Centralized vs. Decentralized

Question 1

What are centralized systems?

Answer 1

Examples
- USA Government
- Coinbase
- Amazon
- Chase Bank
- iMessage, WhatsApp, Messenger, Instagram

Key points:
- most systems we interact with today are centralized
- one single point, one main point, whoever controls that node (point), has control over entire system
- everything in system goes through one main node
- systems controlled by single entity (company, government, individual)
- in a system controlled by a single entity
- can be changed, modified
- people can have certain privledges other people dont
- access to be revoked
- not a notion of fairness
- need to trust the single entity or organization to use it
- people are putting a lot of trust into that system, can only operate because of this trust
- everyday we are using products/services and putting a lot of trust into that, if trust is broken won’t want to use it anymore

Question 2

What is decentralization?

Answer 2

• one person (node) doesn’t have control over entire system
• distributed across multiple nodes
• planning, decision making are delegated away from a central, authoritive location
• need to trust the system, trust math/theory behind network is correct
• don’t need to trust a single entity or individual, allows for trust less interactions
• don’t need to worry about central body controlling system

Question 3

How is data distribution (flow) different in centralized vs. decentralized systems?

Answer 3

Centralized
- fast flow of data
- lower network diameter
- simple to implement, highly performant
- data conflicts are easy (one main place with all the data)
- synchronization is easy and clear
- single point of failure
- central authority can censor/change or delete data
- participation regulated by gate keeper (centralized theory)

Decentralized
- slower flow of data DUE to large network diameter (many edges of participants)
- difficult to implement, lower throughput, lower performance
- complex protocol needed to dispute data conflicts and find resolutions (many different nodes, which one to trust?)
- data synchronization is difficult
- no single point of failure, network can function if large portion are taken down or offline
- censorship harder, almost impossible to change or delete existing data
- anyone can participate in network