There is a set of fundamental trade-offs at the heart of decentralized blockchains. There is this potential for disruption, for the development of trustless and censorship-resistant decentralized applications et al, but that remains a pipe dream for the foreseeable future as these blockchains are limited in their ability to scale up. We are pretty much in the stagnation zone where the three core properties of a decentralized blockchains - scalability, decentralization and consensus - are in a constant state of battle with each other.
As public blockchains require every node to record a transaction to achieve consensus, such a blockchain is only as fast as the slowest node. Centralized databases address this situation by ensuring that all nodes have sufficient computing power to process a transaction quickly because they exert full control over all the nodes. Given the decentralized nature of public blockchains, each node is run independently and the onus is upon the node operator to upgrade the nodes regularly. As the blockchain becomes bigger and bigger, the cost of running a node increases, leading to a centralization of nodes. Bitcoin and Ethereum hit their upper limits on transaction throughput in 2017 and this resulted in higher transaction costs, which led to concerns about their ability to achieve mainstream adoption.
While we firmly believe that scalability will be a problem that will eventually be solved, the tradeoff between scalability, decentralization and consensus varies from blockchain to blockchain depending on their primary use case. For example, micropayment focused blockchains such as Steem that have to process thousands of transactions per day achieve faster consensus by using DPoS consensus mechanism that is more centralized. On the other hand, Bitcoin, which requires a sovereign-grade protection for it to be resistant to a large-state attack compromises on scalability to achieve true decentralization. This explains the Bitcoin Core’s recalcitrance when faced with the prospect/opportunity of increasing the block size in 2017; Such an increase would leave the network more centralized due to the significant increase in the cost of operating a node.
Second layer and off-chain solutions such as Lightning Network achieve scalability by compromising on consensus, without affecting the security of the base layer. Ethereum’s proposed scaling solutions - PoS, Sharding, Raiden Network - are all focused on improving the scalability of the blockchain at the expense of decentralization or consensus. Transition to PoS results in greater scalability but results in a more centralized network of nodes as the minimum deposit to operate a node is 1000 ETH. Sharding involves breaking down the overall state of the blockchain into individual shards where only nodes in a particular shard need to validate a transaction and multiple transactions can be verified in parallel as the load on the nodes are reduced. However, achieving consensus through Sharding is a challenging task as it requires trust between the nodes. In addition, most emerging ‘distributed app’ blockchains that purportedly are building a 'better Ethereum’ in terms of scalability are more centralized than Ethereum, whether it be NEO or Cardano or something else (CHECK THIS). The future of these blockchains is very much dependent on their ability to reach the 'Goldilocks zone’ between scalability, decentralization and consensus.