Translation | On the Network Effect of Storage Tokens

This article is translated from: https://multicoin.capital/2018/05/09/on-the-network-effects-of-stores-of-value/

Translator: Block Chinese Subtitles Block Mercenary


This article is the spiritual successor to Some Fallacies About Smart Contracts .

Bitcoin has a network effect. Many advocates of cryptography assert that the network effect of Bitcoin is so powerful that hyperbitcoinization is inevitable.

But this is not entirely true. Many have asserted that the network effect of Bitcoin as digital gold cannot reach this magnitude. But network effects are subtle and often misunderstood.

In this article, I will explore the network effects of Bitcoin as digital gold and digital cash. I will also explore other non-network effect models for comparison.

For general backgrounds, I recommend the following: this Medioum forum article , this a16z slide , this Techstars article , and this article on data network effects . For long-term reading of network effects and technology platforms, I recommend websites: platform scale and Sangeet Choudary's books.

Basic concepts of network effects

Network effects are a type of "emergency" (see Baidu Encyclopedia) that occurs when:

As more people use the product or service, the value of the product or service to existing users is increasing.

There are several different network effects:

"Direct network effect"-the increase in usage leads to a direct increase in value.

Direct network effects work because existing users can choose to interact with more and more people after a potential product or service is adopted. Basically all closed-loop communication networks exhibit this network effect, including Internet-based services such as Facebook and Whatsapp.

"Indirect network effect" -As the use of products increases, more and more valuable complementary products will be produced, resulting in an increase in the value of the original product. Operating system (OS) is the most prominent product type in indirect network effects. In order to reach consumers, application developers will be attracted to the development of an operating system; by developing applications for a specific operating system, the operating system is more attractive to new consumers and for future application development People create a bigger market.

"Bilateral network effect" -the increase in the use of one group of users will increase the value of complementary products to another group of different users, and vice versa. Notable examples include eBay, Uber and Lyft, AirBnB, and Amazon's e-commerce platform. In these networks, consumers can benefit from the diversity of choices and competition offered by suppliers, thereby attracting more consumers to come in, which in turn attracts more suppliers.

"Data Network Effect" -When a machine learning-driven product gets more data users, it gets smarter. Most of today's cloud-based applications have data network effects, but their comparative advantages vary greatly in use case and complexity.

Quantifying the strength of network effects

How to accurately measure the strength of network effects is quite difficult, this is not an exact science.

This is particularly challenging because the marginal value of the users that the system adds changes over time. For example, many of my friends have started deleting their Facebook profiles in the past few years. Facebook is still as useful to me as it was 3 years ago. And losing 5% or even 10% of my Facebook friends has little effect on my Facebook experience because I have 500 other Facebook friends.

It is often said that business models that use network effects as moats can be quantified using Metcalfe's Law . Metcalfe's Law states that the value of a network is proportional to the square of the number of users. In order to understand more plainly, it can be said that these companies restricted by Metcalfe's Law have network effects of n².

Because Metcalfe put forward a quantitative definition of the value of the network, it was thoroughly exposed . There is no known network that exhibits a network effect of n² during the growth process. Furthermore, the original assumption that contributed to n² was that all connections in the network were of equal value. In contrast, the network effect of most networks may be closer to n * log (n) than n².

Although this (n * log (n)) of course n² makes more sense (nothing can grow forever quadrilateral), n * log (n) is also an eternal superlinear curve. What we see in reality is that not only all connections are not of equal value, but after a certain period, the value of each marginal connection in the system starts to decline (for example, on Facebook, 10 million users in Asia in the future The value to existing US users is very low).

In practice, the best network effect model should be closer to the S curve than n² or n * log (n).

In real life, there are many case studies that prove the S-curve nature of network effects. This is why the Macintosh (Apple computer brand) survived in the 1990s (if the network effect of Windows is actually n², then Apple may not survive); why there are so many messaging applications in the market (Whatsapp, Telegram, Facebook Messenger, signals, etc.); why Lyft can effectively compete with Uber (as long as I can call a car in less than two minutes, then I don't care how many special car drivers are on the road), and why there are so many professional electric drivers Marketplaces can compete with Amazon.

Even with so many examples, why do people still think that network effects should be n² or n * log (n) curves? Because at the initial stage of these three curves, it is difficult to distinguish the difference between each curve:

And the second half of each curve—the part that only appears when the network reaches critical mass—does not deviate substantially from these three curves . The n² curve continues to accelerate upward twice. The n * log (n) curve will also accelerate forever, albeit at a much lower speed. On the other hand, when the network exceeds a certain saturation point, the S curve changes from superlinear to sublinear.

Of course, not all networks follow the same s-curve. Moreover, not all networks fit the S-curve under optimal conditions.

Some network effects can never achieve exponential growth similar to the initial stage of the S-curve. And some network effects show an increase in log (n) from the beginning, making them always sub-linear, rather than s-curves, because the s-curves are super-linear at the beginning.

The most common example of a network effect with log (n) is to provide a liquid, alternative commodity exchange. Even if you can make a very radical assumption that each new user adds daily liquidity, the marginal value of this new liquidity is becoming increasingly valueless to all existing users. This is true even in the early days of the web. This curve was never superlinear; it was always sublinear.

Let us consider a simple case: every new user trades some alternative products, which will increase the daily liquidity of the products by 0.01%.

When there are 100 users, the daily liquidity is 1% of the market value of the commodity.

When there are 1,000 users, the daily liquidity is 10% of the market value of the commodity.

When there are 10,000 users, the daily liquidity is 100% of the total market.

When there are 100,000 users, the daily liquidity is 1000% of the market value of the commodity (10 times the daily turnover).

If a user owns 0.1% of the trading commodity, the liquidity provided by each marginal user will become increasingly valueless. Technically, the decrease in (marginal liquidity value) will become smaller and smaller as the number of users and liquidity increase, but in fact, the marginal liquidity benefit will become very low, so low that it is difficult for all users to perceive This marginal return.

Any network effect that has an approximate log (n) on the exchange of a given alternative asset can be expressed as:

There is ample evidence that this is consistent with practical experience. If the network effects of alternative asset exchanges were superlinear at any point on the curve, we would not have so many cryptocurrency exchanges. We can observe that if an exchange has a certain amount of liquidity-even if it only shares a small portion of the market leader's liquidity-it is usually sufficient to maintain the operation of an exchange itself and provide market participants with reasonable fluidity.

The network effect of digital gold

To answer this question, let's look at how users use digital gold.

The purpose of holding a value store like digital gold is to … store the value for later consumption. Except for the time when digital gold is converted into other things, the rest of the time digital gold only exists in our account and does nothing. It does not benefit from the increase or decrease of new users.

When a user wants to cash out her digital gold to consume other goods or services, she needs to find liquidity first: people who are willing to buy digital gold. This can be done on exchanges that specialize in alternative digital gold.

The utility of digital gold is a function of its liquidity. As mentioned above, this means that the network effect of Bitcoin can be approximated by log (n).

The network effect of digital cash

What type of network effect does digital cash represent?

To answer this question, let's look at how users use digital cash.

The purpose of using digital cash is both to store value and as a medium of exchange. In addition, digital cash can become a unit of account.

Therefore, the overall utility of digital cash depends on how many merchants are willing to accept the goods and services paid by digital cash.

This is similar to the "direct network effect" (schematic diagram of a telephone network) described earlier. The more merchants that accept digital cash payments, the more existing customers can do business with more merchants.

All major common currencies have demonstrated this network effect in their respective jurisdictions. Because businesses and consumers must pay taxes in their jurisdictions in their home currency, they choose to receive wages (for employees) and revenue (for businesses) in their local legal tender. This creates a strong network effect, because few people are willing to take on a currency with a balance sheet risk. After all, this currency is subject to price compared to the common currency used to buy goods / services and pay taxes. The effects of (exchange rate) fluctuations.

Intuitively, the network effect of digital cash may be more in line with the S curve. The first 50% of merchants who accept digital cash payments get much greater value than the last 50% of merchants who enter.

Storage value vs practical value

The arguments put forward by Bitcoin's main proponents are all playing word games. Specifically, they will argue that "new users (additions) will of course make Bitcoin more valuable. Everyone will buy and hold it, which logically shows that Bitcoin will be more valuable! The argument of asset liquidity It's just another way of saying it. (Originally red herring, translated as red herring, meaning a topic that diverts attention.)

Although this statement is correct in a narrow sense, it ignores the competitive reality: if other digital currencies become digital cash and achieve a superlinear network effect, what will happen to (Bitcoin)? This is what I want to express A broader perspective. Looking at the sub-linear network effect of asset liquidity alone, it really is just another way of saying it. But if other digital currencies become digital cash with super-linear network effects, and Bitcoin is still only digital gold with sub-linear network effects, then Bitcoin will be surpassed.

In the digital currency circle, such frameworks often trigger debates about value storage (SoV) and practical value . The value store's point of view is based on reflexivity: the more people hold it, the more valuable it becomes, thus driving more people to hold it.

Of course, reflexivity may strengthen the positive effect, and may also strengthen the negative effect. This will cause excessive price fluctuations and instability, so that the purpose of value storage is completely destroyed. When prices rise, it's easy to believe in the value store perspective. But as prices fall, the potential utility value creates a resilient price base.

It's easy to forget that our knowledge of cryptocurrencies is still in its early stages. There are 7 billion people on the planet, but fewer than 50 million people have cryptocurrencies. This number is less than 1% of the global total. In the software world of open source software, every function can be copied , and the key to winning is to achieve network effects as soon as possible. This is why the advantages of network effects are so important. These advantages depend mainly on the ability to successfully connect tens of thousands of users, which will make a huge difference in the value of the network.

Other moats

The network effect is just a competitive moat. There are many other moats.

Other types of moats advocated by Bitcoin's main proponents are "brand recognition" and integration with third-party ecosystems (such as exchanges, ATMs, other financial products, hardware, and mobile wallets, etc.).

To explore the power of these moats, let's compare Bitcoin with Ethereum. This is not to say that Ethereum has the potential to surpass Bitcoin. Rather, this is just an example to illustrate that Ethereum, as a competitor, can achieve a leap in value in less than three years of its creation.

Brand recognition is indeed a moat. Bitcoin is the leader in cryptocurrencies. But if its brand is unsurpassable, it is ridiculous. Because no brand is insurmountable.

There is currently no good way to measure the value of an open and unauthorized brand like Bitcoin, but we can use Google Trends as a rough measurement tool.

(As shown in the figure above) The blue curve represents Bitcoin and the red curve represents Ethereum. When the gap is greatest, Bitcoin's search frequency is about 11 times that of Ethereum. Even today, the difference is 8 times. Given the strong volatility and rapid pace of development in this area, Bitcoin's leading edge may disappear in a few years.

So what about third-party ecosystem integration? In this regard, Ethereum is almost comparable to Bitcoin:

Exchanges-All major exchanges support bitcoin and ethereum fiat currency trading pairs.

Hardware wallets-All major hardware wallets support Bitcoin and Ethereum.

ATMs-As far as I know, all cryptocurrency ATMs support both types of assets.

Mobile wallets-whether Bitcoin or Ethereum, there are a large number of mobile wallets on iOS and Android.

Other financial products-Bitcoin is leading the Chicago Mercantile Exchange (CME), the Chicago Board Options Exchange (CBOE), and the NASDAQ futures. But considering Ethereum's development trajectory, it seems quite reasonable that it can achieve equal status in 24 months.

Again, my point is not that Ethereum will replace Bitcoin, but that the advantages of third-party ecosystem integration are not insurmountable.

to sum up

Network effects and competitive moats are often misunderstood. Unlike the mainstream idea, there is no n² in the world. In fact, many networks are in line with the network effect of log (n), especially exchanges that can replace assets. When bitcoin is used as digital gold, its network effect will conform to the eternal sublinear log (n) curve. When bitcoin is used as digital cash, the proportion of cryptographic technology in the global population will increase from 1% to 50%. , Then it can achieve super linear network effects. If Bitcoin is to become a mainstream value store, by definition it needs to show a superlinear network effect as its value grows.

In addition, other types of competitive moats-such as brand recognition and broader ecosystem integration-will not result in incremental reports due to scale growth, and will easily be affected by superlinear network effects. Defeated by competitors. We have sufficient evidence to prove this.

The battle to become the big winner of cryptocurrencies has just begun. When all projects are at the first 1% of the network effect, the differences between them are not easily recognized. Therefore, it is easy for everyone to think that these network effects have already occurred, but in fact these judgments are too early.

Thanks to Chris Dixon and Matt Huang for providing feedback on this article.

Blockchain Chinese Subtitles Group

Committed to the dissemination of cutting-edge blockchain knowledge and information, and contributing to China's integration into the global blockchain world.

If you know some technology and some English, welcome to join us and add WeChat: w1791520555.

Click to view project GITHUB, and more translations …

Translator profile

Block mercenary, Master of Science and Engineering of Zhejiang University, head author of Coins, column author of Golden Finance and Gyro Finance. Focus on blockchain technology research and industry analysis, welcome to add WeChat: wxlinzju.

All rights reserved. The above content must be specified in full.