Forward Research

ao Verified account @aoTheComputer · Mar 20

Something is starting to click.

Open Claude. Tell it to fork a permaweb app. You now have a new app: fully hosted, open, decentralized. Forever.

No servers. No devops. No company to found. Just you, an AI, and the largest open data lake on the web.

PermawebOS Seed: a preview of what's coming. 🌱

Build on the permaweb. Win $AR.

• Top 2-4 forks/week: 35 $AR each
• Showcase on X with #PermawebOS

Rules + more info in the AO Discord

Week 1 deadline: Thursday, March 26, 11:59pm ET

View quoted post on X

🐘🔗 sam.arweave.net Verified account @samecwilliams · Mar 10

Huge shift here. Thanks to everyone on the @OdyseeTeam, @dh_association, and @fwdresearch teams who made this possible -- as well as @ar_io_network for all their help in the transition.

But what does this shift mean?

tl;dr: Arweave dot net now runs on AO-Core. So...

1⃣ Every response from .net is now verifiable, right to the end-user. This is the fundamental building block of decentralization. When you get a reply from an AO-Core node its headers contain everything that you need in order to verify the data atomically. No need for consensus, querying multiple nodes, etc., just fully trustless cryptographic verification.

Additionally, all of the useful tags and metadata that have always been attached to content on Arweave is now available to callers. Users can now process this information and act upon it, just like 'body' data. You can think of Arweave as a permanent database, with each item being a row. Now the whole content of each of those rows can be accessed by users, not just the largest field.

2⃣ Data served from .net is now directly sourced from Arweave nodes. Previously, there needed to be caches in between the user and the nodes, which made gateways heavier to run and 'separated' from the dataset. This detachment introduces points of software and operations dependence in the caches themselves. While .net is still importing some of the data from the legacy gateway (and will be over the next few weeks), these caches have now been removed from the data serving flow. This also opens the opportunity for...

3⃣ ...Permissionless nodes operating .net. Because each AO-Core node serves everything needed to verify each response it gives by default, the next step is to let anyone register to provide the data for IDs to the gateway, then verify their responses before relaying them to users.

This alone is a big deal. In time we expect it will provide an additional incentive for Arweave miners to serve (and also store) data, as well as improve performance (by routing to and rewarding the fastest providers) and reliability (by removing points of failure).

Further down the track there is a clear path to even decentralizing the operation of these verification+routing nodes, so that every Arweaver can take part in running those, too. We can achieve this by letting TEE nodes register with one another and share private TLS credentials, allowing them to directly serve end-user traffic routed by the DNS layer. We have tests of this flow working in principle, but principle -> practice-at-scale will take some time.

4️⃣ Compute-Over-Arweave-Data just dropped. AO-Core is a protocol to orchestrate a decentralized supercomputer. In this deployment we are making heavy use of AO-Core's codec devices, but it is now possible to do so much more than that. More on this soon.

Upshot? Decentralization and trustlessness of data access on Arweave just took a huge leap.

Trustless verification of content ->
Permissionless data serving ->
Decentralization of data access and transformation.

Congrats again everyone 🫡.

View quoted post on X