Re: [Bitcoin-development] How small blocks make delibrate orphan mining costly

Re: [Bitcoin-development] How small blocks make delibrate orphan mining costly

[Gregory Maxwell]

On Sat, Feb 23, 2013 at 5:06 PM, Peter Todd <pete@petertodd.org> wrote:
> In the low-subsidy future fees will be the main source of income for
> miners. Thus in some circumstances large miners may even have a reason
> to delibrately try to mine a block that would orphan the current best
> block. A simple example would be what would happen if a 1000BTC fee tx
> was created, but more realistic examples would be just due to a large
> number of tx's with decent fees.

It's come up a number of times in the past that when there is no
subsidy we might expect slow convergence as miners try to orphan each
other instead in order to fee snipe.   What Peter pointed out here
that I had not previously considered and find interesting is was that
if there is a sufficient backlog (or nlocktime immature) of
transactions with fees beyond the maximum block size the incentive to
orphan blocks to take their fees is greatly reduced or eliminated.

[Bitcoin-development] How small blocks make delibrate orphan mining costly

[Peter Todd]

[-- Attachment #1: Type: text/plain, Size: 1465 bytes --]

In the low-subsidy future fees will be the main source of income for
miners. Thus in some circumstances large miners may even have a reason
to delibrately try to mine a block that would orphan the current best
block. A simple example would be what would happen if a 1000BTC fee tx
was created, but more realistic examples would be just due to a large
number of tx's with decent fees.

However, with limited block-sizes such a strategy runs into a problem at
a point: you can't fit more tx's into your block so you can't increase
the fees collected by it even if you wanted too. Best strategy will soon
be to accept it and move on.


The second thing that could help defeat that strategy is if clients use
nLockTime by default. Clients should always create their transactions
with nLockTime set such that only the next block can include the
transaction, or if the transaction isn't time sensitive, possibly even
farther in the future.

Remember that to get ahead, you need to mine two blocks, and with
nLockTime the first block could only gain the transactions in the block
it orphans, so any further transactions could only go in the second.
With limited blocksizes that creates even more pressure in that the
block becomes full.

I don't see any reason why nLockTime in this fashion would harm clients,
so I think it's a perfectly reasonable thing to do and provides some
nice benefits down the road.

-- 
'peter'[:-1]@petertodd.org

[-- Attachment #2: Digital signature --]
[-- Type: application/pgp-signature, Size: 490 bytes --]