, which most blockchains allow. In this way, people loaded photos and other information into Bitcoin’s comments. Then, people started adding real data to represent specific assets, and these are called “colored coins.” To maintain colored coins, you must have a master list of which coins carry which value — otherwise it would be possible to counterfeit them. I can imagine coming up with a code system for the seats in a theater and inserting a 6-letter code for each seat and performance right into the token itself.

In the second approach, the token doesn’t actually store the value. A smart contract keeps track of which tokens are in circulation and makes sure that each token corresponds to the thing it represents. It could be a gram of physical gold, or it could be a pointer to an encrypted photo of your passport. In this scheme, the payload is in the smart contract and the token itself is just a blockchain address.

**Divisibility**If a token has a fixed supply, you want to make it very divisible. The standard is 18 decimals. You should have a very good reason for not making a limited-supply token divisible to 18 decimal places. One important reason to make tokens divisible is that people buy/exchange them with many different currencies, so when one is a whole number the other will be a fraction. Your token should be divisible to accommodate that kind of transaction.

**Rails**Any nontransferable token has a chance to become a rail. A rail is a store of value or medium of exchange that spans systems. The best example of a rail is the UPC barcode that can be used for information, inventory tracking, price lookup, and much more.

I don’t know of an identity system that consists of a token. Most systems let people put their identity documents on the blockchain in the form of encrypted scans of paper documents. But someday, there will be a digital token with an identity format that can carry your data from system to system, from purchasing alcohol to crossing a border to casting a vote.

A rail is essentially a platform-as-a-token. Active tokens that emerge as winners in one ecosystem will probably be adopted to perform their services in other systems. So expect token competition for network effects and tokens hopping from one system to another.

A metatoken is a similar concept. Example: most of the transit systems in the world are now digital — you get a card or an app and you buy your credits to ride the bus or the train or pay your bridge toll. Today there are billions of transit tokens in use, and more and more transit systems are creating their own mobile apps. A metatoken would negotiate with each of those systems to exchange its value for credits in the system. If you have a transit metatoken, you won’t need any more specialized cards or apps, you just pay using a token you already have and ride any bus or train in any city in the world.

We can take this up a level and imagine a metatoken that pays your taxi fare, parking, airplane tickets, airport fees, and lets you buy a cup of coffee at Starbucks. You can see immediately that metatokens border on currency, and that a stable metatoken is a better metatoken. We are designing the Pillar metatoken to do a lot of small, low-level things for you that you could use micropayments for. The token is designed for active micropayments, so it may have a place in several future ecosystems. A stable metatoken is unlikely to appear any time soon.

A note on Bancor: Bancor is a sophisticated smart-contract driven system for creating tokens that represent various combinations of existing cryptocurrencies — a kind of automated currency-pair or ETF generator. Using Bancor, you could easily create a token with 100 underlying currencies in it, which would immediately be more stable. I believe Bancor is less clever than most people think, for two reasons: 1) these tokens could easily be ruled as securities by regulators, and 2) the tokens don’t have a natural market-price discovery function built in, as the creators claim, they rely on market participants on other exchanges to set the price for them.

There is no way to predict which tokens will become dominant in the market. If things play out as they often do, marketing, buzz, sexiness, signaling, and luck will probably play more important roles than features and fitness for the task.

**Forks**This section discusses the very real possibility of a fork that could affect your tokens. There are two types:

A substrate fork is when your underlying blockchain forks. Which fork has your tokens? This is a potential “black swan” event that could have a serious effect on your tokens and all other tokens on the same blockchain.

A project fork can happen to an open-source project, where someone forks your software and issue a completely new token that runs that system, and that system could become more popular than yours. This is always the case with open-source software, and you should think about it ahead of time. If you have a valuable service but your fee prices are too high, someone can fork your project, issue new tokens, and attract people with lower fees. This is why fee prices are almost always determined by the market.

If you are starting your own open-source blockchain, there’s a significant chance of a fork someday in the future. Forks are almost always divisive and contentious. I think most forks can be avoided with better governance.

**Monetary vs Fiscal Policy**This is an underappreciated aspect of token design. Most projects just copy others. Let’s break it down.

Monetary policy establishes the supply and availability of tokens. There could be a fixed number forever, which is very popular, starting with Bitcoin. This is called an inflationary token, because the supply is not only fixed but it goes down over time? How does it go down? Tokens can be destroyed through burning, losing the private key, people die without giving anyone their private key, and people send real money to to bad addresses. Some people estimate that at 1–2 million of the 16+ million bitcoins issued to date have been lost forever.

There are two kinds of token inflation: price inflation, which is driven by market demand, and monetary inflation, which is a change in the supply. If the number of coins in circulation goes down over time, that’s inflationary (has a tendency to drive the price up), regardless of market demand.

The Ethereum creators estimated that around one percent of all circulating tokens will be lost each year. This amounts to hundreds of millions of dollars annually for a currency valued in the tens of billions. Ethereum has a “disinflation” design, with 15.6 million new ether created each year in perpetuity, so as the monetary base grows, the percentage of new tokens being added will decrease over time.

Just as there are more and less effective monetary policies for fiat currencies, there are many other ways to allocate coins and tokens. We’re just beginning to explore the possibilities. One set of tokens, Counterparty, were given to people who provably burned (destroyed) bitcoin, so one project’s monetary policy can affect another’s!

When having an initial offering, a good idea is to put all unsold tokens on ice rather than destroying them. This has a lot of benefits. I recommend putting about a third of unsold tokens on ice for three years, a third for five years, and a third for seven years. This gives everyone a clear indication of the money supply going forward and plenty of time for the price to appreciate before having to think how you will release the iced tokens when the time comes.

**Fiscal Policy**Fiscal policy is what you do with your tokens to encourage market demand or affect the price for your own tokens. Selling your own token inventory, as I mentioned above, is an example of fiscal policy. Here are some others …

Buying back tokens is something you’d only do if you find yourself with a lot of cash. It will push up the price, but at a “retail” cost.

Selling tokens. After an initial offering, any tokens you sell has the dual effect of raising money and dropping the price. The most important thing is to give the market your plan months in advance. Six months before any kind of unfreezing, you should give clear notice to the market what to expect.

Market making is using your cash or tokens to take the bid or the ask on exchanges. This improves liquidity. You can enter limit orders to catch the tokens if they fall too far, and you can sell tokens into a spiking market. It would be a good idea to reserve 5–10 percent of cash for market making to limit the downside of your token supply (though no one does this). Since market-making is generally profitable, we can expect algorithmic market-makers to scale up their efforts as our token economy grows.

Paying in tokens. If a token is a security, you should not pay people using tokens. If you’re paying in a token that isn’t a security, it has a cash value that must be reported as income for tax purposes. If the token is locked up, recipients still have to pay taxes on a token they can’t sell in time to cover the tax payment.

Granting tokens: several foundations have set up grant funds to foster research in their cause. We have decided to donate a significant number of our personal tokens into a common grant fund to sponsor research, experiments, and other projects we think will benefit our ecosystem. Keep in mind that tokens granted will likely be sold on the market soon, so drip them out rather than award them all at once. Don’t do it the way Stellar did it.

Burning tokens is destroying tokens forever. This happens naturally, but you can also do it on purpose. It can make sense if you want to reduce the uncertainty about tokens becoming unfrozen. You could also refreeze them, which probably makes more sense.

Splitting: You could give each token holder more tokens, in proportion to their holdings. As with a stock split, this should change the price of a token immediately and can allow for scaling. The recent experience of Stellar lumens is an example of how not to do it.

**Token Sale Structures**I’ve written about token sales elsewhere, but I’ll mention the main parameters of a token sale.

Presale. Most token sales have a presale. Some presales have exceeded $20 million. My belief is that presales are important and don’t have to be large. A $500k presale is fine. Even though people think a big presale is a big indicator of a big sale, I’m not sure that correlation actually holds. If you have a great project, you shouldn’t need to entice people with gimmicks, you should just have something to announce that builds momentum. A 10–20 percent discount should be a good way to signal the market that you have eager buyers.

Main sale. Many sales are scheduled for a 4-week period, where each week begins with the next discount. What we see is that people all buy at the very end of any discount period, effectively keeping their options open. So there’s a tradeoff between a) encouraging people to “just get it over with” and b) leaving more time for more news and press reports to bring new buyers in. Is a longer sale period better? Probably not. The price of ether will fluctuate over a month, so different people will acquire tokens at different prices. You may want to play with four three-day discount periods rather than four weeks. Do ninety percent of your marketing before the sale.

It’s common to reward a large number of people helping you with your token sale using tokens. Many of these people will want unrestricted tokens as a reward. You can bet these tokens will be on the market immediately after the sale, so think twice before granting unrestricted tokens. Better to have them frozen for at least a month, so these people have skin in the game.

This is all part of token stewardship, which would include managing splits, porting tokens to new technology, giving tokenholders a say in future directions, and more. We’ll explore more aspects of tokens and sale structures below.

Part III: The Token Taxonomy

With that framework in mind, let’s explore token-design space. I’ll break it into eight categories:

I believe 1.2 and 1.4 — the fixed-supply/nontransferable categories — are empty, but I’ll leave space for them in case anyone shows me examples. This leaves six “species” of tokens …

Learn from business leaders, blockchain pioneers, and crypto natives during our Unconference. Hack, build, and experiment for a week with the Pillar team!

Group 1: Fixed Supply

This is where most of the cryptocurrencies and current offerings are.

1.1 Fixed Supply — Passive Token/Transferable Open source: Blockpay for payments Open source: Golem Open source: Powerledger Open source: Cosmos Closed source: EncrypGen genomic data storage Closed source: Mysterium decentralized VPN

In this model, the token “powers” the system only by paying fees, as it does with bitcoin or ether. It can also represent the ownership of a limited commodity, like trees in a forest or hectares of land. In a very small number of cases (so far), it can represent equity in a company. Passive tokens are more likely to be considered securities than active tokens.

System/utility tokens. Looking at the list above, we see many passive tokens that just pay the fees for a system, without any other features. All of them are open source. Because the supply is limited, the value of the token is determined by the market, which means fees constantly adjust to account for the exchange rate. This turns a unit of exchange into a store of value, which is somewhat artificial but is now very common. This almost guarantees volatility, as speculators go in and out and try to move the price. In currency and stock markets, more than 90 percent of the trading happens intra-day. That could eventually be true for these tokens as well.

How to charge fees when the price is bouncing up and down? The answer is to use bitcoin’s auction approach — whoever pays the highest fee gets the best service, and the market adjusts to account for the exchange rate. This will play out in many of the systems being built now.

**1.2 Fixed Supply — Passive Token/Nontransferable**Is there anything in this category? I can’t think of anything.

1.3 Fixed Supply — Active Token/Transferable Open source: Bitclave decentralized search Open source: Pillar wallet token Open source: Blocktix — ticket ecosystem Open source: CoinDash — cryptoportfolio management Open source: IOTA — IoT token Open source: PTOYs — Personal health data ecosystem Closed source: DENT — Mobile data

Active tokens — tradeable API keys — are flooding into the market now. They make for a good token-sale story. While many active tokens should probably be issued on a per-use basis with an unlimited supply, groups are issuing and investors are buying them in large quantities. These tokens will compete against each other. Some will win, many will die.

As I mentioned previously, these tokens will probably evolve to have more built-in features. Some day, the tokens themselves could carry all the data and business logic they need and be autonomous.

I think we’ll see several new token standards emerge. You could imagine a token-scripting language that could make tokens autonomous and executable, possibly on their own blockchain. It’s likely, in my opinion, that many of today’s ERC20 tokens will later be exchanged — probably more than once — for new token models with new features.

Another novel feature of fixing the supply is that the system can have zero transaction cost, because holding tokens is profitable if demand for them increases. When you want your system to handle billions of small transactions per day, a token with zero transaction cost is far better than one with even a negligible fee. So while you can imagine an unlimited supply of Iota tokens, the limited supply may just have better economics in the long run. It’s a good example of where the fee and the token value are quite separate, and it’s an interesting experiment, because most systems charge a fee to incentivize the people who run the system.

**1.4 Fixed Supply — Active Token/Nontransferable**I can’t think of any active tokens here. The identity tokens don’t work this way — they simply provide the money to run the identity system, which is separate from the token.

Group 2: Unlimited Supply

This group is important, because it’s what most tokens probably should be. Unfortunately, there has been almost no work on the theory of unlimited-supply tokens. All blockchains are based on coins, which by definition have to be limited. Today, all tokens have to be issued at once — you can issue a billion tokens, but to add one more is impossible, you’d have to issue another batch, and the two batches aren’t interchangeable, they would be like another currency. Today, we don’t have the ability to issue more identical tokens as they are needed (this is a limitation of the smart contracts that issue tokens without allowing counterfeits).

I can imagine a blockchain someday that has this capability, or we may figure out a token standard based on an existing blockchain, but my purpose here is to work on the economics, not the mechanics of unlimited-supply tokens. A couple of rules apply here …

Single issuer. Unlimited-supply tokens come from a single issuer. This isn’t different from other tokens.

Refunds/Redemptions. If you sell a token at a fixed price, you may or may not be willing to buy it back for a fixed price.

Transferability is rare in this domain. You would only sell or buy an unlimited token if the issuer won’t buy it back. In most cases, like poker chips or bus passes, there’s no reason to buy one from anyone other than the issuer. But, as we’ll see, there are some transferable use cases.

Fundraisability is important. We’re looking for a way to issue an unlimited token but still raise enough money up front to build the system. But in many cases, fundraising won’t be the major function — organizations will easily and cheaply issue new tokens for existing systems.

**The Kickstarter Model**To fund these projects, you could use what I call the Kickstarter model, which I will illustrate with several examples. Suppose you want to build a new open-source system that helps people in developing countries raise money for infrastructure projects that use smart contracts to allocate the money, eliminating the middlemen who skim so much off of these kinds of transactions. You want to build this system, but it’s going to cost money. Fortunately, there’s a community of people around the world who would love to see this system come into existence — individuals and institutions who are already donating money and can see that this system will help ensure that more of their money goes to where it’s needed.

You could have an ICO, but that would limit the supply of tokens and subject them to unnecessary price volatility. Instead, you implement the Kickstarter model:

Step 1: You sell $5 million worth of “introductory offer” access tokens to your future customers (or whatever amount gets your first system built). Those tokens won’t increase in value, because you can always make new tickets later, but your true buyers don’t care — they are interested in using the system as soon as possible, because it’s a good solution for them.

To make the offer better, you could sell twelve tickets for the price of ten, and many would go for that — you might easily be oversubscribed. Since these customers are taking the highest risk, a discount makes sense.

You may even be able to sell them at a premium. For a period of time, I was the world’s only market maker in Tesla Model S delivery slots — people would pay me to connect them with someone who had an early slot and wanted to sell it, and people who had a slot and didn’t want their car anymore came to me to help them find a buyer for their slot. (Unfortunately, the company learned what I was doing and made the slots nontransferable and that business disappeared.) Eager beavers for a much-anticipated new product are usually more than happy to pay a premium for something they are excited about. Make it exclusive and give them perks — this is what they do on Kickstarter.

Step 2: Once the initial tickets have mostly been used and your system is getting good reviews, you sell the next round, at a slightly lower discount/premium. The price is converging toward the eventual open-sale price.

Step 3: After perhaps 1–2 more iterations, you have raised the money you need in advance and have built a sizable customer base, and then you can just go to an open, fixed-price ticket system.

This is the Kickstarter model. It is only for tickets, not for tokens. It appeals to the natural buyers of a system and keeps the speculators out. I hope open-ended token sales will become a viable way to kickstart many new projects.

**Unlimited Token Flow**To look at the token flow, let’s break this down using a toy system. Suppose you issue fruitcoin, and one fruitcoin is priced at $1. Fruitcoin enables people to purchase fruit at any fruit stand in the world. So Bob purchases ten fruitcoin for $10. While on a trip to France, Bob passes by Alice’s fruit stand. Alice sells apples for one fruitcoin each. Bob gives a fruitcoin to Alice, and she gives him an apple. Alice doesn’t want the fruitcoin, she wants .84 euros. She takes the fruitcoin to the cashier and changes them to euros. Now you have the fruitcoin back again, Bob has his fruit, and Alice has her money. This system is designed to preserve stable prices, and neither Bob nor Alice has to think about exchange rates. It’s similar to a poker chip. You can issue as many as you want, and you always get the cash needed to redeem the tokens in advance.

What you don’t have in that scenario is a business model. You need to have enough money to power the system. Visa does this by charging merchants three percent of the price. You could do it by charging a 1 percent fee, for example, paying Alice .83 euros and keeping the extra penny. This works if you don’t need cash up front to build the system, or if you can borrow the money and pay it back over time. This can also work for nonprofit or open-source models. In this scenario, the fruitcoin aren’t traded. They are only redeemed by the issuer. Effectively, you destroy the token when you redeem it. In that case, don’t call it a coin, call it a ticket.

Another version. Suppose you set up a local carnival on a vacant lot. You have the permit, the lease, the advertising, and the infrastructure. You sell tickets to the public, and all the rides are operated by independent business owners. You sell tickets for $1, and the rides charge however many tickets they want for their ride. At the end of the day, the ride owners give you tickets, and you give them 80 cents for each ticket. Then you destroy the tickets. This is a fair system for everyone.

Using the Kickstarter model, you could sell fruit-tickets at a small premium to your enthusiastic wandering fruit buyers and get a bit of cash to build the system. But if 95 percent or more of your token is the cash value of the token itself, you can’t have an ICO to raise much money for it. People won’t pay a premium for what is essentially their own money in a new system (this is why I believe the Lydian token isn’t a valuable token — they are essentially selling people dollars, and you can’t charge a premium for dollars).

Let’s look at a scenario with bigger margins. You and your partners have created a hit video game, and you have an even bigger idea for the next one. So you sketch it out, make a video of yourselves talking about it, have a crowdsale, and sell $7 million worth of video game tickets (this is what your budget says you will need). These tickets are necessary to run the game. You can even throw in some special beta tickets to let your token buyers help with testing and feedback. You can now build the system. When it’s ready, anyone with a ticket can use the system and play the game.

In this scenario, there’s no fruit or other vendor to pay — you get 100% of the money in exchange for the game experience, so you can raise money by selling tickets in advance. This is the real value of the Kickstarter approach. Once your first batch of customers is happy with their gaming experience, you issue the next batch and raise another $2 million. The next raise is $1 million. Then you go to a continuous-issue model and just sell the tickets to anyone who wants to buy them. All tickets are fixed price and expire worthless as they are used. You can make and sell as many as you like.

Note that the ticket can represent the “gas” needed to run your new gaming (or other) system. As with poker chips, customers use the full value of the ticket inside your system. So they can spend it down as they go, potentially win some tickets, and sell their leftover tickets to someone else.

Now we can add vendors. In this scenario, your video game is an ecosystem, with many people offering add-on services (avatars, virtual clothing, virtual gifts, etc.). In this scenario, they can’t sell these things for your token, because the vendors don’t want your tickets. The tickets expire worthless. Your game token doesn’t play the role of money, it plays the role of game credits. There’s no market for game credits, because you’re always selling them at the ticket window. Vendors will have to charge for their own services in another currency, not yours. You could create that currency, but it would be different from the tickets that allow access to the system.

This is an important lesson: an unlimited-supply token is not a currency. It’s only a currency if you’re willing to buy it back at FULL face value. If you buy it back for less than face value, or if it’s used to pay for the system’s services, then it’s a ticket.

With all this in mind, let’s look at the unlimited-supply categories.

**2.1 Unlimited Supply — Passive Token/Nontransferable**Unlimited passive tokens that can’t be traded would represent a shipping container, a unit of blood, a bottle of vaccine, a report card, a doctor’s appointment, a medical prescription, a work of art, etc. These things may have unique names or serial numbers, but the names and numbers go on forever. As far as I know, there aren’t any blockchain-based tokens yet in this category, but I’m sure they are coming.

Let’s understand this category. Is an x-ray a token? It really isn’t, because it’s not an IOU for anything. But a prescription is an IOU, and it can go from one system to another. So a blockchain-based prescription rail will help us build a far more efficient medical system, while x-rays will simply be assets on the blockchain or in encrypted storage somewhere. Your medical file will be made of both records and tokens.

Some things have limited transferability. A doctor’s appointment could be given to someone else. A token that allows you to pick up your child from daycare may be transferred to someone else (a smart contract can maintain a white list of acceptable people). And more. There’s a recurring theme here — there are batches and groups that share affinities but don’t share the same rights with the rest of the world.

We have many real-world tokens for things in this category now. Blockchain will make working with them cheaper, faster, and better. If you know of any tokens in this category, let me know and I’ll list them here.

**2.2 Unlimited Supply — Passive Token/Transferable**Similar to their limited-supply cousins, these tokens will change commodity ownership. In this category, rent-seeking tokens will probably rule. A rent-seeking token is one that represents the ownership and profit from the underlying asset. It may be equity or a derivative or a debt instrument. The way we manage these today is so inefficient that, while these projects may not be financed with spectacular token sales, I can imagine big industry players or venture-backed startups creating these ecosystems and reaping the benefits.

In this category, industry consortia can create an open-source token ecosystem to enable payments for their industry. We’ve seen a few experiments, but it remains unclear whether industry tokens will be popular. Personally, I don’t think company coins and industry coins will be successful — a stablecoin is what’s needed.

A stablecoin is an attempt to create a store of value, a unit of exchange, and a stable unit of account in a single unlimited-issue token. There are two kinds:

A local stablecoin would be like Tether, the crypto-dollar that I hope will be accepted in many many places. It is stable as long as you’re in the United States or anywhere that deals in dollars. As I keep saying, someday we’ll take the “crypto” off these labels, because dollars, euros, yen, and other currencies will be based on shared ledgers. Any token that is stable within a country or currency zone will be a bridge to those future government-issued currencies.

An international stablecoin is and will always be just a dream. There’s no way to have a stable common currency, because it begs the question — stable against what? Currencies are always being traded and their values are changing, so there’s no way to have a stable coin that is stable to everyone. That’s actually a good thing — as we’ve seen previously, we don’t want a single world currency. On the other hand, pricing goods and services in bitcoin always exposes both buyers and sellers to currency risk.

There have been several attempts to create blockchain-based stablecoins. All of them have limited caps on their money supply. Several, like Maker Dai, are based on a standard inflation index (which is much better than basing it on a basket of currencies). Since they are less volatile than any particular cryptocurrency, they should have a place in our token-based economy. And what, exactly, keeps the price linked to the index? It’s beyond the scope of this essay, but the mechanisms are elaborate. Perhaps an index token could be of service as a stable coin (when the regulators allow it). Perhaps an unlimited token will find a place here. We’ll see if one of these stable coins emerges as a common unit of account.

Supply-chain tokens will be common. As I wrote about in my book, Pull, every product will have its own digital birth certificate, which can track it from cradle to grave. There’s no reason these should be limited in number. These tokens could be active (rails) or passive (records).

Imagine you have purchased 1,000 hectares of forest in Myanmar, and you want to grow teak trees on the land. Teak appreciates about 8 percent a year, from sapling to mature tree, so you should be able to raise enough money by selling a token that represents each tree individually to long-term investors. These tickets would be priced at the sapling price, plus the potential for appreciation, discounted by the uncertainty. Then buyers could buy and sell the tokens for the next 25 years, as the trees mature. When they are mature, a token entitles the investor to a share of the harvest each year. The founders of the project would reserve something like fifteen percent of all tokens, which serves as their income for later years. The company has no profits — all the value accrues in the trees. This token is not a security, because it’s selling the product (the tree) not a share in the business and its profits.

Internet of Things needs trillions of tokens. Even though there are 2,779,530,283,277,761 IOTA tokens for the Internet of Things, any fixed number is forced. It would probably be better as an open system.

Commodity tokens have existed for more than a year now, and many more are coming. The Digix Gold token represents a gram of gold in a vault. Tokens are already representing kilowatt hours, a ton of crude oil, and more. There will soon be an explosion in commodity tokens and their integration into trading systems.

Tokens that represent ownership of unique items are going to change their markets. Everything from land to art to jewelry to homes to cars, boats, and even memberships will become tradable tokens, leading to massive disintermediation, far more liquidity, and entirely new systems for managing these assets. Fractionalizing ownership of unique things will create new financial instruments, derivatives, and other products.

Another interesting example is funds. At 20|30, we plan to issue an index token that represents the value of a portfolio of cryptocurrencies. Similar to issuing an ETF, we will make the tokens according to market demand. The price of the token is always very close to the Net Asset Value (NAV) of the underlying portfolio. However, we can’t do that using an ethereum smart contract. Because we want the tokens to be tradable, they all have to be made in the same batch. So we’ll have to make far too many tokens and then issue them one by one and keep most of them on the shelf. This makes them “all the same species” of token, so they can be traded.

A fund token is a security, by the way. So is start-up stock. So would be public equity or debt. Several groups (mine included) are now working on creating special markets where tokenized securities may be traded legally.

There should and will be other unlimited-supply tokens for prediction markets, the internet of agreements, contracts, tickets, travel, and much more. I believe there’s an opportunity to create a new blockchain around open-issue tokens, because today’s smart contracts can’t manage them.

**2.3 Unlimited Supply — Active Token/Transferable**Example: Insure bits

A ticket is an active token. While almost all unlimited-supply tokens are not really made to be transferable, some are, and for some it doesn’t matter. If your system doesn’t issue refunds, then the tokens should be transferable (if allowed by law). The gaming tickets above, for example, should be saleable. In fact, if you sell a limited number to raise money using the Kickstarter method, there could be a market for them, but that premium would disappear as you issue more and more tickets. A half-used ticket is also going to fetch a market price somewhere.

Remember that tradable API keys are active tokens. The only reason so many such tokens are in limited supply is that they were used for fundraising. But in the future, the majority of API key tokens will be unlimited. A gym membership may not be transferable today, but there’s no reason a gym membership can’t be transferable as long as the economics work for everyone. Seriously — what does the gym care if I use it five times a week or five people use it once a week? I think the ease with which we’ll be able to program these things will change the nature of tickets.

**Case study: Energy Markets**Suppose you wanted to tokenize your energy markets, which will reduce costs and increase innovation. You want to incentivize people to save energy, produce and sell energy, and create new applications that don’t exist today. You’d want a stack of tokens that work together; here’s one a made-up scenario:

The KWH token gives you one kilowatt hour, whether you’re selling or buying. There’s an endless supply of kilowatt hours, and they evaporate after use, so they should be represented using tickets. However, these tickets should be priced by the market, not by institutions.

To let the market set the price, you can just sell KWH tokens at a “wholesale” price and let speculators set the retail price. You could also use a betting token, like GNO or REP, and use the information from the betting markets to set the official price. Ideally, the price of energy adjusts in real-time, so you want to connect speculators directly to the price and give them every incentive to do it efficiently.

Another approach is to recognize that market forces should drive the KWH token directly, so the utilities or some governing body could issue, say, about 10x more KWH tokens than are currently used and let them go in a circle, driven by supply and demand. The trick here is matching the KWH tokens to the actual KWH use, because otherwise you’re manipulating the supply side of the equation, and you want the price to be set by demand, not supply. If done well, this eliminates the need for the betting markets described above, since speculators will be able to buy and sell energy units directly.

Then, to foster innovation, companies and start-ups can offer their own tokens that power their own systems, or you could launch a collective ICO for many groups and let them all use the seed money to do experiments and build pilot projects using a common token. In this way, many apps can work with the underlying energy tokens and create a new energy ecosystem. For example, we can have a predictive-analytics/AI/data stack many apps can use to automatically turn things on or off in anticipation of use, store energy at night, accommodate shifting solar and weather patterns, charge batteries quicker, and much more. An ecosystem like this has the potential to vastly increase our energy efficiency without increasing the head count needed to run it, because the blockchain and smart contracts do most of the work.

**2.4 Unlimited Supply — Active Token/Nontransferable**Airline miles, award points, and mobile minutes are active unlimited tokens that we generally can’t transfer to others. Since we have so many digital active tokens today, we can only imagine many will migrate to the blockchain. You may not even know you’re using them.

Keys in general are in this category. Keys have very limited transferability, if any. Every time you sign up for a software subscription, you get a private key that you don’t see, and that key gives you access to your game, anti-virus software, social network, etc. As we saw with the game ticket above, you can use the Kickstarter method to raise funds to create these systems. But most tokens here won’t need a fundraiser. They’ll be made by today’s institutions to replace their legacy systems.

We’re already seeing phone-based keys that can let you into your hotel room, rent a bike, or go into an unmanned grocery store in the middle of the night to buy a snack. We’ll probably see a lot of tokens having to do with your identity and permissions. Your passport or driver’s license could be an active, unlimited token. So far, these tokens have been created in proprietary systems, but the blockchain makes creating and using these tokens much easier.

Let me give an example: to start your car, you don’t need insurance. But to drive it on the street, you do need insurance. So the token that lets you start your car and drive it (or if the car drives itself) will be connected to your identity, your license, your criminal record, your insurance, and possibly even to a breath analyzer or phone app. When the conditions are right, the token stack allows you to drive your car. Otherwise, it doesn’t. The nontransferable token ecosystem allows this.

A refugee identity token could be used to help people move along their journey toward settlement in a new home. I hope a refugee rail emerges, letting people self-establish their own identities (often very difficult due to lack of paperwork) and then building their identities and history and reputation, giving them more and more access to social services, work options, and settlement offers.

**Summary**Some key takeaways:

Token sales are project finance, not start-up finance.
Tokens aren’t really designed to do many of the things we want tokens to do. They are more designed to be like currency than working tokens.
Token sales allocate capital very inefficiently.
Most tokens aren’t needed — they are simply a way to get funded.
Understanding the types of tokens may help us create better ones.
Regulators should look at tokens as new entities and not try to fit them into existing frameworks.

This token taxonomy may be helpful to people planning their token economies. We’ve seen many open-source projects raise money through advance, limited token sales. I hope we develop funding models that help other kinds of projects as well.

In the next essay, I’ll discuss how to fix venture finance — with tokens. Or go to the top of this page to continue.

Part IV: Resources

Fred Krueger on the future of tokens and ICOs

Vitalik Buterin’s thoughts on token sales

The Excellent Autonomous.Next Token Deck

William Mougayar’s Tokenomics

ICO best practices, by William Mougayar

A new paper by Vitalik Buterin and Jason Teutsch on equilibrium pricing for ICOs (technical).

David Siegel is a serial entrepreneur from the United States. He is the founder of Twenty Thirty AG and the Pillar project, both of which have newsletters you may wish to subscribe to. 20|30 provides consulting services and is building the CryptX — a broadly diversified portfolio of cryptocurrencies in a single token. His full bio is at dsiegel.com. Connect to him on LinkedIn.