# Universal Basic Income Funded by Data, Not Taxes Every serious UBI proposal runs into the same question pretty fast: where does the money come from? The standard answers are some version of wealth redistribution - higher taxes on income, capital, corporations, or wealth - redirected into a universal payment. The debate is mostly about the rates, the thresholds, and whether the economic effects of the redistribution undermine the goal. That debate has been running for decades without resolution because the political and economic tradeoffs are genuinely difficult. There's a different funding model that doesn't get nearly as much attention, probably because it requires thinking about data as an economic asset rather than a privacy issue. The premise: your personal behavioural data is being used to generate enormous amounts of economic value. The companies using it are paying almost nothing for it. If that value were properly attributed and compensated, a UBI-scale payment to every data-generating human becomes at least theoretically plausible. --- Let's try to make the numbers concrete, because vague claims about data value don't help anyone. Meta's revenue in 2024 was approximately $164 billion. The global population with internet access is roughly 5.4 billion people. That's about $30 per connected person per year flowing through Meta alone. Google's advertising revenue is larger. Add Amazon, TikTok, the broader surveillance advertising ecosystem, and the training data purchases that AI companies are making - you're talking about several hundred dollars per internet-connected person per year being generated from data that those people created. That's not UBI-level money by itself. But that's the current arrangement, where the platforms take the vast majority of the value and return an extremely small fraction. The question is what happens when you flip the architecture. In a system where users own their data, decide what it's used for, and receive compensation when it generates commercial value - the extraction doesn't go to zero, because data genuinely is more valuable when aggregated and processed by companies with the infrastructure to do it. But the split changes. Instead of the user receiving zero and the platform receiving everything, something closer to a licensing model: you share your data for specific purposes, the platform processes and monetises it, and you receive a share of the revenue it generates. What share? That depends enormously on the leverage the user side of the negotiation has. Right now, zero - because users can't effectively coordinate and have no alternative infrastructure. In a system with verified identity, data ownership infrastructure, and the ability to collectively negotiate terms - potentially significant. --- The AI training data market makes this more concrete. OpenAI, Google, Meta, Anthropic and others are spending tens of millions to hundreds of millions acquiring training data. Verified, consented, high-quality human behavioural data is the scarce input they can't get through scraping. The demand is real and growing. A person generating continuous, verified, quality data about their daily life is producing something these companies will pay for. Not any individual data point - the value is in the aggregate, the continuity, the verification, and the consent architecture. But the aggregate value of a network of verified human data contributors is genuinely significant. The mechanism would work something like this. Data contributions from verified users get processed by AI agents that package them into structured, labelled, high-value datasets. Those datasets get licensed to AI companies under terms that include compensation back to the contributing users. The compensation flows through the karma/reputation system so that higher-quality contributors - those whose data is more useful, more verified, more consistent - receive higher compensation. Working through the economics in a real implementation: a karma-weighted cashback system with tiered subscription rates produces meaningful per-user returns once the network reaches sufficient scale. The ceiling isn't retirement money. For someone in a lower-income country, consistent quality data contribution translates to material daily income. For anyone, it's non-trivial. The honest caveat is that this depends heavily on achieving the scale required to make the aggregated data valuable - something no system of this kind has yet demonstrated at population scale. --- The objection worth taking seriously is whether behaviour changes when you pay people to be observed. It does. People who know they're being recorded behave differently from people who don't. If the economic incentive is to generate more data, there are perverse incentives toward data quantity over data quality - performing behaviour rather than living it. The karma architecture is partly a response to this. If what's valued is genuine quality and consistency rather than raw volume, if the scoring system is designed to detect and discount performed behaviour, if the verification layer is robust enough to catch systematic gaming - the incentive problem is manageable if not entirely solved. This is where the design of the system matters enormously. A badly designed data compensation system would produce low-quality performance data that's worth nothing to anyone. A well-designed one produces genuinely valuable human behavioural data that funds real compensation. The difference is in the details of the verification, the karma mechanics, and the data quality standards. None of those are trivial. But none of them are unsolvable either. --- The UBI funded by data model doesn't replace tax-funded UBI as a policy mechanism. They address different problems - one is about redistributing existing economic value, the other is about recognising and compensating value that's currently being extracted without recognition. What's interesting about the data-funded version is that it doesn't require political consensus on redistribution. It requires building the infrastructure that makes the value visible and the compensation possible. That's a harder technical problem than passing a tax. It's a much easier political problem than changing the tax code. The political economy of "you should be paid for the data that's currently being taken from you for free" is simpler than the political economy of most UBI proposals. Most people understand intuitively that their data has value. Most people would like to be paid for it. The barrier isn't the argument - it's the infrastructure to make it real. Building that infrastructure is the interesting problem.