Greetings RadicalxChanger,
Our monthly community calls, starting tomorrow and from now on, will begin one hour earlier (at 9am ET / 3pm CET)****. This change is reflected in our call calendar (click the link and then “Add to Google Calendar” in the bottom left to subscribe). Thanks to everyone for your cooperation, and we hope to see you tomorrow.
Now some exciting news...
The case for data dignity is no longer coming only from advocates and academics. It’s coming from inside the labs.
In Bartz v. Anthropic, a class action in the Northern District of California, a 2021 internal memo from Anthropic was unsealed as an exhibit. In it, CEO Dario Amodei lays out a framework for compensating the people whose data trains AI models.
Amodei described the goal as building “a quasi-market equilibrium designed to artfully counter the concentration of wealth,” one where data producers would receive a share of model profits proportional to their contributions. Senior researcher Andy Jones compared the economics to farming and forestry: many providers of essential inputs, market structures that prevent any single buyer from extracting all the value.
If this sounds familiar, it should. This is the data dignity thesis — data as labor, producers deserving compensation and voice — arrived at independently by the CEO of a leading AI lab.
So, why didn’t they build it? In Amodei’s own words, this “might not work itself out due to collective action problems or blunt government action.” Even if you believe compensation is right — and it appears Amodei did — doing it alone is a competitive disadvantage. No single company can pay for data when competitors train on the same material for free.
That’s not a technical problem. It’s a collective action problem. And collective action problems have a well-known solution: we have to solve them together.
This is another significant moment for the RadicalxChange community! We’re working on what comes next, and we’ll have more to share soon.
Check out this new Quadratic Voting tool
The quadratic formula can feel intimidating. Our brains think linearly, not quadratically, so asking people to calculate quadratic costs on the fly puts them off before they even engage with the issues.
So Tara Karpinski and Wesley Hartogs designed an interface that moves quadratically but appears linear. The prototype uses emojis, a visual language people already know, to lower the cognitive load and make quadratic voting more accessible.
Try it yourself at iloveqv.org!
From Around the Pluriverse
→ Plurality: 協力技術と民主主義の未来 now available in Japanese!
→ The A.I. Evangelists on a Mission to Shake Up Japan
→ AI May Swell Japan’s Lower-Income Ranks, Party Leader Warns
→ OpenAI Is Making the Mistakes Facebook Made. I Quit.
→ Faith Family Technology Network
→ AI Agents “Talk Amongst Themselves” on Moltbook
Where To Meet Us Next
March 4, Virtual: You’re invited to our March 2026 Community Call tomorrow, on Wednesday, March 4 at 9 am ET (UTC-5) | 3 pm CET (UTC+1). Please note the new call time. Use this Typeform to add agenda items, and we’ll see you there!
Live long and prosper,
The RxC Team