To date, no decentralized app has gained traction despite billions invested. Here are some working hypotheses I’ve developed as I’ve looked at projects in the space. Dapps are far away Web3 infrastructure remains very weak, making it impossible to build a usable dapp. A decentralized application today has fifteen second latency; requires payment for even read-only transactions; risks catastrophic loss of data if a user misplaces a single secret key; cannot effectively store or query data; and has no ability to interface with data in the real world or other services. My advice: Don’t invest in a dapp unless the developers really understand today’s unscalable infrastructure — and have a solution. But Dapps aren’t necessary years away I heard from some investors that “dapps are years away”. I disagree with the implication here — that it’s simply a matter of time. This kind of thinking works in later stages of a technological cycle, when it’s more about following trends, such as smart phone penetration or broadband adoption. Blockchain is early cycle. It’ll follow the pattern of discontinuous leaps in capability based on real technology. Dapps will work once someone builds the infrastructure to make it happen. It’s like Tesla building the gigafactory, Amazon building its own fulfillment centers when Ingram was bought out by Barnes and Noble, or Google inventing BigTable — they actively decided when to build a critical piece of infrastructure to unlock new applications. My opinion: The template to follow is not the last 15 years of Internet investment, but, the first 5 years. On the app side, learn from stories like Google, which invented the infrastructure it needed, and Amazon. Don’t learn from later stage successes like Facebook and Uber. Computer Science Matters The launch of AWS enabled high level developers to build scalable, powerful web and mobile apps quickly without needing to worry about computer science issues. This is a thesis Mark Suster recently published with backing data. But in frontier technologies, there is no AWS to rely on. So you need real deep down, low level computer science thinking. Lack of real computer science is a root cause of the underperformance of every single dapp out there today. My opinion: One of your must haves to invest in a blockchain startup should be computer science skills. This goes for LPs investing in VCs as well. Incentives Matter Prior to Nakamoto Consensus, open protocols like SMTP suffered from abuse like spam because they were not designed to disincentivize bad actors. One of Bitcoin’s core innovations was including economic incentives in a data storage protocol. Real calculations were made on its efficacy (see section 11 of the original Bitcoin whitepaper). And it made properly nuanced claims, such as consensus being probabilistic not absolute. Many decentralized projects make claims about their decentralized protocols without understanding the incentives at play in a complex system of independent actors. A common symptom is when kludgy mechanisms are bolted on to address specific attack vectors, rather than defining broad classes of problems to be solved. My opinion: If you are looking into a decentralized project that requires incentivizing behavior, make sure their description of incentives and game theory are rigorous. Quoting Leslie Lamport, the researcher who defined the Byzantine Generals Problem: This argument may appear convincing, but we strongly advise the reader to be very suspicious of such nonrigorous reasoning. Although this result is indeed correct, we have seen equally plausible “proofs” of invalid results. We know of no area in computer science or mathematics in which informal reasoning is more likely to lead to errors than in the study of this type of algorithm. Infrastructure Should Solve Problems The conventional wisdom in infrastructure is to build something 1000x faster than Ethereum… even though that would not be sufficient to power usable dapps. I think the issue here is that people are hoping to find the next token with the same explosive returns as ETH or Bitcoin. Ethereum is worth a lot… if a new compute blockchain is faster, then it’ll be worth a lot too, right? Bitcoin is worth a lot… if a new payment blockchain can do everything Bitcoin can but also provide privacy (and also has an anonymous founder) then it should be worth a lot too, right? Bitcoin and Etheruem were true zero to one efforts. They introduced brand new capabilities, never before seen. The bar for new efforts is now much higher. My opinion: Infrastructure needs to be 1000000000x faster, cheaper, larger, more connected, more secure, more queryable, and more usable than Ethereum. Even then, it will likely not achieve ETH/BTC level returns. Go Beyond The Echo Chamber With zero dapps finding product-market fit, VCs have turned to the next best measure of success: startup-investor fit. Entrepreneurs who raised money, but have no traction for their product, are treated as the experts. This is leading to an echo chamber which re-enforces thinking that has already led to unused products. My opinion: Triangulate “on the ground” data from blockchain entrepreneurs with the experience of entrepreneurs from previous waves of innovation.