Platform-Risks: why do we need Decentralization?

Hagen Hübel


Most people think of CryptoCurrencies when they hear decentralization being talked about. But this topic is much bigger and more comprehensive than a few fancy financial tricks that produce millionaires and total losses, depending on certain factors.

I want to talk about problems that didn’t exist two decades ago and how we can solve them through decentralization.

Let’s start with a sharp claim:

Platform-Risks are the biggest threat to society and can only be avoided with decentralization.

What is meant by a Platform-Risk?

Platform risk refers to the potential for data loss, financial loss or operational disruption arising from a technological platform's failure or malfunction. This can include things like getting excluded, deactivating a user's account, software bugs, data breaches, or other technical issues that can impact the availability or functionality of a platform.

Platform risk can also refer to the risk of a platform becoming acquired by another company and suddenly taken down, becoming obsolete, or losing market share to competitors. In a financial context, platform risk is often used to refer to the risk of an electronic trading platform or other financial technology infrastructure.

In a social media context, platform risk is often used to refer to the risk of people losing access to the platform with which they often spend a particular part of their lifetime.

Data breaches

Social media platforms collect and store large amounts of personal data, which can be a valuable target for hackers. A data breach could result in the loss of sensitive information, as well as harm to the platform’s reputation.

Privacy concerns

Social media platforms have been criticized for their handling of user data and privacy. Privacy concerns can lead to decreased user engagement, loss of users and revenue, and increased regulatory scrutiny.

Disinformation and misinformation

Social media platforms have been used to spread disinformation and misinformation, which can have serious real-world consequences. Platforms may be held liable for the spread of false information and may face regulatory action as a result.

One of the consequences of all mentioned issues is just to lose data, content, access to other people and all the contributions and effort one has spent on this platform. But there is another one, a bigger challenge we have to solve somehow:

Risk of Algorithms

Social Media platforms rely heavily on algorithms to curate content and personalize user experiences. These algorithms can be opaque and difficult to understand, perpetuating biases and harmful content. This leads us to the term of getting “de-platformed”.

Getting de-platformed:

A Platform Risk also means that a platform, from its very first day on, tends to gain too much power, which allows the platform to de-platform you.

What is meant by getting “de-platformed”? Let’s use a very basic example often happening on Social Media platforms like Twitter or Facebook that everybody has already faced, either themselves or with a friends profile: you write a sarcastic or critical comment to a post or a tweet, and some algorithm “thinks” that this comment is violenting platform rules.

Example 1: Criticism can lead to get locked out of a platform

Once upon a time, a user named John had been using Twitter for years. He was very active in the community, but had recently become more outspoken in his views.

One day, John posted a tweet that contained criticism of a specific company that he felt was unjustly being supported by Twitter. The comment, although not offensive, violated Twitter’s platform rules.

John soon received an email from Twitter stating that his account had been locked out due to the violation. The email also reminded him of their guidelines.

Example 2: Algorithms doesn’t understand Sarcasm

I have a friend who enjoyed posting amusing quips and jokes on his Twitter account. One day he made a comment that was full of sarcasm, but unfortunately, the algorithm used by the social media platform didn’t understand it and misread the man’s humor. The algorithm saw the tweet as offensive and it automatically locked the man out of his Twitter account.

The man was distraught as he was an avid platform user and had many followers. But he was determined to get his account back. He turned to Twitter, but no one answered him. He turned to his community and with the power of his 136,000 followers he was able to get his account reactivated.

Example 3: The power of a (too) big Marketplace

In 2014, Amazon, the world’s largest online retailer, began a dispute with publisher Hachette over e-book pricing. The disagreement between the two companies centered around the prices that Amazon charged for Hachette’s e-books. Hachette wanted to maintain control over the prices of its e-books, while Amazon wanted to set prices as low as possible to drive sales and market share.

In response to the disagreement, Amazon took steps to limit the sale of Hachette’s books. Customers searching for books published by Hachette were suddenly redirected to books published by other publishers. As a result, Hachette lost up to 90% of its current revenues, affecting many young first-time authors.

Example 4: Acquiring a startup and shutting it down

We have seen many acquisitions of startups that were already on their way to becoming a platform where users spent their time, but the companies were closed and taken off the market immediately after the transaction. But why?

Startups may get acquired and then shut down for a variety of reasons. The common ones are due to the buyer not having a strategic plan for the acquired company, or because the target startup was not producing the expected return on investment. The buyer may also not have the resources or expertise necessary to integrate the acquisition into its existing business operations successfully. In some cases, the acquisition may be a pre-emptive move made by the buyer to prevent the target startup’s technology or services from becoming a competitor. Sometimes it happens only to prevent a competitor is buying this company, like ThinkProperty-dot-my which was acquired by iProperty-dot-com for exactly this reason. And sometimes the acquisition happens to “buy” the team or part of its technology and knowledge.

For instance, In 2014, Twitter acquired the mobile video-sharing app Vine for $30 million. However, in 2016, Twitter announced that it would be shutting down the Vine app, citing a lack of success in generating revenue.

More examples like this:

  1. Beamery — Acquired by Oracle in 2019 and shut down in 2021
  2. Juicero — Acquired by Alphabet in 2017 and shut down in 2018
  3. Yik Yak — Acquired by Square in 2018 and shut down in 2019
  4. Nest acquired by Google but shutdown in May 2019
  5. Wunderlist acquired by Microsoft but shutdown in May 2020
  6. HopStop acquired by Apple but shutdown in October 2015

These are just a few examples, and it’s worth noting that startups may also be acquired and integrated into the acquiring company’s existing products or services, rather than shut down completely:

  • In 2012, Facebook acquired the mobile photo-sharing app Instagram for $1 billion. However, in 2018, Facebook announced that it would shut down the standalone app Direct, a feature within Instagram.
  • In 2015, Google acquired the smart home company Nest for $3.2 billion. However, in 2018, Google announced that it would merge the Nest brand with its hardware division and that some Nest products would be discontinued.
  • In 2016, Microsoft acquired the professional social network LinkedIn for $26.2 billion. In 2019, Microsoft announced that it would be shutting down LinkedIn’s “Learning” feature, which allowed users to take online courses and earn certificates.

Nevertheless, the founders and investors were richly compensated for this acquisition in all cases. But what happened to the users? They sometimes lose all the data that they have put onto the platform. They lost their “friends” with whom they probably had no personal contact in real life but communicated regularly on the platform.

Eventually, this is not just about users losing their own data and friends, but the entire community losing access to valuable content and relationships.

Ultimately, the injured parties are the users and their respective communities!

Decentralization: A Potential Solution

So, how can we mitigate these platform risks? The answer lies in decentralization.

Decentralization is the process of distributing and dispersing power away from a central authority. In the context of technology and digital platforms, this means moving away from a single, central entity controlling all data and decision-making power, towards a system where control and decision-making are spread across multiple nodes or entities.

Blockchain technology is one of the most promising tools for achieving decentralization. It is a type of distributed ledger technology that allows data to be stored across a network of computers, rather than in a single centralized database. This makes it much harder for any single entity to control or manipulate the data.

How will decentralization solve this problem?

Decentralization can prevent platform risk by distributing control, data, access rights and decision-making power across multiple entities rather than having a single point of failure. This can make it more difficult for technical issues or malicious actors to disrupt the system, as no central authority or single point of control can be targeted. Additionally, decentralization can help to reduce the risk of data breaches or other security issues, as sensitive information is not stored in a central location that hackers can easily target.

Some examples of how decentralization can help to prevent platform risk include:

  • Blockchain technology, which is based on decentralized networks, can help prevent platform risk by creating tamper-proof digital ledgers protected by cryptography. This can make it much more difficult for hackers to steal or manipulate data.
  • Blockchain technology is fully decentralized, meaning that it is not controlled by any single entity. This makes it more resistant to censorship and hacking attempts.
  • As a very important fact, it must be said that Blockchain Technology is not limited in its most common use case to provide CryptoCurrencies and financial operations. We will go into details later.
  • Decentralized autonomous organizations (DAOs) are digital organizations controlled by a network of users rather than a central authority. This can help to prevent platform risk by distributing decision-making power across multiple users, rather than relying on a single person or group to make decisions.
  • P2P networks, which allow users to connect directly with each other rather than relying on a central server, can help to prevent platform risk by making it more difficult for a single point of failure to disrupt the system.
  • Decentralized Data Storage can solve the risk of storing data on a single entity and instead provide a distributed approach where files will be split into many parts and distributed (think of: “spread”) across the planet on hundreds and thousands of independent nodes.

It’s important to note that decentralization is not a panacea, and it doesn’t necessarily mean that a platform is immune to risk. It just distributes the risk among the decentralized entities making it less likely for a complete failure of the platform.

Advantages of Decentralization

Decentralization offers several advantages, including:

  1. Resilience: Decentralized systems are less vulnerable to single points of failure, making them more resilient to disruptions.
  2. Censorship resistance: Decentralized systems are not controlled by any single entity, making it more difficult for them to be censored or shut down.
  3. Transparency: Decentralized systems often rely on transparent, open-source technology, making it easy for anyone to view and verify their workings their code quality and investigate for corruption in code.
  4. Better security: Decentralized systems can be more secure because they do not have a single point of control that can be attacked.
  5. Higher efficiency: Decentralized systems can be more efficient because they do not rely on intermediaries, which can save time and money.
  6. Inclusion: Decentralized systems can provide improved access to goods, services, and information for people who might not have had it before.
  7. Greater participation: Decentralized systems can enable greater participation in decision-making, which can lead to more democratic and inclusive outcomes.

What is a Consensus-algorithm?

A consensus is a general agreement or a settlement of a dispute among a group of individuals or parties. In the context of distributed systems, a consensus refers to the process of reaching agreement among network participants on the state of the distributed data. In the context of blockchain technology, a consensus refers to the process of reaching agreement among network participants on the state of the distributed ledger.

This is typically achieved through the use of consensus algorithms, which are designed to ensure that all participants have a consistent view of the data and that new data will be recorded in a manner that preserves the integrity of the system. Examples of consensus algorithms include RAFT, Proof of Work (PoW), Proof of Stake (PoS), Proof of Authority (PoA), Proof of History (PoH) and Delegated Proof of Stake (DPoS).

In technology, a consensus is not only required for blockchain. Instead, many other distributed software systems that has nothing to do with Blockchain or “Crypto” already rely on Consensus to achieve a consensus between all nodes that are involved in the system.

High Availability requires that data can tolerate nodes going offline without interrupting service to applications that rely on this data.

The reason is simple: whenever data will be stored on multiple nodes, a single failure on a specific node could corrupt the whole state and thus a mechanism is required that identifies corrupted data or state and that fixes that.

Popular systems using a Consensus are

  • Kubernetes (etcd via RAFT)
  • ElasticSearch
  • Cassandra database
  • CoackroachDB

How is Consensus achieved in databases? Let’s briefly dive into the consensus part of ElasticSearch, a popular search engine and database for storing documents. If this is too technical, please feel free to skip to the next section.

In Elasticsearch, the consensus is achieved through the use of the “shard replication” feature. This feature allows multiple copies of a shard (a portion of the Elasticsearch index) to be maintained across multiple nodes in the cluster. When a new document is indexed or an existing document is updated, the change is propagated to all replicas of the affected shard.

When a replica receives an update, it first checks that it is still in sync with the primary shard (the shard that received the update first). If it is not in sync, it will first synchronize its state with the primary shard before applying the update. Once the update is applied, the replica sends an acknowledgement back to the primary shard.

The primary shard only considers the update as committed once a configurable number of replicas have acknowledged the update. This number is known as the “replication factor” and is set by the user. This replication factor ensures that there is a quorum of nodes that have acknowledged the update, thus achieving consensus.

This feature of Elasticsearch provides fault tolerance and high availability, as it allows for the cluster to continue functioning even if a node goes down, and it also allows for automatic failover in case a primary shard becomes unavailable.

Let’s dive in into Byzantine Fault Tolerance (BFT)

Byzantine fault tolerance (BFT) is a concept in distributed systems that refers to the ability of a system to function correctly even in the presence of malicious or faulty components. In particular, it is concerned with the ability of a system to tolerate “Byzantine failures,” which are failures that can take any form, including arbitrary message values, delays, reordering, or duplicates.

A distributed system is considered Byzantine fault-tolerant if it can continue to function correctly even when some of its components behave arbitrarily. In practice, this means that the system must be able to tolerate failures, such as nodes that may be compromised or that may be sending conflicting or inconsistent information.

One of the main ways to achieve Byzantine fault tolerance is through the use of consensus algorithms such as Paxos, Raft or Practical Byzantine Fault Tolerance (PBFT). These algorithms are designed to ensure that all participants have a consistent view of the system state and that new transactions are recorded in a manner that preserves the integrity of the system, even in the presence of faulty nodes.

Byzantine fault tolerance is particularly important in distributed systems such as blockchain networks and distributed databases, where the integrity of the system must be maintained even in the presence of malicious actors.

Let’s explain the RAFT consensus algorithm

The Raft consensus algorithm is a distributed consensus algorithm that is designed to be more understandable and easier to implement than others like Paxos. It provides a way for a cluster of nodes to reach agreement on the current state of a replicated log.

The Raft algorithm operates on a “leader-follower” model, where one node in the cluster is designated as the leader and the rest are followers. The leader is responsible for receiving and processing client requests, and for coordinating the replication of the log to the followers. The followers are responsible for responding to the leader’s requests and for maintaining a copy of the log.

The Raft algorithm has three main components:

  1. Leader Election: The Raft algorithm uses a leader election process to choose a leader node. This process ensures that there is always a single leader node in the cluster.
  2. Log Replication: The leader node receives client requests and appends them to its own log. It then sends the new entries to the followers, which append them to their own logs. Once a majority of the followers have acknowledged the entry, the leader considers the entry to be committed.
  3. Safety: The Raft algorithm includes several safety mechanisms that ensure that the leader and the followers have a consistent view of the log and that the log is not modified in an inconsistent way.

The Raft algorithm is designed to be robust in the presence of failures and network partitions. It can handle scenarios where nodes in the cluster are unreachable, or where a leader node fails. When a leader node fails, a new leader is elected and the log replication process continues, thus ensuring that the cluster maintains a consistent state.

Raft is widely used in distributed systems and is the foundation of many projects such as etcd, consul, and CockroachDB.

Decentralization: what does it eventually look like?

Most people think of cryptocurrencies when they hear decentralization being talked about. But this topic is much bigger than a few fancy financial tricks that produce millionaires and total losses (depending on certain factors).

Like a cryptocurrency on a censorship-resistant blockchain, that no government can take away from you, the same pattern can be applied to data. Imagine Wikipedia hosted on a decentralized, distributed database. No longer could any single entity delete or manipulate an article that contained some critical parts about a person of public interest.

But what if some malicious writer abuses his ability to write on this database and publishes wrong information? We’ll cover this issue later.

Think of Twitter using a decentralized and distributed Database: not a single entity could remove your tweets anymore or even remove your access to the network, whether for writing purposes or just for reading.

Again, we talk about misinformation and moderation or deletion of problematic tweets later.

Think of YouTube using a decentralized and distributed database: no single entity could ban you from publishing videos or even remove your old ones.

Think of Reviews in your favorite eCommerce store: not a single entity could achieve the deletion of your review of receiving a total broken product (assuming you were telling the truth).

Think of publishing your Website on decentralized web storage so that your web provider can’t remove your website anymore because he was acquired by someone else and shall shut down or because administrators made a mistake, or just because your competitors are claiming something bad that is not true. Or just because of your political views are not welcomed anymore, and the Boss of your web provider belongs to the other political side.


In conclusion, decentralization offers a promising solution to the platform risks associated with centralized systems. By distributing power and control, we can create more secure, resilient, and trustworthy digital platforms.

But how to deal with problematic data?

The downside of decentralization

To be published soon, in the second part of this series.