Eligius POLL: Reward system changes, and new ASIC-ready Eligius-Hu pool

[wizkid057]

(snip - Regarding PPLNS) With N = 8 x difficulty, every single share returned to Eligius within the last year would have been payed at least once. As far as I can tell, only Eligius-De was ever unlucky enough to have [any] blocks go longer then 8 x difficulty.

This is very good to note.

[1714233717]

1714233717

[1714233717]

1714233717

[1714233717]

1714233717

[torusJKL]

Whatever new model will be used, the old EC from the current system should be carried over, as TheSeven already pointed out.
Not doing this will punish the loyal miners and will damage the image of Eligius. Because who is to say that the model will not change again in the near future and will the loyal miners again loose their EC?
I for myself will consider this risk when evaluating the various pools next time.

[Luke-Jr]

Whatever new model will be used, the old EC from the current system should be carried over, as TheSeven already pointed out.
Not doing this will punish the loyal miners and will damage the image of Eligius. Because who is to say that the model will not change again in the near future and will the loyal miners again loose their EC?
I for myself will consider this risk when evaluating the various pools next time.

PPLNS doesn't have EC.

Nor does this fear of losing EC make much sense, rationally speaking:

• New miners are losing out pretty bad under the current SMPPS system, and PPLNS will be a welcome change to them.
• Old miners have already made more from SMPPS than other reward systems would have paid for the same luck, at the expense of these newer miners.

It seems the complaint more-or-less boils down to, old miners want new miners to continue subsidizing them in times of bad luck. I do realize this is probably not the intention of the old miners, but I think they (and everyone) need to carefully look at the logic behind this and realize that's how it works out in practice.

Also, please keep in mind that I am in the same boat. I have quite a bit of EC piled up myself, and am not very happy with recent luck trends or the direction we must take to move forward either.

[TheSeven]

Nor does this fear of losing EC make much sense, rationally speaking:

• New miners are losing out pretty bad under the current SMPPS system, and PPLNS will be a welcome change to them.
• Old miners have already made more from SMPPS than other reward systems would have paid for the same luck, at the expense of these newer miners.

It seems the complaint more-or-less boils down to, old miners want new miners to continue subsidizing them in times of bad luck. I do realize this is probably not the intention of the old miners, but I think they (and everyone) need to carefully look at the logic behind this and realize that's how it works out in practice.

Also, please keep in mind that I am in the same boat. I have quite a bit of EC piled up myself, and am not very happy with recent luck trends or the direction we must take to move forward either.

You got a point there. However this is not entirely correct:
- The group that profits most are the old miners who anticipated that this would happen and left like a month ago, after the pool was already deeply buried in EC, but still able to pay most of their EC until today.
- Those old miners who did stay at the pool might not have lost too much if you consider the pool's luck. But they still partially subsidized the former group, which shouldn't have happened.
- The new miners, who joined the pool after the EC period had started, and are mining until today, obviously got badly screwed. These are group that IMO deserves the EC payback most.

Exactly that is the reason why I prefer an ESMPPS model that ports over the old EC, but not considering old shares, but old users instead, so that the users who joined after the beginning of the EC period will be paid back first. While this model might not be appealing to new miners during the first couple of days, I think this model will provide sufficient incentive to stay at the pool for anyone who has EC because things are very likely to get better for them the longer they stay.

[Luke-Jr]

That's a good point (about the old miners who left), I'll have to ponder things more.

[kuzetsa]

((snip ...))

• New miners are losing out pretty bad under the current SMPPS system, and PPLNS will be a welcome change to them.

((...snip ))

Disagree. Don't put words in my mouth. PPLNS is NOT a welcome change to me.

As a new miner (having joined during the time of bad luck / much backpay owed) I would like to point out that this change (non-backpay / non-active-miner PPLNS) will not do as much for the current active miners in the pool as compared to an "active miner" variant of PPS.

[organofcorti]

((snip ...))

• New miners are losing out pretty bad under the current SMPPS system, and PPLNS will be a welcome change to them.

((...snip ))

Disagree. Don't put words in my mouth. PPLNS is NOT a welcome change to me.

As a new miner (having joined during the time of bad luck / much backpay owed) I would like to point out that this change (non-backpay / non-active-miner PPLNS) will not do as much for the current active miners in the pool as compared to an "active miner" variant of PPS.

It sounds like you're more concerned about extra credit rather than the reward method. Under PPLNS, you wouldn't have extra credit owed, so what Luke-Jr writes is correct - new miners would be better off under PPLNS.

Concerns about the extra credit being unpaid should be a completely separate matter to the new reward method. Don't confuse the two issues.

[Luke-Jr]

I am exploring other options as well. What do you all think of CPPSRB? With this system, the most recent <difficulty> shares get paid full PPS, but shares are only ever paid once. If your shares fall before <difficulty> in a long block, they get put in the "extra credit LIFO stack". So for example, if we have a long block followed by a short one, the long block might pay only the last 90% of the shares submitted, but then the short one will pay all from its round plus however much from the previous round it can. This is the system the BitPenny pool has used for months. CPPSRB can also be reconciled with existing extra credit by just throwing it on the stack initially.

[kuzetsa]

I'd just like to point something out. Regardless of anyone's opinion of "better" (PPS versus PPLNS, vs solo mining VS payout algorithm X) There have been votes... A few weeks ago, I thought that the issue of backpay was a moot point. People seemed to indicate that "active mining" variant is desired (for PPS)

Previously voted and discussed:

https://bitcointalk.org/index.php?topic=97282.0

• (( previous poll )) CPPSBAM is a form of PPS that everyone already voted for (works fine)
• (( previous poll )) SMPPS w/EC limited to active miners (( another payout that keeps Eligius-Ra running with owed backpay, without changing from PPS, though once again, only pays backpay to "active miners" ))
  
• the current vote: Support Ra AM /// 24 (22.2%)

Support exists for a modified PPS variant...

PPS which has extra credit enabled, and extra credit must first pay the active miners before anyone else's backpay.

Between the two votes, is there evidence of an overwhelming majority whom are STRONGLY OPPOSED to PPS with backpay? (active-miner variant or otherwise)

[SleeperUnit]

I am exploring other options as well. What do you all think of CPPSRB? ... CPPSRB can also be reconciled with existing extra credit by just throwing it on the stack initially.

OK, I'm not going to pretend I fully understand the subtle differences between RSMPPS and CPPSRB.  I do feel that either of these should make for an agreeable compromise between the current (demonstrably broken) SMPPS and any of the AM variants. As someone that plans to keep mining here long term, I would of course prefer to cut off the inactive leaches by going full AM, I would be content to switch to one of these "recent" versions.

Also, I thought one of the major reasons to change reward system to PPLNS was to get rid of the infinitely growing shares database. I don't see any of these PPS variants helping in this respect. Then again if keeping the shares database also means we keep the current stats, then that would be a nice plus, at least until we have a new stats system in place, tested and tweaked to work with whatever reward system is in this pool's future.

Whatever the final decision may be, I hope something can be done before even more miners switch to another pool.

[Luke-Jr]

Also, I thought one of the major reasons to change reward system to PPLNS was to get rid of the infinitely growing shares database. I don't see any of these PPS variants helping in this respect. Then again if keeping the shares database also means we keep the current stats, then that would be a nice plus, at least until we have a new stats system in place, tested and tweaked to work with whatever reward system is in this pool's future.

Well, the problem with the share db is that it's HUGE and just gets bigger - it even includes the proof-of-work for every share. The data needed to support CPPSRB is relatively small, and can be sparse-deleted for every block we find (when I upgrade the server kernel to 3.0+).

[sharp666]

I would like to start by thanking luke-jr, i have been mining with eligius for a while now, but not long enough to be one of the lucky ones (i'm around 90% work paid), and have felt well taken care of and respected by him.  i have been watching the last 2 polls, but because of forum restrictions i have not been posting. 

i think that people feel that we need to do active only (was always winning the polls) ec payouts soon, because at this point we are just digging a hole (i've been getting a lot of ec recently), and when asic comes out luke-jr would like to be on a new system.  the pools in the last week have even been going up in favor of the new system, and i feel that this could be just to get us away from mining for people who have left eligius.

[kuzetsa]

Yes. Exactly.

Luke: pretty please? Let's switch to some sort "active mining" PPS for now.

[Luke-Jr]

Luke: pretty please? Let's switch to some sort "active mining" PPS for now.

Someone have code for this?

[kuzetsa]

I already offered to reverse engineer and implement a new one. Please send it in a standards compliant way this time.

xz is fine, but why was it encoded without any headers?

seriously. what is this? it wasn't recognized by any tool I piped it into.

/Td6WFoAAATm1rRGAgAhARYAAAB0L+Wj4FpaGs5dABGIQkeKIzPDdw8z/VhnH14++zRWlqT74u2K
6d9qY+BVnYK2iw3W+YZ5vVEVEgJIIt7TEoOOq780Hbx2heEaWLCaNh/E5sEl21ZILx0kb3T5fBMr
Mibd6Ge9gg6QjnCmN39kQx+GmX6BWZEw2Ly274ERFLxqwbWVmhLM2Ja+l5RTgpEEw4bBiuj3Y4vC
QM2CVsY4hQVNK8+kpIHqKS0L0NkYdiCcmZKmpDxYM9bRtyrOEHlF/l/oxhFNNCgJgMXel7oQ5YLP
+8k6E5nO7xHUEVq4+4vKtQRfJCkXjUgEE5irab4s/+ufTmlFzJ1Xfo9V1ADKb/9fMsFivHinLDLZ
OkuHvjgOjNGpqXRD8lo/L2zq6lt8BiNuzxH0oSSwCxTLB6FDpTm7lbK6sfgC7JVpPahq7FvI1uoQ
0HmHZQEEuH3PIl/MQ6yKLIintBWXnHaaCZ3bB66urgvooPKkvgH4G1H3QUCSqvzsYC+6wi+E/Rwl
e5jONkzxcW (snip)

just like, email it or something. my PGP key is well known (feel free to check with the OTC database if you're not sure what my email address is)

[TheSeven]

Seriously... You can see what that is at the first glance. No idea where you got that base64 encoded blob from, but what you pasted above looks like this if you decode it:

Code:

#!/usr/bin/python
# -*- coding: utf-8 -*-
import fcntl
import hashlib
sha256 = hashlib.sha256
import json
import jsonrpc
import os
import psycopg2
import re
import socket
import sqlite3
import stat
import sys
import time

blkfiledir = 'stats/blocks'
magic_invalid_addr = '16hBeRE2CSDHcHS7CsRKGi8fiixhTg4oo'
magic_generate_keyhash = b'S\x99\xc3\t=1\xe4\xb0\xafK\xe1!]Y\xb8W\xb8a\xad]'

ignoredblocks = (
	'000000000000036050969046f7481e2675c530486daffaaa3a7b18fefeb174e0',
	'000000000000051b48b60613bebbb3168d5d7e1ef898b6efedd240e24d655a45',
	'000000000000070a8205504c865ca8755bf4

[kuzetsa]

what the... how'd you do that?

‎Monday, ‎July ‎23, ‎2012, ‏‎luke "sent" (posted a link to that encoded crap)

at me and I was unable to decode it.

what tool did you use?