Package aware Fee estimator post cluster mempool

Package aware Fee estimator post cluster mempool

Original Postby sdaftuar

Posted on: February 8, 2024 15:24 UTC

In the realm of fee estimation for transactions within a blockchain network, some challenges are being addressed.

One key observation is that neglecting certain transactions—specifically, those with unconfirmed parents—does not severely impact the effectiveness of fee estimation models. This is due to the sufficient number of independent transactions that can be analyzed to produce reliable fee estimates. The majority of these transactions tend to exist in isolation while in the mempool, without any associated unconfirmed children or parents.

However, there is a noteworthy concern regarding the misclassification of transactions' confirmation states based on their fee rates. The assumption that a transaction will always be confirmed because of its high fee rate is flawed, as this might not always hold true. This misconception could potentially lead to inaccuracies in the fee estimation process and needs to be corrected to ensure more precise predictions.

Moreover, the current understanding suggests that most confirmed transactions are solitary in the mempool, which might imply that improvements to tracking methodologies should primarily focus on excluding transactions with descendants in the mempool. Nevertheless, it's advised to verify if this understanding still stands, as it forms the basis for current optimization strategies.

Looking ahead, with the ongoing developments in package relay and validation, there's anticipation that the incidence of bundled transactions—or 'packages'—will increase. In light of these advancements, it becomes imperative to explore and develop a robust theoretical framework for fee estimation that accommodates the evolving dynamics of transaction packaging. This foresight is critical to equip the fee estimator with the capability to adapt to future changes in the transaction landscape.