A proposed solution to adjust for lost Bitcoins: wallet 'heartbeats'

[Garrett Burgwardt]

How many USD are in existence? I'm not sure that there is an exact number due to credit and such. And from people with old bills stored under mattresses, etc.

With bitcoins, you know there are at most 21 million, and by looking at active coins over the last X years, you can get a good idea of how many are in circulation.

I fail to see how adding another layer of complications to things will help address the supposed problem that you've found.

[1714843041]

1714843041

[1714843041]

1714843041

[1714843041]

1714843041

[1714843041]

1714843041

[1714843041]

1714843041

[ascent]

You can get a good idea of how many are in circulation, but as time progresses, you will have less certainty of how many are saved vs. lost.

Let's assume it's the future and all 21 million have been minted. Let's assume that we see 14 million in active circulation and 7 million that aren't circulating. Are they lost or being saved? Valuations will be less accurate over time. I don't see how they cannot be. Imagine valuations based on the assumption that they are mostly lost, and then have them dumped on the market from a dormant account.

Over time, there is a guarantee that bitcoins will be lost, but there will be no real way to measure the average loss over time. Requiring a heartbeat eliminates that problem and allows valuations to be more certain over time.

More importantly, and perhaps most importantly, the accuracy of valuations over time based on circulation will be relatively constant.

[JoelKatz]

I think this is worth considering if a new bitcoin-like protocol and hash chain is ever developed. I think it would be out of the question to try to add this to the current hash chain and client -- disrupting settled expectations like that would be one of the worst things imaginable for bitcoin's reputation.

It would be simple to solve from a technical perspective. All of this would have to be part of the original protocol though:

1) A lifetime would be defined for bitcoins, say 7 years (defined as a precise number of seconds to avoid leap years, leap seconds, and the like). Once a bitcoin had stayed unclaimed for 7 years, it would be subject to reclamation.

2) When each block is issued, any unclaimed bitcoins created or last transferred in blocks with timestamps more than the defined number of seconds ago become subject to reclamation. (Automatically, nothing need be put in the block to indicate this.)

3) To avoid an incentive for miners to fake timestamps a bit ahead, the actual reclamation must occur some fixed number of blocks after the bitcoins become subject to reclamation. (Say 50.)

4) On the block 50 after the block with the timestamp that made the bitcoins reclaimable (or on any subsequent block should a miner fail to reclaim a block), the miner may insert a reclamation transfer, transferring the unclaimed bitcoins to his account.

To avoid reclamation, simply transfer your bitcoins to another bitcoin account that you also control. (Or 'transfer' them to the same account, if that's legal.)

[ascent]

My final point, essentially the nail in the coffin is this:

Assuming nothing else destroys Bitcoin and its success is so phenomenal that it lasts a very long time, it is absolutely guaranteed to self destruct over time when the uncertainty of the ratio of coins in savings vs. coins that are lost is so great, that it will become impossible to realistically valuate them, thus rendering them unusable as a trading instrument.

[ascent]

In fact, it's clear that whole studies will arise, and papers will be written trying to estimate the rate coins are lost, attempting to create models that can valuate Bitcoins. But the studies will be based on hot air, because nobody will really know which coins are lost. Estimations will be made, and then those estimations will be rendered useless when some huge wallet is dumped on the market.

Then papers will be written trying to measure the rate at which coins previously thought to be lost are actually 'discovered' and dumped on the market. Terms will arise to label such events, such as 'sleeper' wallets, etc. Economists will be vexed, to be sure.

[bji]

I have written a solution for the problem you want to address in this thread:

http://forum.bitcoin.org/index.php?topic=20866.0

Some people still will not like the idea of transactions eventually expiring to be reclaimed by anyone, but I think the practical value of guaranteed 0% inflation/deflation is a good argument for the proposal.

[kjj]

I just read this thread again, twice, to make sure I understood where you went wrong.  This post is the key:

You can get a good idea of how many are in circulation, but as time progresses, you will have less certainty of how many are saved vs. lost.

Let's assume it's the future and all 21 million have been minted. Let's assume that we see 14 million in active circulation and 7 million that aren't circulating. Are they lost or being saved? Valuations will be less accurate over time. I don't see how they cannot be. Imagine valuations based on the assumption that they are mostly lost, and then have them dumped on the market from a dormant account.

Over time, there is a guarantee that bitcoins will be lost, but there will be no real way to measure the average loss over time. Requiring a heartbeat eliminates that problem and allows valuations to be more certain over time.

More importantly, and perhaps most importantly, the accuracy of valuations over time based on circulation will be relatively constant.

Your entire argument is based on the notion that someone could calculate in advance the value of a bitcoin by dividing something by the number of coins available.  This notion is false; value is what the market will pay you.

The market performs price discovery, and it uses only participating coins to do so.  Coins that are not participating, whether they are lost or saved, do not factor into prices at all.  The uncertainty over the amount of coins remaining in the system has no particular effect on values.

If the value of bitcoins increases on the market, a feedback system will draw coins out of hiding to reach a new equilibrium.  If they decrease, feedback will push them back into savings.  This equilibrium will be somewhat higher or lower depending on how many coins are actually lost, but this does not indicate uncertainty of value, and it will not destabilize the system.  This change in equilibrium actually represents a transfer of the value of the lost coins into all of the remaining coins.

Oh, and read this, particularly question 6 and the reply.

[ascent]

There may be merit to what you're saying, but are you positive it is addressing this scenario:

The problem is that the ratio of saved coins to lost coins becomes less knowable over time, while at the same time, the ratio of coins known to be in circulation to the coins that are currently not in circulation (lost or saved) becomes smaller and smaller, which leads to a system that becomes, at the extreme, so uncertain that it is unusable.

[Mark Oates]

If the value of bitcoins increases on the market, a feedback system will draw coins out of hiding to reach a new equilibrium.

Oh no no no no no you got it all wrong.  See, someone saves 10,000 coins for 10 years, finds them and then BAM!! instant market imbalance!!  

/joke

which leads to a system that becomes, at the extreme, so uncertain that it is unusable.

In all seriousness though, there are new and interesting economical ramifications with a constantly inflating currency.  Though I can't say for certain, it doesn't seem likely that the above scenario will disrupt the economy in any substantially negative way.  In real-world terms, think "homeless man finds out he's actually rich". http://www.thedailybeast.com/cheats/2011/06/19/homeless-man-finds-out-he-s-rich.html

[kjj]

There may be merit to what you're saying, but are you positive it is addressing this scenario:

The problem is that the ratio of saved coins to lost coins becomes less knowable over time, while at the same time, the ratio of coins known to be in circulation to the coins that are currently not in circulation (lost or saved) becomes smaller and smaller, which leads to a system that becomes, at the extreme, so uncertain that it is unusable.

I don't understand why you think that knowledge of the number of coins available is important, nor why you think that the lack of that knowledge causes instability.  You keep stating it as if it were self-evident, but it clearly is not evident to other people, so you need to explain.

[ascent]

You're missing the point. These aren't issues that will affect you or me. But as a model to endure the ages, it is guaranteed that over time, the ratio of coins known to be in circulation to the coins that are of unknown status will decrease from a number that is orders of magnitude greater than one to a number that, near the end, will be a number that is near zero.

That is the whole point. The dynamics of the system is guaranteed to change radically over a very long period. Why is that a desirable design characteristic? The short answer is, it is not.

[kjj]

If the value of bitcoins increases on the market, a feedback system will draw coins out of hiding to reach a new equilibrium.

Oh no no no no no you got it all wrong.  See, someone saves 10,000 coins for 10 years, finds them and then BAM!! instant market imbalance!!  

/joke

I'd laugh, but this comes up a dozen times a day from people that aren't joking.

Personally, I blame high school science teachers.  If there was one concept that I wish that everyone I met understood, it is the concept of a dynamic equilibrium.  So many people are unable to understand that those 10,000 coins sitting out helped increase the value of the circulating coins, and they will have the inverse effect when (or if) they come back into circulation.  Yes, this causes an imbalance, but in a feedback system, such as an economy, this result is an adjustment, not an explosion.

[ascent]

The stability of a system where 10 million coins are in circulation and no more than 1 million are unaccounted for is much greater than a system where 1000 are in circulation and 10 million are unaccounted for. By unaccounted for, what we mean is, are they lost or being saved? We cannot know.

If 1000 coins were in circulation right now, we know that about 6 million are saved, as opposed to lost. This is a reasonable assumption. This assumption must become murkier over time. We do know that coins will be lost, until every coin is lost. But we don't know the rate, nor will we ever have a more accurate estimate of it than we do now.

What we know for certain is that our estimate of the ratio of coins in circulation compared to the number of coins either saved or lost is guaranteed to become less accurate over time. In its extreme, that ratio approaches zero. And when that ratio is near zero, the currency is unusable.

[ascent]

So, please address the following two points:

1. Show that somehow, the ratio will not approach zero, as I claim it must eventually.
2. Assuming you cannot address point 1, show that the currency is usable when the ratio is near zero.

[kjj]

The stability of a system where 10 million coins are in circulation and no more than 1 million are unaccounted for is much greater than a system where 1000 are in circulation and 10 million are unaccounted for. By unaccounted for, what we mean is, are they lost or being saved? We cannot know.

If 1000 coins were in circulation right now, we know that about 6 million are saved, as opposed to lost. This is a reasonable assumption. This assumption must become murkier over time. We do know that coins will be lost, until every coin is lost. But we don't know the rate, nor will we ever have a more accurate estimate of it than we do now.

What we know for certain is that our estimate of the ratio of coins in circulation compared to the number of coins either saved or lost is guaranteed to become less accurate over time. In its extreme, that ratio approaches zero. And when that ratio is near zero, the currency is unusable.

Why?  Why is it unusable?  You keep saying this, but you offer no reasons why it might be true.

[ascent]

Why?  Why is it unusable?  You keep saying this, but you offer no reasons why it might be true.

Because when the ratio is near zero, the number of Bitcoins in circulation is small. It is so small, that some discovered dormant wallet can be huge by comparison, which can have a drastic effect on the wealth of all current users of Bitcoins. There is no way to know what might happen, but it is certain that it could happen. Because it is certain that there is extreme uncertainty when the ratio is near zero, there is little inclination to put faith in Bitcoins.

But conversely, when the ratio is much larger than one, as it is now and as it is in the beginning stages of any protocol that does not allow reclaiming of inactive wallets, then we operate with the knowledge that most coins are not lost, and thus can be put into circulation.

This is not difficult, but it is important to understand.

[TECSHARE]

Since Bitcoins are (nearly) infinitely divisible this is a moot topic. Even IF enough of them could be destroyed to change the market - you can always just form a new block chain, which is inevitable anyway.  
See https://en.bitcoin.it/wiki/Units
You keep coming in here telling all the pros to listen over and over again, but you never actually ready any of the replies.

[bittersweet]

Sure, knowing the total number of bitcoins that have been lost makes for interesting trivia, but is it really worth the hassle and the inconvenience?

Refreshing coin can be 100% automatic, when you start Bitcoin client, it can check automatically if some coins weren't used for a year or two and refresh them with a simple automatic transaction (or ask user). How is that inconvenient?


I think ascent is right and just because Satoshi didn't think about that 2 years ago doesn't make it bad idea. I think there are many bad design decisions Satoshi made, for example the rate of Bitcoin generation that made the system look like a pyramid and made people who generated blocks with cheap CPU when nobody used this money the most wealthy ones (please don't tell me about investment risk of mining with CPU on your home PC).

But I agree that Bitcoins should stay like they are. If better cryptocurrency will be developed, Bitcoins will slowly lose value and disappear eventually, though some hobbyists will probably still generate blocks just to keep it alive.

[ribuck]

From day one it has been understood that coins could be and would be lost.

Absolutely so. Coin loss, even delibrate coin loss, has always been part of Bitcoin.

Several times, Satoshi proposed deliberate coin burning as a way to use Bitcoin for escrow ("deliver the goods, or else I burn the coins that would otherwise be your payment") or document timestamping ("burn some coin to record a document's hash in the block chain").

Here are some references:
http://forum.bitcoin.org/index.php?topic=750.0
http://forum.bitcoin.org/index.php?topic=645.msg7712#msg7712
http://forum.bitcoin.org/?topic=2162.msg28533#msg28533

The number of coins is totally arbitrary, as they can just be subdivided to the necessary level of precision. Bitcoin could function with any number of coins (even less than one, if the number of decimal places is increased).

Coin loss is nothing to worry about (provided that it's not your own coins being involuntarily lost, of course).

[Pieter Wuille]

Assume 1% lost coins per year. Probably, it was a lot higher initially, but we're talking from now on.

That means it'd take 229 (log(0.1)/log(0.99)) years before we lose one decimal of precision.