« first day (1728 days earlier)      last day (2572 days later) » 

01:22
@MaxVernon there is no 'the mempool'
@MaxVernon my node has > 1M transactions in it.
@eponymous fair enough. "My" node's mempool has over 250k.
I do know that each node is not guaranteed to have any particular transaction, but surely that is a very large discrepancy?
I.e. the 1,000,000 vs 250,00
01:55
@MaxVernon @eponymous I imagine the discrepancy would come from a difference in how long unconfirmed txs are stored in a node's mempool, as well as the option to ignore zero or low fee txs? Still seems like a large discrepancy though, are there other factors I'm not considering? I can't imagine there just happens to be 750k txs that one node just "didn't happen to hear about"
02:11
@chytrik some of these are 24 months old.
my point was that it's totally meaningless to shout out numbers like that.
the only thing that matters is how many people are paying a fee higher than you.
 
1 hour later…
03:20
@eponymous gotcha, that makes sense. Although in choosing a fee, while you wouldn't care how many txs have a lower fee rate than yours, you would care about how high the fee rate of the tx next in line after yours is. Obviously it is best for you to beat the next highest fee that is included in a block by as small a margin as possible. But once your tx is broadcast what you said holds true - the only thing that matters is how many txs are paying more than you.
 
14 hours later…
17:49
@eponymous - it's problematic when there are more than approximately 5 transactions per second, regardless of the actual fee paid by each transaction, since the capacity of the network is only around 5 transactions per second regardless of fee.
i.e. it's possible that some transactions will never get processed if there are consistently over 5 transactions per second happening, regardless of how high the fee is. I can imagine a state where a transaction might need to pay a fee of 50% of the value of the transaction just to have that transaction be processed at all. Of course, it's arguable that the network will have failed long before that.
and the mempool number I was quoting is based on a combination of observation of my own node, and looking at the number of outstanding transactions via blockchain.info
clearly if your node is not dropping "old" transactions, it will have more outstanding transactions in its own mempool.

« first day (1728 days earlier)      last day (2572 days later) »