How Cryptocurrencies/Tokens Accrue Value And The Purpose Of Digital Crypto-Assets Including Utility Exclusive Network Access

19 min read

WAX news

You can’t go anywhere, watch anything or read anything without hearing about crypto currency, bitcoin, NFT’S and the metaverse.  You are starting to feel left behind.  Things sound so complicated though, and technology can be so confusing.  How can a simple person off the street learn about this stuff and catch up to the masses that seem to be getting rich off of this new trend?  Well, you have come to the right place. This article will cover the basic and break down the important things you need to know, so read on.

Cryptocurrencies and tokens are a completely new digital asset class never before seen in financial systems. It’s why one of the first and most commonly asked questions about crypto-assets is what is their purpose and why are they valuable?

Answering these fundamental questions requires a thorough examination of three separate dynamics:

  • Identifying what purpose cryptocurrencies/tokens serve within underlying blockchain networks
  • Understanding why cryptocurrencies/tokens are preferred over traditional monetary instruments
  • Determining how cryptocurrencies/tokens accrue value

We aim to answer these questions, as well as provide examples of how some of the most popular cryptocurrencies/tokens currently function today.

Defining Cryptocurrencies and Tokens

Before diving deeper, it’s important to define the terms cryptocurrency, token, and crypto-asset. Generally, cryptocurrencies are defined as digital assets whose main purpose is to serve as a medium of exchange (MoE) and/or a store of value (SoV). Hence why the word “currency” is denoted in the name, and why cryptocurrencies are often thought of as being a new form of money. The most obvious examples of cryptocurrencies are Bitcoin (CRYPTO: BTC) and Litecoin (CRYPTO:LITE), which aim to be used as digital money for goods and services (MoE), as well as being a scarce digital commodity similar to gold and silver (SoV).

On the other end, tokens are typically defined as digital assets whose main purpose is to provide some type of utility outside of being a MoE or SoV. We dive deeper into the various utilities in subsequent sections, but some of the most common token use cases include exclusive network access, cash flow equity, staking insurance, protocol governance, and more.

It’s important to note that the line between cryptocurrencies and tokens is not always clean-cut, with most digital assets having both properties. For example, nearly all tokens both store value and serve as a medium of exchange. So while most tokens will likely never be used commonly as a digital currency or make for a better SoV than a digital asset like Bitcoin, they still have the main cryptocurrency properties. Similarly, it can also be argued that nearly all cryptocurrencies have token properties too since cryptocurrency is used as a miner incentivization mechanism to generate and maintain network security. This is by definition an expanded utility that goes beyond MoE and SoV.

Given the extensive overlap in terms, we will use the word cryptocurrency, token, and crypto-asset somewhat interchangeably for ease of use, which all comprise any digital asset cryptographically secured and stored on a blockchain network.

The Purpose of Crypto-Assets

With a proper definition established, let’s extrapolate out the purpose of crypto-assets. Doing so requires unpacking several layers, particularly the function, incentives, and bootstrapping of both blockchains and smart contract applications, which we will refer to collectively as decentralized computation networks.

Defining the Function of Decentralized Computation Networks

In order to comprehend the purpose of crypto-assets, one must first understand the underlying function of decentralized computation networks. This can be most easily understood by comparing decentralized computation networks to traditional businesses.

Businesses are centralized entities that typically own and/or license the Intellectual Property (IP) of the product and/or service they provide. Businesses are legally designed to maximize profits for their shareholders by extracting as much value as possible from their products and services. So while they may aim to provide consumers the lowest price and at times even engage in philanthropic causes, that decision is almost always made with the goal of generating more profits for shareholders.

Alternatively, decentralized computation networks are not businesses; they have free and open-source IP, with the product/service itself maintained by a decentralized network of independent operators. Thus, decentralized computation networks do not have owners nor do they have legal mandates to maximize profits. Instead, they can be thought of as public goods that offer services equally accessible to everyone, without built-in privileges for any set of users.

These digital public goods are operated through the use of Minimally Extractive Coordinator (MEC) protocols — self-running systems of logic that connect buyers and sellers of a particular asset or service together, with the goal of allowing those buyer/sellers to retain as much value as possible during their transaction by minimizing excessive rent extraction. In many ways, MECs are similar to companies like Amazon and Uber, except the company is replaced with a decentralized computation network that automatically matches supply with demand based on preset parameters that all parties can verify, yet no one can tamper with.

Centralized Business versus Decentralized Computation Network; (source).

MEC protocols are fundamentally designed to facilitate a business process for the minimal cost possible. For example, users of blockchain networks like Bitcoin and Ethereum (CRYPTO: ETH) only need to pay a transaction fee to use the network; no additional upcharge is added given there is no central coordinator to rent seek. The cost to use a MEC protocol is typically determined by the users themselves through an open auction, where supply and demand meet at equilibrium (such as users bidding for scarce blockspace).

On the other hand, when a centralized company facilitates a business process, it owns the facilitation mechanism and runs it as a for-profit operation. This gives the business, which serves as a facilitator, the power to act in their own self-interests, such as raising costs when they establish a monopoly, censoring transactions to favor a particular party, or selling users’ data discretely to turn an additional profit.

As a result, MECs are designed to capture the large network effects that facilitators often do (e.g.; banks, social media, e-commerce, etc) without the negative downsides that often accompany large businesses-based facilitators who become “too big to fail.” By minimizing rent extraction, MEC protocols direct more value back to the users and provide a superior service long-term. The next logical question then is how do you finance and maintain the incentives of a decentralized computation network without a built-in rent extraction mechanism?

Incentivizing the Growth of Decentralized Computation Networks

Decentralized computation requires incentives to bring individual infrastructure providers (nodes) together to perform a shared objective (coordination services) in a highly secure and reliable manner. The incentives have to be sufficiently high too because decentralized computation is purposely inefficient in order to lower the barrier to entry and generate strong determinism.

For example, the Bitcoin Network has approximately 10,000 independent nodes that all verify the validity of each block of transactions on the network to ensure the ledger of who owns Bitcoin is highly trustworthy, tamperproof, and available to everyone. Without incentives, users would have to trust the benevolence and altruism of node operators, which is not a security model anyone would rely on to secure anything remotely valuable, let alone a market worth over $900B dollars (current Bitcoin market cap at the time of writing).

In businesses, incentives to act fairly are driven by profits, legally binding contracts, and brand reputation. The idea is that it is long-term profitable and legally necessary to act honestly. However, extensively large companies can use their network effects and opaque backend processes to protect themselves in situations where they act unfairly, resulting in them never experiencing any negative repercussions for their actions. A few examples of this incentive misalignment include the 2008 bailout of financial institutions, Facebook’s harvesting and monetization of personal information, and Apple’s monopolistic and rent-seeking App Store policies. Thus, if decentralized computation networks are to provide superior services, they require a better financial gain/loss system that properly rewards positive performance and punishes negative performance.

For decentralized computation networks, the most obvious place to start then is financial incentives, which require a source of capital. To even get a decentralized computation network off the ground, there is a chicken and egg problem that must be overcome: users will not pay to use a network that doesn’t exist or is insecure, and node operators will not secure or operate a network if there are no paying users or revenue. Without a financial subsidy to jumpstart network operations, each side of the market will remain in limbo waiting for the other side to make the first move.

Both supply and demand within a common network depend on the existence of the other; (source).

Traditionally, centralized companies receive outside capital to fuel their growth by raising funds from venture capitalists (VCs) or other fundraising means. While this model can work fairly well for providing the initial capital to fund the development team of a minimally extractive network, it is nearly impossible to support a sustained stream of financial incentives required to subsidize the network to the point of long-term self-sustainability. For example, the Bitcoin blockchain still has a block reward ten years after its initial launch of 6.25 Bitcoins (≈ $306k), which is issued roughly every 10 minutes to help fund the mining nodes securing the network (≈ $44M a day and ≈ $16B a year at current rates).

Decentralized computation networks that attempt to rely upon VC funding for long-term subsidization require some type of value extraction mechanism from users (such as an upcharge on network fees) in order to pay back the debt they take on. This would remove the very value proposition the network set out to generate in the first place, being a minimally extractive coordinator. It would also create misaligned incentives where time and resources are spent catering to the demands of the network’s largest investors as opposed to what may be better for the long-term success of its actual users. Thus, the network could not offer any credible neutrality, as the entities providing the capital for subsidization would ultimately have excessive control over the future direction of the network’s development.

Additionally, by extracting value from users, the decentralized computation network’s competitive advantage will weaken in comparison to protocols that do not take on VC debt, particularly because their competitors can undercut them in network costs by being less extractive. It also makes the network less secure by reducing its security budget, as some of the value that would normally flow to nodes who secure the network is rerouted to investors to pay back the debt.

**It’s important to note that VCs are not inherently bad and this isn’t meant to take a shot at them. They play a key role in providing initial capital to development teams of MECs, however, VCs as the source of perpetual funding for network subsidization is likely unprofitable for VCs and antithetical to the ultimate goal of a MEC.

Instead of relying exclusively on outside capital to grow a decentralized computation network long-term, a more advantageous approach is to create a debt-free native crypto-asset (token) specifically for the network. This native token can then be used to fund the network’s growth by making it a required component of the network’s usage and security. Upon doing so, the value of the token on the open market can be tied to the value the network provides to users, which rewards highly adopted projects and allows them to grow the network long-term. It also creates a scenario where the network operators have a direct financial stake in a token specific only to that network, meaning the network’s performance/security is tied directly to the nodes’ own financial well-being.

Native network tokens benefit all parties in the value chain:

  • The development team can raise funds in a debt-free manner to support the network’s development by allocating an initial portion of the token’s supply to be sold to users (including VCs) in a token sale (e.g. Initial Coin Offering).
  • The MEC protocol is able to bootstrap its own growth by setting aside a large portion of the token’s supply to be paid to network operators over time as a subsidy/block reward for securing the network.
  • The users receive the lowest cost for network services through built-in subsidies and zero rent-seeking.
  • The nodes securing the network receive the highest rewards possible without value extraction by non-value-producing investors.

Ultimately, newly created capital in the form of a native token allows decentralized computation networks to avoid rent-seeking middlemen, retaining their valuable property of being minimally extractive. However, the only way for those newly minted tokens to actually work in support of the network’s growth and security is for them to have financial value on the open market.

Capturing The Demand of a Decentralized Computation Network in The Value of its Native Token

While the issuance of a native token allows a team to raise funds for development and create a subsidy allocation to bootstrap the network’s growth over time, it is only effective if the token has value on the open market. The only way for the token to have value on the open market is for it to have some type of way to capture the value generated by its underlying decentralized computation network. If it doesn’t capture any of the network’s value, then the token has no intrinsic value outside of speculation or an expectation from holders that the token-economic design will eventually change to capture value. If the token is financially worthless, then the allocation set aside to subsidize the network’s growth is worthless too, as nodes will have zero incentive to spend money to run network infrastructure that earns valueless rewards.

However, when a token’s value is directly tied to network demand from users, the value of the subsidy allocation increases alongside network adoption. An increase in the subsidy allocation results in a larger budget for the network to leverage as a means of generating additional security/utility for users and incentivizing more adoption. This generates a virtuous cycle of growth:

  1. A native token is issued by a development team. Alongside its initial distribution method (mining, public sale, airdrop, yield farming), a subsidy allocation is created and held by the protocol, development team, or community.
  2. A portion of the token subsidy allocation is used to bootstrap the network’s growth by rewarding infrastructure providers (e.g. liquidity providers, miners, validators, etc) with those new tokens coming into circulation.
  3. As a result of the network subsidy, the utility of the network increases for users (e.g. lower slippage trades, more secure network, additional services, etc), leading to an increase in adoption and additional fees paid by users to infrastructure operators.
  4. The increased network usage generates more market demand for the native token (though the methods described in sections below), culminating in a higher valuation of the native token’s market capitalization.
  5. The rise of the token’s value on the market directly leads to a rise in the value of the remaining subsidy allocation, extending the per-unit ability of those tokens to further grow the network. This enables more reinvestment into the network as a means of stimulating additional network utility, greater user demand, and larger pools of user fees; accelerating the virtuous cycle once more.

Virtuous Cycle of Growth Enabled from a Token Subsidy Allocation

The key benefit of token subsidization is being able to bootstrap the supply side of the ecosystem in a debt-free manner before the demand side exists. Once the supply side of the network is sufficient, then the demand side will naturally arise if there is real network utility. As the demand side rises via paying users, the subsidy can gradually be reduced until eventually, the network becomes self-sustainable completely from the aggregation of user fees. The remaining subsidies can then be redirected towards other network initiatives to generate more adoption such as expanding services or growing network security.

Fundamental to this entire virtuous cycle is driving demand for the native token, which in pursuit of this goal, has resulted in a wide spectrum of different token economic designs. Below are some of the most effective methods in which decentralized computation networks today generate token demand via creating token utility, which serves to tie the token’s value to network demand.

Network Access Through Exclusive Token Payments

The most recognizable way to tie network demand to the native token is to require payment for all network services to be made exclusively in the native token. By doing so, all users must acquire and gain exposure to the native token itself before being able to use network services. Having a standardized payment medium for utilizing the network ensures that demand from users must flow through the token. It also means that nodes have a direct incentive to uphold the value of the token via maintaining the health of the network, as their future revenue streams depend upon a well-functioning network that users want access to.

The most noteworthy example of a native payment design is the Ethereum blockchain and the usage of its native token ETH. In order to have a transaction validated and finalized by the Ethereum blockchain, users are required to compensate network service providers (miners) via a “gas fee” that is paid exclusively in ETH. This makes the ETH token a “first-class citizen” on the Ethereum network as all transactions, including interactions with smart contracts and movements of other tokens like stablecoins, require fees to be paid in ETH.

Since every Ethereum block only contains a limited number of transactions, as network demand rises so do transaction fees, requiring users to purchase more ETH on secondary markets to pay for gas. The rising market demand for ETH also increases the value of the subsidy already being paid to miners via its block reward, further strengthening the network’s security and utility as a global settlement layer for financial assets. Even as layer-2 solutions begin to emerge and batch transactions, the per-user transaction fee will decrease, but the total amount of ETH being paid to miners remains the same (or even increases as layer 2 attracts more paying users).

The amount of fees paid to Ethereum miners per day continues to grow alongside accelerating network demand; (source).

The Bitcoin Blockchain also operates in a similar manner where the native asset BTC is required to make transactions on the network. While Bitcoin’s primary value is derived from its “digital gold” Store of Value narrative rather than smart contract utility, users will need to continually transact on the network to generate enough fees to support the miners that keep the network secure. This is due to the fact that Bitcoin’s block reward halves every four years, meaning user fees must supplement the decline in block rewards over time if the Bitcoin network is to retain its high security.

An important caveat, however, is that while exclusive native token payments increase market demand from the user side, it does not necessarily increase market demand from the infrastructure provider side.* The reason being is that nodes may sell their earned tokens on the open market to pay operational costs, dampening the price appreciation from user demand. Therefore, exclusive payment utility is most effective when combined with an additional form of value creation that requires nodes themselves to acquire and hold the native token such as through some form of staking (e.g. Ethereum moving to Proof of Stake consensus, creating supply-side demand) or a strong social consensus around being a store of value (e.g. Tesla buying $1.5B of Bitcoin).

*Many infrastructure operators are also long-term believers in the network they secure, thus, they will have natural incentives to hold a large portion of their profits, leading to reduced sell pressure. For example, many miners use crypto-earnings as collateral for loans that are used to pay for expenses, allowing them to maintain greater exposure to cryptocurrencies.

Cash Flows Through Dividends and Burns

Another common way to generate value accrual for native tokens involves redirecting some or all of the fees paid by users to token holders. As a result, an increase in network demand from paying users directly leads to a proportional increase in the revenue rewarded to token holders. This provides token holders with a form of passive income and allows for the usage of more formalized valuation models such as discounted cash flow and price-to-earnings ratios.

The method through which network revenue is distributed to token holders can be achieved in various different ways. One approach is to use some or all of the user fees generated by the protocol to automatically purchase the native token on secondary markets and burn it, thereby reducing the total supply of tokens. This method increases the scarcity of the native token through deflationary pressure and is often used in combination with a hard-capped total supply (no inflation). The advantage of such an approach is that revenue is distributed to all token holders equally by increasing everyone’s percentage ownership of the total supply. The most well-known DeFi protocol following this model is MakerDAO, a decentralized stablecoin protocol, which has a native token called MKR. All the interest paid by borrowers is used to purchase MKR tokens off the market and burn them. In return for receiving the network’s cash flows, MKR holders act as the lender of last resort (e.g.; MKR tokens are minted to re-collateralize the network, as seen during black Thursday).

The second variation of token cash flows involves issuing dividends, wherein some or all of the fees collected by users are awarded directly to the token holders. These fees can also be used to purchase the native token in the open market and then distribute it to token holders, providing both price appreciation through market purchases as well as dividends to token holders (who may sell those earnings or keep them to earn even more dividends). One example of this dividend model is the decentralized exchange protocol SushiSwap and its native token SUSHI. Every trade made on the SushiSwap exchange incurs a 0.30% fee, with 0.25% going to the liquidity providers and 0.05% used to purchase SUSHI tokens in the open market and distribute them to xSUSHI token holders (the staked form of SUSHI).

Cumulative Sushiswap protocol revenue (in blue) paid to xSUSHI token holders; (source).

Another example of this dividend model is the decentralized derivatives protocol Synthetix and its native token SNX. Synthetix allows users to stake SNX as collateral and mint the synthetic stablecoin sUSD (500% overcollateralized). sUSD can be sold on the secondary market or converted at zero slippage into various other “synths” that track the value of different cryptocurrencies, commodities, fiat currencies, US equities, and indices. Stakers receive dividends from the fees generated from synth conversions (0.3% of trade value), as well as inflation rewards to compensate for the fact SNX stakers have short exposure to every circulating synth (akin to a clearinghouse).

While in theory, a token burn and issuing dividends should have an equivalent effect on the market value of the token, in reality, market psychology must be taken into account. A token burn occurs in the background, meaning the value accrual is not always immediately apparent to token holders and often cannot be differentiated from market speculation. With a dividend, users directly receive additional tokens, making the economic incentive of acquiring and holding a token with cash flows more apparent. However, how much this difference in perception of cash flows matters for the long-term valuation of a native token is still unclear.

Security Through Staking and Token Lock-ups

Staking is a method through which token holders are incentivized to lock up their tokens in exchange for the rights to provide and/or receive network-specific services. While staking mechanisms greatly vary in purpose and implementation from one protocol to another, the common denominator involves users/nodes taking native tokens off the market and putting them in a state of illiquidity, reducing the circulating supply of tokens available within external markets. Staking is often combined with dividend and network fee rewards, where users provide token-based capital as a form of crypto-economic security and in return receive some form of passive income generated by the network (e.g.; Synthetix).

The most recognized form of staking is Proof-of-Stake consensus, which powers various blockchain networks like Etherum 2.0, Polkadot, Tezos, Cosmos, Aavalance, etc. In the case of Ethereum 2.0, any entity that wants to participate in validating transactions and producing blocks on the Ethereum blockchain is required to lock up 32 ETH. Stakers can have their ETH tokens slashed if they perform malicious activities that attempt to corrupt the network (signing conflicting attestations), resulting in those tokens being permanently burned and the staker’s node kicked out of the network. Thus, staking in this format creates crypto-economic security that incentivizes the honest performance of network services. In return, ETH 2.0 validators are paid via a block reward subsidy and network transaction fees. This has already generated a large token sink, with over $5B of ETH locked in the Ethereum 2.0 beacon chain (as of writing).

A different form of staking involves the creation of an insurance pool that can cover any potential losses of a protocol. The most prominent example is the decentralized money market protocol Aave, which has approximately $2B of its native token AAVE locked in the Safety Module. 30% of this insurance pool can be used to absorb any black swan shortfall events, such as protocol under-collateralization. Stakers are incentivized to lock up their AAVE tokens through a reward in the form of an inflation subsidy as well as the rights to any fees generated by the protocol. This ensures that any users who want access to the protocol’s cash flows must have their AAVE tokens staked as insurance. Aave’s Safety Module covers a much different category of risks when compared to ETH staking, however, it has the same effect of taking tokens off the market and creating an incentive to hold tokens long-term to the benefit of the protocol’s security.

Aave Security Module Flow Diagram used to protect users from downfall events; (source).

It should be noted that many tokens have some staking utility in that they can be staked as liquidity within automated market makers such as Uniswap and SushiSwap. This means a user can stake their tokens in an AMM as a liquidity provider and in return earn a percentage on the swaps executed using the tokens they provided (albeit, not taking into account impermanent loss and double-sided pools). However, such staking is more of a product of AMMs and not a built-in mechanism for tying a decentralized computation network to its own token. If the token had no intrinsic value on its own network, then it wouldn’t be worth anything in an AMM.

Protocol Governance Through Voting

With the rise of Decentralized Autonomous Organizations (DAOs) — a structure for distributed social coordination — we have seen an increase in the number of native tokens that include an aspect of governance. Governance tokens allow holders to directly vote on proposals to change/upgrade the network itself. In most implementations, each vote is weighted by how many tokens a user holds, meaning anyone who wishes to gain significant influence over the direction of a network’s development is required to acquire tokens off the market to increase their voting power. However, token-based governance from one network to another varies greatly in its ability to influence the network, ranging from simple parameter adjustments to large sweeping changes of its underlying infrastructure

The most direct form of token-based governance is through binding on-chain votes. For example, in Aave, proposals are codified as smart contracts and can be executed immediately on-chain if approved by a sufficient quorum of token-weighted votes. Aave has used this form of on-chain governance for large changes such as the launch of the protocol’s v2 version, as well as the onboarding of new collateral types to its market. A more indirect approach to token governance involves off-chain signaling, such as in Synthetix, where token-weighted polls are created to gauge token holder sentiment and see if changes should be implemented by the DAO. These votes are not binding, meaning acquiring a large number of tokens is not a guarantee of influencing the direction of the protocol away from community consensus.

The value of network governance from one holder to another is highly subjective, making formal valuation models for “pure governance tokens” a near impossibility. Therefore, governance is almost always an additional form of utility for a token, and not its driving value proposition. However, there are always exceptions and this could change as these decentralized computation networks grow in value. Additionally, it’s becoming common to see tokens start out as a pure governance token and only later evolve to become a revenue-generating token after a community vote has been approved. An example of this is the decentralized exchange protocol Uniswap and its native token UNI. Currently, UNI is only a governance token, but it is broadly expected that the community will vote at some point in the future to add additional cash flow utility in a similar vein to Sushiswap.

On-chain governance allows token holders to vote on binding changes to a protocol; (source).

Tokenomic Fluidity

Most token designs being used in-production don’t implement just one method of linking network demand to the token’s value. Instead, they combine two or more mechanisms together to provide value creation through multiple forms of utility. There is no one-size-fits-all approach to value creation within minimally extractive networks, as each seeks to provide a different service to users, resulting in the diversity of implementations we see today.

Networks are also not locked into using a specific token economic model forever, but can evolve over time given there is enough consensus from network stakeholders. The LEND token (Aave’s previous token ticker) initially had a hard-capped supply and used a buy-back and burn model. However, that was later changed during the token migration to AAVE (100:1 conversion), along with a switch to an inflation-based token supply for subsidies and a new distribution mechanism where protocol fees are issued as dividends to token holders (revenue currently goes directly to the Aave DAO which is controlled by AAVE token holders). Similarly, the ETH token started out as solely being a utility for payments to miners but has since added more token utility through the recently launched ETH 2.0 staking. It’s also incurring a potential third addition to its utility via growing community support for a token burn implementation as outlined in EIP-1559.

Conclusion: Through Tokens, Decentralized Computation Networks Become Public Goods

Decentralized computation networks serving as minimally extractive coordinators (MEC) provide humanity with an unprecedented set of technological primitives that, if implemented correctly, can completely redefine how humans interact with one another both socially and economically. Such backend infrastructure, which replaces centralized for-profit institutions with decentralized non-profit facilitators, brings about open agoras where buyers and sellers can freely exchange value without warlords exercising monopolistic control or leeches sucking out value.

Realizing the power of MECs requires the use of native crypto-assets. Crypto-assets allow MECs to be just that, minimally extractive, as properly deployed tokens can generate large network effects without taking on any debt. This empowers networks to bootstrap themselves to the point of self-sustainability, allowing them to remain focused on servicing users as opposed to appealing to special interests.

The end result is the creation of market facilitators as public goods, where financial, insurance, gaming, social media, and various other markets yet to be imagined are run purely by user input. The benefits of this are not fully understood or realized yet, but it’s bound to re-architect the way we create and manage the value within social groups and economic markets. If the Internet is any indicator, the change we are about to undergo will be profound, and it’s up to all of us as a collective society to use token-based decentralized computation networks to harness human input in a way that generates equal output. In other words, the value you put in is the value you get out; no unnecessary extraction.

This article was originally published on SmartContent and appears here with permission.

Via this site