- vulnerabilities created by building features on top of tx relay (https://github.com/bitcoin/bitcoin/pull/33106#issuecomment-3155627414 )

- mempool policy 🤮

A better design would be to relay transactions over some other decentralized network and submit them to mining node mempools via an API. This moves the DOS risk to another layer of the stack where it can be better isolated and mitigated. This design protects the most essential functions of a bitcoin node: producing, relaying, and validating bitcoin blocks. It removes the social attack vector that is mempool policy by removing the need for policy entirely. I wonder if anyone is building out a truly decentralized network that would work for transaction relay... 🤔🤔🤔

Endogenous transaction fees introduce unnecessary complexity to trustless protocols built on top of bitcoin (layer 2s). See the sole remaining valid criticism of CTV: when you need to construct transactions ahead of time, endogenous fees ruin your day. This problem is handily solved with a robust and decentralized network capable of carrying transaction fee payments. I wonder if anyone is building out a truly decentralized network, perhaps including bearer asset digital tokens, that would work for exogenous transaction fees... 🤔🤔🤔

"But vnprc, without a global mempool state, how are we supposed to predict when a transaction is going to get mined?"

Great question, anon! Nodes would be better served to take their transactions from weak blocks. This is a much stronger signal than endogenous fees because it indicates which transactions have real hashrate behind them. You no longer need to look at a secondary indicator (fee rate) to estimate what you are primarily interested in (time to confirmation). In other words, weak blocks account for out of band transaction accelerators. Bonus: a robust weak block network is a necessary condition for a decentalized mining pool. I wonder if anyone is working on a mining pool concept using a blockchain or block dag of weak blocks... 🤔🤔🤔

The security budget problem is better framed as the mismatch of blockspace demand and blockspace supply. Demand is variable and, due to the 1MB block size limit hack introduced by Satoshi, block space supply is fixed. When demand is low (paper bitcoin summer) miners struggle to earn tx fees due to an oversupply of blockspace. When demand is high users are forced to pay exorbitant fees to get their transactions mined with any urgency. The economic problems this creates for miners are concealed by the block subsidy, which vastly overshadows miner profit from transaction fees. The subsidy is going away so the logic of a variable block size will become apparent in time. It's inevitable, really.

A block size adjustment algorithm akin to the difficulty adjustment algorithm would allow the block size to change over time to match demand. So when suitcoiners dominate trading with offchain receipts, blocks will shrink and miners can extract more profit from less transactions. Miner profits will revert to the mean in time. When the hottest new on-chain metaprotocol drops and tx fees go parabolic the block size will adjust upward and users can get some relief from astronomical fees. Again, miner earnings will revert to the norm over time.

Satoshi created the transaction relay network because there was not an alternative at the time. It was the simplest solution to get out the gate. But this solution necessitated a bunch of ugly hacks on top of transaction validation. We have been tip toeing around policy limitations ever since. If you have never attempted to build unusual bitcoin transactions before, then take it from me: policy sucks. It's terrible and everyone hates it.

Endogenous fees were a necessary condition for bitcoin to launch because before bitcoin there was no other way to transmit trustless digital monetary value. Today, we have alternative solutions that would better solve this problem.

Satoshi yolo'd the block size limit out there as a quick and dirty fix to a long term existential problem. He likely knew he wasn't going to be around for the long haul so he built a suboptimal but permanent solution.

The bitcoin network has evolved substantially since Satoshi vanished. We don't need to use the same architecture going forward. Continuing to do something just because that's how it was done in the past is often a terrible idea.

We need to reimagine what the bitcoin network should look like from first principles without blindly adhering to the legacy architecture that we inherited. To do otherwise is to invite failure. Bitcoin is too important to let it fail.