Blockchain Series Part 2: Blockchains vs. Centralised Systems

Like any game, blockchain deals with rules applying to many participants.

How do blockchains differ from the centralised systems used by such services as Paypal and Airbnb? And why does it matter? Provenance blockchain engineer Thibaut Schaeffer explains.

In terms of predictability

  • Centralised systems
    In this system, the owner decides how the system can be used, what rules govern it and when these rules change. At any time, they can increase fees, ban users, and even cease to operate. This doesn’t necessarily always happen, but it makes the future quite unpredictable.
  • Blockchains
    You can think of using a blockchain app like playing a game. The game has rules and anyone can see which moves other players have made in the past, or whether they conformed to the rules. From financial transactions to copyright, to the certification of diplomas, many applications treat events written to the Bitcoin blockchain as being as close as possible to being “written in stone”. But blockchains can also make statements about the future, although it can only guarantee the registration of actions that conform to the rules. This makes the future more predictable.
  • So what?
    Unpredictability is risk. Do you like risk?

In the aspects of latency and finality

  • Centralised systems
    Centralised systems are often optimised to write information quickly. This means that the time between the moment you press the “send” button on your app, and the moment that sent data is available in the system, is very short. This is possible because the data only needs to be saved on a few computers controlled by the organisation, but this also means that you have to trust the organisation to guarantee that the data will still be there in the future.
  • Blockchains
    With public blockchain systems, data needs time to reach, and be saved on, a lot of computers all over the world. These computers then need to agree that the new data is correct, and whether it should be recorded in the chain. Once they have agreed, it becomes increasingly expensive to remove or modify that piece of data over time.
  • So what?
    Public blockchains comes with a new way of thinking about finality. For situations where an instant confirmation is not necessary, the data can be considered permanent after just a few minutes.

Confidence in data submitted to a centralised system is immediate but static, whereas confidence in data submitted to a blockchain is delayed but increases over time.

In terms of throughput

  • Centralised systems
    Centralised systems are optimised for throughput, meaning many pieces of data can be processed in a small amount of time. As a consequence, the database used to save the data can grow very fast and require specialised hardware. As an example, VISA can process 10000 transactions per second.
  • Blockchains
    Such volumes are not possible with blockchains. Validators need to be able to store the entire database, which forces the amount of data written to the ledger to be as small as possible. And since all that data needs to be replicated on all nodes of the network, it needs to flow through the internet fast enough to reach them. Blockchains thus typically have low throughput, with the Bitcoin blockchain able to process… 7 transactions per second.
  • So what?
    While a lot of work is being put into increasing the throughput of public blockchains, this constraint forces developers to focus on what blockchain does best: enforcing rules. And this does not necessarily require a lot of data.

With regards to integrity

  • Centralised systems
    Centralised systems win hands-down on data quantity. But let’s talk about data quality. Think about how many times you’ve entered your phone number into a form: that’s at least how many different records of your phone number there are out there. If you get a new phone number today, all those systems will have no way of knowing you did, so you would have to update them one by one. This is the same for bank settlements. When transferring assets from one party to another, updating records in all systems can take multiple days.
  • Blockchains
    With blockchains, it’s simpler: there is only one database. Update something and if it follows the rules, it’s updated everywhere within a few minutes. At least in most cases. This simplicity is due to a consensus method that enables all nodes of the network to agree on the “right” version of history. In the case of public blockchains, a validator profits if they play by the rules. There are some caveats, but in practice, this guarantees integrity.
  • So what?
    Centralised systems or “silos” bring inconsistencies that are expensive and time-consuming to resolve. Thanks to well-designed consensus algorithms, organisations that use blockchains can potentially cut costs by relying on a single source of truth instead of multiple inconsistent databases.

Remember this? Availability is an issue with centralised systems.

In terms of availability

  • Centralised systems
    So you trust a centralised system with some important data. What if the datastore catches fire? Is the data backed up somewhere safe? Maybe, maybe not. Your only choice is to trust the system operator.
  • Blockchains
    With the blockchain, data is replicated by design, so there is no way, voluntarily or not, to stop the system or wipe out the data. To do so would require going to every single node in the network and shutting them down.
  • So what?
    Remember that time Twitter was down? The blockchain is never down.

In the case of fraud

  • Centralised systems
    This might sound paranoiac but in centralised systems, anyone participating is implicitly making the assumption that the central operator will not cheat. If Facebook wants to delete your account, it is entirely up to them. Your bank could refuse to let you withdraw your money.
  • Blockchains
    Blockchains don’t necessarily prevent fraud, as anything written to the ledger is entirely up to the validators – who are exempted from the rules. But if the rules are defined and shared in advance, anyone could “audit” the system and check that the actions in the ledger follow the rules. In this case, fraud is said to be self-evident. In the specific case of public blockchains, users can prevent fraud by becoming validators themselves and opt in to the “right” version of history; the one that respects previously-defined rules. Most importantly, validators make money if they submit actions that respect the rules. This is not the case with consensus blockchains where it’s easier to cheat.
  • So what?
    In some cases, cheating is a good thing. If the rules of the game are not well designed, being able to go and change them for the better can come in handy. This makes centralised systems reactive and evolutive. The problem is that there is no way to prevent malicious cheating. On the blockchain, the approach is to define all rules right at the beginning, to reduce the chances of fraud.

Going further: Cooperation

A highly predictable environment with self-enforcing public rules and open accessibility produces a unique environment for cooperation. Have your ever heard of mechanism design? It’s the “field in economics and game theory that takes an engineering approach to designing economic mechanisms or incentives, toward desired objectives, in strategic settings, where players act rationally.”

This piece might liken building blockchain applications to designing the rules of a game, but key to Wikipedia’s definition is “toward desired objectives”. On top of giving players an incentive to play the game, well-designed rules can enable everyone to cooperate towards a common purpose. Cooperation is much more powerful than collaboration because it does not require the players to “help” each other. They can simply play towards their own interests and still pursue a shared goal. At Provenance, coordination means that consumers, certifiers, manufacturers and producers can all come together, in spite of conflicting interests, and still act to, say, have more ethical supply chains.

“Incentive schemes, whereby people who have done the most good for humanity are rewarded 20 years into the future would create the expectation that doing long-term good is valuable.”
Jaan Tallinn, founder of Skype

Centralized systems can also be considered games with rules, except they always have an incredibly powerful player at the center that makes it impossible for other players to cooperate. Blockchains accept tradeoffs in speed and flexibility. In exchange, they enable anyone to invent schemes for letting different users cooperate without the need for a central party.

Of course, such mechanisms are hard to get right. But this is the first time we have such a playground for experimenting with cooperation at a global scale.