Just tested RBF - it already works much as I was suggesting it should, much 
different to what I saw described.


Regards,

Damian Williamson

________________________________
From: Damian Williamson <[email protected]>
Sent: Friday, 1 December 2017 5:40:10 PM
To: [email protected]
Subject: TLDR; summary


Thank you for the discussion in reply, it is a complicated concern.


I realize now with further consideration of the article that I originally 
posted, uncapped block sizes may introduce a problem. Another person posted a 
similar idea as a proposed BIP to the bitcoin-dev list yesterday. My 
realization was that there may be nothing to stop a rogue or greedy miner from 
scooping all of the transactions in the transaction pool into the next block, 
scooping all transaction fees for themselves. Then, with all transactions 
processing into the next block, there would be no need to pay fees. That would 
be bad for Bitcoin. However, the other person seemed to suggest that it would 
not be economical to scoop all transactions. Due to the extra computational 
bandwidth required their block would not likely be the first solved so, there 
would be an equilibrium point where you scoop some transactions to solve the 
block first and get paid. If that is correct, fully uncapped block sizes at 
miners choice would still not quite work as the next block would then always 
contain one transaction only to be the first solved for the least computational 
power. What I suggested was that each transaction has a likelihood of being 
included in the next block and, what I supposed is that all transactions whose 
lucky number comes up then get included. This would still provide a random 
distribution of solved blocks amongst all miners. How what I foresaw could be 
enforced amongst all mining software to stop block underskimming by rogues and 
just solving one transaction (or one or, two transactions less) for a block 
which would likely be an issue, I do not know. If there was a block minimum 
size, why would anybody risk expending more computational power with less 
likelihood of solving the block?


A slight divergence on the topic of bad. IMO the current RBF implementation is 
bad. If store merchants operate on instant approval, as they would with EFTPOS 
and as we have seen already in implementations of Bitcoin at POS (not waiting 
for any confirms) then, a dishonest Bitcoin user could redivert the BTC back to 
themselves using RBF before there were any confirms and after they had already 
left the store. A solution would be to only allow RBF to replace the fee and 
not the whole transaction (replace with a higher fee). Another possibility 
would be for POS implementations of Bitcoin to not approve transactions with 
RBF set until there are confirms. The first does not introduce any additional 
issues but limits the possible uses of the RBF feature. The second would likely 
be problematic at POS - what to do if the customer pays with RBF set, have them 
take a seat for three to thirty days? Not an issue for online merchants where 
it is simple to wait for confirms - is Bitcoin actually seen as a live-in-store 
retail solution?


Regards,

Damian Williamson
_______________________________________________
bitcoin-discuss mailing list
[email protected]
https://lists.linuxfoundation.org/mailman/listinfo/bitcoin-discuss

Reply via email to