So I was trying to walk my Mom through why crypto may well be a valuable asset class, and why certain elements make it so hard to tell thoughtfulness from rashness in the space. I started typing up what I thought would be a page, and ended up with a bit more. To amortize wrist-strain, I thought your Mom might enjoy reading it too (mine will say she did regardless :b).
Disclaimer: I freely editorialize and will make corrections as people point out my (many, I’m sure) mistakes here. Read at your own risk (like everything else online).
Much of this is debatable, I’ve strived to make sure my facts are not.
1. The technology
1.1 Some history
The basis for a lot of the work in this field comes from Distributed Systems. This is an area of Computer Science that really took off along with networks (and the advent of the Internet). The idea is more or less that as you are able to connect more and more machines on which to spread out work (computational tasks, storing large amounts of data, etc.), you want to make sure that this network is resilient to machine failure. Simply said, if your Macbook dies every 2–3 years on average, a data center with 10,000 machines will experience a dozen or so failures a day. The idea is to make sure the system as a whole can work even if some of its parts fail. Out of this comes the question of how to coordinate work across cooperating machines.
From this work came the idea of coordinating work across machines that don’t belong to a single entity (e.g. Google), but rather to a bunch of different parties with varied interests. This is the rise of peer-to-peer protocols in the 90’s and early 2000’s , used, for instance, to create file-sharing apps. This is a good use case for peer-to-peer sharing because downloading a file can be a bandwidth intensive task, so rather than hitting a single server to ask it to give you the file, you can spread the file across machines and ask a group to each send you a small bit of the file. Plus, by co-opting various machines to do this, you get high availability for your files: a machine will always be on the network to get you your content. Lastly, this is a good application because people started distributing free music and movies through these networks, which is of course illegal, but the decentralized aspect of this made it hard for authorities to crack down on any single culprit (the files are spread across the Internet and the people sharing them connect and disconnect at will from the network).
The key development here is a shift from cooperative protocols to byzantine ones (which means machines can act arbitrarily, even adversarially on the network): when Google owns and runs all of the machines, they can centrally ensure all of them are doing useful work; when you pool together a bunch of different peoples’ resources even though they might have orthogonal interests, you need to figure out how to ensure that people are all working toward a common goal. For instance, I want to download movies, but I don’t necessarily want to upload them for other users — a waste of my bandwidth — the peer-to-peer protocol may try to ensure I upload to allow me to download (tit for tat). And so, the protocols started focusing on a notion of consensus: how do you ensure that in spite of orthogonal or opposing interests, all parties provide utility to one another? What is the basis for this trust?
Throughout this time, and as the Internet grew, people started thinking about how to build a trustless electronic money system. We’ll get to what trustless means, but what is money? A means of exchanging value through a proxy: in order to have value, a currency needs to exist in finite amounts (otherwise people could just create it out of thin air), and needs to be widely recognized as being valuable (otherwise I can’t get someone else to accept it). That first condition is hard to ensure with digital currencies: this is the double spending problem (if you’re familiar with this, just save yourself some time and skip the rest of this section).
N.B.: We are going to skip over all questions of cryptography here. If you want to read more on it, Google “Public Key Cryptography,” “Merkle Trees,” and “Hash Functions.” Basically just know this is research from the ’70s onwards (based on millennia of numbers theory) that creates certain safety guarantees on the network (like you are who you say you are) that most everyone believes to be true (though we can’t quite prove it — if this turns out to be false, then cryptocurrencies being worthless won’t be your biggest problem: anything you have ever put online will be open for the world to abuse: global catastrophe and end of all connected digital systems — look up “P=NP” for more info).
To illustrate the double spending problem, let us take three participants: Alice, Bob, and Eve. Eve has $20. She can transfer $20 of digital assets to Bob in exchange for a pizza. But how does Bob know that she hasn’t already transferred these $20 to Alice for beer and double spent?
In the real world, it’s simple enough: we use physical money (once I give it away, I no longer have it) and assume that getting forged money is unlikely (hard to do, and highly punishable). These systems rely on a trusted third-party (in the case of paper money, the US Federal Reserve). But in fact, we also use digital money already: Visa and Mastercard are trusted third parties that work with multiple banks to allow you to pay someone else who might bank elsewhere. Whilst you can’t see the money leaving hands as with paper, your bank tell Visa “Eve is good for $20,” which tells Bob’s bank “Bob now has $20 more,” and by trusting both banks, the payments network, and the government’s power to prosecute everyone if there is wrongdoing and not to devalue the dollar, etc, Bob trusts that Eve paid him. An issue is made evident: lots of trust is needed on all sides, and each of the trusted parties takes a fee for the work they provide, so you get charged to pay someone.
And so this question: Can this double spending problem be fixed without a trusted intermediary?
There were a lot of attempts to do this over the ’90s and ’00s but one ended up being particularly clever and elegant: Bitcoin. The idea was proposed in a paper published online in 2009, by a pseudonymous researcher (Satoshi Nakamoto). The basic premise is simple: keep a distributed list of transactions, a ledger (the blockchain), that everyone has a copy of and that keeps track of who owns what, what payments are made, etc. That way, when Eve pays Bob, the transaction is added to the ledger, and Bob can just check his copy of the ledger to make sure he’s been paid.
But how then can Bob trust that his ledger is the same as everyone else’s and that Eve hasn’t just sent him a version saying he was paid to screw him over? How can there be consensus around what the state of the system really is?
This is where something called Proof-of-work comes in. There are these things in CS called “one-way functions” that are super hard to come up with for a given input, but very easy to verify.
Nakamoto thought, “what if we made it very hard to add a transaction to the list, and got some third party to use one of these one-way functions to generate some special sign that he/she has authorized the transaction before putting it on the list?” We need only ask that this verifier check that Eve is able to pay Bob what she promised before doing the one-way function and adding the transaction to the list. In a sense this computation is an authorization of the transaction. They’re hard to authorize but it’s easy to check whether they have been.
You might say but “Bob can’t trust that any verifier is doing their job:”
· Why on earth would a verifier waste their time doing this?
· What if they work with Eve?
· What prevents Eve from sending a transaction for Bob and one for Alice to different verifiers (both of transactions are independently correct), and getting her shit from Bob and Alice before the different versions of the blockchain are noticed by the verifiers?
To answer these questions I should explain: Nakamoto thought of incentivizing a verifier by giving them a reward (the coinbase) every time they successfully verify a transaction. This is called mining, the verifiers are called miners. Interestingly this is how money is minted in the network: miners get paid with brand new Bitcoins.
So miners are doing this to make money and they’re all racing to be the first to verify and add a transaction to the chain so they can earn their coinbase.
In this race, there’s a chance multiple transactions might get mined at the same time (Eve is asking as many people as she can to verify her transaction so she can prove to Bob she paid him), and this is the last piece of the puzzle: how do you deal with conflicting chains?
Nakamoto decided that when multiple chains exist (because of simultaneously mined blocks, this is called a fork), which is bound to happen from time to time, people should only pay attention to the one that has the most transactions. In a way it’s the one the most work has gone into.
And this is what prevents a miner from working with Eve: as rational economic agents — in fact a multi-billion dollar industry has grown out of that, from people doing this proof-of-work for money. It’s more or less like converting electricity (to run your one-way function) into $$ — , the miners will always try to add transactions to the longest list. Otherwise their work will be useless (anybody who refuses is going to be on the smaller chain and will basically earn Bitcoin that no one else accepts as real money). And this is the one assumption Nakamoto makes: most of the miners need to be honest here, so that as they race to approve transactions, any miner who notices their colleague is working with Eve and has approved a bad transaction immediately drops this list and get back to work on a proper one (where Eve isn’t cheating) since he/she knows that most of the miners are honest and so the longest list will not contain falsehoods. This looks a bit like a circular argument, it’s made easier to understand by what’s called Nakamoto consensus: since miners are all earning Bitcoin, they all have an incentive to be honest because if they aren’t and allow Eve to cheat (something the entire world can see), than Bitcoin has lost its value as a currency. No one will want to use it since it can’t be trusted. And so, since they’re working so tirelessly to earn Bitcoin, the miners have a vested interest in protecting its value by ensuring no one cheats (everybody and their mother has a different definition of Nakamoto consensus, so please correct me if this isn’t precisely what it refers to).
So we know that eventually, truth will win out. Now Bob just needs to make sure his transaction approval is actually on the longest chain: he waits to hear from enough people around the world to believe that his approval is most likely on the longest chain. Bob can now rest easy, his transaction is confirmed: if Eve had already spent the money, he would know it by now. She gets the pizza.
This notion of consensus is the key that opens new doors in distributed systems. We can now build an entirely transparent source-of-truth that represents the state of some system in the world. What else can be represented using this technology?
To some people, a lot of the early adopters of the space, this is a long-awaited solution to a rational fear: what happens to me if the trusted powers lie?
· If visa says I don’t have the money I have? I go to my bank
· If my bank says I don’t have the funds I gave them? I go to my government
· If my government strips me of my rights? I quietly go away
It’s about trust: and ensuring society is not turtles all the way down…
To others, this is the opportunity to build new sets of technologies based on trustless networks. It’s not so much about trust but about the efficiencies that come from cutting out the middlemen — traditional purveyors of trust, replaced by miners now.
People started thinking “what else can we do with this blockchain”: what does this scale allow us to build that didn’t exist before? What forms can mining take? Could it be some other operation?
They started building alternatives to bitcoin (altcoins), optimized for certain uses. Some examples:
· Bitcoin transactions take too long to clear (10 minutes +), we can build a better currency tool -> Litecoin, Navcoin
· This ledger can be used to track more complex transactions across large financial systems like Visa -> Stellar, Ripple
· Bitcoin isn’t really anonymous, transactions can easily be tied to economic entities ->ZCash, Monero
· We should use this blockchain stuff to store gigantic files all across the world -> Filecoin, Sia, Storj
· We should use this blockchain stuff to give people a perfectly trusted computer, where all of its state (files, operations, etc) is public and people can reliably prove that they are doing what they say they are à Ethereum, Tezos
· And so many more…
A lot of these new protocols work as described: powered by miners, and some token that they work for. Those need tokens to run, and so each give rise to a new currency — if the network is to be used, presumably it serves some purpose to someone, and so the miners’ input has some value. The tokens are the embodiment of that value. (Note that a ton of interesting work is build outside of the token space altogether, tokens are simply one way to incentivize an ecosystem).
1.2. Some nomenclature
· Block: A bundle of transactions (transactions are actually authorized in bunches to help make the system more efficient).
· Blockchain: the trustless ledger, or list of that records the state of the world (all transactions) under a given system.
· Miners: the folks who keep the blockchain running by authorizing transactions and adding blocks to it, earning tokens in exchange.
· Token: All of these blockchains need to be verified by miners, the tokens are what is used to reward their work.
· Cryptocurrency: It can mean the token for a given blockchain, but I use it to refer to the entire system of token + blockchain + miners. A cryptocurrency is the result of a set of choices (a protocol) made around what a transaction should symbolize, what miners have to do to approve transactions on the blockchains, how rewards are managed, etc.
· Trustless: the set of protocols which can operate on a basis you can verify for yourself: no need for a trusted third party to vouch for you or anyone else.
· Decentralized: This refers both to
- The fact that the systems built on blockchains are distributed by their very nature: everybody on the network should have a copy of the blockchain, everybody can mine, etc.
- The related fact that there is no central third-party to provide a copy of the truth.
· Bitcoin: One application of the blockchain, the first really. It is a blockchain built to create a currency.
· Network: This is widely used to refer to the set of people participating in your system as a whole. It usually refers to a set of miners powering a cryptocurrency, or to the larger set of people who own and use your token.3. The market
2. The Market
Here is where we move into debatable territory.
2.1. From a product standpoint
Some people (me most of the time) think this stuff is super over-hyped. We’ve built trustless databases… It’s very cool, so what? The world of engineering is that of tradeoffs, and the cost of trustlessness is high. With it comes poor performance, latency, and so many other problems that make a computer system really shitty.
However incentives to try and build something here is huge: by its very process, with a new trustless protocol comes a need to incentivize miners and so a new token. Provided the underlying thing that is being done is valuable, then the tokens they are paid with are worth something. Thus, in building a new blockchain, you are creating a currency, and since you’re building it, you get to decide how tokens are generated and you control the money supply. This is pretty much akin to if everybody started drawing their own paper money assuming that if enough people use it, then you’re rich (which is true). But people don’t do that, because drawing is hard to scale (think carpal tunnel), and no one really gives a shit about your art (if you’re not Rubens). Except that the blockchain allows you to scale token creation, and if you’re smart about what you design it for, potentially allows anybody in the network to do something cooperatively they could not have done before.
Picture the wild-west: we’ve found this new process for extracting minerals from the ground and everybody is rushing in to use it, without necessarily caring about what they are extracting: some people are pumping shit out of the ground, others might be pumping gold… But most everybody is too busy admiring how the pump works to worry about telling them apart right now.
Launching your currency is now called an Initial Coin Offering (or ICO), and people are so utterly enthralled by the idea of being part of the next revolution that many will literally back anything with the word crypto in it, no matter how ill-fitting with decentralization it is (look up “Doge Coin” for an early example of satire around this), or the stage of the product in question. How does it work? Publish a whitepaper outlining your vision for a blockchain use (a protocol) and sell a right to the future currency through a SAFT (Standard Agreement for Future Tokens) in order to fund the process through which you will build the system that will create said currency and endow it with value.
As an investor, hope that it gets created so you receive your token some months later, and hope it’s actually a useful protocol otherwise the token is unlikely to have value (once the speculation pops).
Enter the world of terrible ideas: i.e. every popular product you’ve ever seen spun on the blockchain. As an investor in the space said to me:
Look at the realm of ICOs and ask yourself
1) Should this process really be decentralized? 98% of ICOs out.
2) Does the team have the technical ability to build it and the community needed to power it? Take out another 1.5%.
3) Is it valued appropriately? You can go home and save yourself some money now.
This is where a lot of people will lose a lot of money because reality will catch up in time and useless protocols will die, and with it the value of their tokens. For others, the hope is that the underlying utility will catch up to the speculative price, speculation will increase, and so on.
In the latter case, the ICO has created this amazing opportunity removing virtually all barriers to funding a product through a public process. What’s more is that the token’s value is itself inherently tied to the value of the underlying network. In a sense, this is the most unmediated way to invest in an idea and space! In this sense, this is an amazingly fruitful space in which so much value will be created. A big part of the question deals with your risk-profile and the returns you seek as an investor.
2.2. From an investor standpoint
There are three broad sets of investors involved in this market now, so far as I can tell:
These folks are doing a lot of what they’ve always done and investing in technologies they think have a meaningful chance of creating large outcomes. An added layer of complexity has been added to their work though: how do they price these investments since a lot of these technologies are so far away from being useful to end-consumers that they have to invest in the underlying protocols themselves, which, as a kicker, are given monetary value directly by other people in the market. So hypothetically some company may raise funding from traditional VCs and then launch an ICO in which they sell 10% of a token for $100m (valuing the currency at $1.0B). When it comes time for the VCs to distribute capital to their LPs, how do they price that huge capital infusion into the company?
· Hedge Funds
With the plethora of these altcoins launching, there is actually a market of a few hundred tokens that are liquid right now, and so these funds are trading on them based on the underlying technology, perceived value of the protocol, market sentiment, or just dumb luck… A lot of funds are also charging LPs just to hold a couple of assets long, because doing this is still so difficult right now; but perhaps good custodianship is worth this price. Most of their LPs right now are VC funds, high net-worth individuals and now some family offices.
But the thing to wrap your head around is that much of this is still speculative. People assume these technologies will have value in the future and so there is an aggressive land grab now to own the future means of production, but none of these currencies have found mainstream uses just yet.
People with literally no experience in technology or investing are anointing themselves “blockchain experts” since any expert has fewer than 8 years of experience anyways and going out to raise absurd amounts of money to invest in unproven protocols themselves.
NB: To be fair to a lot of folks though, there is some clear defensiveness in the market from early crypto purists who rely on how long you have been interested in the space for as a proxy for legitimacy. I think this is short sighted: ultimately, everybody in the space should want it to reach as many people as possible since newcomers can build valuable things too. But because money is so directly involved in the system, people get a bit emotional.
Within the hedge fund crowd, there are two sets of actors:
· People coming from the hedge fund world
· People coming from the tech world.
I have obvious biases and tend to favor the latter here as they are usually looking at these things from a first principle perspective and thinking about the underlying value of the network being built, but the former have clear understanding of markets and trading!
There are going to be cons in both sets: some traders eager to apply their skills to a highly unregulated space since 2008 took the fun out of derivatives, and some technologists who can now raise money on literally nothing by talking to LPs who assume they won’t understand the subject matter anyways.
The amount of noise in the market is horrendous.
3. Some rational theses
There’s obviously a lot to unpack here and it can get pretty hairy to tell meaningful innovations from fluff in this market. I usually think common sense and first principle questioning of the underlying value of a protocol works well. A fair rule of thumb: if someone can’t explain to you why the protocol should be valuable in plain English, it probably isn’t (though writing this challenges my notions of what is easy to explain).
With that said, here are simple heuristics I have found helpful when looking at this space.
A first clear value of the blockchain is its ability to create self-organizing marketplaces. That is, you can organize liquid markets without a need for a middleman. The prototypical example here is crypto-Uber. Ultimately, Uber’s value proposition is seemingly being able to bring together drivers and passengers under one roof and create a liquid market around ride sharing. Driver availability brings the price down; excess demand brings it up (this is surge pricing).
Uber takes a big cut of each transaction in order for this service and thus creates inefficiency in the market: what if you could pay your driver directly without needing Uber to find him/her? Ubers tend to be cheaper than cabs because they are able to leverage a larger network of drivers (not just people with medallions), and in much the same way as the market has taken over the oligopoly, so too can the blockchain replace the market maker. This would mean, for instance, a cryptocurrency where the proof of work for miners is driving, and passengers pay tokens to get a transaction (ride) on the blockchain. There are obviously a number of questions that remain unsolved here (most obviously: how do you prove you’ve driven someone somewhere?), but it’s reasonable to believe that this service would be 15–25% (Uber’s cut) cheaper than centralized alternatives, and could incorporate a broader set of value-add services.
And so too with most other products that require market dynamics or benefit from network effects: the blockchain allows for these markets to organize and police themselves and so can run more cheaply.
Now I think people take this a bit too far and generally forget a huge number of issues that have to be figured out in order to make crypto-Uber work: Uber isn’t simply a market maker, they also provide a reputation system for drivers and passengers, a sleek UI, nice integrations with GPS apps, financing for their drivers’ cars, etc. This is all really hard to build on the blockchain, and some of these features may in fact justify Uber’s 20% cut. It’s not because things can be decentralized that they should.
Fat Protocol (Joel Monegro for USV, now at Placeholder)
The notion here is rather simple, it is simply that value from a blockchain’s use accrues at the protocol level, with the token. The argument is basically that whereby the monetary value for creations enabled by the Internet was concentrated at the application layer (the founders of Google have made a lot more than the inventors of the http protocol), due to the nature of how the blockchain works, useful applications will give value to the underlying tokens: value is amassed at the protocol layer. Specifically, the point Joel Monegro makes in explaining this is that the value of a given protocol will be more than the combined value of all of the applications built on top of it, since there will be speculation as to the future value of the protocol (in much the same way that people trade stocks with varied Price to Earnings ratios).
And so an argument can be made that investing in tokens directly is often superior to investing in the companies that created the protocol: the token is where most of the value will accrue, and while the company can fail, the token may live on (Satoshi Nakamoto disappeared in 2014 and yet Bitcoin’s value has grown 7x since then).
I think this point can be taken further to argue that lower-level protocols (things lower in the stack that are bound to be used for many applications) will be more valuable than the applications that sit atop them.
For instance, filecoin, as a decentralized hard-drive, might be used by cryptoUber to store driver reputations, ride histories, and other such things needed to build a good product. It may do the same for cryptoAirbnb and any number of other decentralized products in need of trustless storage. Since value will accrue downwards through the layers of abstraction, invest in middleware and core protocols rather than apps right now.
Usage vs asset tokens (Nick Tomaino for 1Confirmation)
This notion tries to distinguish the types of tokens created by blockchain application, and broadly creates classes for what gives them value. Nick makes a deeper distinction but basically usage tokens are those which are used as part of the consumption of a decentralized service (that you buy into using the token), asset tokens instead are a store of value (much in the same way that Bitcoin could be seen as an alternative to gold). The former provides access to a clear service, the latter operates based on a common agreement that they hold value. Prefer the former, although clearly much of the value in the market has accrued to Bitcoin (and so much has been invested in hardware there that it seems unlikely to go away).
On-chain or off-chain
One should always ask “why should this be decentralized?” There is a more subtle question here as well regarding does this entire product need to be built in a decentralized fashion, or could it be made more efficient or simpler by building part of it off of the blockchain.
4. How this could be a valuable asset class
Such as it is, the market is extremely immature right now: we basically have a technology that we believe can possibly yield value on par with the Internet, but only limited infrastructure for it and tons of missing services (at the protocol level and the app level). But whilst a big portion of Internet investments only became possible after the market matured (and a search engine made it possible for someone to find your product for instance), the nature of the technology here allows us to invest in the protocols themselves…
As Chris Dixon (a16z) explains, it’s as though you had the opportunity to invest in http, ssh, tls and other Internet protocols back in the 70’s and 80’s. In this way, due to a contingent way of incentivizing miners, there is some sense in seeing this space as a new asset class of technology-enabling assets.
Because this technology is protocol-fat rather than app-fat, we get to invest in core infrastructure before we know exactly what it will be used for.
And so people are speculating on future usage. This is why some companies are raising at valuations of $B before launching a product. However, I think some of the investments into core technologies make sense here: whilst most everyone lost money in the dot-com bubble, interesting opportunities to invest in the Internet protocols themselves and reap the value of all future companies built on top of it exists. This must be the bull case for crypto long-term. Put another way, investing in the space as a whole may be like investing in the Apple App Store, but with Apple making far more than 30% on everything sold in-app.
This to me is the rationale for investing in this space. Assuming there is indeed clear value in these protocols, there is a fair chance that that future value built on usage rather than speculation will outstrip today’s pricing. There is a chance that the market will remain irrational long enough for the value to be created and catch up to the prices (for instance, too much money has been poured into Bitcoin mining hardware for Bitcoin to ever fully go away I believe), but ultimately, I expect there will be a pretty violent market correction in the coming years. And so, I believe one should invest on a 10-year time horizon, in the hopes that this turns out to be a real asset class.
3. Further Reading
There was an article on the cyclical movements in crypto pricing, turning assets from speculation-heavy pricing to utility-heavy pricing and back. I can’t seem to find it. Please let me know if you do.