avatar
Upside GM 👋
77,313 Twitter followers
Follow
Upside is Vietnam’s largest community of long-term crypto investors — powered by @WeTheIvy, a Web3 Media Company. @TheIvyNFT is our official NFT collection
Posts
avatar
Upside GM 👋
🌟 The Internet Has No CEO Systems that belong to no one are often more durable and secure than those controlled by a single company. The internet has proven this for over 50 years. Recently, thousands of Americans discovered that their home solar panels were remotely shut down. Not broken, but locked down by a foreign company because the installer hadn't paid. Even though you bought and own the equipment, you don't really control it. This doesn't just happen with solar panels. It happens with most of the software and equipment we use every day. American farmers own machinery worth hundreds of thousands of dollars but can't repair it themselves, having to wait for the manufacturer's permission. A single software bug can cripple the world, like the incident that caused millions of computers to crash in 2024. The problem is: we are building an economy on closed systems, where a single party has the power to decide. When you use someone else's platform, you think you're a customer, but in reality, you're dependent. 📍Why is the internet different? The internet has no CEO, no parent company, no "shutdown" button. If one part fails, the data automatically goes elsewhere. If one service goes down, another continues running. The core of the internet is open standards; anyone can use and build upon them. Because no one owns the internet: - No one can raise your price - No one can lock you out of the system - No one controls the entire game Thanks to this, the internet has become the platform for tens of trillions of USD in economic value. 📍Conversely, closed platforms are "sucking" value Today, platform companies stand between you and everything digitized: software, data, transactions. They can raise prices at any time, and you still have to pay because the cost of leaving is too high. Businesses are spending more and more money on software, not because the software is better, but because they have no other choice. This system isn't faulty. It's working exactly as it was designed: creating dependencies. 📍Is there another option? Yes: open systems. Open-source software like Linux or Git shows a different model: no one owns it, everyone can test it, fix bugs, and continue using it even if one party disappears. When a bug occurs, the entire community sees and fixes it together. No need to wait for the "manufacturer to handle it." Open systems can fail, but they fail publicly and are fixed quickly. Closed systems fail in the shadows. 📍But what about value? The internet handles information transmission well, but money and value still have to go through banks, card companies, and intermediaries. Every transaction can be blocked, frozen, or controlled. This is where Ethereum becomes important. Ethereum is not a company. It has no CEO. No one owns it. It's an open protocol that allows the transfer of value, not just information. No one can “brick” your assets or lock down the system at will. Just like the internet, Ethereum exists because no one controls it. 📍In short, it's very simple: There are only two types of infrastructure: - One that no one owns - And one that owns you The internet has proven the first model effective with information. Ethereum is trying to do the same with money, assets, and economic agreements. And the final question isn't "which technology is better," but rather: do you want to build your future on infrastructure that no one controls, or on infrastructure that can be shut down at any time? By James @Snapcrackle - Head of Ecosystem at Ethereum Foundation twitter.com/gm_upside/status/2...
ETH
1.3%
avatar
Upside GM 👋
💎 The Trillion-Dollar Opportunity for AI: What are Context Graphs and Why Are They Important? The biggest trillion-dollar opportunity for AI in businesses may not lie in making models smarter, but in documenting how humans make decisions. Previous generations of software created immense value by becoming "truth-recording systems"—stores standardized data about customers, employees, or operations. However, these systems primarily recorded the final outcome, not why that outcome was accepted. When AI agents begin to enter real-world workflows, this limitation becomes very apparent. Businesses don't lack data, but rather the trace of decisions: why this deal received a deeper discount than stipulated, why that ticket was prioritized, why an exception was approved this time but not another time. Those answers are often found in Slack, Zoom calls, private messages from leaders, or in the memories of a few long-time employees—not in any formal system. Rules and policies only tell agents what to do in general situations. But operational reality is full of exceptions and precedents. People make good decisions not just because they know the rules, but because they remember: “how we handled a similar situation last time.” The same is true for agents. Without access to the history of previous decisions—who approved them, in what context, what exceptions were accepted—an agent is just a rigid, judgment-lacking execution machine. When an agent is placed directly into the workflow, it can record the entire decision-making process at the moment the decision occurs: where the input data comes from, which rules are applied, which exception branch is triggered, who approves it, and why. Over time, these traces connect to form a contextual map—a “living memory” reflecting how the business actually operates. It not only tells what happened, but also explains why it was allowed to happen. This is something that current systems struggle to build. CRM or ERP systems only store the current state, not the context at the time of the decision. Data warehouses only receive information after everything is done, when the reason is no longer relevant. To preserve the trace of a decision, the system must be at the point where the decision was made, not looking back afterward. This is the structural advantage of startups building a orchestration layer for AI agents. Therefore, the greatest opportunity may not lie in replacing the entire old system, but in the emergence of a new decision-making record system. Initially, it only supports automation, with human involvement in the approval process. But gradually, it becomes a place where businesses can look up: “Why did we do it that way?” As decision-making traces accumulate, precedents become searchable, and automation can be safely and controllably implemented. Ultimately, the question isn't whether the old record-keeping systems will survive, but rather: will the next trillion-dollar platform be built by attaching AI to old data, or by documenting how humans make decisions so that data truly becomes actionable? “Context mapping” is the foundation for the second approach. By Jaya Gupta @JayaGup10 - Entrepreneur & Partner of Foundation twitter.com/gm_upside/status/2...
avatar
Upside GM 👋
🔎 The Core Paradox of Prediction Markets: Pricing by Imperfection Prediction markets, in theory, should price probabilities across an extremely large space of states. Even a seemingly binary contract like "Candidate X wins" is actually conditioned by countless variables: macroeconomics, legal, media, funding flows, voter behavior… If there are n variables involved, then there exist up to 2ⁿ possible world states. However, prediction markets never fully price the entire space. Traders are limited by time, perception, and information, so they only choose a very small subset of variables to trade. Therefore, market price is not a "pure" probability, but rather the cost of buying exposure to a package of implicitly correlated scenarios, with a series of assumptions that are not isolated for pricing. More importantly, prediction markets are reflexive: price not only reflects reality but also impacts the underlying conditions – influencing media, funding, voter behavior, etc. This creates a feedback loop where belief and fundamentals are endogenously interconnected. 🚨 The paradox lies in this: - The market tends towards an impossible state-space to encompass. - But ignoring much of that space makes price both computationally feasible and capable of influencing behavior. - Exploitable inefficiencies often lie in the “forgotten tail” – valid scenarios that fall outside the narrative that most traders are paying attention to. This isn't a flaw, but a structural characteristic: prediction markets need asymmetry in research and belief to survive. If everyone understood things the same way, uncertainty would collapse into consensus, leaving no edge or reason to trade. Because it's impossible to list every 2/4 of the scenarios, traders are forced to compress information. In reality, most sentiment fluctuations are usually explained by a few key parameter classes: 📍Structural baselines – slow-moving fundamentals (partisanship, demographics, institutions). 📍Macro-directional indicators – macro-directional indicators that guide the overall narrative. 📍Catalytic events – highly elastic disruptive events (legal, scandal, geopolitical). 📍Behavioral & narrative priors – media, endorsement, money flow, voter sentiment. 📍Model-free momentum – price movements themselves acting as proxies for future possibilities. In short, the edge in prediction markets doesn't come from listing every scenario, but from the ability to identify the most effective layers of variables that compress uncertainty into narrative – where much of the market is missing. Source: @polyfactual - A market analysis channel with 14k followers twitter.com/gm_upside/status/2...
avatar
Upside GM 👋
Thread
💯 Why does DeFi lending offer greater sustainable advantages than you think? Many believe that in the DeFi lending landscape, power and value are increasingly shifting toward vaults, curators, and distributors, while lending protocols are facing increasingly thin profit margins. However, if we look at the entire on-chain credit value chain and the actual cash flow, the picture is completely different: lending is the layer that holds the biggest sustainable advantage. The reality on AAVE and SparkLend clearly shows one thing: the interest that vaults pay to the lending protocol is higher than the revenue that the vaults themselves generate. This directly refutes the "whoever controls the distribution wins" principle in lending. Not only does AAVE earn more money than the vaults built on it, but it also generates more value than asset issuers used in lending such as @LidoFinance or @ether_fi. ------- The reason lies in the position of lending within the system. The lending protocol is where the supply and demand for Capital meet, and every leverage strategy is forced to go through this layer: - Users: deposit assets and find vaults to optimize returns. - Vault/curator: strategic packaging, loop management, and risk management. - Lending protocol: provides liquidation and infrastructure, and collects fees directly on loans. - Asset issuer: issues stETH, weETH… and receives a portion of the yield. - Blockchain: the infrastructure layer for processing transactions The key point is that lending protocols charge fees based on the loan size, not on the net profit of the strategy. The more leverage used, the more value flows into lending. 🌟 The clearest example is the http:/Ether.fi vault on AAVE. This vault borrowed approximately $1.5 billion, but its actual net Capital is only about $215 million. With a platform fee of 0.5%, the vault only generates about $1 million per year, while paying AAVE approximately $4.5 million in interest. In other words, the lending protocol earns many times more than the vault, even with a large and efficient looping strategy. This pattern is repeated with several other vaults: - @0xfluid: borrowed approximately $1.7 billion, vault collected around $4 million, while the lending protocol collected around $5 million. - @mellowprotocol: Small TVL but high leverage, lending protocols continue to earn more per unit of TVL. - On SparkLend, large vaults also show similar results when compared by % TVL. Even when compared to asset issuers, lending protocols still hold the advantage. With the same amount of ETH pledged as collateral, the value generated from ETH interest and stablecoin interest through lending is significantly higher than the performance fees earned by issuers. According to Silvio - Researcher from Blockworks Advisory x.com/SilvioBusonero/status/20...… twitter.com/gm_upside/status/2...
AAVE
2.03%
avatar
Upside GM 👋
🧐 Mert Mumtaz and the crypto 2026 picture In late 2024, Mert predicted that by the end of 2025, the dopamine of the masses would run out. Crypto would become less speculative, and "safe and legal" options would prevail. That's exactly right. Payment chains, on-chain stocks, Canton Square… all became the focus of attention. But when that wave passes, where will crypto go next? According to @mert, crypto has always consisted of three parts: commerce, casinos, and cypherpunk. For many years, cypherpunk (private money, an uncontrollable system) was almost forgotten due to legal fears. But privacy is starting to become appealing again. For Mert, this is the biggest dopamine generator left in crypto. Therefore, 2026 will be the year of privacy. (Remember this, guys: if 2026 isn't the year of privacy, then you can come back and laugh in Mert's face!) At the same time, Robinhood, Revolut, and Stripe started going on-chain, capturing the "easy and safe" segment. This forced crypto builders to be more innovative in order to survive. Products that run right on the legal borderline, but cleverly capitalize on the crypto-native niche, will thrive. Perps for stocks and commodities. The prediction market is directly linked to newspapers and social media. Completely new ways of trading. Mert also believes that X and at least one other major app will try to go head-to-head with Robinhood and Coinbase. 👉 And starting with Solana, more teams like Pump or Axiom will emerge, truly competing with traditional fintech. The reason is: Fintech requires years to build its infrastructure. - Crypto builders just need to plug it into the blockchain and then focus entirely on the user experience. ✍️ Qing twitter.com/gm_upside/status/2...
SOL
0.69%
loading indicator
Loading..