In this article, I have discussed a viewpoint: the key to decentralized social is to lower its threshold. Once a zero point is reached and the threshold for decentralized applications becomes as low as that of centralized applications, it will grow rapidly - because no one dislikes being able to control themselves, whether it is the wealth, the data, or even the destiny.
However, how can we lower the barriers? There are many aspects that need to be worked on. Improving user experience is an important one, and this is something that ArcBlock platform has been continuously promoting since 2017. Another aspect is the cost of usage. When there is no longer a centralized service provider to foot the bill for users (actually using their data for profit), users need to pay for their own services. Just like in blockchain, where every transaction requires a gas fee, the cost of this service must be low enough, which is also part of the barriers.
Four key functions that are not easy to implement in decentralized social networks
The above-mentioned issues have gradually become a reality, but there is still some crucial features - how to solve the several essential problems that social networks need after achieving service decentralization.
- Spreading - The basic purpose of social networks is to ensure that the sound one produces can be effectively transmitted to the appropriate audience.
- Aggregation - Social networks need 'town squares' and ' courtyards,' as well as 'groups' and 'tribes.'
- Discovery - Social network is about engineering serendipity, it needs can "unexpectedly" discover people and things that interest themselves, and cannot be trapped in an information cocoon.
- Search - Social needs to be able to find the people and things one needs.
The more centralized the four important functions mentioned above are, the easier they can be realized and the better user experience can be achieved. These reasons are likely to be the key reasons why Blog and RSS declined and were defeated by completely centralized platforms like Facebook and Twitter. In decentralized social networks such as Mastodon and Nostr, the lack of these four functions is an important reason why they cannot become popular.
AI Agent serves as an automated coordinator.
Based on LLM, AI could soon become a basic service available to everyone. I believe LLM AI service will become a fundamental service like cloud computing and storage services, with many options and significantly lower costs. And LLM-based personal agents will become assistants for each of us in the always-on network.
These Agents are not our 'digital avatars' or our masks, they are our 'assistants'. Just like today's political elites and big entrepreneurs have their own assistants or personal secretaries, AI empowers each of us and enables everyone to have their own intelligent personal assistant.
With these forever online, tireless and understanding agents, the four key functions that are inconvenient to achieve decentralization can be easily solved: AI agents can automatically help us with dissemination, aggregation, discovery, and search, and can effectively filter out content that we may not like.
Although the LLM behind it may still be some public services, the service logic of Agent and the service object are completely user-oriented. That is, these Agent serve their owners completely loyal, not some operational service providers!
In the past centralized social networks, operators often used recommendation systems to recommend advertisements that users do not like or even mislead them; in the past history, whether it is Facebook or Twitter, they have done unsavory banning of political views, censorship, or inducing users.
However, AI Agent belongs to everyone. Each person's Agent is only responsible for themselves and cannot be manipulated by anyone else. Therefore, decentralized social networks based on these Agents have incomparable advantages compared to traditional social networks.
In the future, I will continue to discuss why blockchain, decentralized identity (DID/VC), and cryptocurrency will have important impacts on decentralized social networks after AI agents make decentralized socialization thoroughly feasible. The significance of this importance may be equally important as the meaning brought by AI agents.