Subspace Network 是Web3上支持大规模可扩展，同时提供存算一体化底层基础设施服务的模块化Layer-1，在此基础上，Subspace Network将致力于增强和连接其他区块链生态系统。
2018年-2019年：获美国国家科学基金会 Grant 资助以及 Candaq Fintech Group, OneBoat Capital 等的早期支持
2021年：发布 Whitepaper，获得 Web3 基金会 Grant，$4.5 million fund raising led by Hypersphere and Stratos Technologies（Seed Round）
2022年：$33 million fund raising led by Pantera Capital & CoinBase（Strategic Round）
以下为Subspace Network中英文对照版白皮书 (Ⅰ) 。
Subspace: A Solution to the Farmer’s Dilemma
Nakamoto-style blockchains, such as Bitcoin  and Ethereum , , combine the longest-chain fork-choice rule with a proof-of-work (PoW) mining puzzle. These systems are provably secure, with respect to safety and liveness, given an honest majority of miners.
Unlike legacy Byzantine Fault Tolerant (BFT) consensus algorithms, participation is both permissionless and scalable. These properties are the standard against which all new blockchain consensus protocols are measured.
Unfortunately, the security afforded by PoW comes at a massive cost in electricity. Collectively, miners on Bitcoin and Ethereum consume the energy budget of a medium-sized country, with these numbers steadily increasing as more capital flows into the system.
This raises the critical question of whether cryptocurrencies can reach wide scale adoption without adding more fuel to the fire of global warming. Moreover, while mining was originally envisioned as a democratic and egalitarian process, as expressed by one-CPU-one-vote, it quickly became a highly commoditized and centralized enterprise.
Today participation in Bitcoin mining instead follows one-ASIC-one-vote, assuming a miner also Research conducted with support from NSF-SBIR Grant 1844037 has access to low-cost electricity. Ethereum mining sought to circumvent this by adopting one-GPU-one-vote, but this too has proven susceptible to special purpose hardware and still has the tendency to concentrate in regions with low-cost electricity. This raises another key question of whether or not existing cryptocurrencies are actually decentralized, or if we have simply substituted one trusted third-party (financial institutions) for another (mining pools).
如今，比特币采矿的参与已经被“一ASIC（专用集成电路）一票”给取代，假设一个矿工也在NSF-SBIR Grant 1844037的支持下进行研究，可以获得低成本的电力。以太坊采矿试图通过采用“每GPU一票”来规避这一问题，但这也被证明容易受到特殊用途硬件的影响，并且产业仍然倾向于集中在电力成本较低的地区。这就提出了另一个关键问题，即现有的加密货币是否实际上是去中心化的，或者我们是否只是简单地将一个受信任的第三方（金融机构）替换为另一个（矿池）。
These challenges have served as a rallying cry for a di-verse group of hackers, researchers, and engineers who have sought to design a sustainable blockchain that holds true to Nakamoto’s vision for a more democratic and decentralized future. The most well-known solution to this problem is proof-of-stake (PoS), which employs a system of virtual mining based on one’s wealth, under the adage one-coin-one-vote. While PoS clearly solves the ustainability problem, it doesnot hold true to Nakamoto’s vision. It instead reflects a permissioned and plutocratic alternative, which also exhibits strong tendencies towards centralization.
上述这些挑战成为了一群不同的黑客、研究人员和工程师的战斗口号，他们试图设计一个可持续发展的区块链，以实现中本聪对更加民主和去中心化未来的愿景。这个问题最著名的解决方案是权益证明（PoS），它采用基于个人财富的虚拟采矿系统，俗称“一币一票”。虽然 PoS 清晰地解决了可持续性问题，但它并不符合中本聪最初的愿景。相反，它反映了一种获得许可的财阀式的替代方案，它也表现出强烈的集中化趋势。
In fact, PoS systems serve to magnify the existing wealth disparity in cryptocurrencies, which are already significantly larger than historically high disparities in global fiat wealth distribution, effectively serving to make the rich even richer. What is instead needed is a cryptographic proof system based on an underlying resource that is already massively distributed and which does not lend itself to special-purpose hardware.
Enter proof-of-capacity (PoC), which replaces compute-intensive mining with storage-intensive farming, under the maxim one-disk-one-vote. Disk-based consensus seems like an obvious choice, as storage hardware has long been commoditized, consumes negligible electricity, and exists in abundance across end-user devices. As it turns out, implementing a PoC such that it does not devolve back into PoW, without resorting to a permissioned model, is highly non-trivial, as witnessed by the paucity of live chains to date. Moreover, all existing PoC blockchain designs fail to address a critical mechanism design challenge, to which we turn next.
Observe that in any PoC blockchain a farmer is, by defifinition, incentivized to allocate as much of its scarce storage resources as possible towards consensus. Contrast this with the desire for all full nodes to reserve storage for maintaining both the current state and history of the blockchain.
These competing requirements pose a challenge to farmers: do they adhere to the desired behavior, retaining the state and history, or do they seek to maximize their own rewards, instead dedicating all available space towards consensus? When faced with this farmer’s dilemma rational farmers will always choose the latter, effectively becoming light clients, while degrading both the security and decentralization of the network.
This implies that any PoC blockchain would eventually consolidate into a single large farming pool, with even greater speed than has been previously observed with PoW and PoS chains.
Recall that in any Nakamoto-style blockchain, a new consensus node must synchronize the chain state from genesis, in order to be assured they are actually on the longest valid chain, which implies the availability of the chain history. If a large fraction of nodes stores the history, this data will be readily available, and the network may be considered decentralized.
However, as time goes by and the history grows, the storage burden on all full nodes grows as well, and some nodes may choose to prune the history, instead only storing the current state of the chain. This trend was already clear in the Bitcoin network as early as 2014 . If full nodes do not store the history, new nodes must instead rely on altruistic archival nodes or third-party data stores for initial synchronization, resulting in a more centralized network. In a PoC blockchain farmers have nothing to gain by storing the history, but clearly stand to lose out on block rewards, especially as the history grows, consuming a larger fraction of their available disk space.
In order to extend the longest valid chain and collect fees for valid transactions, a farmer must maintain the memoized state of the chain. As the state is often too large to reside in memory, it too must compete with consensus for precious disk space. While perhaps negligible for low-throughput UTXO style chains, state storage is signifificant for any EVM style chain, or any chain which seeks base layer scalability. Furthermore, all farmers are also required to compute the state transition for each new block as part of the ongoing verifification process, imposing a non-negligible computational overhead, which conflflicts with the desire for farming to be a lightweight task. The farmer’s dilemma then serves to exacerbate the well known verififier’s dilemma, by further raising the opportunity cost of verifification .
If a farmer is willing to adopt a weaker security model, they may instead join a trusted farming pool, whereby they delegate transaction verifification and block proposing functions to an operator, while the farmer focuses solely on evaluating the block challenge against their plots. This has the added benefifit of drastically reducing the computational overhead required to participate in consensus, which fifits with the ideal of many small farmers pledging unused disk space on their home computers. When a farmer fifinds a valid solution to the block challenge, they send it to the pool operator, who forges the new block in return for a portion of the block reward. As long as the fee is lower than the opportunity cost of local block production, a rational farmer would always choose to join a pool. In PoW blockchains this choice is largely dictated by a desire for a smoother reward function, since, unlike joining a farming pool, joining a mining pool does not increase one’s total rewards.
The chief problem with this model is that it is not decentralized. Although the actual consensus hardware is highly distributed, compared to existing PoW mining pools, the operators still present a point of centralization, more akin to validators in delegated or nominated PoS protocols. However, PoS systems at least provide strong penalties for misbehavior, which have worked in practice so far. As farmers in the pooled model are at best acting as light clients, the scope of action for malicious or colluding operators is much higher than in a typical blockchain. The honest majority farmer assumption becomes an honest majority operator assumption. If that assumption does not hold, farmers, and most users, will be unable to distinguish between valid and fraudulent transactions which appear in the longest chain, allowing operators to create coins out of thin air or spend farmer and user funds at will.
PoC blockchain design appears to stuck on the horns of a dilemma. On the one hand, we may abandon the goal of having farmers retain the history, while doing everything possible to minimize the burden of maintaining the state such that that the opportunity costs of running a full node remain negligible, giving farmers little incentive to pool. This leads to a much more limited construction, ruling out stateful smart contracts and even modest base layer scalability. On the other hand, we can abandon Nakamoto’s vision and accept pooled consensus as a necessary evil, as has largely been done within the PoW and PoS communities, while at least rejoicing in the fact that participation is now fair and sustainable.
In this work we present a third option, which circumvents the farmer’s dilemma without sacrifificing the security or decentralization of the network, organized as follows:
1) To prevent farmers from discarding the history, we construct a novel PoC consensus protocol based on proofs-of-storage of the history of the blockchain itself, in which each farmer stores as many provably-unique replicas of the chain history as their disk space allows.
2) To ensure the history remains available, farmers form a decentralized storage network, which allows the history to remain fully-recoverable, load-balanced, and effificiently-retrievable.
3) To relieve farmers of the burden of maintaining the state and preforming redundant computation, we apply the classic technique in distributed systems of decoupling consensus and computation. Farmers are then solely responsible for the ordering of transactions, while a separate class of executor nodes maintain the state and compute the transitions for each new block.
4) To ensure executors remain accountable for their actions, we employ a system of staked deposits, verififiable computation, and non-interactive fraud proofs.
For concreteness, we present this approach within the Ethereum model of a fully-programmable, account-based blockchain, which periodically commits to the state of all accounts within the block header, though we believe many of the proposed techniques could be applied more generally for any Nakamoto-style blockchain.