Topic: [1500 TH] p2pool: Decentralized, DoS-resistant, Hop-Proof pool - page 732. (Read 2591916 times)

I'm glad you volunteered to write the documentation, now get cracking, I expect it done by the end of the week.

Lulz yes I know that 2nd paragraph was directed at me
I guess using about 6 times the blockchain space compared to DeepBit when mining is nickpicking for some
Also note that it means that all that 6 times is in a single transaction that needs to get around the network as fast as possible - rather than multiple transactions queued up before the blocks actually appear - hmm I wonder how that affects latency on p2pool blocks? Have you thought about that? I don't know because ... my next paragraph ... yeah ignorance is bliss for you?

It actually relates directly to an issue that I've brought up a few times already.
Lack of technical documentation.
It's all well and good to run a pool and tell everyone to connect to it with a miner, but when that pool is software you must run yourself and thus there are many issues (not all of them the same of course) that other pools spend a lot of effort dealing with to help with performance - that you the miner must consider yourself, then a lack of documentation showing the technical details and the issues and how they are dealt with is certainly necessary long before bitcoin should risk destruction by having everyone switch to p2pool (well as the software stands at the moment the network would crumble and die and that would be a BIG problem for bitcoin in general if everyone switched to p2pool without being able to switch back when the network dies)

Stability is the next issue.
I've no idea why people in here were telling those with issues to run RC releases of bitcoind to solve their p2pool problems.
It seems the answer was to simply get the latest stable 5 release of bitcoind - as it should have been.
If p2pool suddenly has a dependency on a non-stable release of bitcoind then that is indeed a major p2pool stability issue.
What vetting process is there when changes go into p2pool?

But anyway - regarding that first paragraph - hard drive technology will obviously not be the first issue in the near future - RAM will be an issue long before HDD space is.
Think of the obvious, 2TB HDD's are cheap and easy to get - that's over 1000 times the current blockchain size (well closer to 2000 but anyway)
So then think of that in terms of how much memory bitcoind will need to be responsive compared to how much it uses now - which with p2pool becomes an issue since p2pool needs bitcoind running also ...
Obviously to be able to run bitcoind you will need a lot of RAM compared to now if the number of transactions on the network should ever explode ...

Sorry if I wasn't clear on my previous post.  What I meant was: If there are ways to not have to use the whole block chain for everyday transactions, then it shouldn't matter how big a block is nor how big the block chain itself is.

Based on your post and Holliday's I have some research to do into these light/alternative clients!  IMO, if the majority of users won't need the whole block chain, then we are moving in the right direction.

  so we are working...

Panda Mouse

ragnard there are lite clients and there are ways to prune the blockchain of dead end transactions.  Still saying don't worry about the size of the block worry about the blockchain is kinda silly.  The blockchain is the sum of all the blocks.  Larger blocks = larger blockchain.  On a large enough timeline most users won't use the blockchain.  Miners, merchants, banks, wallet services will form the blockchain network with a variety of light and indirect clients/wallets.

It seems this may be getting a bit off-topic regarding P2Pool, but here goes.

I remember when I was first introduced to Bitcoin that a big selling point was that micro-transactions would be possible with little to no fee and how BTC is divisible down to, what is it now, 8 decimal places?  Anyway, if BTC goes to some astronomical value in fiat currency, I can see such transactions as a cup of coffee costing milibitcents.  This is exactly what we want to to encourage if Bitcoin is to succeed as the "Currency of the future!".

So, I would think that rather than worrying about the size of a block, focus should be on how to efficiently deal with such a large block chain. Right now, if I'm new to Bitcoin, downloading a 1Gig+ blockchain is a daunting thing.  There are still people on dialup and what about smartphones?  I've read there are ways to mitigate this but they aren't implemented yet or are in infancy still.

So, it seems to me, the ultimate goal of Bitcoin is to have as many transactions as possible, including the extremely small ones.  The protocol needs to be able to handle that if we are to be successful.  And, P2Pool miners with their multiple small payouts is a great way to test ideas and code to handle such a future.

I just wonder - what is the reason of 89.6% of happiness in such a long period (90 days) in our pool.
Is everything ok with the mining program?

Panda Mouse.

Network security is directly related to global miner revenue.  If global miner revenue falls by half we can expect network security to fall by half.  Nobody said quickly declining but the reality is bitcoin was ALWAYS intended to be supported by fees.  My opinion is my opinion only.  p2pool doesn't and can't enforce any fee policy.  Each miner is free to include or exclude any tx they see fit.  The best analogy is p2pool is solo mining with group rewards and that applies here.  tx selection is no different than solomining.

As far as tx "can't" make up any significant portion of miner's reward I disagree.  If tx volume doubles over the next year, after the block subsidy at an avg fee of 0.01 BTC fees would make up 4% of global miners rewards.  Now 4% isn't 50% or 100% but it is still significant.  It is a starting place.  Eventually fees will need to make up significantly higher % of global rewards.  Still if most miners don't enforce fees then most users won't increase fees and fees will remain low. 

No one person can FORCE fees upon the network.  The avg fee paid depends on the supply (aggregate of miners fee policy) and demand (how much users are willing to pay).  Some will say we don't need to worry about it and we likely don't but building robust protocols where clients are aware of miner's fee policies via broadcast system and can make fee recommendations won't happen overnight.  Bitcoin is an experiment and part of that experiment is how the fee marketplace is going to work.

I am sorry if my actions scare or concern you but:
a) p2pool has no control over me anymore then they have control over you
b) moving your hashing power from p2pool to deepbit wouldn't affect me or my fee policies and simply centralizes bitcoin for no benefit
c) My 15 GH/s represents about 0.15% of global hashing power.  My actions are simply a declaration of intent.  Unless a majority of miners (p2pool, solo, and major pools) start enacting fee polices (at various price points) it won't have any effect on the network.  As it should.

Thats exactly my point! It halves every 4 years only! We are still at the full blockreward (until end of 2012). There is no shortage of bitcoins. If there is any problem with mining, it surely is not that there are too little fees in the blocks! Does anyone seriously believe suddenly the fees would make any substantial part of per-block-earnings? I find it much more possible that tx will drop dramatically.


Thats what I was commenting on with

Ente

Hmm just in case you didn't realise (though I'm sure most people do by now with the number of times this has come up on the forum)
The block reward halves every 210,000 blocks (4 years)
So in 210,000-173,094 blocks (= 36,906 blocks = 256.3 days) it will drop to 25 BTC per block.
Then again 4 years later to 12.5 BTC etc ...
So that also means that Bitcoin is getting close to being 4 years old soon

+1 (except for the "join deepbit"-part)

p2pool cannot enforce fees. every miner in p2pool decides for himself what transaction he includes in his blocks.

I am strictly against enforcing fees at this point.
Too early, way way too early, we should focus on to get people knowing and joining bitcoin first. Enforcing fees, now, would give a few bitcents per block only, almost nothing. It would harm Bitcoin, because then its "just another obscure internet payment system where someone earns money with my transfers".

edit:
Are you seriously suggesting "blockreward is quickly and dramatically reducing, we have to act now to still get any reward on mining blocks!!1!"? Really?
/edit


I find the plan to have enforced fees more dangerous than the deepbit situation. And would abandon p2pool, and any other pool who enforces fees. If all other pools enforced fees I would rather join deepbit. If all pools enforce fees I would consider selling my hardware, buying Bitcoin, bury them in an offline wallet, and go along to other hobbies for the next few years.


Ente

I meant number of outputs in the coinbase.  I don't think it will be a problem, was just curious.

8640 ~= 290K if 8640 people mined and each got a block ... though of course unlikely.
However, that means if P2Pool ever actually had more than 2880 miners, some would certainly be be unhappy
(and at 2880 it could ~= 99K)

There is only one coinbase per block.  The number of outputs in the p2pool coinbase increases with the number of shares earned*, but the p2pool difficulty adjusts too, so the number of outputs won't grow without limit.  Also, there is a hard limit to the number of shares counted in p2pool.

* Actually, with the number of addresses to be paid.  There can be fewer addresses paid than shares, but never more.

Um - so that's assuming the number of people mining P2Pool will never increase from the number it is today.

Not a problem at all.  The # of coinbases is fixed.  As tx volume grows the coinbase as % of block size will continue to fall. 

How much of a problem is having a large number of small coinbase transactions?

If blockchain bloat becomes a problem maybe pools like p2pmining.com will gain more popularity.

It is already prioritized in the default bitcoin transaction selection algorithm.  When there are too many transactions to fit in a block, bitcoin will choose transactions with the highest priority and tx fee is one way a transaction can get a higher priority.

Followup of my above post with some DeepBit numbers:
so I asked DeepBit and he said that the auto payout feature is 1 payout per day (but 3 manual payouts are possible but most people use auto)

If we calculate based on the 1 payout per day, then since DeepBit is around 3500GH/s that's about 48 blocks a day so most people are getting at most 1 payment per 48 blocks so using 1/48 of the previous calculated ratio: P2Pool is using 6 (rounded down from 6.4 to compensate for some manual payouts) times as much blockchain space as DeepBit is using per miner.

Even if you make the wildly ridiculous claim that everyone on DeepBit is using manual payments 3 times a day - P2Pool is still using twice as much blockchain space as DeepBit is using per miner.