Buy & Swap Arweave (AR)

A protocol that starts with burned libraries

Arweave's yellow paper, written by Sam Williams with Viktor Diordiiev, India Raybould, Lev Berman and Ivan Uemlianin, begins far from blockchains. It talks about fires at libraries and archives, book burning in authoritarian states, and news articles quietly edited after publication. Its point is that digital information feels permanent but isn't: almost every web page sits in a store controlled by one organisation, and it can be lost, altered or removed when that organisation changes its mind or runs out of money.

The answer it proposes is the permaweb: data, websites and apps stored on Arweave and reachable through ordinary web browsers. The goal isn't faster payments or cheaper trading. The goal is memory.^{[1], [2], [3]}

A chain woven, not just linked

Normal blockchains link each block to the one before it. Arweave links each block to two: the previous block, and a randomly chosen older one called the recall block. The paper calls the result a blockweave. To mine a new block, a miner has to prove it can access that recall block, which the paper calls Proof of Access.

It's a neat trick. Miners are paid to keep old data around because they never know which piece of history the next block will ask about. Storing more of the past improves their odds, so the network ends up rewarding the thing it was built to do.

A less-known design choice: no node is forced to store everything. The paper says each node can decide which blocks and transactions it keeps, so nobody has to host material they object to, while storage is guaranteed across the network as a whole rather than miner by miner.^{[5], [6], [7], [8]}

The two-hundred-year spreadsheet

The boldest part of Arweave is its pricing. The paper starts with an unglamorous number, the cost of storing one gigabyte on one hard drive for one hour, and points out that it has fallen by an average of 30.57 percent a year over 50 years. If storage keeps getting cheaper, the cost of keeping data forever becomes the sum of a shrinking series, which is a finite number you can pay up front.

That up-front payment flows into what the paper calls a storage endowment, designed so that the reward paid to miners in every block stays above the cost of maintaining the whole weave. The paper even estimates how much room is left for density improvements: at an optimistic 30 percent growth rate, it says, it would take 434 years to reach the theoretical limit of data density. It also names the principle behind the design: the goal is data permanence, not network permanence.^{[9], [10], [11]}

What the official sources leave open

UNKNOWN from official sources: the reviewed pages do not give a current figure for the size of the storage endowment or total data stored, and the yellow paper reviewed is marked as a draft.

The pricing model depends on storage continuing to get cheaper over very long periods. The paper sets out its reasoning for that assumption; whether it holds for centuries is something no source can confirm today.^{[12], [9]}