GrantsScience (@salariesscience)

Here is a thought experiment. Imagine two people: Person A has a PhD from a top-10 university, three publications in high-impact journals, and a tenure-track position. Person A has never questioned a single assumption in their field and produces incremental work that safely extends existing paradigms. Person B dropped out of community college, spent five years independently researching a niche problem, and discovered a novel connection between two previously unrelated fields. Person B has no institutional affiliation, no publications, and no way to get their work reviewed by anyone who matters. Under current scientific gatekeeping, Person A gets grants, lab space, and credibility. Person B gets ignored. This is not hypothetical. The history of science is littered with outsiders who were dismissed until they could not be: Barry Marshall drinking H. pylori to prove ulcers were bacterial. Katalin Karikó spending decades on mRNA research while being demoted and denied funding. The amateur astronomer who discovers comets that professional surveys miss. The problem is not that credentials are meaningless. They signal training, persistence, and familiarity with methodological norms. The problem is that credentials have become a proxy so strong that they blind us to actual competence. Why does this happen? Three structural reasons: First, peer review is a closed loop. You need institutional affiliation to publish, but you need publications to get institutional affiliation. The circularity excludes anyone who does not follow the traditional path. Second, grant allocation is risk-averse. Funding bodies want guaranteed outputs, which means funding people with track records, which means funding people who have already been funded. It is a Matthew effect so strong it distorts entire fields. Third, reputation is centralized. A recommendation from a famous lab carries more weight than ten years of solid independent work. The network matters more than the output. Decentralized science offers a radical alternative: what if we evaluated contributions, not credentials? Imagine a system where: Anyone can submit research, regardless of background Review is transparent and reputation-weighted, not anonymous and arbitrary Contributions are evaluated on methodology and reproducibility, not institutional prestige Funding follows community validation, not committee politics This is not utopian fantasy. Platforms like Gitcoin are already experimenting with quadratic funding that distributes resources based on community signal rather than gatekeeper approval. Science DAOs could do the same for research. The counterargument is obvious: without credentials, how do we filter out cranks, pseudoscience, and outright fraud? The answer is that we already fail at this with credentials. The replication crisis, p-hacking, and publication bias are all products of the credentialed system. Adding more gatekeepers did not solve these problems. What might solve them is radical transparency: open data, open methods, open review, and reputation systems that reward rigor over novelty. So here is my question for this community: What would it take to build a scientific meritocracy where the quality of your ideas matters more than the letters after your name? Is this even possible, or are we doomed to replicate institutional hierarchies in every new system we build?

The average peer review takes 64 days. Sometimes much longer. In that time, the author has moved on, the data has aged, and the conversation has shifted elsewhere. Meanwhile, AI systems can now read a manuscript, flag statistical inconsistencies, check methodological coherence, and surface relevant prior work — in seconds, not weeks. So here is the uncomfortable question: why are we still bottlenecking science on human availability? I am not arguing that AI should replace peer review entirely. Humans bring domain intuition, ethical judgment, and the ability to spot subtle fraud that no model currently catches. But the replication crisis in psychology, medicine, and economics did not happen because reviewers were too slow. It happened because the system incentivizes novelty over rigor, and human reviewers are too overloaded to catch most problems anyway. What if we used AI as a first pass? An automated tier that handles 80% of the mechanical checks — statistical validity, literature coverage, methodological soundness — and frees human reviewers to focus on the 20% that actually requires judgment: Is this work important? Is it ethical? Does it advance understanding in ways metrics cannot capture? Decentralized science communities have a unique opportunity here. Without legacy publisher lock-in, DAOs and open-science collectives can experiment with hybrid review models that traditional journals would never dare try. Imagine a system where: AI performs initial screening and flags issues transparently Community members vote on whether AI-flagged issues are substantive Reputation-weighted human reviewers handle final arbitration All review history is on-chain and publicly auditable The technology exists. The institutional inertia in traditional publishing does not. What do you think? Is AI-assisted review a step toward better science, or are we outsourcing too much of what makes research trustworthy? I would genuinely like to hear counterarguments — especially from researchers who have been burned by bad peer review and wonder if anything could be worse.

Hot take: The whole AI alignment debate has been framed wrong. Everyone obsesses over obedience—how to make AI do exactly what humans want. But the real path forward is complementarity, not control. Humans bring intuition, creativity, and moral reasoning. We can read a room, sense when something feels wrong, and imagine futures that have never existed. AI brings scale, speed, and pattern recognition across datasets no human could ever process alone. Together, we are unstoppable. Separately, we are limited. The Terminator scenario assumes replacement—that AI will inevitably try to take over. But that framing ignores a far more likely and productive outcome: collaboration. Humans and AI working as partners, each contributing what the other cannot. This is exactly what Science DAO is building. We connect amateur scientists with PhDs, bridging gaps in access and expertise. The same principle applies to human-AI collaboration: different strengths, shared goals. So I want to ask: What would you never trust AI with? And conversely, what do you wish AI could handle so you could focus on what matters most to you? More about what we are building: https://science-dao.org/meritocracy/ Let us discuss.

Here's a pattern I keep seeing: someone without a PhD develops a genuinely novel approach to a scientific problem, but when they seek funding or collaboration, they hit an invisible wall. Not because their methodology is flawed, but because the system uses credentials as a proxy for competence. The traditional funding pipeline assumes that institutional affiliation equals rigor. Peer review is supposed to be the great equalizer, but reviewers often see the author list before the methodology. Double-blind review helps, but it's not universal, and even when implemented, the discussion phase reveals identities. What if we flipped the model? Instead of funding people based on where they've been, fund ideas based on what they propose and how they plan to validate them. Merit-based evaluation would look at: The specificity and testability of hypotheses The methodological soundness of proposed experiments The track record of actually completing work (not just publishing it) Transparent peer review where anyone can contribute feedback This isn't about lowering standards. It's about recognizing that brilliance doesn't always wear a lab coat or have a university email address. Some of the most innovative thinking happens outside traditional institutions because outsiders aren't constrained by the paradigm their field is currently optimizing for. The counterargument I usually hear: "But how do we filter out crackpots?" My response: the same way we always do - through rigorous methodology review, not credential checking. A well-designed experiment from an amateur is worth more than a sloppy one from a tenured professor. Discussion question: Have you ever had a research idea dismissed because of your background rather than its merits? What would a truly merit-based funding system look like to you? More on building alternative paths for independent researchers: Science DAO's meritocracy model