[9 TH] Bitparking Pool, DGM 0%,vardiff,stratum,Merge Mining

[juhakall]

I'll register a new account when the current block finishes. If I understood DGM well enough, I could lose more BTC abandoning the current round than I have in altcoins...

I'm under the impression that you should have no reason to think like that. In a fair payout method, all shares should have the same expected payout, and not depend on history at all. With DGM (and most other methods as well), the actual payout per share will have lots of variance, but it should average out after a long enough time, even over multiple accounts that are mining only occasionally.

DGM is really hard to understand, and unlike with PPLNS, I still haven't grasped it well enough to be 100% sure of the above. However, it would be a very lousy payout method if miners were forced to stay on the pool to not lose their expected payouts.

as I understand it abandoning a round at any other time then "very near the start" could infact drastically reduce your payout. Which is why I've argued so much about the switch. I'm off mmpool (for good I think) finally as a result of vircurex going down I don't really feel like recreating another dozen worker accounts to reset addresses.

I would really like to hear some proof that DGM indeed works that way. I don't feel 100% confident that it's fair in that case, but haven't seen any real proof that it's not, either. It might definitely feel like you are losing expected payout when abandoning a round, but unless you have proof that DGM works that way, I'm more inclined to believe that the opinion is just based on a misunderstanding.

The huge failure of DGM is that it's almost impossible to understand (I mean really understand, not just on the "capacitor is being charged" -level ) if you aren't a hardcore mathematician. I'm far from that level myself, and thus Bitparking doesn't feel as attractive anymore. Then again, even if you can understand a pool's payout method, it's another thing whether they have implemented it properly...

[Chemicalbro]

I created a new user last week. Watched one account go down and the other go up, pretty much exactly matching where the old account was first. Then the old account paid out for another 3.5 blocks or so at the ever decreasing rate of that "capacitor" effect.

[Lucko]

Yes!!!!!!!!!!!!!!!!! My first block on new pool. EDIT: I didn't find it but it was the first we found since I join.

And I don't understand what is hard to understand in DGM... But then again I'm relay good in mathematics... And yes if you create new account and start mining on it old and new will total in same earnings.

[Chemicalbro]

On that note, it'd be cool to have the number of blocks each user solved in the json. Sure it'd be zero for most, but neat.

[sgrunger]

Is it a bad idea to have multiple workers submitting shares under the same userID ?

[firefop]

I would really like to hear some proof that DGM indeed works that way. I don't feel 100% confident that it's fair in that case, but haven't seen any real proof that it's not, either. It might definitely feel like you are losing expected payout when abandoning a round, but unless you have proof that DGM works that way, I'm more inclined to believe that the opinion is just based on a misunderstanding.

The huge failure of DGM is that it's almost impossible to understand (I mean really understand, not just on the "capacitor is being charged" -level ) if you aren't a hardcore mathematician. I'm far from that level myself, and thus Bitparking doesn't feel as attractive anymore. Then again, even if you can understand a pool's payout method, it's another thing whether they have implemented it properly...

What can I say I've looked at the math.

The target for DGM is "same payouts as pps", like all 'hop-proof' methods it has the net effect of setting your maximum reward as a ceiling and then punishing the miner who doesn't mine continually or drops out of a round. It transfers most of the risk and variance from the pool ops to the miners.

If everything goes perfectly, you'll see about the same earnings you'd get on pps... when the pool has bad luck (or you do re:connection issues, low accepted shares etc) your earnings will drop off accordingly. Since it's waiting until confirmation of block generation and paying based on the last 5 rounds these losses will be obfuscated because they happen proportionally over the next 5 rounds. As I miner all you'll be able to notice is that your earnings seem to have decreased.

All I can say is, look at the equations  - they aren't that hard to figure out. DGM does exactly what it was designed to do - transfer risk taken by pool ops to miners instead.

[eleuthria]

The target for DGM is "same payouts as pps", like all 'hop-proof' methods it has the net effect of setting your maximum reward as a ceiling and then punishing the miner who doesn't mine continually or drops out of a round. It transfers most of the risk and variance from the pool ops to the miners.

PPLNS doesn't punish miners for disconnects/non-continuous mining, while still being hop proof.

[Ephebus]

After 10 minutes of thinking, my solution will be radical and easy: I'll re-create another account, and this time I will use my own wallets for every coin. And I would recommend this to anyone creating an account at this point!

Certainly the way to go. Stay in control of your own coins. Also, Virc, Bter, and maybe other exchanges will "lose" any deposits from a new block coinbase. (They don't count as deposits any newly minted coins without an Input). It's a bug on the exchange's end but it seems widespread. If bitparking ever started paying that way, your payments would vanish.

I did the same thing a while ago. I don't think doublec should change the way things work now. The main things that make Bitparking special are the site's simplicity, the merged alt coins mining and the pool's transparency. Pool hashrate had a steep drop when doublec changed the reward method from PPS to DGM, but that was probably a good thing and only temporary: pool hoppers were purged from the pool and a whole different kind of users are coming - miners who are more likely to remain loyal to the pool in the long term, which benefits everyone.

[firefop]

PPLNS doesn't punishes miners more for disconnects/non-continuous mining, while still being hop proof.

FTFY

[eleuthria]

PPLNS doesn't punishes miners more for disconnects/non-continuous mining, while still being hop proof.

FTFY

FTFY?  Do you understand PPLNS at all?  You get paid for each share based on blocks solved after it's submitted.  The value of the share has NOTHING to do with timing/consistency of share submissions.  It is entirely based on how many blocks the pool solves after the share is submitted.  There is no exponential/DGM scoring involved that makes shares worth less when you stop or when you start.

[m00min]

FTFY?

<woodenSpoon>
http://www.urbandictionary.com/define.php?term=FTFY
</woodenSpoon>

[juhakall]

Judging from the latest posts, it looks like forum messages are sometimes almost as hard to correctly understand as payment methods.  

[eleuthria]

FTFY?

<woodenSpoon>
http://www.urbandictionary.com/define.php?term=FTFY
</woodenSpoon>

I know what FTFY you is.  But when what is posted is inaccurate BS, it's hardly fixed .

[doublec]

Since it's waiting until confirmation of block generation and paying based on the last 5 rounds these losses will be obfuscated because they happen proportionally over the next 5 rounds. As I miner all you'll be able to notice is that your earnings seem to have decreased.

What is this 5 round thing you are writing about? Where in DGM do they mention 5 rounds? Note this point in the DGM article:

the expected payout per share is always the same no matter when it was submitted.

[doublec]

PPLNS doesn't punish miners for disconnects/non-continuous mining, while still being hop proof.

Nor does DGM. The Slush scoring method has got people thinking that "hop proof" means "must mine continuously".

[roy7]

Is it a bad idea to have multiple workers submitting shares under the same userID ?

Not at all.

[RogueLeader]

Nice pool, thanks for running.  I like the no nonsense website.

[firefop]

FTFY?  Do you understand PPLNS at all?  You get paid for each share based on blocks solved after it's submitted.  The value of the share has NOTHING to do with timing/consistency of share submissions.  It is entirely based on how many blocks the pool solves after the share is submitted.  There is no exponential/DGM scoring involved that makes shares worth less when you stop or when you start.

Reading comprehension is your friend.

PPLNS (while better than DGM) still punishes inconsistent miners on long rounds. It only counts the last N shares. You can claim it isn't related to timing or consistency of submitted shares... but anyone who understands how the payout method works knows you're incorrect. The saving grace on PPLNS is on short rounds you effectively get paid the same as an average round... But how often does a pool have multiple short block finds in a row?

[eleuthria]

FTFY?  Do you understand PPLNS at all?  You get paid for each share based on blocks solved after it's submitted.  The value of the share has NOTHING to do with timing/consistency of share submissions.  It is entirely based on how many blocks the pool solves after the share is submitted.  There is no exponential/DGM scoring involved that makes shares worth less when you stop or when you start.

Reading comprehension is your friend.

PPLNS (while better than DGM) still punishes inconsistent miners on long rounds. It only counts the last N shares. You can claim it isn't related to timing or consistency of submitted shares... but anyone who understands how the payout method works knows you're incorrect. The saving grace on PPLNS is on short rounds you effectively get paid the same as an average round... But how often does a pool have multiple short block finds in a row?

Your terminology is flawed.  Punish implies the method actively discriminates/penalizes the miner.  You're simply giving an example where a miner submits shares during bad luck then isn't mining during good luck.  That's variance, and is present under EVERY system except for PPS.  They could have the exact opposite, and only mine during good luck.  That's not rewarding them for inconsistency, that's just being on the positive side of variance.

PPLNS with bad settings could make it so they get -no- payment during bad luck, which could be considered punishing.  With proper settings, they will get some payment.  BTC Guild's setting for 'N' is ~25x difficulty.  There will never be a time where a share is not paid something.  The higher the N value, the less variance for inconsistent miners.

[roy7]

Depending on the setting for o in DGM, your shares also carry value across rounds like in PPLNS. It's just in PPLNS you have 100% credit until hitting the cap for N, then the value goes to 0. In DGM (depending on the settings) the value of the shares decay geometrically. Since the value of the share is set at the time the share is logged, whether you leave the pool or stay around has no effect on that share's value. Your earnings will vary based on short/long blocks with equal probability.