[1200 TH] EMC: 0 Fee DGM. Anonymous PPS. US & EU servers. No Registration!

[cengique]

I wasn't in when the block was found in rounds 19 and 20, and hadn't been mining here much into 21. And Meni algo mentions nothing about time, only the decay of shares as the total shares grow. What you described is more like a PPLNS variant.

So I understood you wrong, so you were hopping *in* at the start of the round. In the algorithm there is no mention about time, but the number of shares do increase with time. Think about the 50 GH/s coming in every second. I thought that was obvious

Maybe it's not clear when was the start and end of the rounds? The web interface has been quite flaky about that IMHO. It doesn't show immediately when the block is found.

[1714125585]

1714125585

[1714125585]

1714125585

[1714125585]

1714125585

[Inaba]

I've sent a message to Meni to see if he has any advice.  I can't see anything wrong at the moment, but I'm sure I've screwed something up somewhere.

[Shevek]

I've sent a message to Meni to see if he has any advice.  I can't see anything wrong at the moment, but I'm sure I've screwed something up somewhere.

Somewhere has been discussed the possibility of having a long-fixed (say, 100,000 items) list of valid shares to divide reward with. At the beginning all shares are accepted. But when the list reaches its limits, new share is accepted only if its value is better than the worst one in the list, and this last is pissed off. Actually, this means fixing acceptance target (and so, assorted difficulty) to the worst share in the list.

If some hopper leaves the pool after scoring, say, 10,000 shares, It will be possible that resting miners will continue generating shares with better values than many of these 10,000. When the block is solved and the hopper come back to get the reward, then she would be surprised if only 3,000 shares count toward her income...

[Inaba]

Wouldn't that then reward luck as opposed to work?

[Shevek]

Wouldn't that then reward luck as opposed to work?

For miners with really low hashrate (compared to the global pool one), yes. Let's say, CPU-miners.

On the other side, this fixes also the minimum worth money to pay. For N=100,000 it is 50,000 satoshi. Perhaps 10,000 is a fairer limit (N=500,000).

Only a proposal...

[Inaba]

Looks like we found another block.  It is processing now - I am having trouble tracking down why it's taking so long to process, though it is processing. 

[Inaba]

The block processing is being picky...  I am working on it, but block stats and share display may not be right for a few hours. 

Nothing is being lost from the shares, etc... so anything you submit is still counted towards block 23 properly.  It's just a display issue at this point and share allocation.  The update is stopping halfway through for some unknown reason and I'm about to just chuck the current code and start from scratch, but that will take a few hours to complete.

[WebMonkey]

just don't forget the line of code:

WebMonkey_reward = WebMonkey_reward x 10

'preciate it.

=]

'monkey

[Meni Rosenfeld]

It would be nice to have Meni in here to comment (hint hint)

Well, if you insist .

If you stopped mining at share 700000 and the round lasted 1.9M your payout should be virtually 0 (and I mean something like 0.0000000001, give or take a few places), if it's not it definitely means there is a problem.

I found the bug which caused this and PMed Inaba about it.

[village.idiot]

So, does this mean if I can't mine 24/7, I don't get paid?

[Inaba]

Thanks Meni, it's much appreciated.  I am trying to beat this thing down with a stick but it's proving feisty.  With the bug, it was still essentially calculating proportional shares, I'll update when I have it fixed.

village.idiot: Not necessarily, it only means you won't get paid if you stop submitting shares a long time before the round ends.  If you submit shares near the end of the round you get paid full price on them. 

Per Meni's recommendation, I've also adjusted c to .01 instead of .00001, which will pay out for much older shares going forward, though it was largely irrelevant due to the bug anyway.  But now the decay rate won't be so fast once it's calculating properly.

[Reuef]

So, does this mean if I can't mine 24/7, I don't get paid?

To me it seems to mean, if you stop mining before the end of a block, you get less than if you did for an equal amount of shares.  This is to prevent someone from going in when the pool is fresh and mining with 60 gh/s and then leaving once it hits 300k shares, leaving the rest of the pool to solve the block while they have a large portion of the shares.

Their 300k shares would be worth way more if they mined at the tail end of the round, but since no one can predict when that would happen it is not likely anyone would do that.

This type of scoring gives the incentive not to pool hop, and has the unfortunate side effect of negatively impacting the part time miner as they will need to stay until the round finishes if they hope to be paid.  The part time miner would probably become a thing of the past if this scoring method takes root, but pools would probably stabilize as pool hopping becomes unprofitable.


Just my .02 BTC

[Meni Rosenfeld]

So, does this mean if I can't mine 24/7, I don't get paid?

Your average payout per share is the same no matter when you mine or in what pattern. But your variance will be higher for intermittent mining. For example, if you mine 1 hour per day, if you're lucky and a block is found at or near this hour you'll get a "jackpot" (reward much higher than the normal reward for mining an hour), but if not you'll get nothing.

This type of scoring gives the incentive not to pool hop, and has the unfortunate side effect of negatively impacting the part time miner as they will need to stay until the round finishes if they hope to be paid.  The part time miner would probably become a thing of the past if this scoring method takes root, but pools would probably stabilize as pool hopping becomes unprofitable.

Not really, part-time miners are welcome, it's only the variance that increases. The long-term average is the same. In particular, if you mine for some time and no block is found, there's no reason to stick around longer than you planned - that would be a version of the sunk cost fallacy.

PPLNS is an alternative hopping-proof scoring method which has less variance, but with its own disadvantages.

[village.idiot]

I'm still trying to get my rig set up right, and only run it when I'm home.
When I get everything right, I plan to take it to the office where it can run full-time.

[btcboston]

In the case of block 19, I was mining from the start of the round and then left at approximately share 2.5 million or somewhere near there.  I didn't rejoin the pool again during that block (which ended close to 5mil shares).

I saw that my reward was actually slightly higher than it would have been on proportional and figured that something must be wrong with the scoring, because I expected to receive almost nothing.

[Inaba]

Block 22 payout is incorrect in My Account and Block Stats.  I am going to reset block 22 and re-run the payout algorithm with the fixes from Meni.  This will result in several people (mostly people with usernames from A - C) getting a lower payout (because they've been paid a decaying rate X6 while I worked on the code) and other people getting a higher payout (because the decay rate should be correct).

I just wanted to let everyone know that what you see is not correct and will be corrected as soon as block 21 validates or I move the validation forward manually before we hit 120.  Once that happens I will completely reset block 22 and start the calculation over.

[WebMonkey]

perhaps the WebMonkey scoring system might be a better "fit"

instead of applying weight to shares and then devaluing said weight as the round goes on, a simpler, lighter approach:

based soley on time in round.

100% time spent mining in round = 100% of miner's estimated payout.
10% time spent mining in round = 10% of miner's estimated payout.

not only does this discourage people from leaving during the "1st half" of the round, but also discourages entering during the "2nd half" of the round.

now, i realise that is very simple (and very light) but a percent here or there could be tweaked to allow a tiny variance of miner reboot or even server outtage.
(round = round - outtage) BEFORE calculation of payout.

* WebMonkey crawls back into his way out of touch hermit hole and turns on the 80s pop music

'monkey

[Inaba]

The problem with that method is how do you count time in round? 

For instance, if I were to find a pool that did that, I would start a hopping algorithm to hop out after X many shares and have one miner submit 1 share every 5 minutes or so to keep me "active" in the pool and thus gain my hopping advantage without having to actually participate in the pool.

[cengique]

not only does this discourage people from leaving during the "1st half" of the round, but also discourages entering during the "2nd half" of the round.

I agree with you that it would be simpler, except that if you don't encourage people to stay in the pool, then there is nobody left to solve the block?? That's why I like Meni's method, it makes sure people stay there to until the block is solved.

[WebMonkey]

there is always a catch

=]

* WebMonkey writes a note to himself to stay out of the deep end of the pool