Everything in the universe, from galaxies and solar systems to rain-forests and colonies of simple-cell organisms, behaves as a peer to peer network. Cultures, societies, and families form, survive and endure on the peer to peer framework. Not uncoincidentally, the human brain operates on the peer to peer model. A neural network model to be precise - neurons, the basic unit of the brain, connecting to form a vast mesh network. The brain is peer to peer at its finest. Nature does not operate on a centralized or hierarchical system of organization, data collection, dissemination, and communication.
Contrast this with today's most prevalent computer architectures that are still based on the constraining centralized client-server frameworks. Dumb clients rely on this centralized architecture and its binary communication construct. The ramifications are at once obvious – sharing is slow, learning is slow, and idea proliferation is slow.
When we think about cognitive artificial intelligence, we expect it has the ability to think and learn on its own. Thinking and learning require a neural model - a peer to peer configuration allowing for the unlimited potential to form infinite sequences of learning and execution models. Neural models mimic nature's design of peer-to-peer networks that habitually co-exist in endless forms.
Regression to the Nature Mean
The current explosion of mega client-server environments such as Amazon Web Service, Microsoft Azure and Google Cloud Services, is not due to their superior design technology. It offers what businesses believe they require based on the current model of architectural design. However, in no small part the growth of AWS/ Azure / GCS reflects the realities faced by organizations around capital costs and the need for control.
In contrast, a good example of peer-to-peer model is the Napster story. Napster, one of the most legendary peer-to-peer networks failed, due not to its network architecture, but to copyright and intellectual property infringement (capital, control). Architecturally, Napster was brilliant and highly effective in its design.
We believe crypto-token (the latest version of peer to peer network’s - distributed blockchain networks) - the new and improved Napster model - is the future of computing and will influence how we architect solutions, including how much and where we store data. We view today's data ‘sources' as a node, rather than a peer, in an ever-expanding distributed computer network model. Blockchain, hyper ledger, and other derivatives of peer to peer are the technologies of the near future. They probably will be called something else in 10 years, but you’ll recognize them in the core framework.
With the growth in "crypto-coin/token/currency/contract/apps" technologies, data lakes and data marts become mere peers on the network. As more IOT comes online and the endpoints get smarter and smarter, the need for IOT to talk to a central server ceases to exist. The IOT will communicate directly with whatever device they need. Witness the evolution of the phone system when all traffic traveled through an operator at a switchboard. Today cellphone calls are peer to peer.
We see Application Programming Interfaces (API's) as a primary modality for peer-to-peer connections. The API itself becomes a peer. However, APIs are not as abundant and pervasive yet and when available, they are used as connectors and not peer systems. This perpetuates the need for collecting and storing copies of data. In the future, we don't see data copying as a sustainable or intelligent model of operation.
In peer-to-peer architectures physical constraints fade and give way to limitless cognitive possibilities that mimic the complexity and beauty of nature and our universe.