I expected a question to be asked at Brooklyn Tech Week 2024 about GDPR and how it applies to Arweave's permanent storage, especially given the presence of legal minds and the venue at Brooklyn Law School.

Alas, no one asked it, but it got me thinking about one of the core principles of Arweave: its "right to be remembered." This concept is fundamentally different from GDPR's "right to be forgotten." Both are essential for society, but I believe the best system is built on permanence.

The right to be forgotten empowers individuals to erase personal data that’s outdated, irrelevant, or harmful. It’s about regaining control in a world where our data is everywhere and often persists longer than it should. It embodies personal dignity, forgiveness, and second chances. Europe’s cultural emphasis on privacy and individual autonomy has pushed their lawmakers to enshrine the "right to be forgotten" in law through GDPR. However, laws like GDPR also introduce challenges: high compliance costs, ambiguous legal interpretations, and potential conflicts with innovation, leaving both businesses and consumers struggling with its complexity.

The right to be remembered, by contrast, ensures that information—especially that which holds cultural, historical, or societal significance, is preserved indefinitely. It’s about creating a collective memory and safeguarding against loss and revisionism. In the U.S., the First Amendment prioritizes freedom of speech and access to information, placing a strong emphasis on permanence and transparency. However, freedom of speech, while vital, can also bring societal challenges like the spread of misinformation, hate speech, or conflicts with privacy.

Both rights aim to protect individuals and society from harm, but the tension lies in permanence vs. deletion. Should we have systems that allow for the past to be erased to protect privacy, or remember it to preserve truth?

The rise of artificial intelligence adds new dimensions to this debate. AI datasets are massive, and temporary or mutable storage introduces significant risks like data inconsistencies, loss of provenance, and the inability to verify the authenticity of training inputs. These issues can lead to biased or unreliable models and erode trust in AI systems. By contrast, models trained on permanent, time-stamped, and tamperproof datasets can ensure the integrity of training inputs, outputs, and even the large language models themselves. However, these benefits also raise ethical concerns about data usage, licensing, and privacy, highlighting the urgent need for thoughtful governance in AI development.

Permanent cloud networks like ar.io offer a novel middle ground, enabling an internet that never forgets while respecting local moderation principles. Gateways in the network face two key choices: they can censor ArNS names or other pieces of data. This might happen if a name or data falls outside their content moderation policies or if the original uploader signals they no longer want their data served. However, there are still economic disincentives for these actions, ensuring the network remains resilient and incentivized to preserve and serve content.

Gateways are incentivized to resolve and serve all ArNS names because doing so maximizes their potential rewards. If a gateway chooses to censor certain names, it risks receiving less incentive compared to gateways that serve the full spectrum. Similarly, while gateways can choose to censor the data itself, Arweave nodes remain incentivized to store and replicate all data permanently. Nodes that censor data risk reduced mining rewards, reinforcing the network’s commitment to permanence.

This dual-layer system ensures that the underlying data and naming infrastructure remain immutable and permanent while giving gateways the flexibility to apply local content moderation standards. The result is a system that balances the permanence of data with ethical flexibility, making it resilient and adaptable to a whatever humanity may throw at it.

Perhaps a foundation of immutability with layers of human and AI-driven moderation on top is the best of both worlds. It feels like a meaningful step toward a system that honors both the right to be forgotten and the right to be remembered.

But what do you think? Can a permanent cloud network built on Arweave balance the tension between remembering and forgetting? How do you see these rights evolving in a world increasingly built on data? I’d love to hear your thoughts.