In his post, Andresen writes that the new code will enable something he calls “smarter” fees which govern the length of time it takes for transactions to be confirmed on the blockchain. The idea is that the new code will determine transaction priority, and ensure all transaction confirmations are done so more efficiently.
Basically, floating fees will act as a kind of smart tool that determines which fee amount is appropriate for each transaction. These floating fees will also determine if the transactions’ priority is high enough that it should be confirmed free of charge, and still receive that confirmation in a short time frame. According to Andresen, there are dozens of needless complexities in Bitcoin’s transaction fee code that leads to inconsistent and time-consuming confirmation periods.
Currently, the Bitcoin Core code can lead to severe headaches for those who send large bitcoin transactions. As Andresen explained, the new code eliminates some of the hurdles that slowed down transactions in excess of 1,000 bytes in size.
“Instead of using hard-coded rules for what fees to pay, the [new] code observes how long transactions are taking to confirm and then uses that data to estimate the right fee to pay so the transaction confirms quickly – or decides that the transaction has a high enough priority to be sent for free but still confirm quickly.”
Andresen adds that the current situation for free, high-priority transactions is even worse, as the “high-priority” constant is far too low. This means that free transations typically take a very long time to confirm – something the new code will also address.
However, he notes that the possibility of paying small, fixes fees is out of the question, for anyone who was wondering. On the topic of future transaction fees, Andresen adds that he expects these to rise unless a “good solution” for optimizing the propagation of blocks across the network can be deployed. He says it’s likely that transaction volume will increase in future, and that miners are unlikely to include more transactions in their blocks until someone fixes the “bigger blocks take longer to broadcast” problem.