Gold collapsing. Bitcoin UP.

[Erdogan]

Some future day when the subsidy is near zero miners will refuse to mine a block unless it at pays for itself.

Interesting observation, never thought about it.

Only if there's enough transaction fees to be had from the mem-pool would miners power up. In fact they might even mine a different sha-coin that offers some fees in the meantime. But disregarding that for a moment, what would the effect be? Since the overall effective hashrate is lower, difficulty would drop at some point and then the network would find blocks faster when tx load is high, slower when it's low. Would security suffer? My first thought was: yes, but then again you have the option to fend off an attack by making some high-fee transactions and thereby releasing more hashpower.

I think the number one priority for a large miner is to keep his machinery going as smooth as possible at all times. It is too risky to stop the machines and wait for transactions. With too few transactions with fees, rather the aggregate mining power will suffer.

[shmadz]

your doom and gloom is tiresome.  here's the greatest conspiracy of all:  is Anonymint a paid NSA shill?

I didn't want to be the first to say it, but I increasingly believe that he does indeed have ulterior motives.

Like an infiltrator sent to sidetrack or co-opt a group from the inside. The way he can deftly deflect criticism while demanding citations and blathering on about slanderous comments, all the while pouring out links to various rabbit holes just to keep you of balance and waste your time, it's all a little too convenient.

Also the repeated use of the term "DEEP STATE" always in all capitals, over and over, it just reeks of mk-ultra style programming.

- - -
Anyhow, I can't be certain of any of that, but one thing I am certain of is that barring a white paper outlining his "super coin" I won't be wasting any more time on him.

[cypherdoc]

right now there are 675 tx/block on average.
each pay around 0.0001 BTC/tx in fees.
that equates to 675*0.0001= 0.0675BTC/block in tx fees.  that is miniscule and accts for why miners construct small blocks mainly to capture block rewards.

if the cap is removed, i figure that to be economically attractive for a miner to create a bloated block just for the sake of mining for extra fees they would have to be at least worth 50% of the current 25 BTC block reward or 12.5 BTC in fees, give or take.

that equates to an extra 12.5/0.0001= 125,000 tx's required per block.  where are those extra tx's going to come from?

answer:

1.  they won't magically appear until the Bitcoin economy grows itself to that point which will take several years.
2.  an attacking miner can try to manufacture them himself to torment small miners.  would he do this?  NO.  the creation of an extra 125,000 worth of tx's would cost time and energy to create, sign, broadcast and then hash the POW for all those extra tx's and for what purpose?  some fuzzy theory about driving unspecified small miners out of business as a result?  no, he will continue to mine small, efficient blocks according to the avg output of the rest of the miners mainly to chase the block rewards.  this dynamic will continue for many years.

here's how i answered theymos on Reddit:

[–]theymos 2 points 2 days ago*
The network needs to completely agree on this. Say that there's no limit, but most miners are producing 1 MB blocks for whatever reason. Then someone mines a 3 MB block. If half of the economy accepts blocks only up to 2 MB in size and half of the network accepts blocks up to 3 MB in size, then these two halves of the network will split forever and never again be able to transact with each other (as long as they don't change). This breaks the network.
There are various things that you might think of to solve this problem, and they've probably all been thought of before. The most common first-impression idea is that you'll have miners decide it. (In fact, you may have been implicitly assuming that this is how it would work, but currently even if 99% of miners mine too-large blocks, they will be ignored by the Bitcoin economy.) The problem with this is that miners nearly always have an incentive to create larger blocks because it gives them higher fees, but they don't have much incentive to be worried about the long-term consequences of larger blocks. You should not be prevented from running a full node (the only type of node with any real security) because a handful of miners in China decide to mine blocks that are too large for your bandwidth/hardware to handle.
permalinksaveparentreportgive goldreply

[–]cypherdoc2 1 point just now
the way i would answer this is where is the tripling of tx's and the fees required to mine a 3MB vs 1MB block going to come from?
if that miner is an attacker, would he manufacture his own fake tx's to create a 3MB block? no. it would be too expensive and risky for him to do so, not only from time and energy needed to create them and hash them into a valid POW but also in terms of foregone block rewards that he might miss from being orphaned and screwing around with the system.
permalinksaveparenteditdisable inbox repliesdeletereply

edit:  the only caveat to my argument, in the 3MB vs 1MB case, is if the bloating miner can pull enough unconfirmed tx's that have been constructed by ordinary users out of his current set in his mempool in order to create that 3MB vs 1MB block.   which is why i asked this:

What % of the unconfirmed TX set gets included into every block?

but it would still be a risk in terms of increased network latency with orphans and the foregone honest blocks.

[cypherdoc]

can they pull the $DJT out of the ditch yet again?  i doubt it.  looking weak with the divergence growing ever larger:

[Adrian-x]

Some future day when the subsidy is near zero miners will refuse to mine a block unless it at pays for itself.

Interesting observation, never thought about it.

Only if there's enough transaction fees to be had from the mem-pool would miners power up. In fact they might even mine a different sha-coin that offers some fees in the meantime. But disregarding that for a moment, what would the effect be? Since the overall effective hashrate is lower, difficulty would drop at some point and then the network would find blocks faster when tx load is high, slower when it's low. Would security suffer? My first thought was: yes, but then again you have the option to fend off an attack by making some high-fee transactions and thereby releasing more hashpower.

I think the number one priority for a large miner is to keep his machinery going as smooth as possible at all times. It is too risky to stop the machines and wait for transactions. With too few transactions with fees, rather the aggregate mining power will suffer.

No it is easy even now to keep your equipment running and adjust the hashing rate with software. The result is you are always on but only consume loads of electricity when mining at a high hash rate. Effectively you are always mining and only mine effectively when it's profitable.

[thezerg]

Some future day when the subsidy is near zero miners will refuse to mine a block unless it at pays for itself.

Interesting observation, never thought about it.

Only if there's enough transaction fees to be had from the mem-pool would miners power up. In fact they might even mine a different sha-coin that offers some fees in the meantime. But disregarding that for a moment, what would the effect be? Since the overall effective hashrate is lower, difficulty would drop at some point and then the network would find blocks faster when tx load is high, slower when it's low. Would security suffer? My first thought was: yes, but then again you have the option to fend off an attack by making some high-fee transactions and thereby releasing more hashpower.

I think the number one priority for a large miner is to keep his machinery going as smooth as possible at all times. It is too risky to stop the machines and wait for transactions. With too few transactions with fees, rather the aggregate mining power will suffer.

@molecular: yes you get the concept.  And yes hash rate would suffer.  However, the actual security may not.  Block subsidy is tiny so 51%ing the network can't be financially motivated.  Also, unlike today, imagine if 99% of presumably honest miners are turned off (on average).  If something starts to go wrong due to misbehaving miners people with stake in the success of the system (coinbase, bitpay, etc) would convince these miners to turn back on -- they could just pay them per ghash for example.  And you couldn't use your "paper" 51% to deny a txn.  Submitter just resubmits with a high txn fee and all the sudden you have 99 times more hash power mining your txn!

@Erdogan: Right now it makes no sense to turn off miners.  Miners today are high priced based at the lowest on NRE (non-recurring-engineering) costs and at the actual price is what the market will bear (i.e. how many BTC the miner is projected to produce).  But making another batch of ASICs puts the NRE at zero and the actual cost of production of a QFN32 packaged chip has got to be < a buck because you can buy microcontrollers in that package for $2 in volume.  So in a future where miners are priced a reasonable margin above actual chip fab costs, miners cost very little and electricity costs become paramount.

@cypherdoc & @anonymint:

Cypher, he's not full of shit talking about txns being orthogonal to blocks.  He's on the right track.  I mean I would not have used the term orthogonal but the real confusion here is that a reformulation needs to tweak the definition of what a block and txn is, so using those terms creates confusion and mental blocks.  Think about it this way:

Today every node needs to see the full information of every other part of the network.  This is inherently not scalable.  What if every mobile phone in the world received everybody's email and chats?  The idea is silly.

In fact, the ONLY way to scale is to limit the information propagation across the network.  But you need to preserve the essential security of having blocks record transactions, so I would have said that blocks and txns should be not-quite-entirely-orthogonal.

A long time ago I wrote that Satoshi's genius was to realize that disks have gotten so large that you could put the entire world's transactions on a single $100 drive.  But in fact he was wrong -- you might be able to store them all, but you can't send them all.  On the other hand, its quite possible that he was not trying to replace credit cards, only bank accounts and wire transfers... in which case my hunch is that Bitcoin at 20MB or 100MB blocks works fine.  And also it may not be ideal but its OK if full nodes exceed the home hobbyist's budget.  If it takes some $ to rent a small data center to become a full node we will still have "permissionless innovation" for small startup companies.  This is what really matters.

[NewLiberty]

Found an interesting game-theoretic idea on blocksize no one seems to have mentioned before:

http://www.reddit.com/r/Bitcoin/comments/35kxdu/mentor_monday_may_11_2015_ask_all_your_bitcoin/cr5fqz


It's a bit rough, but the core idea seems like it could have potential. It's kind of like a Nash equilibrium that disincentivizes miners from tormenting others by creating too-large blocks when the cap is removed.

I agree this has potential. It'd be a clean, permanent solution with nothing arbitrary about it. Sometimes we just don't see the simple things?

Too-large-blocks are already disincentivized by their slow propagation and resulting higher orphan risk. I read that many pools already use a "soft limit" well below 1 MB because of this. It would probably make most sense for a miner to calculate the cost of the orphan risk (needs some assumptions, I guess) and contrast that with the potential additional tx fees being earned and make tx inclusion decisions based on that calculation.

what will happen when block propagation will be O(1)?
(see: https://gist.github.com/gavinandresen/e20c3b5a1d4b97f79ac2)

The scenario was discussed at length.

The most obvious risk is that the larger pools could easily drive others out by pumping up the sizes creating more centralization.  Essentially a bandwidth attack.

[cypherdoc]

Found an interesting game-theoretic idea on blocksize no one seems to have mentioned before:

http://www.reddit.com/r/Bitcoin/comments/35kxdu/mentor_monday_may_11_2015_ask_all_your_bitcoin/cr5fqz


It's a bit rough, but the core idea seems like it could have potential. It's kind of like a Nash equilibrium that disincentivizes miners from tormenting others by creating too-large blocks when the cap is removed.

I agree this has potential. It'd be a clean, permanent solution with nothing arbitrary about it. Sometimes we just don't see the simple things?

Too-large-blocks are already disincentivized by their slow propagation and resulting higher orphan risk. I read that many pools already use a "soft limit" well below 1 MB because of this. It would probably make most sense for a miner to calculate the cost of the orphan risk (needs some assumptions, I guess) and contrast that with the potential additional tx fees being earned and make tx inclusion decisions based on that calculation.

what will happen when block propagation will be O(1)?
(see: https://gist.github.com/gavinandresen/e20c3b5a1d4b97f79ac2)

The scenario was discussed at length.

The most obvious risk is that the larger pools could easily drive others out by pumping up the sizes creating more centralization.  Essentially a bandwidth attack.

can you explain O(1) and O(1)²?

[cypherdoc]

@cypherdoc & @anonymint:

Cypher, he's not full of shit talking about txns being orthogonal to blocks.  He's on the right track.  I mean I would not have used the term orthogonal but the real confusion here is that a reformulation needs to tweak the definition of what a block and txn is, so using those terms creates confusion and mental blocks.  Think about it this way:

Today every node needs to see the full information of every other part of the network.  This is inherently not scalable.  What if every mobile phone in the world received everybody's email and chats?  The idea is silly.

In fact, the ONLY way to scale is to limit the information propagation across the network.  But you need to preserve the essential security of having blocks record transactions, so I would have said that blocks and txns should be not-quite-entirely-orthogonal.

A long time ago I wrote that Satoshi's genius was to realize that disks have gotten so large that you could put the entire world's transactions on a single $100 drive.  But in fact he was wrong -- you might be able to store them all, but you can't send them all.  On the other hand, its quite possible that he was not trying to replace credit cards, only bank accounts and wire transfers... in which case my hunch is that Bitcoin at 20MB or 100MB blocks works fine.  And also it may not be ideal but its OK if full nodes exceed the home hobbyist's budget.  If it takes some $ to rent a small data center to become a full node we will still have "permissionless innovation" for small startup companies.  This is what really matters.

what you claim sounds reasonable but i'm not entirely convinced given the growth in storage and bandwidth capacity over the last few years.  nonetheless, i was very excited by the pruning release.  that could and should solve the storage issue completely.  as for bandwidth, i'm hopeful for something along the lines of IBLT.  we'll see.

[thezerg]

answer:

2.  an attacking miner can try to manufacture them himself to torment small miners.  would he do this?  NO.  the creation of an extra 125,000 worth of tx's would cost time and energy to create, sign, broadcast and then hash the POW for all those extra tx's and for what purpose?  some fuzzy theory about driving unspecified small miners out of business as a result?  no, he will continue to mine small, efficient blocks according to the avg output of the rest of the miners mainly to chase the block rewards.  this dynamic will continue for many years.

I think that you are right in general.  But there is a small hole.  If a miner creates 125000 fake transactions he could not broadcast them, keeping them for the block he mines and therefore collecting his own txn fees.  Therefore the net loss is only the CPU effort to make these fake txns.  It would be better if he loses txn fees.

An elegant solution (probably not possible for Bitcoin due to politics) would be to require that half of the txn fees be paid to the address in the prior block.  Given honest transactions, presumably a miner would make on average what he's making today because the miner of block N pays to the block N-1 but receives from block N+1.  BUT the miner that holds fake transactions until he mines a block must pay half of those fees to some other miner.  

To stop abuse in the 2-in-a-row situation, maybe pay txn fee/2^(N-X) txn to block X (i.e. halve the fee and pay that to your block, halve it again and pay the prior block, halve it again paying block-2, etc)


I'm a fan of this kind of idea in general, because I think something similar can be used to discourage miners from hopping between alt-coins.


EDIT: On the other topic, I also don't think scalability is this huge doom and gloom issue for bitcoin, so long as we allow it to happen!  Worst case, when BTC is adopted worldwide it may not be usable for small payments, and you may need to rent a kick-ass cloud computer to run a full node.  But that's an effect of too much success... I wish we had that problem today! :-)  This is why I'm not in a huge rush to post my ideas on a scalable blockchain. [today I'm running a full node at home on an old junker laptop I bought for $25 on ebay.  We have a long way to go]

[cypherdoc]

I think that you are right in general.  But there is a small hole.  If a miner creates 125000 fake transactions he could not broadcast them, keeping them for the block he mines and therefore collecting his own txn fees.  Therefore the net loss is only the CPU effort to make these fake txns.  It would be better if he loses txn fees.

can you repackage this argument and clarify the point?

[thezerg]

I think that you are right in general.  But there is a small hole.  If a miner creates 125000 fake transactions he could not broadcast them, keeping them for the block he mines and therefore collecting his own txn fees.  Therefore the net loss is only the CPU effort to make these fake txns.  It would be better if he loses txn fees.

can you repackage this argument and clarify the point?

It would be really painful for a malicious miner if he had to pay TXN fees on those 125000 txns.  And he DOES have to do so because the txns would trigger the spam detector (actually I'm not sure if the spam detector is a rule-of-thumb or a requirement -- but they'd be a requirement if we used my txn-fee expands block idea). 

But unfortunately the miner will pay the transaction fees to himself, so no loss.

By requiring miners to pay half of the txn fees to the prior block a malicious miner loses half of his txn fees...

[cypherdoc]

I think that you are right in general.  But there is a small hole.  If a miner creates 125000 fake transactions he could not broadcast them, keeping them for the block he mines and therefore collecting his own txn fees.  Therefore the net loss is only the CPU effort to make these fake txns.  It would be better if he loses txn fees.

can you repackage this argument and clarify the point?

It would be really painful for a malicious miner if he had to pay TXN fees on those 125000 txns.  And he DOES have to do so because the txns would trigger the spam detector (actually I'm not sure if the spam detector is a rule-of-thumb or a requirement -- but they'd be a requirement if we used my txn-fee expands block idea). 

But unfortunately the miner will pay the transaction fees to himself, so no loss.

By requiring miners to pay half of the txn fees to the prior block a malicious miner loses half of his txn fees...

thanks.  now i understand your point.

but the attacking miner still won't do this b/c it is highly expensive (CPU wise) to have to construct all that garbage, broadcast them (full nodes won't validate/relay the blocks if the fake 125K tx's are not in their unconfirmed tx set), hash them into a Merkle tree, & POW the bloated block.  and the risk of orphaning is way high.

[thezerg]

I think that you are right in general.  But there is a small hole.  If a miner creates 125000 fake transactions he could not broadcast them, keeping them for the block he mines and therefore collecting his own txn fees.  Therefore the net loss is only the CPU effort to make these fake txns.  It would be better if he loses txn fees.

can you repackage this argument and clarify the point?

It would be really painful for a malicious miner if he had to pay TXN fees on those 125000 txns.  And he DOES have to do so because the txns would trigger the spam detector (actually I'm not sure if the spam detector is a rule-of-thumb or a requirement -- but they'd be a requirement if we used my txn-fee expands block idea). 

But unfortunately the miner will pay the transaction fees to himself, so no loss.

By requiring miners to pay half of the txn fees to the prior block a malicious miner loses half of his txn fees...

thanks.  now i understand your point.

but the attacking miner still won't do this b/c it is highly expensive (CPU wise) to have to construct all that garbage, broadcast them (full nodes won't validate/relay the blocks if the fake 125K tx's are not in their unconfirmed tx set), hash them into a Merkle tree, & POW the bloated block.  and the risk of orphaning is way high.

yes, the block subsidy and relay delay basically makes all these shenanigans unprofitable.  But doing my suggestion really closes the hole, and closes for the future when the block subsidy is low.  It closes a whole class of loopholes we haven't yet considered. 

I mean, txn fees are the basic way to discourage spam generated for any reason (by definition its not spam if someone is willing to pay the going rate for it).  The design relies on that and it relies on the idea that although the real cost of a block is socialized (ie borne by all full nodes), its ok for the subsidy and txn fees to just benefit a single miner.  This idea is that the cost to the spammer will keep spam down so all nodes benefit.  But that does not happen in the case where a miner is generating spam blocks.

[TPTB_need_war]

Cypher, he's not full of shit talking about txns being orthogonal to blocks.  He's on the right track.  I mean I would not have used the term orthogonal but the real confusion here is that a reformulation needs to tweak the definition of what a block and txn is, so using those terms creates confusion and mental blocks.  Think about it this way:

Today every node needs to see the full information of every other part of the network.  This is inherently not scalable.  What if every mobile phone in the world received everybody's email and chats?  The idea is silly.

In fact, the ONLY way to scale is to limit the information propagation across the network.  But you need to preserve the essential security of having blocks record transactions, so I would have said that blocks and txns should be not-quite-entirely-orthogonal.

A long time ago I wrote that Satoshi's genius was to realize that disks have gotten so large that you could put the entire world's transactions on a single $100 drive.  But in fact he was wrong -- you might be able to store them all, but you can't send them all.  On the other hand, its quite possible that he was not trying to replace credit cards, only bank accounts and wire transfers... in which case my hunch is that Bitcoin at 20MB or 100MB blocks works fine.  And also it may not be ideal but its OK if full nodes exceed the home hobbyist's budget.  If it takes some $ to rent a small data center to become a full node we will still have "permissionless innovation" for small startup companies.  This is what really matters.

Excellent. So now if an altcoin is released with this direction, then no one will know if it was me or not, because it is clear that others visualize the concept.

In fact, I was not even the first person thinking along this direction. At least bytemaster and the author of Decrits were doing delegation in their designs.

[Peter R]

Found an interesting game-theoretic idea on blocksize no one seems to have mentioned before:

http://www.reddit.com/r/Bitcoin/comments/35kxdu/mentor_monday_may_11_2015_ask_all_your_bitcoin/cr5fqz


It's a bit rough, but the core idea seems like it could have potential. It's kind of like a Nash equilibrium that disincentivizes miners from tormenting others by creating too-large blocks when the cap is removed.

Yes, and this is why I was in favour of Gavin's original 20MB + 40%/year growth.  Many people were concerned about the exponential growth of the blocksize limit under this proposal, but in my mind it was just a sensible way to slowly transition from a hard limit to this "Nash equilibrium" limit.  Whether 40% was the *perfect* number to match BW and storage growth was never really that important in my mind.

Let's allow the network to grow!

[rocks]

The age old model is that when the banks get out of control, a few fail (or a few hundred) and it resets as the survivors scoop up the remains.  Now with the use of state power, they have taxpayer funded bailouts, bail-ins and all sorts of new ways to destroy wealth.

It was the emergence of the "Too Big To Fail" policy which created this and was effectively codified into law with the LTCM bailout in the late 1990s.

To Big To Fail = disabling the clearing mechanism which is necessary for a functioning capitalist society. If you are too big to fail, then you can do anything and be as inefficient as possible, but that inefficiency will never be cleared out of the market.

The creation of the FED in 1913 was essentially to formation of a too big to fail policy at the government level (before the US gov would have to go for bailouts itself). But as you pointed out banks and other industry were still allowed to fail. This stopped in the 1990s and TBTF was extended to corporations, which is why every corporate entity from banks to auto manufactures (i.e. GM/Chrysler) have tried to position themselves as too big to fail.

This will continue until the dollar fails, effectively destroying the too big to fail enabler.

Perfectly stated.

TBTF has also been added on the other side of the dollar though with the IMF SDRs (XDR), so we get to break the world now, not just the United States.  
Here's how:
As you noted, LTCM was the catalyst for the smaller side TBTF.  Since this happened after we already had the fall of USSR from bond failures, the USA certainly has had a similar risk.  XDR are still a small percentage of FOREX trade, but as the USD gets closer to a potential fail point we should see that percentage grow.  

Jim Rickards postulates that the US might do a gold bail-in and initiate a new gold standard, but I suspect that would be a last resort after the XDR TBTF pops.  He details the process the US could use to accomplish this quite well at the end of his last book.  Its feasible.

Bitcoin could also potentially fill that spot if investment continues to advance.

Foreign governments are draining their gold out of the FED and back to their home countries, it's dropped below 6,000 tons for the first time since WWII. Coincidence? I think not.

Also notice that the last set of withdraws starting in 2007 before the big crash.

[cypherdoc]

EDIT: On the other topic, I also don't think scalability is this huge doom and gloom issue for bitcoin, so long as we allow it to happen!  Worst case, when BTC is adopted worldwide it may not be usable for small payments, and you may need to rent a kick-ass cloud computer to run a full node.  But that's an effect of too much success... I wish we had that problem today! :-)  This is why I'm not in a huge rush to post my ideas on a scalable blockchain. [today I'm running a full node at home on an old junker laptop I bought for $25 on ebay.  We have a long way to go]

precisely.

[cypherdoc]

EDIT: On the other topic, I also don't think scalability is this huge doom and gloom issue for bitcoin, so long as we allow it to happen!  Worst case, when BTC is adopted worldwide it may not be usable for small payments, and you may need to rent a kick-ass cloud computer to run a full node.  But that's an effect of too much success... I wish we had that problem today! :-)  This is why I'm not in a huge rush to post my ideas on a scalable blockchain. [today I'm running a full node at home on an old junker laptop I bought for $25 on ebay.  We have a long way to go]

precisely.

another way of saying it is "you can't have a universal money if it's not universal".

[Zangelbert Bingledack]

Yes, and this is why I was in favour of Gavin's original 20MB + 40%/year growth.  Many people were concerned about the exponential growth of the blocksize limit under this proposal, but in my mind it was just a sensible way to slowly transition from a hard limit to this "Nash equilibrium" limit.  Whether 40% was the *perfect* number to match BW and storage growth was never really that important in my mind.

Let's allow the network to grow!

So... we just need a great visualization of this. If an idea is good enough AND people can understand it easily and quickly, it spreads faster than others and takes over.