February 17, 2018
6 min. read
Lets first take a step back, discuss the concept of metaphors itself and why they are important.
The human brain is composed of a large neural network. Neural networks are basically association machines. They constantly match new input against already known patterns, look for similarities and update/refine existing patterns. This happens with a bias to stick as closely as possible to existing patterns, because building them is hard work and evolution has optimized for energy efficiency — with the side effects of making us susceptible to all kinds of cognitive biases and of making us suffer of cognitive dissonance when the mismatch grows too large.
Whenever new input significantly updates/alters our stored patterns, we refer to this as learning. In order to dive deeper into that topic, I recommend a look at the works of Jean Piaget.
Metaphors are essential for humans to bridge knowledge gaps. In the times of Galileo Galilei it may have been possible for individual brains to accumulate most of the existing knowledge. That times have long passed, making metaphors essential for navigating the world.
The rapid advancement of Information Technology and the inability of most education systems to keep up have created a serious gap of common understanding even on the level of metaphors. As a result, modern technology is often not only perceived as, but also marketed as magic, ina way regressing us back into a pre-enlightenment state where understanding is replaced by myths and believes.
Historically, coins have played a central role (although by far not being the only form of money before paper — see The Origins of Money by Nick Szabo) in forming our understanding of the concept of money.
For a long time, the value of a coin corresponded to the value of the material it was composed of — usually gold or silver. It’s worth noting that techniques to inflate the monetary base — leading to the Cantillon Effect — already existed and were frequently employed (e.g. by applying excessive Seignorage). Fiat money isn’t a prerequisite for that, it just makes it easier to obscure the inflation.
Over time, coins tended to become more standardized, making the handling of them more convenient (no need to weight individual coins). However the fixed denominations — usually multiples of 1, 2 and 5 — as we’re used to today — are a pretty recent development, closely related to the switching to Fiat money.
As the name suggests, the account model has a notion of accounts which have balances. Despite of that, it has become common to refer to their units of account (e.g. Ether) as coins (or altcoins) anyway, adding to the confusion.
Originally, the coin metaphor is closely related to the UTXO model, thus the following observations will focus on how the metaphor fits for that.
Crypto coins usually don’t come in fixed denominations as we’re used to. In the UTXO model it is possible to design crypto coins with similar fixed denominations (see e.g. Chapter 26 of the Byteball Whitepaper), but in general it’s not an efficient way to implement crypto-currency this way and thus rarely done.
A coin in Bitcoin could have any denomination between 0.00000001 and 21000000 (at least in theory).
When transacting a crypto coin, the receiver is defined by an address.
The concept of address suggests that we’re talking about some kind of identifier which is associated to a human or a human’s persona/role — like an Email address.
In the context of a payment system, we’re tempted to relate such an address to the concept of a bank account (which works pretty well in the account model).
A Bitcoin address is a strange animal. It expects to be used only once (which isn’t enforced) and is related to the coin metaphor in a complicated way:
Whenever the value of one or more coins (inputs) is transacted to an address, a new coin (more technically, an unspent transaction output aka UTXO) is created.
A Bitcoin address encodes a set of rules which govern how and under what circumstances a coin can be spent. While this is typically just a check of private key ownership (see P2PKH address), it could also be something much more complicated (e.g. temporal restrictions — whatever one can build with Bitcoin script, typically implemented with P2SH addresses).
The point is: the semantics of an address in Bitcoin and other UTXO systems goes way beyond what the address metaphor can communicate.
Coins can be spent.
In the analog world, that happens by taking specific coins out of the pocket and handing them over to somebody (or to a machine).
Ownership change here is reflected by changing the physical location.
Crypto coins have no physical location. Spending a coin (an UTXO/unspent transaction output) happens by creating a transaction message containing a cryptographic signature which proves that the sender knows the secret required to spend that specific coin. This transaction message is then broadcast to a P2P network and hopefully included in a block by a miner.
Once that transaction was mined, the sent coin ceases to exist. Its value was moved to one or more new coins created by that transaction.
There is however a second way to transfer ownership of crypto coins which more closely resembles transacting with analog coins:
Instead of transacting on-chain, the secret (private key) needed to spend a coin can be transferred. This requires that this secret was not revealed to the prior owner (or anybody else).
This is typically implemented by some kind of physical object — often resembling traditional physical coins (examples) — containing the secret such that revealing it invalidates the object in a visually obvious way (basically, protected by a physical seal).
Implemented such, crypto coins can indeed be transferred the “old way”, but such off-chain transactions inherit the limitations of non-crypto physical coins like not allowing partial spending and being constrained by the physical location. Also, such physical crypto coins require trust that whoever created them in the first place didn’t keep a copy of the secret.
Overall, this properties and the variations they allow make it difficult to understand the concept of ownership and how it can be transferred, where the digital asset is stored, how it is represented etc. That’s probably were the limitations of the coin metaphor are most obvious.
Lastly, a look at the property of scarcity — a prerequisite for being money.
The second most important metaphor introduced by Bitcoin was that of Gold for explaining its scarcity.
Gold derives its scarcity from a physical limitation. There’s only so much of it and getting hold of it is expensive (for now).
Lets assume the supply of Bitcoins were also physically limited. Even so, the Gold metaphor wouldn’t really hold, because the total all-time supply of Gold is unknown, because its supply curve doesn’t really resemble that of Bitcoin & Co and because the production (that is, inflation) of Gold is influenced by its market value (Bitcoin sticks to a pre-determined inflation schedule by continually adapting the difficulty to the network hashrate).
I would instead claim that Bitcoin & Co derive their scarcity from a social contract, implemented as an Internet protocol.
Unlike with Fiat money, fidelity to this contract does not rely on a central governing institution, but on the network itself (the exact meaning of which is still disputed). Furthermore, crytoeconomic systems usually (there are exceptions) allow easy and exact verification of this contract at any time and allow networks to partition in case of disagreement. E.g. if half of the Bitcoin network (however you measure that) decided they wanted to increase money supply from 21 to 42 Million units (e.g. because 42 is the answer to everything), they could do so without forcing the other half to accept that — there may however be a dispute about which one is the real Bitcoin (as seen before).
Despite of this limitations, the early introduction and successive repetition of the Gold metaphor created a strong inter-subjective reality — reinforced by the metaphor of mining — around the notion that the scarcity property of crypto coins is somehow more similar to that of Gold than to that of Fiat money.
As long as this believe has broad enough consensus, this metaphor will keep working and protect crypto currencies against inflation.
It does however not protect them against a new kind of inflation, lets say meta-inflation: while the number of precious metals suited to become money is quite limited (see bimetallism), the same is not true for crypto-currencies. We’ve yet to see how this plays out in the long term.
Metaphors are important! They create a cognitive interface to new concepts.
Metaphors can do more harm than good when they create leaky abstractions, when users interacting with a system based on intuitions related to deployed metaphors are likely to make mistakes.
In order to minimize such mistakes, we can’t just blame users for not understanding technical details. User interfaces hiding details shouldn’t just be easy to use, they should also be safe to operate for users acting according to their understanding of employed metaphors.
If new concepts don’t fit potential metaphors well enough, it may be better to introduce new terminology, making the requirement of learning something new more explicit.