and the multi will leave as soon as the difficulty goes up
and if the difficulty goes down fast, they will just jump back on until it goes up again
but, I don't know enough to actually have an informed opinion, I am just guessing
it stops crazy spikes
so when profit = high
multipools jump on and get all the blocks
then doesn't retarget for 4 hours
KGW retargets every block
so it's more fair once the multipoolers leave
won't the multi just come back again?
and get all the next blocks?
I don't get how it works, really.
what does the targeting use to set the difficult of the next block?
if it's the current block, then it will still be hard for 1 block
then if it goes down
the next block will be low
and the multi will jump on that one
raising the next block, but leaving when that happens
so the multi will get all the easy blocks
and leave the hard blocks for everyone else.
again, just guessing how targeting works
also, how multis work
the only thing that would work is creating some kind of way to make it impossible to jump in and out of mining
if the point of mining is to make transactions work, then why not make the reward for mining the ability to do transactions
then you don't have to artificially make mining any harder than it needs to be.
still, I don't understand this, in effect.
I don't know enough about targeting vs. multipool tactics to know what the right answer is
I'll try to read about it, but I expect that every scheme will be susceptible to manipulation by big pools to some degree
If I was really optimistic I would hope that there's a serious mathematical analysis of these targeting methods complete with bounds on how effective various pool tactics might be
I do agree about moving to fixed block rewards, for what that's worth
fixed block reward, there are a couple of PRs so we're getting there
the difficulty retarget is the biggy
hey just catching up on everything here, I forgot about this to be honest... I actually did a little work on an alternative difficulty algorithm when I was playing with a test coin I made. It was an exponential moving average I think, but I bailed on it at some point
had concerns about floating point determinism because I'd recently run into a problem with that in a game I'm working on.
FWIW I think dogecoin has a strong enough network that 90% of the problem could be addressed by changing the diff retarget interval to a smaller number. like 10? 20?
Interesting. Netcoin recently switched to KGW and I see this morning that the coin appears to be broken. Not sure it's related though.
I think I would default to keeping the current simple algorithm with increased interval. The fancier the algorithm is, the more likely it has some weird dynamical behavior you never thought to check for.
and somebody smarter will come along and do some clever variational calculus thing to find an exploit that makes small miners sad
Can I be a wet blanket and ask if it would be better to just make the reward fixed and then re-evaluate whether the difficulty retargeting is still insufficient? Changing two behaviors--especially if we're going to come up with a new retargeting algorithm--seems fairly risky.
But I don't know if the current feeling in the community is that both things must be changed "right now."
At any rate, if I can help by running a testnet client or two, please post a message here and I'll be glad to.
Well, for the community it has to be done yesterday and we should also integrate ethereum, scrypt-n, scrypt-jane, KGW and whatnot
"yes please just cram all these features in all at once, I'm a developer and I know nothing could go wrong"
I agree though that changing two things at once is not very representative in the outcome. but random block rewards is just for the "block gambling" problem
yeah I definitely think random rewards should go away first
The multipool jumping problem is well known. I remember the Fedora community asking for help cause a Multipool jumped off them in mid diff cycle, leaving them with 5 times or so higher blocktimes.
Yeah mooncoin was stuck with multi-day confirm times not too long ago because of that too
although I think our base hash rate is high enough right now that we may not suffer as much from jumping