Bitcoin Forum

Imagining a hypothetical scenario in the future, where all 21 bitcoins have been generated, and no more are possible. Giving this more time, users, one way or another, will end up making mistakes and losing their wallets from formats or faulty hardware. Since this results in permanent loss of access to those bitcoins, will people gradually and steadily lose more and more access to the blockchain's coins until no more can be retrieved? This is currently not a problem since we can keep generating new coins, but once the limit is reached, I imagine the supply won't be stuck at 21m but will in fact start to go down as these mistakes gradually make coins unavailable.

Yeah, that is already happening. Computer crashes and such. Other forget their coins.

If that were to happen, the code would change to expand the limit, and everybody would upgrade.
But only if the whole network agreed...

A grisly thread full of cautionary tales...

https://bitcointalk.org/index.php?topic=7253.0

Is that even possible without everybody starting over from scratch?

If real money is lost or damaged, somebody eventually finds it, or more can be printed in its place. If bitcoins are lost, new ones can be mined. But this will one day not be possible anymore. As for expanding the limit, it's a slippery slope. Nobody can prove they lost their bitcoins, and the network can't tell the difference between bitcoins left for safekeeping for 10, 50, 100 years and bitcoins whose wallet has been permanently lost.

If new coins are generated beyond 21m "to compensate" for the lost ones, how can one possibly gauge the loss? And with losses constantly occurring, this "compensation" cap rise would very well mean that it will be raised indefinetly, allowing for perpetual creation of new BTCs, which breaks one of its own rules brings back the inflation problem.

On the other hand, if we do not generate BTCs beyond the cap, we will slowly move towards owning # gigabytes worth of unreadable hdd-occupying gibberish. When the last wallet containing coins is lost, the entire blockchain - by then probably in the terabytes if not petabytes - will be worth absolutely zero.

I feel like the system needs to make coins orphan-proof, or reminable. In its current state that simply isn't possible since again - it's impossible to tell between money left for safekeeping and money without a wallet. If a rule would be made to force all coin blocks to be read at least once a year before being deemed orphan and made reminable, that would solve the problem, and introduce a whole new set of issues in turn. I'm not sure it's even possible for the network to check the birth date of a block once it's encrypted and only the private key can read its content, not to mention nothing of its current structure stores last read timestamps. So in other words, there's no easy fix...

Assuming coin loss is slow the system can be adjusted .   The current 8dp limit is arbitrary and can be changed with the consent of the majority of the hashing power(they all agree to update their software to allow 16dp say).
             

Just nitpicking about symantics: bitcoins don't exist in the first place, there's only a list of transactions between addresses. This history of transactions is never lost. But sure, coins can become unavailable (effectively "lost") if people lose their private keys.

This is never gonna happen. It will come arbitrarily close, but the actual 21 million will never be reached.

Way, way before year 2140 (or thereabout) when the block reward halves down to less than 1 satoshi, the smallest subdivision of 1 BTC will be reduced to less than the 1/100,000,000th that current clients support (probably to unlimited pricision, by simply allowing floating point BTC values of arbitrary length).

So no matter how many coins are destroyed, new fractions are always being mined, even in hundreds or thousands of years from now.

Good point Jace

As bitcoins are lost, the remaining bitcoins become more valuable.

As such the average person has less of them and therefore can't lose as many.

This cycle continues.

So while 3 years ago someone might have lost 10,000 bitcoins, it is far less likely to happen now.

Today someone might lose 100 bitcoins, but if the exchange rate continues to increase like it has,, in another 4 years is will be unlikley for someone to lose much more than 1 bitcoin.

If a lot of those are lost and the value increases even more, individual losses greater than 0.01 bitcoin will become unlikely.

If enough such loss continues, it is possible that we could get to the point where 0.00000001 BTC is to large a value to be used for small purchases.  At that time (how many hundreds of years from now is that likely to be?  Do we really care how they choose to deal with such an issue 5 centuries from now?), they could modify the protocol to allow more than 8 decimal places, or a variety of other solutions that have been suggested.

has someone done a detailed exponential forecast model including mining and how many btc are lost each day predicting the supply of bitcoins for the next x years?

There's no way to tell if bitcoins are actually lost or not.

Unless you can see an address is obviously fake. I mean valid, but generated as an address only, without having the corresponding private key. Such as 1BitcoinEaterAddressDontSendf59kuE (https://blockchain.info/address/1BitcoinEaterAddressDontSendf59kuE) - that's 0.41196588 BTC stuck in a black hole for the length of time :)

Sure, you can lose your Bitcoins out of existence, just remove your wallet.dat. It is highly doubtful that all the users would suddenly do the same though. So the answer to your question is no.

It is highly unlikely that ALL bitcoins would ever be lost, as when approaching that point the value of the remaining coins will increase exponentially - and thus be looked after with greater and greater care. If there were only one bitcoin left its value would be in the billions or trillions of dollars, and people would be doing business with satoshis. Additionally the bitcoin software could allow division beyond the current 8 decimal places.

Very good point - I did not realize it only "approached" 21 million. Makes more sense...

Do you know what the actual coin-limiting algorithm is?

Is there a possibility to see how many bitcoins have been truly "lost"?

I can't think of a way, but maybe there is.

50 BTC (stored internally as an integer value of 5000000000) per block for 210,000 blocks, then perform a binary right shift on the subsidy value to get 25 BTC (2500000000) per block for the next 210,000 blocks, then binary right-shift again to get a subsidy of 12.5 BTC (1250000000).  Continue every 210,000 blocks until the reward has been 0.00000001 BTC (1) per block for 210,000 blocks.  The next right-shift results in a reward of 0 from then on.

Are you looking for the actual C source code?

Depends on what you mean by "truly lost".

There is no difference between bitcoins that have been intentionally unspent and bitcoins that are unspent because the private key is lost. So in that case, no there is no way to know.

There is another method of "losing" bitcoins that can be seen in the blockchain.

If a miner is running buggy software and therefore doesn't take the full subsidy+fees the difference between what they could have taken and what they actually took is permanently "lost".  It vanishes from the blockchain forever as of that block.  This can be seen/identified in the block where it happens.

A lost wallet in a crashed computer can definitely be a problem, if no keys are avalible. Also, i don't know if they can be intentionally destroyed - i don't think so. Not considering why would anyone do that :)

Yes, they can.


Well they might do it accidentally if they are mining and have a bug in their program, but as to why they would do it "intentionally": some people do dumb things.

Not the whole network, only the major part of it. There are many "new" and only a few "old" miner. Maybe the new one want a greater piece of the cake? Why shouldn't they outvote the early adopters?

i could imaginage one day someone figures out how to make a new coin that is "backed" by bitcoins

so if there were only 100,000 coins left in existance they could be turned into billions of "
newcoins"

I don't think it's a problem though

What good would that be? By definition there is never a shortage of units with Bitcoin, no matter if there are only 21 million or only 100,000 or only 3 in existence. They're infinitely divisible.

Obviously, for practical reasons, people would not be dealing with 0.0000065 BTC but with "6.5 nanoCoins" or some similar convenient unit.

I feel bad for all those people who formatted their hard drive, losing their wallet.  I originally installed the client in the middle of summer, and turned it off since it was already hot.  Just wish I had left it running longer!   :(

they are only divisible by .00000000 as far as i know

In the current version of the software, yes.  However this is something that can be changed if the need ever arises (such as BTC1 = US$1M for example).

While not impossible, this wouldn't necessarily be an easy change.

There are significant considerations about the largest value an integer variable can hold in the program.  More importantly, it would change the way the blockchain represents the values and would therefore create a forked blockchain unless you could get EVERYONE to switch to the new programming and all of them to do it at the same time.

There are other possible solutions. A couple of examples are an alternate cryptocurrency for small amounts, a new cryptocurrency that is backed by bitcoin, or a cryptocurrency that is designed to smoothly replace bitcoin.  Other solutions may also present themselves in the future (necessity is the mother of invention).  Seeing as possible solutions exist, and even better solutions could be discovered, I just don't see a reason to be concerned right now about the possibility that 1 Bitcoin could be worth more than $100,000,000 someday.

Ignoring this, which nails it IMO

You're right that changing from 8dp to something else would require a fork.  However if memory serves this has already been done at least once, with clients maintaining backwards compatability until a majority of nodes are running the new version, at which point clients automatically move up to the new version of the protocol.

Correct.  As I said, it can be done, but it isn't "simple".  It requires more than just a "majority", you need to get as close to a 100% adoption of the new software as possible.  Any number of users at all on the old software would cause a fork.  Changing the software is a possible solution, but not the only possible solution.