Topic: cryptocurrencies and monetary supply (growth rates) - page 2. (Read 5034 times)

MtGox will take any change in bitcoind.

Let's not pretend Gavin & co. cannot introduce this change. Let's just admit the truth: it would be a dangerous precedent and ought to be avoided.

Not everyone, just a big enough majority, can introduce basically any change you can think of. Obviously it won't happen against the will of the big players though.

AFAIK only one has, and that was because of a bug in the software that allowed someone to create several billion coins. This was also very early on when a breaking change was easy to introduce.


D&T is right though that everyone, everyone, has to agree. Mt.Gox has to simply say "I reject your code change" and nobody in their right minds would use that fork since Mt.Gox is currently the center of the bitcoin universe. It can be done, but something like changing the coin distribution is never going to have a consensus and it is better to just create a new currency if you feel very strongly about such a thing.

It's not the precision problem, but the extreme long term reward towards not using the coins. I think 0% is radical enough, -0.01% (let's say) sounds just wrong. Granted, it's not a deal breaker, but I don't think it's desirable.


Compatibility-breaking changes have been introduced already if I'm not mistaken. No need to do a Rube-Goldberg sync operation, it suffices with agreeing the future block # that makes the change effective and broadcast an alert to upgrade with enough time in advance, this is something Gavin and a couple client devs can single-handedly achieve. We shouldn't pretend this cannot be done, we can just give our honest recommendation to avoid it.



The resource is measured in KB, but the economic significance of the reward in BTC. It probably should be a mixture of the two, but a % big enough on value would be simpler and practical (expenditures are not decided on KBs, but on cost, if you want the user to worry about internal implementation of the blockchain I think we're fu**ed).

There are roughly 20,000 nodes.  Most aren't miners.  You can't simply change the miners you would need to convince all the nodes to switch.  First that means convincing enough trusted developers to code the changes.  It also requires enough users, exchanges, merchants to implement the new code.  If there are nodes with incompatible codebases on the internet at the same time each will produce their own valid network and reject blocks/tx from the other nodes.  Essentially a permanent hard fork.  If you only convince some of the miners to switch they will simply produce blocks (and coins) that nobody uses.   The users who don't switch will produce tx (and fees) accepted by other users and merchants and exchanges who don't switch.   That means there is economic value for miners who don't switch.  Difficulty will fall and thus the revenue per MH/s will rise simply by not switching. Essentially miners will go where the money is.

While in theory it is possible it is far more complex than just convincing some pools to switch.  Even relatively non-controversial breaking changes have taken a very long time to get accomplished (and involved some spririted debate).  You can't force a change like this.  It has to be accepted willingly by a super majority of all users, developers, merchants, exchanges, and miners. 


Merchants don't need to update if their customers are running the old nodes.  Customers don't need to update if their merchants are running the old nodes.


Clients =/- miners but if the number of miners is small but still a significant fraction of total (say 25%) in time difficulty will fall and the miners who remain on the old fork will see their revenue per MH/s rise.  Miners on the new fork are taking a risk they will simply be mining worthless blocks (and coins) accepted by nobody.


The bitcoin network doesn't know the intended amount of the tx only the sum of the inputs and outputs.  If for example the client chooses a 100 BTC input to pay a 1 BTC tx you would be charged on the 100 BTC.  0.01% is likely too low to produce the revenue necessary to protect the network.  For example PayPal has ~$100 B in transaction volume per year.  At 0.01% that would only produce ~ $10M in revenue for miners.  Also the "cost" of a tx is based on its size not value.  Charging more for higher value but smaller tx and less for lower value but larger tx creates a break between price and cost.

The critical resource of the blockchain is KB.  A 1,000,000 BTC tx is just as easy to process as a 1 BTC tx however a 100 kb tx will require 100x as much storage into perpetuity as a 1 kb tx.

Overall, I wouldn't change much about bitcoin.  Unfortunately, there's no ideal way of distributing 21 million units of currency.  The initial high rate of inflation is necessary to distribute & promote trade.  If anything, I would have created a shorter window to distribute all 21 million units.  That way transaction fees would be more important by now.  But no one could have predicted how fast bitcoin would take off.  In either case, the shrinking block reward will be a good thing.  Inflation will decrease, causing prices to rise.  Higher transaction fees may be required preventing groups like Satoshi Dice from creating such small/unnecessary transactions that create unnecessary work for everyone else when validating transactions.


This is true, but we can always move the decimal point if a satoshi isn't small enough.

This "introducing changes in the future" is a double-edged sword. It also ends with the perception of long-term predictability of supply. It would be the most radical thing to ever happen to the system.


This is intriguing but it introduces an unpredictable element, as transactions could end up being very substantial long term, and could also be minimal as payments got centralised in "bitcoin bank accounts" with even lower commission for payments inside of their system. So what happens if transaction fees are exploited as a means of multiplying supply by a cartel of miners? (talking about newly generated coins proportional to previous fees). This probably deserves experimentation but it would take a long while to see potential problems and exploits.

I was rude because D&T is a non-stop sycophant that can't imagine that there is something that bitcoin maybe didn't get right. It's not as if this has been our first exchange. "No bitshift? Must mean floating points!" Asinine. And he calls into question my ability to think like a programmer. His posts are almost always dishonest when it comes to bitcoin's shortcomings, so I often have to set him straight for the non-initiated that may be reading.

I am for b) [no reward halving] [or d) as explained later] in favor of increased inflation. I don't see a dramatic problem with changing the code.

We have to ask pool operators about their opinion. We can ask them to present their view and miners will just switch to the pool with the view they share :-)
If the pool operators agree with the change we have the majority of new blocks generated with the new software. All other solo miners will also update to get the 50BTC reward. Merchants will update to receive incoming transactions [otherwise they will see no transactions, only transactions generated by a minority of clients that were not updated].
If the number of not updated clients is small, they will not have enough hashing power to generate new blocks.
The change will also teach the network about the feasibility to introduce changes and how to do it in the future.

The better change (d) is as follows:

there is a minimum required transaction fee of 0.01% (much better than the current unpredictable fee). [And here comes the new part] The miner collects all transactions fees in the current block + a reward equal to all fees generated 6 blocks ago. The current block fees are paid from the "account" that sent the funds. The reward (second part) represent newly generated coins that facilitate inflation at a rate proportional to the volume of transactions. The delay of ~6 blocks is needed to prevent miners from profiting from fake transactions. The delayed "confirmation" can be used also to introduce theft protection [the owner of the sending account can cancel the previous transaction in the time frame of 6 blocks]. Of course "6" is just an arbitrary number AND the transaction canceling is a completely separate issue.

I agree is a choice and it has a sensible base, but it's not infeasibly hard to provide a binary spec that would allow bit-for-bit precision. Simply leaving the block reward fixed is even simpler and safer (it would take ages to achieve low inflation rates though), or decreasing every x blocks and up to a block n. It was simply a choice and all things considered I think he got a good deal of choices right. I think this choice is not bad either, I just think it would make a substantial difference to have something a bit less dramatic.

Pros: simple, won't cause trouble with unintentional forks, won't have to wait fu*****  50 years to see how it works with low inflation (to "normal economy" standards)
Cons: dramatic a couple of times (1st and 2nd halving), will relatively abruptly reach a ~0% growth economy that hasn't been tested ever

Don't be silly!  You can't use subtraction because eventually you'd get negative numbers which means miners would have to PAY bitcoins to mine a block!!! 

But seriously your proposal would result in a probably trivially different curve, you were extremely rude in proposing it, and who cares nobody is going to move to an alt-coin for this trivial issue. 

And most importantly, that solution would not be nearly as FUN!!! 

You heard it here first folks, subtraction is complicated and can't possibly be accurate across all systems. There is NO POSSIBLE WAY the coin distribution could be determined by subtraction because that means we would eventually resolve to a whole number rather than an ever decreasing amount that goes beyond the divisibility of the currency in 2140. Oh the huge manatee. Or if this really bothers you, I'm sure we could figure out a way to bitshift the amount of subtraction to come up with the exact same curve except without the stark halvings.

Thanks for remind me why I rarely waste my time responding to you.  I must have forgotten this morning.

Your arrogant attitude is why 10 years from now you will have 20 theoretically superior currencies which don't have a single line of code written and never will.  Accuracy and simplicity is important in a peer to peer network because OTHER PEOPLE will be writing clients.  Those clients could contain bugs not discovered until years later.  The more complex the system the chance there is for errors.  Satoshi couldn't stop, modify, or control Bitcoin now if he wanted to.   There is VALUE in simplicity.  Something you fail to understand and thus lash out like a child.

Feel free to rant on, the ignore button ensures I won't need to listen to it.

I agree.  I think a system which reduced say 1% every difficulty adjustment would be an interesting alt-coin.  Likewise an alt-coin which used the traditional "bitcoin half" but only down to 1.25 BTC then went to a permanent 1 BTC into perpetuity would also be interesting.

I would point out the post above.  Cutting the reward in half (technically a right shift) reduces the chance of errors.  Making one client error proof is hard enough.  Making multiple clients by multiple teams on multiple hardware and using multiple programming languages is more difficult. 

Right shift is "precise".  You can't get it wrong.
Right shifting 11011101011 will always be 01101110101.

Still there are a lot of thing which Satoshi could have done differently.
why 8 digits when the data structure supports 11.  Just make it 11 from the start.
Why 32 bit nonce (and needing all the junk of extra-nonce).  Just make it 64bit.

For the most part we are "stuck" with these quirks.  It would be nice to see some innovation in the alt-coin space though.

Remember when you first heard about bitcoin and tried to comprehend it? Yeah, took me several months too.


Did you just not read my post and fill in your own words? Because this has nothing to do with what I said. Your twin's argument is that 25->12.5 won't matter because of the price increases, but that is ridiculous since one day they will mine 25xprice, the next it will be 12.5xprice. That is quite significant.


Do I give a damn whether it's cut in half or it is a bit-shift (which btw, I never said cut in half)?


Apologizing, apologizing, apologizing. "Good lord suh, wesa cannah figga out how to lowah an award withoutta bitshift, suhh! subtrackshun hasna been invente yet" Give me a fkin break.


S-S-S-S-STRAWMAN

I kinda assumed the words "doubled OVER THE NEXT 4 years" implied that price will increase over time.  No intelligent miner looks at 1 day revenue and bases decisions on that.   Sure the network's global hashing rate will change significantly but the network will compensate and life will go on.  Long term the network' hashing rate will be the same regardless of the generation rate.  Daily volatility in the global hashing rate isn't a major concern.  We have seen major changes in difficulty on multiple occasions and none had anything to do with a subsidy change.

Spoken as a non-programmer.  Technically the reward isn't cut in half it is a binary right shift.   There is no issue of implementation bugs. Right shifting a particular binary value will always produce the same binary value.  The OS, underlying hardware, and programing language can't influence that.  The same can't be said about floating point math.  Lots of things to get wrong and in a p2p project there will be multiple clients running on multiple hardware platforms written by multiple teams.   The risk of someone getting it wrong and causing a hard fork is very real.

Sure a periodic reduction in the reward would be better. A ~0.5% reduction in subsidy every difficulty adjustment (2016 blocks) would generate roughly the long term generation rate as a right shift (50%) every 210K blocks with a much smoother curve.  However it would have been error prone.  Incompatible implementations on a p2p network present a real risk.  

To say Satoshi made a mistake is arrogant.  He made a choice.  He chose simplicity and accuracy over a smoother curve.  It remains to be seen if it is a mistake. 

There are permanently lost coins though. If halving goes on indefinitely as soon as in 30 years time we might have more coins lost than generated, because generation will be extremely low.

The point about criminals is that coin concentration is so extremely concentrated in the early years that hacking is strongly encouraged, and it also coincides with the community and software being the most immature. It's a rather unfortunate nature of things


This is something I've mentioned a couple times in other threads, but in reality you only need a big enough economic majority to get away with a fork. Obviously this won't happen if the main client developers don't agree, which is feasible now but ever less so as there are more diverse wallets and clients. Naming would go with the obvious winner. But yep, this would be immensely traumatic and can only be possible if it's decided soon. Maybe too late now.

As for the incentive, it's a long term one. If most people were strongly convinced that it would really make the system more viable, a small growth rate (say 1%) would be a perfectly fine price to pay even for hoarders. Effectively if the underlying economy grows more, because the system is perceived as better, then BTC valuation would over-perform for a good while. Lost of ifs and buts here though.


I agree, I also think it's mostly a matter of alt coins, but there are mechanisms things like fee enforcement that only need a couple of big pools to be strongly encouraged (or else your transaction might take forever to complete). And although I disagree that the mining rate is infeasibly hard to change, it would indeed be traumatic to show you can actually change it.

So yeah, a change in the mining rate probably merits its own coin, although achieving the traction bitcoin has as a newcomer would be a great achievement in itself and looks hardly possible. A number of improvements would be necessary to justify it.

I've seen Etlase2's proposal but it looks ungodly complex and relying on many things to be actually tested IRL. We probably wouldn't be through early testing stages in our whole lives. If anything I think a system with less arbitrary numbers than bitcoin would be better (for instance, without the halvings, or adjusting per block instead of dramatically every four years, ...).

The above "nothing dramatic" is not well considered.  People know about it but are miners hoarding coins now waiting for the halving?  I don't think many are, its still too early.  They have bills to pay.  But if you were a miner, would you sell ANY coins 1 day before the drop?  Might as well wait to see what happens... how about 1 week before?  So expect no ask depth in the days leading up to the reward halving.

Econ 101 people think supply vs demand means a reward drop in half = a doubling of price.  But this assumes a linear x=y supply demand curve.  Depending on the flexibility of demand (i.e. gasoline is inflexible while M&Ms are very flexible) this could be very much not x=y, but x=100y or x=y/100.  To figure out what the curve really is, you have to ask yourself, how flexible is demand for BTC?


But, the above argument is faulty.  It focuses on instantaneous supply.  But in fact the supply is not changing (due to miners, it will change due to anticipatory hoarding), it is the RATE of supply increase that is changing (because bitcoins are not a consumable).  So a lack of supply on day 0 of reward halving cannot be fixed in the classical econ 101 manner by an increase in production tomorrow.  There will be a similar lack on day 1,2,3, etc.  The issue must be fixed by a reduction in demand.  

Given a constant economy, this reduction in demand can only be addressed by an increase in the price of BTC compared to other commodities.  In other words, until SR merchants, etc lower their prices in BTC to compensate for an increase in BTC price (or sell less product) we'll see a constant increase in price to meet existing demand.  And if the BTC economy is growing, we'll see great pressure on the price due to lack of supply.  

The above is the normal negative feedback mechanism that keeps prices sane.  But if the BTC economy grows BECAUSE of an increase in BTC price (i.e. due to new interest from investors),  we could see an unsustainable positive feedback mechanism.  This will cause the hockey stick bitcoin price appreciation until the positive feedback switches back to negative (i.e. investors see BTC as overpriced).

So the block rate halving is more likely to affect (increase) the long term appreciation of the value of a BTC, combined with a spike leading up to the reward halving which could turn into a massive price appreciation event.  And then (regardless of whether we see a small spike or a big rise), we'll see a possible dump afterwards as short term hoarding is sold.

C'mon bro, the exchange rate is going to rise over time, and it certainly isn't going to double right after the switch from 25->12.5, so miners are going to take a big hit and the network TH/s is going to drop a lot, again (ignoring ASICs and what affect they might have). Satoshi made a mistake in having such stark drops. It would be much better if the exchange rate could rise gradually as the mining reward drops gradually. Oh well.

I'd prefer B since it keeps the money supply relatively constant in the long term (lost coins), but it's such a minor issue that for a deflationary currency Bitcoin is good enough as-is. Bitcoin has convinced me that deflation is not a huge problem, but it still might be suboptimal.

However, both EnCoin/decrits and Freicoin have some good ideas, so I'm keeping an eye out for those. It's unfortunate that much of the Freicoin community supports mandatory payment to a central Freicoin Foundation, though.

Make ALL the coins! Let the market decide, because I'm sure not smart enough to.

Miner's costs are primarily in USD.  Say x fees are needed to replace a 12.5 BTC reduction assumes a fixed USD:BTC exchange rate.  If USD:BTC exchange rate doubled over the next 4 years miners under 12.5 BTC subsidy could be more profitable than under a 25 BTC subsidy (and current exchange rate).  Yes the day the subsidy changes miner will see their daily profits change but miners will be anticipating it well in advance (not buying new hardware, scrapping obsolete rigs, etc).