The human heart is not a mining machine, and socializing is not mining. If the wrong fuel is used, the higher you fly, the harder you fall.

Written by: Wealth

From July 15 to 16, Jesse Pollak posted a series of long articles on X, stating that he was "definitively wrong" in the realm of socializing and creator tokens — in Chinese, it means wrong with no room for excuses. He said the first quarter of 2026 would be a "punch in the face," and that the past few months have essentially been chewing on the bragging from his prior two years. Two days ago, Coinbase CEO Brian Armstrong also conceded, admitting that content coins "didn’t work," leaving behind the statement "We messed up, time to turn the page." Just a "turning the page" closed the coffin on Base's narrative of betting on Farcaster, Zora, and Frames mini-apps over the past two years.

The Base App was handed back to Coinbase’s parent company and taken over by Jordan Fish, known in the community as Cobie. Pollak himself stepped back to write code, focusing entirely on making Base a "blockchain for global finance." This shift wasn't unexpected; what is surprising is that before this turn, they spent two whole years pondering "socializing," and in the end, the result was something even they couldn't bear to see.

Why initially focus on socializing

First, let’s clarify why they focused on social as a business. There are two layers of logic here: one is the businessman’s calculations, and the other is the idealist’s obsession — fundamentally, they lead to the same destination.

The businessman’s calculation is expressed most bluntly by Musk. He spent $44 billion to buy Twitter, claiming he wanted to turn it into a "digital plaza," a place where everyone can speak freely; what he likely thinks is simpler — whoever controls the plaza controls the agenda-setting power, which is worth more than any advertising space. Socializing has never been a traffic business; it's a discourse power business, and everyone in Silicon Valley understands this.

The idealist’s obsession can be traced back more than a decade. In 2013, Snowden exposed the massive surveillance conducted by the U.S. government in collaboration with major tech companies, awakening the global internet to the fact that the email, search, and social networks they used daily were already someone else’s telescope. In 2014, Ethereum co-founder Gavin Wood wrote an article calling the next internet that needs to be rebuilt the "post-Snowden network." Vitalik Buterin has also repeatedly stated that what the crypto industry really needs to do is not to issue tokens, but to eradicate the structural flaws of Web 2, such as censorship, surveillance, and data monopolization, by addressing them at a fundamental technical level. This ideal boils down to several buzzwords that have been talked about extensively but still remain unachieved: censorship resistance, open source, privacy, security, and decentralization — no single company can unilaterally delete your posts, no backend can observe your private chats, and your fans and articles will follow you across platforms, no longer mere entries in a company database.

These two layers of logic combined explain why Base invested heavily in socializing. On one side is the ambition for genuine traffic, a billion-user engine, and on the other is the grand narrative of "this time, we will give the public plaza back to ordinary people." Grand narratives are the cheapest to talk about; the real expense comes when trying to implement them.

What resulted was an impure casino

Base should have had the best hand in the entire industry. With Coinbase's compliance license, over 100 million verified users, seamless fiat entry, and being one of the smoothest L2 networks in the ETH ecosystem, these conditions are top-tier in any field. Yet, two years later, what came out of this good hand is a collection of hybrid speculative tools.

Zora's core gameplay is minting a post or an image directly into tokens that anyone can publish. Its token ZORA has dropped from a peak market cap of about $550 million to around $30 million, losing 95%. Farcaster reached a peak of 104,000 DAUs in July 2024 thanks to the Frames mini-app, but over a year later it fell to a range of 40,000 to 60,000. The truly active Power Badge users dwindled to about 4,360, monthly revenue dropped 99%, and new registrations fell by 95.7%. These numbers did not decline slowly; they fell off a cliff.

The behavior pattern behind this is not complicated. People flocked in not to chat but to mine and speculate on tokens. The so-called "mini-apps" were mostly just rebranded lowbrow launchers; posts with a bit of information were scattered by pump and meme imagery. Posting transformed from "I want to say something" to "How can I post to earn more tokens?" Once the incentives stop, and token prices drop, these people turn and leave. The so-called social graph turns out to be an empty shell.

Ironically, this result is even less straightforward than pump.fun, which plainly tells you it’s a casino, and the currency is attention. The comments and real-time transactions there naturally cultivate a thin yet real layer of social relationships. The setup supported by Base stuffed speculation into the shell of "social," disappointing both ends — the financial experience is worse than that of specialized perpetual contract platforms, and the density of social relationships approaches zero. Even community modules with transaction scenarios like Binance Square, which thrive on real position discussions, have left Farcaster far behind in scale and influence.

The problem lies in misplaced incentives. Using tokens to leverage the early growth of social networks sounds like a shortcut since tokens can create immediate noise in data. However, socializing inherently requires identity recognition and relational connections, which hold a degree of non-instrumentality — if a comment first asks "Can I earn tokens?" it ceases to be socializing. Tokens turn relationships into transactional objects, turning fans into investors; once they lose money, these "relationships" can be liquidated at any time. This is not a social network; it's a leveraged position that can collapse at any moment.

The old world of web2 is not clean either

It’s unfair to place all the blame on Web3. The centralized social networks also have their defects, just in different forms.

Platforms like X, Meta, and TikTok operate on a business model built on mining attention: browsing time, likes, location data, and even contacts are all collected to feed recommendation algorithms. The algorithm detects a simple truth — the more conflictual, the more extreme, and the more it can make people lose control of their emotions, the longer it can keep you engaged, so such content naturally gains higher weight. Trends can be artificially manipulated, and hot topics can be boosted by bot matrices, MCN agencies mass-producing accounts and creating divisive topics; ordinary users, at the base of this pyramid, can only generate some natural exposure by creating emotions. Familiar relationship chains are responsible for spreading everything — a rumor passed through family and class groups gains more credibility than news media.

This is the real issue that Web3 social was initially aiming to solve: it’s not that socializing itself is flawed, but that the commercial model of "exchanging privacy for algorithmic recommendations, exchanging emotions for traffic distribution" is flawed. The framework supported by Base was supposed to be the antidote to this problem, but it failed to heal the sickness, instead sending the patient into a casino — the original illness remained untreated, and an additional layer of token speculation was added as a complication.

On the same track, others walk differently

If decentralized social networks all turned out this way, it would indicate a fundamental direction issue. But the reality is, on the same track, other projects have taken entirely different paths, leading to completely different results.

Bluesky, led by Twitter co-founder Jack Dorsey, uses the AT protocol, eliminating the token layer altogether. You can register with just a phone number or email, and posting and liking are entirely free. Its underlying architecture adopts the Personal Data Server (PDS) federated model, where each user’s data is hosted on their chosen servers, while Relays handle aggregation and distribution. This system does not go down the blockchain route; instead, it forges a new path through internet protocols on the road to decentralization. By 2026, user numbers exceeded 43 million; although it doesn't match the scale of X, it has at least established a foothold among ordinary internet users. Its user base includes tech workers, media, brands, and many ordinary discussants who migrated from X, with content types mainly focusing on brief news and public topic discussions, all free from token speculation. This provides a counterexample: while both claim decentralization, parties can completely remove the crypto barriers and token speculation at the product level to serve those with genuine social needs.

Hive offers another solution, presenting a more pointed contrast. It forked from Steem in 2020, built as an independent social blockchain based on Graphene, where every post, like, and vote becomes a permanent on-chain transaction record. Thousands of nodes across the network ensure that nothing can be unilaterally deleted. Its token system is tiered — HIVE is a circulating governance token, HP is a staking certificate, and HBD is a blockchain stablecoin pegged to $1, with holders earning fixed annual returns. In the inflation pool, 65% is allotted to creators and curators, with 50% paid in HIVE and 50% in HBD. This implies that during both bull and bear markets, a creator who consistently posts can always secure stablecoin income as a safety net. The game Splinterlands is another tool for retention — users come primarily for the fun of playing cards; the chain serves merely as the behind-the-scenes accounting engine; it isn’t the sole selling point. Users remain on the platform because of the enjoyment from matches, and social behavior naturally occurs during gameplay, presenting a retention mechanism that is entirely different from token incentives.

When comparing these three paths, the divide lies not in technology but in how tokens are viewed. Farcaster treated tokens as an end goal, and social behavior became a medium for mining; Hive considered tokens as rewards, with content creation being the primary goal. In the former, once the token price drops, the entire system loses its reason for existence; in the latter, token price fluctuations only impact the amount of additional income, while the motivation to write articles, chat, and participate in communities remains.

Here’s an angle that is rarely touched upon in industry analysis but is extremely critical — the evidential value in terms of search engines and AI scraping. Content on Hive possesses on-chain hashes and timestamps, enabling platforms like Google to index it steadily after scraping. Although rendering requires RPC to pull block data and generate pages, and the inclusion speed is not as fast as standard web pages, once included, long-tail knowledge content achieves long-term stability in search results. Moreover, in terms of AI scraping, large models value verifiability and immutability of factual sources, and Hive's on-chain verification happens to meet this standard, making it easier for tools like AI Overview and Perplexity to treat it as a citable source.

In contrast, posts on Farcaster are stored in third-party Hub server clusters, are not on-chain, and can be deleted or modified by operators. The pages lack permanent URLs, and the presence of numerous low-quality memes and promotional content has lowered the site's E-E-A-T rating, resulting in far lower visibility in search engines and general AI models than Hive. Base’s heavy investment in Farcaster equals an active surrender of a project's long-term discoverability within internet infrastructure. Once content lacks the properties of being retrievable and citable, its presence online approaches zero — this isn't an issue of user numbers; it’s a limitation imposed by the underlying architecture.

In 2024, Farcaster indeed captured far more attention and capital than Hive, but during the past two years of the SocialFi collective downturn, there has been almost no news of Hive experiencing mass user departures. In this shared wasteland, it's not difficult to guess who will survive to the next decade.

What should be retained

At this point, it’s time to collect thoughts: decentralized socializing is not a false proposition; Bluesky and Hive have each proven this point. What really needs reflection is that Base initially relinquished the old baseline principles of censorship resistance, privacy, security, and decentralization to the KPI of token issuance.

A healthy Web3 social platform should still adhere to these baselines — no single entity should be able to unilaterally delete your account, no backend should control your entire social graph, and the code should be open source so anyone can review it or take their data to start anew. Viewed through an economic lens, these issues are essentially public goods: non-excludable and non-competitive, a market is naturally unwilling to spend money on them, as they don't yield short-term profits; however, they are foundational facilities that everyone needs. This is precisely why they should be treated as public goods — funded through protocol treasuries, community donations, and a small number of individuals willing to pay for the ideology, rather than pulling a batch of retail investors to engage in a game of pass-the-parcel.

On a larger scale, this is also a battle for commercial freedom of expression. Algorithmic recommendations, privacy collection, and the industrialized production of content by MCN have turned public discussions into a business monopolized by capital and platforms. The significance of decentralized socializing is to carve out a space that isn’t manipulated by singular capital — even if it’s currently small and poor, it at least proves that a different way of living is possible, one that doesn't rely on selling your privacy and emotions to get by.

Governance challenges; no one has done it right

It’s not difficult to articulate ideals beautifully; the challenge lies in governance. Nearly every decentralized social protocol has been stuck in the same unresolved contradictions up to this point.

The first hurdle is the content moderation paradox. Projects like Hive and DeSo, with all posts stored on-chain, cannot delete content once it’s on-chain; if there’s a violation, the only recourse is to block it locally on the client side, yet it can still be viewed through a niche client, and regulators trying to assign responsibility can’t figure out whom to hold accountable. In contrast, Farcaster’s content is stored in Hub server clusters, nominally decentralized, but the real moderation power is entirely in the hands of a few leading Hub operators, making it no different from Web2 platform monopolies.

The second hurdle is the oligopoly of token governance. Nearly all blockchain social platforms rely on token-weighted voting, naturally causing early investors and big holders to monopolize discourse power. In Hive, HP staking weight determines witness seats; Lens's governance token is controlled by the Aave team and early VCs; Farcaster itself lacks a token, but large holders of community meme coins effectively control the thresholds for channels and the weight of rewards. Ordinary users who post without holding tokens have no voice regarding rules, yet those rules directly determine if they are visible.

The third hurdle is unclear responsibilities. Protocols, nodes, clients, and DAO treasuries have ambiguous delineations of authority, causing mutual shirking of responsibilities when issues arise — the protocol claims it only sets rules, nodes say they are not bound by the protocol, clients assert they are merely local filters, and DAOs state they cannot manage the behavior of nodes.

The fourth hurdle is misaligned incentives. Necessary "dirty work" such as anti-spam, running nodes, writing documentation, and developing clients tends to earn far less than mass posting, so no one willingly engages in it; governance and operations remain chronically understaffed. The speed of on-chain governance inherently conflicts with the real-time needs of socializing — a proposal voting period in Hive takes over seven days, and by the time it is resolved, the violating content has already spread across the entire network.

Currently, all projects offer compromises as answers to these hurdles: hybrid moderation, quadratic voting, staking juries — fundamentally, these all seek a suboptimal middle ground in a dilemma. No one has truly unraveled these knots, and this is the root of why decentralized social hasn't scaled significantly to date.

To survive, learn how to transform

While the deadlock in governance may be unsolvable for now, there are still opportunities to rethink the economic model. If the aim is to retain creators and ensure real content dissemination, the core focus should be on redesigning attention transformation, rather than merely leveraging token airdrops to force a crowd into the scene, creating a false sense of prosperity.

The first step is to decouple money from socializing. Hive's approach is worth emulating — earnings shouldn’t rely on token speculation but on a stable revenue share bolstered by stablecoins, providing creators some fixed income regardless of their activity, rather than fully exposing them to the risks of speculative markets. The second step is to establish clear rewards for "dirty work" — reviewing, running nodes, and developing clients should not be unpaid; otherwise, governance will forever be in a loop. The third step is client diversification. Referencing the protocol logic of email — a single identity across any client can log in, allowing users to change interfaces without losing fans or content, enabling genuine fluidity between different screens and social layers. Transformation should naturally result from visually attractive content and an open protocol, rather than being artificially generated through incentives.

Finally, intertwine these segments into a long chain — from creation, to curation, to distribution, to monetization, every link should have its own legitimate earnings without excessive cuts feeding the platforms. Email exists as an infrastructure of the internet not because of any single company's ownership, but precisely because no one has ownership over it. Similarly, socializing should perhaps head in that direction.

Burning through the budget, it’s time to wrap up

Base's admission of error is fundamentally not about the contraction of a single business line, but rather that the whole industry spent two years and burned hundreds of millions of dollars to realize a truth that should have been clear from the outset: using tokens to leverage a network that requires relational density is using the wrong fuel. This aircraft took off fiercely, but all its lift stemmed from hype, not friendship; once the fuel runs out, it crashes.

Decentralized socializing is not a false proposition; Bluesky and Hive have already restored some dignity to this industry. The remaining question is whether anyone is still willing to act not for token issuance but to diligently walk through the principles of the "post-Snowden network" once again. The business of the human heart cannot be hurried; everyone wants to take shortcuts, but if enough people take shortcuts, it becomes the most crowded road in front of the casino.