Topic: Delay BlockReward=0 Forever (Read 1953 times)

I already answered this question but the short review is that, if it were the case that optimistically high BTC value made it so that .5 satoshi was indeed a mining incentive, by definition it would also be the case that transaction fees tolerable by users would be orders of magnitude beneath 1 satoshi. Ie insufficient mining incentive would only be a problem if it became a problem.

Currently the block reward and transaction fees stabilize each other.

Thanks again Moon, kjj, and retep.

WTF is this shit? Why is this an issue? You're really arguing about a hard fork for a few millionths of a BTC? Trivial much?!

By the time we get to mining fractions of a BTC per block, TX fees will more than likely offer enough incentive for miners to keep hashing.

It's very probable that future CPU's will never directly support integers greater than 64-bits. The width of a native integer is set by the width of a pointer, a memory address. 2^64 bytes of memory is simply an enormous number.(1) Now bus widths have even already been made larger than 64bits wide, but that's just how much data the CPU can move in one clock cycle, it has nothing to do with what numbers are actually supported.

Having said that for what Bitcoin does just using existing "BigNum" libraries is perfectly adequate, and Bitcoin already has one internally for a variety of purposes.

1) Modulo some pretty remarkable advances in computing of course... 2^64 atoms of carbon works out to be 0.3grams, so if we one day build computers that truely are built on a 3d atomic scale it's conceivable. (current computers are essentially 2d due to how silicon chips are built layer by layer) Building such a computer has a lot of problems though, for one, with something that dense, how do you get the heat out?

Technical quibble...  The radix is not in the middle of the 64 bit value.  The 64 bit value is just an integer, and it is scaled by 100,000,000.  100,000,000 is not a power of 2, so there is no clean break between the bitcoin part and the fractional part in the value integer.

log₂(21,000,000 * 100,000,000) is about 50.899, so we need 51 bits to represent the full range.  We could change the scaling factor with a hard fork, and version x transactions could have more room at the bottom end for precision, but doing so means less headroom.

Did I say headroom?  Yup.  We can do accounting beyond the limits of the money supply, just like we can estimate that the net worth of the US is into the hundreds of trillions while there exists nowhere near that many dollars.  In the same way, an accounting package can use 64 bit integers to give satoshi-precise figures up to a bit more than 2¹³ times the total bitcoin value.  This is still, sadly, a small number, only about 2 trillion dollars, but that is (hopefully) an accident of history, not a permanent thing.

I still think that we will switch up to 128 bit values to match future CPU-widths long before it makes any economic sense to do so.

While I can't verify my own statements, I was not ignoring your question.  It's just not one that can be easily answered by the vast majority of the forum membership.  Only the developers can respond to my claims, for or against, with any certainty; and most of them don't much participate in the forums.

But let me qualify that, while it's a bit more complicated than even I implied, that is basicly the solution posed by others in the past when this kind of issue has come up.  I'm fairly certain that it's all but decided that such an address type is to be developed, but it's just so far down the priority list (being decades away from need) that it's pretty much ignored.  For additional detail, let me describe how a transaction stores coin value into an address. 

Technnically, bitcoins don't actually exist, even as a digital construct.  This is a concept that tripped me up early on, but there is no text file, digital object, or portion of any transaction that we could point to and say "that's a bitcoin"; like we could open up a filesystem and highlight a file and say "that's an email" or "that's an MP3 song".  The concept of the coins themselves is entirely an abstract idea.  How it works is; the blockchain works like a huge, massively replicated, distributed LETS ledger containing only details of transactions from one or more addresses to another set of one or more addresses.  The varialbes that are used to contain the values that are transfered in these transactions are 64 bit integer values, which contain no decimal points.  As a matter of convention, Satoshi established that the clients should display those totaled values with a decimal point in the middle of those 64 bits (32 bits form the value shown to the left of the decimal point, 32 bits show the value to the right of the decimal point).  Said another way, the protocol has no concept of BTC's, but only deals in satoshis; but humans have a problem with dealing with numbers quite so huge, and this is why teh clients display the totaled values the way that they do, resulting in the (technically false) meme that there will only 21 million currency units.  However you choose to look at it, the bits of those interger variables with the greatest values will never be used, because the entire monetary base can be stored into a single 64 bit integer using only 48 bits, as the actuall 21 M BTC in binary would look like this...
(111,011,101,011,111,000,001,011,010,000,001,110,100,000,000,000,000)

So there are 16 bits (or so) to the big end available to place a binary token, that would tell the clients & miners that this address value isn't in the standard "whole" satoshis, but instead expressed as a more fine value, say with a decimal point in the middle (of the 48 bits in use) thus permitting the protocol to advance the coinbase division to another 24 bits of fractional satoshi units.

Only on a quantity of transactions basis. If half a satoshi were to hypothetically constitute a mining incentive, then fees would need to be 1/100000 a satoshi or less to be affordable. Again, I hope enough people want to pay fees and it all works out fine as much as all of you do.


In saying '30 years from now' you did not read what I wrote carefully enough. I was claiming that in >150 years we could start to experience such a thing...150 years before 2012 the world's layperson did not even know what economic growth was, let alone experience it during their lifetime, and 150 years before that even the concept itself would have been met with ridicule.


Thanks for your information...although I enjoyed the latter half of your post I think that the logic you used here actually implies that GDP is the 'right' metric. If the money supply remains unchanged, but twice as much mining equipment is produce (implying such equipment is half as expensive), then the same block reward should buy more mining equipment and the mining subsidy has increased (it is twice as easy to conduct mining as before). If the money supply doubles, and output remains unchanged, then it DOES impact the real value of $0.05.

Although I now see how you did your calculation, the block reward will not drop to sub-satoshi for another 130 years, and do you not think that the TRANSACTION VOLUME and MONEY SUPPLY will increase over the next 130 years? Is that increase not relevant to the calculation?


Now I would like to thank MoonShadow and Rudd-O for answering my question, and koin for his counterquestion which actually seemed interesting but that everyone ignored...if it really is true that...


...then I consider my question to be answered (although I would like to learn more about the "+" in "24+", and about how impossible it would be to use koin's idea).

That being said, I'm happy to continue the conversation as it shifts the thread inevitably from technical to economics.

I would also point out GDP is irrelevant for the discussion.  We are discussing money supply in inflation adjusted terms.   $0.05 (circa 2012).  If the money supply doubles it doesn't doesn't double the real value of $0.05.  So hypothetically if the money supply jumped 100x as much then 1 satoshi would be worth $5 "future dollars" but those $5 wouldn't buy more than $0.50 today ("circa 2012").

Looking at planetary GDP or planetary total wealth is the wrong metric.  Bitcoin isn't going to replace domestic production, Bitcoin isn't going to replace real estate.  At "best" Bitcoin would replace all planetary currency which is ~$100T (in USD equivelents).  $100T / 21M BTC = ~$4,761,905 per BTC.  1 saothis (1E-8 BTC) = ~$0.05.  


Under any smaller but still insanely optimistic scenario (say BTC money supply is larger than US Money Supply) a satoshi is worth significantly less and tx volume (and thus fees) are almost hard to imagine. Say BTC eventually had an "economy" larger than the US.  Many people look at GDP but that is the wrong measure.  What matters is TRANSACTION VOLUME and MONEY SUPPLY.  It is very difficult to find an estimate of transaction volume simply because it normally isn't a very useful stat.

Here is one cite:
http://www.clevelandfed.org/forefront/2012/winter/ff_2012_winter_13.cfm

Total annual transaction volume is 216 billion transaction and total transaction value is $1,119 trillion (not a typo), lets just say $1,200T to make round numbers.  US M3 money supply is ~$15T.   Each dollar (unit of measure) is exchanged 80x annually.

So let pretend Bitcoin economy someday was that large.

Money Supply   21M BTC = $15T USD (US M3)
Exchange Rate: 1 BTC = $714,285 USD
Transaction volume: 216 billion = 216 billion
Transaction value: 1.68 B BTC annually = $1,200 T USD annually

Per block:
4.11 million transactions with a value of 32,000 BTC ($22B USD equivelent)
At 1 satoshi per tx average fee (0.7 cents) miners fees per block would be 0.04 BTC ($29,357 USD).  

Annually miner gross revenue:
At 1 satoshi per tx average fee (0.7 cents) annual miner revenue would be (216B * 1E-8 BTC ea) =  2160 BTC ($1.54B).
I would point out that represents less than 1/800th of 1 percent of the total value transferred making it magnitudes more efficient than any other transaction system.


Note for the record I don't think that Bitcoin will ever become that large it just illustrates this if Bitcoin did become that large the block reward going to zero is an utterly meaningless concern.

17% annual GWP growth??  No wonder you're getting these crazy ideas with assumptions like that... 30 years from now, anyone could flip burgers and still be a multi-millionaire (inflation-adjusted) in a year!  Sounds good to me!

I think at the point where we start talking about block rewards measured in Satoshis the transaction volume will be sufficiently high that the block reward is going to be irrelevant to attracting miners, because the fees will greatly exceed any possible block reward.

What MoonShadow allows for is that what you propose might happen if the protocol is changed to allow it. It's very possible that could happen. Other changes to the protocol in the next century like a move away from ECDSA key pairs are going to be debated and potentially forced by sufficient advances in quantum computing (granted no good quantum resistant ECDSA alternative yet exists which doesn't fall to normal cryptanalysis).

In the big scheme of things Zero block reward is a non-issue.

Yes.

I am struggling to replicate the $0.05 figure of TC.. most important of the problems is that it is highly sensitive to assumptions of growth in Gross World Product (which I think would probably work better) and/or the money supply of the world's countries over the next 130 years, which is pretty much an unforcastable nightmare. Also, economic output has lots of positive derivatives the rate of the rate of growth is increasing...were annual output growth to ever reach ~17% (seems possible to me) the block reward would start increasing relative to the purchasing power of those BTC (economic output more than doubling, reward halving), such that, after a period of possible irrelevance, as you suggest, eventually each block reward would buy more and more mining equipment and say 200 years from now the mining subsidy again becomes a huge incentive, and stays that way.



MoonShadow seems to think "my" proposal will actually just implement itself anyway. Sounds good to me.

(Thanks to MoonShadow for answering my technical question. Can anyone confirm/disagree with the answer?)

on the list of things that matter this isn't

I suspect that in a couple of decades (or less) we will have cheap common 128 bit CPUs.  At that point, we might decide to widen the value field for aesthetic reasons.  (Daddy, why is the bitcoin value field only half a word?)  As already pointed out, such a change will have no noteworthy change in mining reward or total coin supply.  It could even be exactly zero change, if we modify the algorithm to clip bits beyond the current unit place.

And you really think 0.023 BTC in rewards over the course of decades is going to keep a significant number of miners still mining?  Like Tangible said, a satoshi would only be worth $0.05 in the best-case scenario where Bitcoin took over the whole world, which means 0.023 BTC would only equal $115,000.  Divide that over the course of a few decades and you're left with $3,833/year.  That'd finance three, maybe four ASIC miners at today's costs.  Compare that with the billions of dollars that will be given annually to miners via transaction fees if Bitcoin indeed took over the world, and it is less than a drop in a bucket.  Completely negligible, and not worthwhile to pursuit in any way, shape, or form.

This is one of those things that I look at and ask myself, why is this even a discussion?

In short: because your proposal, while sensible, would make Bitcoin hard and costly or computationally infeasible.

can i donate 1 btc to be held in reserve so that about 100 years from now i alone can be responsible for matching the block reward and doubling the bitcoin currency inflation rate for another quarter of a century?

The Block reward is simply a special transaction with no inputs, using an address supplied by the miner.  The variable that is bitshifted is the actual 64 bit integer of that address.  The block reward would drop to zero in 2130 only because it would bitshift off of the 64 bit integer of the provided addresses.  If a bit-sifted address is developed and recognized by the protocol, the block reward would be able to continue to bit-shift the block reward for another 24+ bits.

Re: MoonShadow

For starters I do not claim the ability to foresee the implications of a nonexistent block reward ("the right answer"). No one can honestly make such a statement, in any direction, about events 100+ years in the future. I am just saying "Why change anything?", and in fact the best part of this idea is that it ignores the impact of 0 BR, whether that impact be small, negligible, good, catastrophic, whatever, and just sidesteps it. I personally am just not comfortable saying that everything will be fine when 0 BR not only potentially fails to guarantee a secure hashrate but (worse) represents a change from the existing incentive-structure.

Secondly, if you have a non-hard-fork way of adding decimal places, it sounds great to me. By all means, expand or contract that as required.

Lastly, though, I disagree with this 'we have years to figure it out' line of reasoning. People are interested in the endgame for obvious reasons: if it wont work in 2150, it will be worthless in 2150, making it a big risk in 2149, recursion, 2148, etc. However, I assume your 'simple solution', while very convenient, would not automatically allow for <1 satoshi block rewards? I thought the BR code just takes bitshift of some constant until it becomes an actual 0. Is that true? What, if anything, would let the BR continue to shift without a hard fork?

I just want the block reward to keep doing what its doing.

It's not a great answer because you don't really know what the right answer is.  


So say it.  That won't occur till about 2130, and only if the bitcoin protocol isn't updated to use a coin variable greater than 64 bits.  And it could happen even if 64 bits is all that is ever used.  Some of the bits are not used to store teh coin value at all, and it has been proposed years ago that greater division can be achieved with the 64 bit interger variable by using those unsued bits as a 'flag' to tell the clients that this value is "bit shifted" into an extremely small range.  For example, bit shifted addresses might not be able to hold more than .99_ BTC, but then would be able to divide down to the 16th decimal place or more.  So while a normal address would be able to hold up to the whole volume of the Bitcoin monetary base (21 M BTC) without risk of a buffer overflow, a bit shifted address would max out at just shy of 1 full bitcoin.  One of each type might be required for some transactions in the future, but the vast majority of addresses would be of the bit-shifted type anyway.

Thisis just one simple solution that fits within the current protocol, and wouldn't break anythin, and it doesn't even have to be decided for another 100+ years.   Let our grandchildren worry about this, should the need arise.

I define the money supply as the actual money supply not the theoretical max.  Currently there will be only  20,999,999.97690000 BTC ever mined.  Increasing the number of digits of precision infinitely would at best expand the amount minted by 2,310,000 satoshis (0.0231 BTC).  Thus 0.0231 BTC / 20,999,999.97690000 BTC = 0.00000011% more coins will be minted.

There is no value in doing that.  Hard forks likely will only occur for the most important and necessary of changes.  Of course to mine all those coins (0.0231 BTC) would require an infinite number of decimal which will never happen.  There may not ever even be any need for more digits but there certainly isn't any need for dozens more digits representing worthless amounts of currency.

Even if BTC replaced all global currencies we are talking something on the order of ~$100T which would make 1 BTC =  $4,761,904.77 and 1 satoshi (current 8 decimal place precision) worth $0.05.  Given this is the "ultimate end game" which is unlikely to ever happen it is very unlikely more decimal places will be needed.   However  the precision may be increased once (say from 8 to 12 digits) however if the minting rules were changed bitcoin minting would still end roughly 40 years after the current end of minting with a less than 1/1000th of 1% increase in the total number of coins minted.

The idea that the blockchain size (and bandwidth, and computing resources) would be increased further to add more and more utterly useless digits just so minting can continue at a negligible rate seems a huge waste of resources for no benefit.

Even if you we use the 1 BTC = $4,761,904.77 the block reward (if expanded past 32 halvings) would be one current day satoshi which means about 2.5 cents (USD circa 2012) added to the mining reward.  Under any more "realistic" scenario it is worth even less.  Say Bitcoin's money supply is someday worth more than the US M1 money supply (~$2.7 trillion).  That would make half a satoshi worth about (1/16th of a US cent circa 2012).



 

Well the 'why' is simple:
People ask "Where do they come from?" and, then ultimately, "What happens when they are all mined?" and I have to say "Well, even though this new thing will eventually happen (precedence of mining fees, decoupling of BTC value and security value), hopefully it will all still work out from then on."

Which is not really a great answer. I'd rather say "It will continue to work exactly the way it does, pretty much forever."

The "costs" you bring up:
Hard fork - if we ever want to add decimals, this would be required anyway. as it stands a human needs to constantly decide this update with more hard forks
Constant updating - not only do we already experience constant updating, but this one update would prevent subjective human-introduced future updates
Transaction size - likely to be a complete non-issue wrt to technology/hard drive size in the 2200's

Vs the "benefit":
Money supply - actually the money supply will not increase at all FROM 21m...in fact it would just more closely approximate this amount...depending on how you implemented it, it could even decrease the money supply (as Zeno's calendar never quite reaches xmas in the cartoon, all 21m coins will never be mined). Moreover, assuming log-growth of BTC value, even these tiny satoshi-cents would continue to finance vast quantities of computer hardware/electricity in the future (and maintain a very strong network hashrate), but more importantly there would always be a reliable source of financing for miners, and of hashrate for users. Thus future expectations can be a little more stable.

So, contrary to what you write, I believe that there are reasons to schedule a future increase in precision (not increase precision today, as you've straw-man'ed). Thus would be better done as soon as possible so we can lock it in and iron out the details.