The Economic Solution to The Blockchain/Web3 Problem
The founder of one of the most well used applications of cryptography wrote a measured review of what is purported to be Web3, and smartly highlights how the important gateways and repositories into that space are mediated by centralized business.
https://moxie.org/2022/01/07/web3-first-impressions.html
The problem of centralization forming around crucial interactions with blockchain is fundamental and present in pretty much any traditionally designed chain, even Bitcoin, and the issues all boil down to economics.
Resource Allocation
The first thing to notice is that blockchains distribute funds, and how they distribute funds determines what that blockchain is proficient at. It's easy to start with Bitcoin: Bitcoin only rewards hashing, because hashing is what makes it expensive to rearrange blocks (the basis for attacking the network). If Bitcoin started naively rewarding nodes for spreading data around, those rewards would have to come from somewhere, and thus would naturally decrease security.
All Proof of Work and Proof of Stake systems operate under the premise that they should not move funds out of security, so they all have this basic problem with resource allocation. Chains which scale tend to make other compromises which almost always make the unpaid infrastructure of the chain more expensive to run, which is exactly the issue which leads to Moxie's complaint - more expensive infrastructure means centralized businesses offer it to users as a service; now your chain access is mediated by a Web2 company.
Transaction Fee Hoarding
The second thing to think about is transaction fees. A block producer earns the fees in a transaction, so when one party has access to transactions, they essentially have money; to share those transactions for free would be giving away money - the rational action to take is to keep them for your own block production or to sell them to block producers. There is no penalty for hoarding transactions (like the penalty for delaying block production in PoW or PoS) because the network doesn't compete to include transactions, it competes to add new blocks; you don't get left behind by ignoring transactions.
Centralized Middlemen
This combination of transactions being treated as second class combined with high data-throughput chains not having any of the money which funds PoS or PoW at its disposal for infrastructure (like Infura) means that the act of serving the actual users requires external funding (like Infura, which monetizes and tracks users) to pay for infrastructure which the chain cannot pay for, while also encouraging the hoarding, rather than spreading of transactions i.e. Infura's most profitable strategy is not to share Metamask transactions for free.
These middleware companies extract value but also condense the parts of the chain which connect users to it into centralized hot-spots which are anti-competitive and easy targets in comparison to a truly decentralized network. You won't really spot these issues with Bitcoin because Bitcoin is completely unwilling to make such compromises and thus remains at such a small scale of data-throughput that it avoids inflaming those sore spots.
Saito Fixes The Underlying Problems
The core solution in Saito revolves around treating transactions as the source of work, rather than hashing or staking. Doing this securely invokes some of the more difficult to understand mechanisms (though the end result is a chain which cannot be 51% attacked).
In Saito hoarding transactions is a losing strategy - the work which allows one to build a block comes from transaction fees, but the nodes which are rewarded are not block producers, instead they are the nodes which best route/share transactions (weighted by the fees the user paid). Basically you get a transaction from a user, it has your routing signature on it, and when the version of the transaction with your signature on it is included in a block you have a higher chance of getting paid. Any nodes which are second, third, fourth, etc. to get the transaction will get half the chance of reward as the node before it - so being closer to the actual user is better, and spending more time and resources to get a transaction into a block is worse.
When you build a dApp on Saito, rather than needing to rent out Infura nodes as a service and fund that centralized service with token ponzis, you instead get rewarded directly from the network based on the fact that you get first access to the transactions from the dApp you host. What mining or staking is to other chains, routing transaction data across the network is to Saito. Where Ethereum nodes are incentivized to hoard transactions, Saito nodes are incentivized to, as fast as possible, collectively arrange them all into the next block and publish.
Providing users with up to date block information is a required step for nodes to entice users to send transactions to them - as users need that information to interact properly with the chain. The beauty in infrastructure being paid for from consensus this way is that the normal properties of blockchain apply fully to all of these interactions. If a node has poor service you can, for no extra cost, send the same transaction to many nodes and force them to compete for it. All the fervor that goes towards mining and staking can now be directed at providing users with the right service such that they send their transactions are sent through the highest quality infrastructure nodes. The ability for new nodes to take advantage of poor service from others is easier than spinning up a miner or collecting stake.
In Saito you cannot build a business off of hoarding transactions like Infura can - if nodes outside your fleet can offer an improvement to transaction routing and you exclude them from participating and earning some share of reward, then other nodes will embrace them, and they will out-compete you. In networks where you are encouraged to hoard transactions you cannot have massive scale; this is only possible when the consensus forces you to share data and compete to do it as efficiently as possible - after all, the purpose of a distributed ledger is to distribute and converge on data. The previously centralized node providers now operate squarely within the rules of consensus. As soon as they start slipping up according to the standards of the blockchain (don't exclude, don't censor, don't tamper, don't delay, don't hoard), they lose money to those who don’t.
The kicker of all of it is that there is no 51% attack, primarily because there is no block orphaning (the basis for the attack), there is only transaction orphaning (which is the basis for forcing competition to include them). If Amazon decided it wanted to try and reverse a transaction, or perpetually earn 100% of the rewards (as you can do in a 51% attack) knowing it routed 90% of all Saito transactions, it would, attempting to do so, publicly burn money on-chain until it went broke. It is the same principle as a Bitcoin miner burning money attempting and failing to make a new longest chain, but with Saito the attacker burns money even if they can produce more than half the blocks - in fact as long as they do not have first access to 100% of all transactions they cannot sustain or profit from an attack. But that deserves another article.
Conclusion
Saito is difficult to explain in a short manner, partly because it solves very deep and nuanced problems in blockchain people have been trained to ignore. The techniques are simple enough to grok, but figuring out why certain choices are made will require very abstract thinking around economics and game theory.
Almost all blockchains do not pay for infrastructure, because it would take away from security. Saito found a way to align security and infrastructure such that you can pay for both at the same time - all while being non-inflationary (no question of if it remains secure after block rewards are gone, as they already are non-existent) and so secure that it has positive cost of attack even under malicious majority. Saito is a revolution in public goods.