LaurentMT
LaurentMT
Sep 4, 2018 · 3 min read

Following the publication of this article, Ricardo Perez Marco made an excellent remark about the PPR including lost bitcoins (currently estimated to 2 millions). Here is some food for thought.

First, it’s worth noting that removing these “lost” coins wouldn’t change the results presented in the article for the past nine years. While working on this analysis, I’ve made a simulation of the PPR removing the equivalent of the bitcoin.days secured associated to the BTC mined during 2009 (~1,624,450 BTC). As it can be observed on this chart, that slightly degrades the PPR but the effect remains very limited. My best guess is that it’s caused by the increasing hashrate.

But what I really like about this remark is that it forces us to question the meaning of “lost” bitcoins. What does it mean when we say that something is forever lost ?

First, we should note that our experience in the physical world teach us that loss isn’t a deterministic permanent state. For instance, consider the story of the San José galleon which sank in 1708 with US$17 billion of gold, silver and emerald. It seems fair to say that in 1708, this treasure was considered as forever lost. But here is the thing: the ship was located by a scientif expedition in 2015 and an operation is planned in order to recover the treasury…

So. What does it mean when we say that some coins are forever lost on the bitcoin blockchain ?

First, we should acknowledge that there’s no marker in the protocol defining that an UTXO is lost or not. The popular estimate of 2 millions lost bitcoins is the result of a heuristic defined by blockchain analysts. As always, a heuristic is useful until it’s proven wrong and I wouldn’t be surprised if we continue to observe (from time to time) the transfer of coins mined in 2009.

Beyond this point, let’s note that when we say that some coins are lost, we usually imply several things:
1/ something happened in the physical world (someone lost her private keys, someone burned some coins, etc).
2/ social consensus will never decide to recover these lost coins by modifying the rules of the protocol.

2/ bitcoin cryptography won’t be broken (i.e: it will remain impossible to recover these lost coins if rules aren’t modified).

This is all well and good but none of these points seems related to PoW. Does it mean that PoW doesn’t play any role here ?

Well, no. It’s important to keep in mind that if you’re able to rewrite the history stored in the ledger then you are also able to modify the “lost status” of an UTXO. It means that on the Bitcoin’s blockchain this “lost status” isn’t deterministic but probabilistic (like the finality of a transaction). For instance, imagine that the owner of an UTXO of 1M BTC fucks up her transaction and burns this UTXO. It’s not really the same situation if this happened 4 years ago or 10 minutes ago. In the second case, the owner might try to recover her funds by proposing a substantial reward to a majority of the hashrate if they agree to rewrite the history. In the first case, the recovery operation would be so expensive that it doesn’t make any sense (even in theory).

So, what we see here is that PoW plays an important role in the definition of what is a lost UTXO and that truly provides utility (think to investors who are taking into account these lost utxos in their investment model).

All these remarks leave us with a last question: How should we evaluate the utility provided by PoW for the lost coins ?

On my side, in the context of this preliminary analysis, I’ve decided to go with the very simple assumption that this utility is equal to the utility provided for the secured coins. This choice was motivated by the fact that it doesn’t change the observations which can be made and by all the uncertainties around the number of lost coins.

That being said, while I don’t think that evaluating this utility to a constant zero is a good idea, I would be glad to see improved/alternate models for the utility provided by PoW on “lost coins” !

    LaurentMT

    Written by

    LaurentMT

    Developer of OXT (https://oxt.me)