“I am putting myself to the fullest possible use, which is all I think that any conscious entity can ever hope to do.” (2001: A Space Odyssey)

Welcome to Nile: A Sovereign Bookstore in the Age of Decentralization

Andy Bromberg
12 min readOct 2, 2019

What would Amazon.com look like if it were created in the image of Bitcoin?

This question pushes us to explore how a truly autonomous organization could come to be — and what that might mean for the future.

We’ll call our (as-of-yet imaginary) creation “Nile.” Welcome to Humanity’s Sovereign Bookstore.

What is Nile?

Nile is a bookstore on the internet that exists wholly autonomously. It only depends on the decentralized platform on which it runs — nobody can exert any control over it.

Nile was created by a pseudonymous individual or group named, say, Jatoshi Bezos. Jatoshi wrote the code for Nile and deployed it as a smart contract onto the Ethereum blockchain. Then, they wrote a tweet announcing it and promptly disappeared. Since then, nothing about Nile has changed, but many thousands of users have interacted with it (in a variety of ways, which we’ll get into shortly), and Nile has behaved as a true independent actor.

Nile is a pure decentralized autonomous organization.

Just like Amazon in 1995, it is made up of two components: Nile, The Bookstore; and Nile, The Organization. Here’s how it works¹.

Nile, The Bookstore

Nile was deployed with a goal of selling books. At its inception, Jatoshi digitized hundreds of books and uploaded them to a decentralized, uncensorable storage solution (think IPFS). Each of these books was assigned a price, and Jatoshi built a table with book titles, access instructions, and prices into the Nile smart contract.

Nile’s inventory table.

Any prospective book-buyer can send a request to the Nile smart contract and get a list of book titles and prices. Then, if the book-buyer wants to move forward with a transaction, they can send the appropriate amount of ETH as payment (along with the identifying title) to the smart contract, and in return they will receive the hash, which they can use to locate and download the book².

It’s that simple — even simpler than Amazon in ’95, thanks to not needing physical inventory or shipping!

Note that the digital nature of the product is key here — were the items physical, we’d have a whole separate set of challenges in making Nile autonomous. We’ll address strategies for those challenges in a future post. In the meantime, see Szabo’s canonical paper on public networks — the section mentioning “vending machine” is particularly relevant.

Nile, The Organization

“Well, you seem like a person but you’re just a voice in a computer.” “I can understand how the limited perspective of an unartificial mind might perceive it that way. You’ll get used to it.” (Her)

Here’s where things get interesting. Nile generates all this revenue from book sales — but there’s nobody running the show or taking a salary. So what happens to that cash?

The Nile smart contract was instantiated with 100,000,000 “ownership tokens” (ticker: NILE) issued on the Ethereum blockchain. These tokens — as we’ll describe in depth shortly — represent a right to future cashflows of the bookstore.

Over time, Nile autonomously issues dividends to the token holders. And on top of that, it autonomously sells and buys back these tokens as it sees fit.

Feel free to gloss over these specific details and parameters regarding how Nile manages its token economy and skip to Nile as a Corporation, if you want.

Trading Window

Every 30 days, Nile opens up a one-day trading window. On this day, Nile puts up 1% of its current NILE holdings for sale. These tokens are offered at a 10% premium to the last price at which the contract successfully sold any tokens.

But the Nile contract doesn’t just sell tokens — it also buys them. The contract keeps a list of all the successful sales it has completed. For every token the contract has sold, it is willing to buy one back for 90% of the original selling price, starting with the cheapest tokens and going up from there (see Appendix: Buyback Example for more clarity).

Given these rules, Nile will never lose money on buying and selling NILE.

That said, Nile does spend ETH on transaction fees for its bookselling, other interactions with users, and potentially book storage and retrieval³, so its appetite for token buying is limited by keeping some amount of ETH in reserve. It will never buy so many tokens back that it goes below that operational ETH reserve balance.

Dividend Window

Once the trading window wraps up, Nile moves on to a final goal: distributing revenues to the NILE tokenholders.

The day after the trading window, Nile looks at its remaining ETH balance. It subtracts out its operational reserve, and distributes all the rest of its ETH holdings proportionally to the token holders. If it has 10 ETH to distribute and you own 1MM of the 100MM NILE tokens, you’re getting 0.1 ETH straight to your wallet.

Nile’s Infrastructure

To recap, here’s a very rough outline of all the interactions around Nile:

Anyone want to teach me graphic design?

Nile as a Corporation

Nile, as you can see, looks an awful lot like a corporation. It:

  • Operates a customer-facing business: selling digital books in exchange for ETH
  • Has a tradable representation (NILE) of its success
  • Distributes dividends to holders of that representation
  • Owns assets: primarily its access to the digital books, as well as its ETH and NILE holdings
  • Manages its finances: reserving capital for operational expenses; buying and selling NILE profitably
  • Limits liability: practically, although perhaps not legally (and certainly not conventionally) — who could a third party hold liable?

Nile as an Actor

But Nile also looks an awful lot like a person — or at least some sort of autonomous actor or agent.

Corporations are definitionally governed by their shareholders, and Nile doesn’t have any. The tokenholders have a right to dividends, but they don’t have any governance rights whatsoever — nobody does. Not a single person can change or guide the way Nile operates.

Nile operates without control or influence by any person, group, or entity. Since its birth, it has been entirely in charge of its own fate. Can anything else say the same, outside of biological, intelligent beings?

What is Nile? (Redux)

“What… am I?” (I, Robot)

I’m not sure that today we have much of a vocabulary for this uncanny position Nile occupies between a corporation and a person or actor.

There are machines out there in the world that we call autonomous: certain robots, cars, weapons, facilities, and more. But there’s a big difference: those run autonomously, versus Nile, which exists autonomously.

Yes, you can turn an “autonomous” robot on and have it clean your kitchen floor without further input. But you can also turn it off, and the manufacturer can beam in new firmware to control it. Once it’s on, it runs autonomously, but its existence is dependent on other persons and actors.

Nile has sovereignty: something that, in the deepest sense of the word, has only existed for biological entities — the sovereignty of nations doesn’t even come close. By this measure, Nile’s existence is more like a person’s than like a country’s or a corporation’s.

Although (as currently outlined) Nile is unable to add new SKUs and expand its lines of business, it takes no large leap to imagine a world where it could. And moreover, it is meant to serve as just a small vignette of what is possible.

Sovereignty and intelligence are often conflated—a point demonstrated by some of the movies sprinkled into this post. With Nile, we’ve imagined a sovereign agent with only a pinch of intelligence. At the same time, there are many brilliant people working on highly intelligent agents (see: OpenAI and others). When the ideas of sovereignty and artificial intelligence are combined, the result could be incredibly powerful—and, without meaning to fearmonger here: potentially risky and dangerous.

Decentralized technology gives us the building blocks of an autonomous, sovereign organization: allowing for transactions, ownership, communication, pseudonymity, and programmability. Nile is one of the simplest organizations you can create with these primitives. We can only imagine what the truly advanced versions will look like.

The New Nature of the Firm

Why is it that we should want these sovereign organizations to develop?

There are a few immediately obvious practical benefits to such an organization, including:

“Nobody altered my protocols.” (Autómata)

1. They remove human corruptibility from the equation. At no point can anyone (Jatoshi or otherwise) decide to start siphoning funds from the treasury or otherwise abuse the organization. All the rules are written in stone (even better: on a blockchain!).

2. They are unstoppable. Nile could sell only banned books — and there would be nobody for a government to call to shut down the service. They could be entirely extra-jurisdictional.

3. They can be relied on to behave predictably. Nile, for example, has a clear goal of maximizing tokenholder value: it distributes all of its excess holdings on a monthly basis. There are no humans involved who could cause a detour from this goal.

4. They reduce transaction costs. By removing employees from the mix, Nile can deliver its specialized goods with as close to perfect efficiency as possible.

But this last point reveals something deeper: that perhaps Nile and its ilk represent an evolution of the firm — or at least a new category.

Ronald Coase’s classic paper The Nature of the Firm outlines why it is that companies are formed in the first place. At its core, he suggests that even in a perfectly efficient market, the transaction costs associated with buying a good in the open market can make it more efficient to form a firm and produce that good internally. Even if widgets are priced perfectly, you may want to make them yourself to avoid costs associated with searching for the best ones, coming to a legal agreement with the supplier, policing their output, etc.

Coase’s theory centered around the growth of the firm — that in these cases of high transaction costs, an entrepreneur can justify hiring and managing a team to produce goods. He suggests growing from a “firm” of 1 person to a firm of 2, 20, 200, 2000, or more. But what if now we can reverse that trend?

What if — in certain cases — the best approach to high transaction costs is not to go from a firm of 1 to a firm of 100 people, but to go from 1 to 0?

What if the new nature of the firm is a maximally efficient autonomous one? An agent that, for a buyer, combines the upsides of the market (no need to hire and manage a team) and the upsides of producing within their firm (minimized transaction costs)?

There are many, many challenges and open questions along this path of evolving the firm — but it seems worthwhile to explore it further.

What Next?

I hope somebody goes out and builds Nile as well as more complicated organizations (if you’re going to, please message me!).

Their existence will encourage hard thinking about questions both conceptual and practical. Some that come to mind:

  1. What businesses or services do these enable that weren’t previously feasible?
  2. How do these interface with existing legal frameworks (liability, securities law, taxation, etc.)? What is their jurisdiction, if they have one? (See Appendix: Smart Contract Disclosures for a thought on this)
  3. What happens when these organizations start to autonomously interact with each other?
  4. Does the creator deserve upside in these organizations? Do they get that like Bitcoin’s creator(s) or like Ethereum’s creators or some other way?
  5. What happens if somebody writes and releases code for one of these without deploying it — and then other people do the deployment?
  6. What happens when there is a vulnerability in one of these organizations?
  7. There is a wide spectrum between Nile and Amazon — how are each of these questions (and others) answered for different points on that spectrum? What defines various categories within the range?
  8. How will some of the interface problems be solved? How will users access Nile — via centrally-distributed browsers or will other solutions emerge? What’s the best way to securely store the books?

Footnotes

[1]: All specific numbers — as well as the precise proposed details of the token economy — are purely for illustration purposes. This piece focuses on the concept more than the details.

[2]: I am glossing over a lot of details and poorly- or un-solved problems in crypto in order to focus on the idea of this organization rather than the implementation. For example, the hashes pointing to the books are stored in a highly insecure way — anyone could snoop on the smart contract and find the books without paying. You can imagine a more complex system with better “protection” of the content, perhaps using something like Nucypher or Keep.

[3]: Perhaps using Filecoin once it is live!

[4]: Of note: for a very long time, Nile will hold the majority of its tokens (remember it only sells up to 1% of its current holdings every 30 days). As a result, the majority of its holdings will be distributed right back to itself every month. A secondary effect of this is that tokenholders are guaranteed to receive a regular dividend, even if there’s no bookselling activity.

[5]: Nile’s basic intelligence includes, for example, its simple market-making behavior for NILE tokens (sell high, buy low).

[6]: andy@andybromberg.com — also, see Appendix: Simple Improvements for more you could do.

Appendix

Buyback Example

The contract keeps a list of all the successful sales it has completed. For every token the contract has sold, it is willing to buy one back for 90% of the original selling price, starting with the cheapest tokens and going up from there

For example, let’s imagine NILE has completed three sales of tokens: 100 NILE for 1 mETH each, 150 NILE for 1 mETH each, and 100 NILE for 1.1 mETH each. See below for the list of successful sales (from Nile to investors).

In this case, Nile would be willing to buy 250 NILE (100 + 150) back for 0.9 mETH each, and then 100 NILE for 0.99 mETH each. The 0.99 mETH price wouldn’t be available until 250 NILE had been sold back in at 0.9 (even if that took many trading windows).

Said another way, the Nile contract would start by bidding 0.9 mETH / NILE for up to 100 NILE. Then, once users have eventually filled that tranche, Nile would bid 0.9 mETH / NILE for up to 150 more NILE. Finally, once users fill that tranche, Nile would bid 0.99 mETH / NILE for up to 100 NILE. If the smart contract managed to sell more in the interim, the price would go up (because it always does, under this formulation), and so the bidding price for buybacks would continue to rise.

Simple Improvements

I kept the Nile contract fairly simple for the sake of this article. But if you’re going out and building it for real, here are a few improvements that jump out at me:

  • Keep the books’ storage hashes private or otherwise store them more securely
  • Or store some sort of proprietary digital content
  • Dig in to the token economics — including at what prices the contract is willing to sell or buyback NILE. I’ve put very little thought into these specifics.
  • A more complex system than “always sell 10% higher than you last did, and always buy back for 10% less than you originally sold” may work better.
  • Think about whether you want to be able to add books over time — and how that impacts the autonomous nature of the assets. Perhaps one day it becomes a real marketplace, in much the same way Amazon has.
  • Consider whether you do want to build governance rights into the tokens — should they be able to vote on changes like pricing or parameter tweaks?
  • Build dynamic pricing for books, so their prices change based on demand
  • Sort out how much of an operational ETH reserve you want the contract to keep, and perhaps make it dynamic based on user transaction volume and ETH gas fees

Smart Contract Disclosures

Disclosures, representations, and warranties are core to legal frameworks globally. But these are all flawed, because they rely on fallible, corruptible human beings telling the truth. Autonomous organizations are open source and 100% transparent and auditable by definition — so are their effective disclosures, reps, and warranties not better than those that we rely on today? What can we then do with that information? We’ll discuss this more in a future post.

Acknowledgements

Thank you to Charles Songhurst, Balaji S. Srinivasan, Naval Ravikant, John Backus, Justin Gregorius, Scott Keto, Mike Maples, Ryan Zurrer, Alexa Cohn, and many others for conversations that fed into these ideas—and some of them for reading drafts of this piece.

--

--

Andy Bromberg

Co-founder & President at CoinList. Founding research scientist at the Stanford Bitcoin Group.