Author

Topic: Eligius: New payout method POLL (Read 6000 times)

hero member
Activity: 658
Merit: 500
June 27, 2011, 04:22:31 AM
#25
I don't understand if that method deals with pool hopping at all
if not, Multipool will be happy with this
The whole point of this discussion was that the old method is susceptible to pool-hopping, so yes, all the methods suggested above deal with it.
I still don't understand how, can't you solve easier shares and leave?
member
Activity: 98
Merit: 10
June 26, 2011, 08:16:42 AM
#24
Then comes as week of bad luck. None if any blocks are found. Shares pay out less and less. The rational thing would be for everyone to stay, but people just say they get more per share at other pools and leave until the pool has caught up. Hashrate goes down and it takes even longer to recover.
In the extreme, everyone would leave, leaving behind a pool where shares are worth less than expected value. Why would anyone start mining again?

As I understand it, account balance increases only after a block was found (correct me if I'm wrong).

So when a longer-than-average round takes place AND the pool has not enough savings to payout the full PPS after the block was finally found, this basically switches to proportional (with more than 50BTC total reward, because the pool still had some savings left from a lucky round) and everyone gets (their_shares/total_shares)*(50+pool_savings) BTC.  [NB: Even though this looks pool-hopping-vulnerable, it probably is not, because it only applies to unlucky rounds]
Now compare this to proportional, where miners get even less reward for unlucky rounds, because there are no pool_savings. And I think most proportional pools don't have any problems with too many miners leaving on unlucky rounds.

Because miners got underpaid in this round, the pool remembers that and later on, when a lucky round takes place, adds the missing BTC to the account balance.
full member
Activity: 168
Merit: 100
June 26, 2011, 07:54:50 AM
#23
Let me give an example.

Imagine everything was going well for a while and blocks were found according do expected number of shares. Everyone is happy, because everyone got the expected number of bitcoins. The pools does not have any coins held back or owes any. (Same here as if pool just started)

Then comes as week of bad luck. None if any blocks are found. Shares pay out less and less. The rational thing would be for everyone to stay, but people just say they get more per share at other pools and leave until the pool has caught up. Hashrate goes down and it takes even longer to recover.
In the extreme, everyone would leave, leaving behind a pool where shares are worth less than expected value. Why would anyone start mining again?

Think of it differently. There are two situations:
  • Pool has above average luck: just regular PPS
  • Pools has below average luck: Proportional with round starts when pool starts to go below expected number of blocks and round ends when pool again is above average number of blocks. (with early partial payouts). Pretty sure Luke will run into that, because that is how long share info would need to be stored.

Anyways ... look at the big pools and their payout methods. People seem to stick to what they understand without having to think much about it. Just look at deepbit's fees. People do not seem to mind.
newbie
Activity: 5
Merit: 0
June 26, 2011, 06:04:47 AM
#22
With proportional you have a 'clean slate' after each round.
With shared MaxPPS you potentially accumulate a big 'debt' across many rounds. Which, of course, would average out eventually, but people may not want to stick around that long.
That's a misconception you have. With either proportional (current method) or SMPPS, the pool will not pay out more than it has actually earned. The difference is davka on shorter rounds, where the proportional method pays out immediately, whereas the SMPPS holds some of the funds until longer rounds. This is precisely how it deals with pool-hopping, BTW.

So, if there is any 'debt' in comparison with the current method, it's actually only accumulated on shorter rounds, and payed back on longer ones.

Hope this clarifies.
full member
Activity: 168
Merit: 100
June 26, 2011, 05:29:30 AM
#21
So you may potentially be setting yourself up for a dead pool if a long dry spell hits.

What am I missing?
That is true for the current proportional method, as well as for solo mining. And if it was not, then the pool operator would take the risk on himself, with such low fees, and then you would definitely be left without a pool..

With proportional you have a 'clean slate' after each round.
With shared MaxPPS you potentially accumulate a big 'debt' across many rounds. Which, of course, would average out eventually, but people may not want to stick around that long.
newbie
Activity: 5
Merit: 0
June 26, 2011, 12:48:45 AM
#20
So you may potentially be setting yourself up for a dead pool if a long dry spell hits.

What am I missing?
That is true for the current proportional method, as well as for solo mining. And if it was not, then the pool operator would take the risk on himself, with such low fees, and then you would definitely be left without a pool..

I don't understand if that method deals with pool hopping at all
if not, Multipool will be happy with this
The whole point of this discussion was that the old method is susceptible to pool-hopping, so yes, all the methods suggested above deal with it.
hero member
Activity: 658
Merit: 500
June 25, 2011, 11:23:40 PM
#19
Well, Shared Maximum Pay Per Share has obviously won by a landslide... now it's up to me to code it!
I don't understand if that method deals with pool hopping at all
if not, Multipool will be happy with this
newbie
Activity: 21
Merit: 0
June 25, 2011, 09:23:05 PM
#18
I can't seem to vote, but if I could, I'd vote for Shared Maximum Pay Per Share.
full member
Activity: 168
Merit: 100
June 25, 2011, 08:47:31 PM
#17
Maybe I do not understand MaxPPS correctly, but ...
what happens if there is a long dry spell?

The missing solved blocks will lead to the pool being short on funds.
Short funds means less than expected PPS.
If the current PPS goes too low, people will leave.
With less people mining it will take longer for things to average out.
The longer it takes, the more people leave.

So you may potentially be setting yourself up for a dead pool if a long dry spell hits.

What am I missing?
legendary
Activity: 2576
Merit: 1186
June 25, 2011, 05:09:05 PM
#16
Well, Shared Maximum Pay Per Share has obviously won by a landslide... now it's up to me to code it!
legendary
Activity: 2576
Merit: 1186
June 25, 2011, 10:08:08 AM
#15
I just wish I didn't have to wait until 1 BTC for payout. I moved to another pool because they will payout above 0.01 BTC. That works better for me because I'm doing low end mining just for fun. Having to wait for 1 BTC is like having to wait for $100 on Google Adsense. Lots of small ops never get there.
The new version should support requesting a payout with signmessage.
newbie
Activity: 42
Merit: 0
June 25, 2011, 09:08:22 AM
#14
Not that the money matters so much but it's discouraging to know you'll never get it.

You should get it after one week of inactivity. You might just use two different payout addresses and change between them every eight days!
hero member
Activity: 784
Merit: 1009
firstbits:1MinerQ
June 25, 2011, 08:22:11 AM
#13
I'm not worried about whatever method you choose.

I just wish I didn't have to wait until 1 BTC for payout. I moved to another pool because they will payout above 0.10 BTC. That works better for me because I'm doing low end mining just for fun. Having to wait for 1 BTC is like having to wait for $100 on Google Adsense. Lots of small ops never get there.

Not that the money matters so much but it's discouraging to know you'll never get it.
legendary
Activity: 2576
Merit: 1186
June 24, 2011, 07:35:03 PM
#12
What's the drawback of geometric? It seems completely reasonable
For me, the main drawbacks are that I can't understand it, and if I simply implement it as-is, Python errors because of insane exponents.
hero member
Activity: 658
Merit: 500
June 24, 2011, 06:49:32 PM
#11
What's the drawback of geometric? It seems completely reasonable
full member
Activity: 123
Merit: 100
June 24, 2011, 03:50:55 PM
#10
  • Slush Score -- rejected because arguably copyrighted, and unfair to legitimate miners

Also, still vulnerable somewhat to pool-hopping.
legendary
Activity: 2576
Merit: 1186
June 24, 2011, 03:08:17 PM
#9
Why does the poll have fewer options than are presented on that Payout Methods screen?
  • Slush Score -- rejected because arguably copyrighted, and unfair to legitimate miners
  • Daily Proportional -- rejected because it doesn't solve the problem (just changes the hopping algorithm)
  • Proportional -- rejected because it has the problem
  • Pay Per Share -- rejected because I'm not willing to run a pool at negative income
newbie
Activity: 17
Merit: 0
June 24, 2011, 03:04:25 PM
#8
Why does the poll have fewer options than are presented on that Payout Methods screen?  I've already voted but I'm just curious.

I'm in the same boat as you.
newbie
Activity: 10
Merit: 0
June 24, 2011, 02:32:23 PM
#7
Why does the poll have fewer options than are presented on that Payout Methods screen?  I've already voted but I'm just curious.
newbie
Activity: 70
Merit: 0
June 24, 2011, 02:24:05 PM
#6
Btw we are in a very unlucky round Tongue 18 hours and still no block. Sad
The pool was down for about 4 hours, but still it is an unlucky round.
legendary
Activity: 2576
Merit: 1186
June 24, 2011, 01:43:04 PM
#5
The problem with Shared Maximum Pay Per Share is that in short lucky runds you get payed per share , instead if you get a long rounds streak there is the possibility that the credit isn't enough to pay all the shares and it switches to proportional method !!
This goes in favour to the pool hopping cheat!!
1. Pool hopping isn't cheating.
2. SMPPS never switches to proportional.
newbie
Activity: 27
Merit: 0
June 24, 2011, 12:18:21 PM
#4
The problem with Shared Maximum Pay Per Share is that in short lucky runds you get payed per share , instead if you get a long rounds streak there is the possibility that the credit isn't enough to pay all the shares and it switches to proportional method !!
This goes in favour to the pool hopping cheat!!
Btw we are in a very unlucky round Tongue 18 hours and still no block. Sad
newbie
Activity: 70
Merit: 0
June 24, 2011, 11:52:55 AM
#3
I would like to add Other: 100% goes to me. Grin
full member
Activity: 123
Merit: 100
June 24, 2011, 11:32:46 AM
#2
Shared MaxPPS looks really good. Very simple too, I like it.
legendary
Activity: 2576
Merit: 1186
June 24, 2011, 11:29:59 AM
#1
Ok, new difficulty is here way too soon... let's get this over with ASAP!

Please vote your preference(s) for Eligius's new payout method. Please only vote if you at least intend to use Eligius after the upgrade (assume your vote is chosen for payout method). Please only vote if you understand the payout methods, and shortcomings of each method. Note that you can vote for more than one option, if you want.

Notable misconceptions:
  • MaxPPS: While you do lose "credit" if you change addresses, you do not lose out on being paid for mining work you did.
  • MaxPPS: Both variables converge on average of once per day based on graphing real-world data, which means you can reasonably time an address change to not lose even any "credit"
  • PPLNS: If a block is shorter than diff*2 shares, it will go back and pay shares from the previous round(s) again

Summaries:
See also: Real-world example graphs for most payout methods
Jump to: