31 March 2014

About trust

In my last post, I used the term "trusted ownership" to describe the "thing" managed by the Bitcoin protocol. Following this post, Oleg Andrev pointed out that trust is no more a required concept.

I guess that what Oleg means is that for the first time, a protocol allows to manage ownership without requiring trust in a human third party (like a bank...). The protocol replaces this need of trust in a human third party by a mathematical proof. I agree that it's a very important message. For many people, bitcoin is still just "weird money" over internet and not so different from a system like paypal. Of course it's wrong. The "beauty" of the protocol is really in its distributed aspect and in the fact that it disrupts an old model based on trust in a human third party.

I've been lucky to attend a presentation by Oleg a few weeks ago. He's undoubtely a very smart guy with a great knowledge of the Bitcoin protocol. On my side, I'm a noob in this domain. Thus, I try to always think twice about what the experts say. The thing is that I don't believe we can get rid of the concept of trust so easily. My take is that we trust in the proof of ownership managed by the protocol because we trust in the maths behind the protocol.

About the maths
I can hear the cries of mathematicians. Trust has nothing to do with mathematical proofs. A proof is demonstrated or it isn't. Trust has no role in this. Yes. Sure. You're right. The problem is that I don't believe in platonic idealism and the existence of a pure world of ideas.

Let me state it differently. You can do 2 things with crypto:
- you can use it to secure data, communications...
- you can crack it (some examples here, here, here)

Of course the latter is much more difficult and requires some serious resources (big brains, big computers) but it's not impossible. Don't get me wrong, I don't say that bitcoin crypto is not secure. I truely believe it is very secure and I hope it will be for a long time. But I don't believe in a "natural law" saying that it's secure till the end times because it's something of a different kind, coming from the non corruptible word of ideas described by Plato. Thus, it's because I trust in the security of the crypto behind the system that I trust in the proof of ownership provided by the system. Trust has not disappeared. It's trust in something different, may be more more secure and less corruptible than human nature, but it's still trust.

Why does it matter ?
Cryptocurrencies gain more and more visibility. I'm convinced it's a good thing and I hope more and more people start to use them. But let's be honest. How many people in the world have the mathematical background to check by themselves the security of the crypto behind the system, taking into account the last researchs in crypto ? I guess not a lot. All we can expect is that people (like me) will use cryptocurrencies because they trust professionals (mathematicians, cryptologists...) saying that the crypto behind the system is still secure. Of course, there's a huge improvement: trust relies now on a open system which can be checked by independent teams. But trust in human people is still there (just replace bankers by mathematicians and cryptologists).

The "marketing problem"
[ConservativeModeStarted] We live in a world in which everything must go faster and faster. If you want to promote something it's better that you come with the right message. Few people are ready to spend time to dive deep in the understanding of a protocol.[ConservativeModeCompleted]

The problem with short messages and analogies is that they have their limits and trust is always challenged when these limits are reached. This is why finding the right message is so difficult. In my previous post I wrote about an analogy with a system made of tubes and valves. To be honest, I think this analogy is too complicated to be really usefull. This is what I liked in the land analogy presented by Richard G. Brown. It's very concrete and simple to understand. My "concern" is that I fear people could infer some wrong properties of the bitcoin system from this analogy. 

For sure, finding the right message is a difficult task and I still don't know the best answer except that bitcoin is bitcoin and that if you really want to understand it, you must invest time (a very bad marketing message :)

For example, let's consider the concept of ownership. I totally agree that bitcoin is a protocol of ownership and that it's an important message explaining why bitcoin matters. The problem is that if you explain that to somebody without any background in the domain, her first understanding will be that the protocol associates ownership of some coins to a person. But it's wrong. There's nothing in Bitcoin stating "these coins are owned by Mr X". Bitcoin doesn't care about the identity of the owner. The protocol does not deal with the fact that private keys can be stolen and used by a thieft. It's out of its scope. This is why I really like the often repeated message: "If you're not the only one to control / own the private key, you don't control / own the coins". It's a very simple message but you can infer very important things about the ecosystem from this message (importance of secured and cold wallets, risks associated to exchanges, mixers, black markets...).

Just a few random thoughts... Comments and criticisms are welcome !

Note : Trust can also be found in many others places in Bitcoins if you consider the "larger" picture englobing the periphery which interfaces the system with our physical world. I won't discuss them in this post. May be in another one.

30 March 2014

About the nature of cryptocurrencies

A few days ago, Richard Gendal Brown published an interesting post describing Bitcoin as a land territory. There is some very good points in this article (importance of UTXO, ...) and it's great because like many things in Bitcoin, it invites you to think about the nature of money, cryptocurrencies... But I must confess that I feel a bit unconfortable with this metaphor, mainly because of its implications about the fungibility of bitcoins.

About fungibility
Fungibility is the property of a good or a commodity whose individual units are capable of mutual substitution. It's often said that cash is fungible, i.e. a 10$ bank note is interchangeable with another. Diamonds are not perfectly fungible because diamonds' varying cuts, colors, grades, and sizes make it difficult to find many diamonds with the same cut, color, grade, and size (more here).
One of Richard's axiom is that bitcoins are not perfectly fungible : they're more similar to diamonds (and others physical objects like lands) than to dollars. This is where I start to feel unconfortable with the metaphor of the land.

Bitcoin system is not about objects but about flows
Trying to figure what are the core concepts of Bitcoin, I noticed there's a common mistake made about the nature of bitcoins. We like to imagine bitcoins as objects (properties): it's in the name (bitCOINS), it's in the visual imaginary...

It's sometime the cause of "funny" situations like TSA looking for bitCOINS inside luggages. The problem is there's nothing in Bitcoin protocol like an object (property) which we could call a bitCOIN. Bitcoin protocol is all about flows.

Let's have a look at a bitcoin transaction. We have 3 inputs (i1=10btc, i2=5btc, i3=5btc) and 2 outputs (o1=10btc, o2=10btc). A bitcoin transaction acts like a multiplexer/demultiplexer plumbing fitting.

It's impossible to say if coins received by o1 come from i1 or i2 or i3. This information just does not exist in bitcoin protocol. The best we can say about this transaction is known as taint analysis and states that:
  • all outputs have received 50% of their flow (value) from i1, 25% from i2 and 25% from i3
  • all inputs have sent 50% of their flow to o1 and 50% to o2.
It's equivalent to consider bitcoins as a liquid flowing in tubes, not as objects. Larger tubes receive larger volumes of the flow.

Bitcoin as a system made of tubes and valves
I like to picture bitcoin as a complex system of tubes and valves. The liquid flowing through this system is what I call "trusted ownership".

Here's the full picture:
  • Coinbase transactions (bitcoins generated by miners) are the sources of the flows.
  • Transactions join tubes with different diameters (values of inputs/ouputs). Diameters of output tubes are defined by the person creating the transaction.
  • Outputs of transactions are valves. A valve can be opened if you know how to open it (you own the private key associated to an address,...).
  • The "thing" flowing in this system is not a kind of object (bitCOIN, satoshis,...) but is trusted ownership. Trusted ownership is not a object transmitted from one address to another. It's a continuous flow going from the sources to the latest UTXO.
  • If you "cut" the flow (basically what happens when a fork/reorg of the blockchain occurs with double spendings) the valves victim of double spending don't receive the flow anymore. The flow of trusted ownership going from the sources to these valves is broken and there's no more liquid reaching these valves : the coins have gone ! (TM MTGox)

Why the metaphor used is important ?
My take is that the metaphor used has a direct incidence about how we think about the cryptocurrencies systems and how they can be regulated (or not). IRS says bitcoin is property. IRS is not alone and it's often said that bitcoins are less fungible than cash. 

The problem is that technically speaking it's wrong. We consider that a bank note is interchangeable with another even if bank notes are physical objects with serial numbers. Thus, saying that bitcoins are less fungible than dollars is like saying that water is less liquid than pebbles. 

IRS can consider that bitcoins are properties but it's conceptually wrong and this mistake will be more and more problematic for them in the future. For now, transactions are most of the time issued by one person and that's why the metaphor of the land (or bitcoins considered as properties) seems ok. But nothing in Bitcoin protocol restricts the system to this scheme and mixers are a good example of a different scheme which breaks the metaphor of the property. 

According to the future decisions made by regulators Bitcoin could become a system of properties but it would be a distortion of the true nature of the protocol. As pointed out by Richard : "it is perhaps no surprise that the fungibility issue is so hot right now."

At last, a good news for Newsweeks
The metaphor of bitcoin as a system made of tubes and valves brings a goodie. It reveals the real identity of Satoshi Nakamoto. The mystery is over !