Latest / TFTC: A Bitcoin Podcast / The Difficulty Adjustment (and Target) | Pierre Rochard, Center of Hash E006

Transcript

The Difficulty Adjustment (and Target) | Pierre Rochard, Center of Hash E006 · TFTC: A Bitcoin Podcast

• 0:00
  all right pierre thanks for doing this on short notice of course always happy to come chat feel
• 0:10
  like um as i was telling you before that feels like a hold my beer type episode but i'm also
• 0:16
  really excited for it we're going to talk about the difficulty target yep and the difficulty
• 0:21
  adjustment what it is its significance to the network and how it works and we're going to
• 0:28
  describe it like we're five so that everybody can understand um before we get into that though
• 0:33
  if you could share in your words how you think about the function of mining and the mining
• 0:44
  function relative to other network nodes that are non-mining yeah great question so um i think that
• 0:51
  the mining function is really dual purpose. So there's two functions to it. And I think that
• 0:58
  sometimes they get conflated a little bit. So first of all, the biggest from an economic perspective
• 1:05
  is about issuing new Bitcoin onto the ledger. So there's this issuance schedule with the halvings
• 1:13
  that looks at what block height are we at? And so how much new Bitcoin are we creating every 10
• 1:19
  minutes. Now, I use these words loosely. I know that our friend Phil would say that they're
• 1:24
  incorrect words because the issuance schedule was set on day zero back at end of 2008, beginning of
• 1:33
  2009, whenever Satoshi coded it. He released it in January 2009. The issuance schedule hasn't
• 1:39
  changed since then. And so really from an accounting perspective, you could think of it as
• 1:44
  a transfer from this retaining pond,
• 1:51
  this set of 21 million Bitcoin
• 1:54
  onto spendable UTXOs.
• 1:59
  Yeah, so Phil has this idea that,
• 2:01
  and we recorded with Phil two episodes ago,
• 2:04
  this is episode six,
• 2:05
  he recorded episode four,
• 2:07
  that all 21 million Bitcoin already exists.
• 2:12
  I personally, it's like, I agree with that,
• 2:14
  But I also think it's easier to conceptualize that the Bitcoin are being issued like the term that you used or that they're entering circulation, however you want to describe it.
• 2:24
  I think entering circulation should be compatible with Phil's view.
• 2:28
  Yeah. Because so now the other part of it, too, is that the if we I think it is he's making a very important point, because when you hold one Bitcoin, you hold a maximum of one out of 21 million.
• 2:44
  So you own a fixed percentage of the supply and can't get diluted out of that.
• 2:48
  And so when people say they talk about like Bitcoin inflation, like I don't think that's the correct term to use.
• 2:55
  in fact, I would argue Bitcoin is neither inflationary nor deflationary. It just has a
• 3:00
  fixed supply of 21 million. And so then, you know, if we think about deflation more loosely of,
• 3:09
  okay, putting aside the monetary policy, what happens in practice? Yeah, some people lose
• 3:14
  Bitcoin. And so there's, you know, deflation in that sense, but that's unverifiable. We don't know
• 3:21
  how much Bitcoin has been lost exactly. There's heuristics of estimating that. But in any case,
• 3:26
  that's the first function of Bitcoin mining is, and doing it in a way that is competitive.
• 3:32
  Because if we look at the history of monetary economics, even if we look at gold, the way gold
• 3:38
  enters into circulation, historically, has generally been through a minting process. And so
• 3:43
  you take raw metal, and you put it through this minting function, and then it comes out as
• 3:50
  spendable coins. And that historically, what rulers do, what states and sovereigns do is they
• 3:57
  want to monopolize that. And so they take over the mints. And they decide when to mint. Yeah.
• 4:03
  And how much to mint. And whose face to put on it. Yeah. And yeah, how much and what the
• 4:08
  percentages should be of the coinage. And, you know, that's where we get into currency debasement.
• 4:13
  So it's really important to have, if we're going to have a decentralized system for the issuance
• 4:20
  of the asset to also be decentralized.
• 4:23
  And the only way to have that in practice
• 4:26
  would be to have it be competitive.
• 4:28
  You could also look at more modern alternatives,
• 4:30
  like, for example, what Ethereum did with their pre-sale.
• 4:33
  Competitive in the sense that anybody can participate.
• 4:36
  Correct, yeah, permissionless.
• 4:37
  Yeah.
• 4:38
  Not that everybody is participating,
• 4:41
  which would be more of an egalitarian spin on it,
• 4:43
  but really it's about equality of opportunity
• 4:45
  that from day zero,
• 4:48
  all the Bitcoin that have entered into the UTXO set
• 4:52
  have done it through a process
• 4:53
  that anyone could have participated in.
• 4:56
  And entering into the UTXO set
• 4:59
  also for more common parlance,
• 5:01
  entering into the circulating supply.
• 5:03
  Yeah.
• 5:04
  Yeah.
• 5:05
  And then beyond the issuance,
• 5:10
  go forward, say like when all 21 million Bitcoin
• 5:13
  are in circulation,
• 5:14
  what function are miners providing in that world but they're doing it today as well but then how
• 5:20
  does that also distinguish say from how a non-mining node secures to help secure the network
• 5:26
  so uh yeah to your point we don't really have to go i don't think that we have to go far in the
• 5:31
  future to to see this function because this has also been around since day zero which is about
• 5:35
  essentially providing a sequence function for transactions, meaning that how do we figure out
• 5:46
  if somebody tries to spend the same Bitcoin twice, which one is valid and which one is invalid?
• 5:52
  And so if you look at the today's banking system and you go and try to spend the same dollars twice,
• 5:59
  they will decline the second charge based on a clock that they have in their system.
• 6:05
  And so the bank is the one that's ordering the transactions to figure out, okay, which one are we going to decline and void?
• 6:13
  So obviously we can't do that in a decentralized system.
• 6:17
  So what the mining function is, is akin to a decentralized clock that is ticking every 10 minutes and creating batches of transactions such that if a transaction already spent an output in the past, going forward, if somebody tries to spend that same output, it would be marked as invalid.
• 6:39
  And that transaction would fail to be included in a block and fail to be broadcast out to the network.
• 6:47
  functionally, how do we coordinate all the activity that might be happening in the network
• 6:52
  at any point in time when people don't know each other and they're in different corners of the world
• 6:57
  but need to know which transactions happen first?
• 7:00
  Exactly right.
• 7:01
  And then within there, how do you think about the mining function specifically to final settlement?
• 7:09
  Because I think that they're related.
• 7:10
  Yeah, they are.
• 7:11
  So the way that miners get paid for including a transaction is through the transaction fee.
• 7:18
  So if you're sending Bitcoin, you have to send a little bit less than you are unlocking.
• 7:26
  And that little residual difference goes to whichever miner includes it in a block.
• 7:33
  And so the transaction fees are really the sum of all from all of the transactions included in a block get added to that new issuance, which if you go into the source code, it's called the subsidy.
• 7:47
  I don't like that terminology.
• 7:48
  I think that's one of the worst words Satoshi picked up, but because that has all sorts of connotations in economics.
• 7:56
  What would have been a better word?
• 7:56
  um you know i would have put new issuance or uh um yeah you know then then we could get to into
• 8:06
  maybe phil would have better uh terminology for not significant to this conversation but i'm just
• 8:10
  curious yeah um because uh you know subsidy implies like the government like subsidizing
• 8:15
  something but i do think that um if you look at the white paper the the reason why satoshi used
• 8:22
  that word subsidy is because in his mind, it was subsidizing people running Bitcoin nodes
• 8:30
  because at the time, those two functionalities of mining and operating a node were so intertwined
• 8:37
  that they were essentially a bundle in Satoshi's mind. And when you read the white paper,
• 8:44
  it seems like, you know, he is bundling those two functions together. Not completely though,
• 8:50
  because he does in the white paper talk about network nodes that are distinct from mining nodes.
• 8:55
  Already he was starting to draw a distinction between those two functionalities.
• 9:01
  But if we think about today, how is the software architected?
• 9:09
  You've got people will talk about a Bitcoin mining company that has lots of mining rings.
• 9:15
  And is that a miner?
• 9:17
  Well, in the parlance of Satoshi, it's interesting because for the most part, they don't even need to operate a Bitcoin node.
• 9:26
  Why? Because that mining rig is connecting to a third party mining pool and that mining pool can be operating a Bitcoin node.
• 9:38
  And that mining pool actually is under no obligation from a technical perspective to be operating a mining node.
• 9:44
  in fact if we go look at the history of bitcoin there was this thought of what was called spv
• 9:51
  mining where you're not running a full node and uh it was like headers first spv mining
• 9:57
  and the idea was to to make it so that you could um it be as efficient as possible in constructing
• 10:06
  your block template so that you're not having to to have the full verification um and then uh
• 10:13
  that got exploited uh and so how was it exploited this is something i don't i don't remember this
• 10:19
  or maybe it predated my time yeah um i i'd have to refer back to greg maxwell post on this because
• 10:25
  uh you know he's obviously og legend and in uh bitcoin history but um it to get into specifically
• 10:32
  how it was exploited but he was pointing out that like here's why you shouldn't do this is because of
• 10:39
  this happened in the past. But I think it proves an important point, which is that Bitcoin mining
• 10:47
  pools operate a Bitcoin node for the same practical reasons as anybody else, which is to verify that
• 10:56
  the history of the chain is valid and that they have valid transactions and that their reward is
• 11:04
  valid. And so, you know, every Bitcoin block has a transaction that is very special. It's
• 11:14
  called the Coinbase transaction. Brian Armstrong stole that name to, you know, benefit his company.
• 11:20
  But that guy actually confuses a lot of people. When you're explaining technically how Bitcoin
• 11:24
  works and you talk about the Coinbase transaction, they think that it's a transaction from Coinbase.
• 11:29
  Good name for a company, but, you know. Yeah, I still think it shows a lack of ethics.
• 11:34
  just clobbering the namespace.
• 11:38
  And that transaction is special because it doesn't have any inputs.
• 11:42
  So it's not unlocking any existing outputs, but it does have outputs.
• 11:48
  And so those outputs are, if you sum up the quantity of Bitcoin being locked up in those
• 11:55
  outputs, that's going to equal or be less than, which is interesting, to the sum of
• 12:02
  the transaction fees and the subsidy.
• 12:05
  It can actually be less than, and there's cases where miners have left Bitcoin on the
• 12:09
  table due to bugs in their software.
• 12:11
  And so that's where actually we're going to have less than 21 million Bitcoin because
• 12:16
  of that.
• 12:17
  Historically, some miners have accidentally not minted the correct amount of Bitcoin they
• 12:22
  were supposed to.
• 12:23
  The asymptotic top was always going to be slightly less.
• 12:26
  Is that correct?
• 12:27
  And then-
• 12:28
  For a number of reasons.
• 12:28
  And then this, what you mentioned specifically, lowers it slightly.
• 12:32
  Yeah.
• 12:32
  So one of the reasons is if you just model out the ideal issuance curve and you assume full supply, it is not 21 million. It's like 0.99999997, right below 21 million.
• 12:48
  Yeah. And then a lot of people, and I don't think that you agree, but I want to just ask the question. You don't think of that, what Satoshi called the subsidy as a security budget?
• 12:59
  or do you? Well, it depends because I've heard compelling arguments that it adds to the security
• 13:09
  budget and I've heard compelling arguments that it subtracts from the security budget.
• 13:13
  And I fall more into the latter camp. Now, when we think about security, first of all,
• 13:20
  that terminology is so broad because there is security of private keys, which has nothing to do
• 13:27
  with transaction fees and the subsidy, right?
• 13:31
  It has to do with, you know, what Unchained does,
• 13:33
  what lots of, you know, people self-custidating do
• 13:36
  and all this.
• 13:37
  So, you know, there's that.
• 13:39
  And then there's also the security of the network
• 13:42
  from DDoS, for example, from, you know, malicious actors
• 13:47
  where really the role of Bitcoin nodes is very important
• 13:52
  and the role of transaction fees
• 13:54
  and the subsidy are secondary.
• 14:00
  So really when people talk about the security budget,
• 14:03
  what they should be talking about is transaction finality.
• 14:06
  And they just use that terminology interchangeably
• 14:09
  when really I think that then it causes,
• 14:13
  it's basically like a source of FUD
• 14:14
  because now you're playing up transaction finality
• 14:17
  to be the security of the system,
• 14:19
  kind of broadening the scope of it
• 14:21
  where I'm like, that's not really appropriate.
• 14:23
  Right. And transaction finality is so important because in the example you were given before where it's important to know which transaction happens first, it's also important if I send you a Bitcoin and you give me your car, then I couldn't end up with my Bitcoin and your car.
• 14:40
  Right So like classic double spend would be that yeah I received I thought I received the Bitcoin And let say three blocks have gone by So I got three confirmations So it was included in a block three blocks ago I give you the car you drive away you drive to your mining facility you rewrite the ledger history to remove that transaction and ideally put in a different one that you know still goes to your wallet and then that means that when I look at my wallet that balance is now zero And so now you have the car and the Bitcoin
• 15:17
  And so I think that's called a Finney attack
• 15:19
  is kind of the word for that.
• 15:22
  So it's really important to not have that happen
• 15:27
  because then that would undermine people's ability
• 15:31
  to use this as a monetary system.
• 15:33
  um and the so when we think about the subsidy versus transaction fees really in the security
• 15:41
  budget we're asking the question of how do we prevent that from happening um there's other
• 15:46
  attacks so you could imagine uh there there's attacks where uh you just mine empty blocks and
• 15:53
  you don't include any transactions and so you're essentially preventing anybody from transacting
• 15:58
  on the network and that maybe that would cause the price to crash or something like that
• 16:02
  And then there's attacks where, and so that's really a denial of service type attack on transactions.
• 16:11
  There's also attacks where you're stealing revenue from other miners.
• 16:17
  And so you could have a minority of the hash rate that is selectively revealing their chain to take revenue away from miners.
• 16:27
  But all of these things, I think, are very much more theoretical than they are practical.
• 16:36
  And we can get into that.
• 16:38
  But just in terms of the security budget, the second point I want to make is that not only is the word security overblown, the word budget is also very misunderstood.
• 16:50
  Because people think of, they'll post a chart of transaction fees historically.
• 16:56
  and the transaction fees are going down
• 16:59
  and the subsidy is going down in Bitcoin terms.
• 17:02
  Yeah, in nominal terms.
• 17:04
  And then they'll say,
• 17:06
  well, look, Bitcoin's security budget is decreasing.
• 17:08
  Well, a budget, typically when you think about budgeting,
• 17:12
  you're thinking about the future.
• 17:13
  You're not looking at the past.
• 17:15
  And so one is that if we did have a budget back then
• 17:21
  and we said, okay, the budget is 50 Bitcoin every 10 minutes,
• 17:24
  right that's what it was in 2009 and so now we with the first halving in 2012 uh now we need
• 17:32
  25 bitcoin of subsidy and 25 bitcoin transaction fees we didn't get that right so even back first
• 17:39
  halving like there was no way the transaction fees were making up for the decrease in the subsidy
• 17:44
  um and so somebody could say oh well bitcoin is less secure at that point uh i would say you know
• 17:52
  if you come under budget, that's probably a good thing because that means that the transactors
• 17:59
  got what they needed, transaction finality, at a lower cost than they expected, whether it's in
• 18:04
  the form of dilution of issuance or in the form of transaction fees. And so I really see like
• 18:11
  mining from a Bitcoin user perspective is a cost center, right? Like we have to put up with the
• 18:19
  fact that we have to give miners money uh it's not like to get transaction finality yeah it's not
• 18:24
  something to be celebrated of oh we wasted a ton of money on mining to get the same amount of
• 18:30
  finality yeah i look at it as like an efficient network should be able to do the same function
• 18:34
  for less and less of the overall output over time and that and i'm curious here is that
• 18:41
  what it like because i i personally think the idea of security budget is um like in a in a
• 18:49
  nominal term or an absolute term is is a misnomer because it's like well what ensures transaction
• 18:56
  finality it's just a nominal amount of currency the value of that is dependent on how much people
• 19:01
  value it so it ultimately becomes how much and how many people value bitcoin that determines
• 19:08
  the value of what miners are being paid at any time,
• 19:12
  as well as the amount of transaction activity.
• 19:15
  But then it's what ensures transaction finality.
• 19:20
  And that, to me, it seems like that is a function of decentralization.
• 19:25
  I view it this way,
• 19:28
  is that the true security budget
• 19:31
  is how much would people be willing to pay
• 19:35
  to get into the next block?
• 19:37
  and so it really is about contention at the tip and it's about potential fees and you can't see
• 19:46
  that right so you can't see that unless you were to say all right i'm going to start mining empty
• 19:52
  blocks and i'm going to see what happens because the way transaction fees get set is in the mempools
• 19:57
  so each bitcoin node or you can actually run a bitcoin node without a mempool so you know but
• 20:03
  if you're running a mining pool, you're going to want to know what are the transactions competing
• 20:09
  to get into the next block in order to make your block template. And you're going to rank order them
• 20:15
  by what is their fee rate. And so if there's a lot of demand for block space, there's a lot of
• 20:21
  transactions and the network is getting, you know, quote unquote, congested, fee rates go up. And
• 20:28
  historically during times where, for example, end of 2017, during that bull market up to $20,000.
• 20:35
  That was the top. We thought we were euphoric at the time. Today, we'd be depressed at $20,000.
• 20:40
  We're depressed at $108,000. We'd be depressed at $20,000 for sure.
• 20:46
  Yeah. So at the time, transaction fee rates went sky high. And so it was just because of
• 20:54
  how much demand there was for block space. And that is kind of the revealed preference,
• 21:01
  but that preference is hidden in times where you don't have a lot of demand to get into the next
• 21:06
  block. And so the real security budget is actually unknown because it's really about what are people's
• 21:14
  subjective preferences about transaction finality in the extreme scenario of like, they're all
• 21:20
  bidding against each other. They're all bidding against each other. And then what prevents them
• 21:27
  from essentially redoing the work? Well, so then on the hasher side or on the mining
• 21:35
  side, they're looking at, okay, well, what maximizes my expected value? Assuming they're
• 21:42
  rational, right? And so if they're going to look at two different block templates,
• 21:47
  one has lots of transaction fees by including transactions the other is empty because uh you
• 21:56
  know that would be censorship right they're they're trying to exclude transactions the rational one
• 22:00
  would try to include as many transactions as possible um and so then if there is let's say
• 22:07
  there was like 60 of the hash rate that was mining empty blocks all of these transactions
• 22:14
  are accumulating in the mempool
• 22:16
  and causing transaction fees to increase
• 22:19
  until basically until you have either defectors
• 22:23
  from that 60% of hash rate being like,
• 22:26
  that's a lot of money to leave on the table.
• 22:28
  Like we're, you know,
• 22:29
  we're going to point our hash rate to an honest node
• 22:32
  or new hash rate comes online.
• 22:36
  And that's actually really important
• 22:38
  that the entry into the set of hash rate
• 22:42
  is permissionless.
• 22:44
  And so even if somebody has 99% of the hash rate, they can't stop new entrants from coming in.
• 22:52
  And so that's really a huge difference with proof of stake, for example, where they actually can stop stakers because staking itself is a transaction.
• 23:02
  Hashing is not a transaction.
• 23:04
  And so that's another huge difference when we think about proof of stake versus proof of work.
• 23:09
  um but uh the so the there's a huge cost attached to being an attacker i think is the the underlying
• 23:20
  premise of proof of work um and that uh that's really what then if you look at it from a game
• 23:27
  theory perspective if you as the attacker know that your attack can be undone by others you're
• 23:34
  just not even going to bother you're going to find other ways to attack and there's lots of
• 23:38
  other ways to attack Bitcoin than to just do the brute force method of 51% attack that are less
• 23:45
  expensive. In fact, like people talk about, oh, what's the cost of renting hash rate, right?
• 23:52
  Well, it would be less expensive to do like a cyber attack or to have like insiders at pools,
• 23:59
  you know, collude. See, but even there, like what ensures that the collusion,
• 24:05
  you know because like that's where i get to to decentralization necessarily being part of like
• 24:11
  what ensures the finality of like there there might be an an incentive to potentially attempt
• 24:17
  to collude but if each individual actor represents a smaller share their own expectation of what they
• 24:23
  could influence yeah diminishes and diminishes do you think that that so i i think bitcoin mining is
• 24:30
  is always decentralized.
• 24:32
  Let's take the reductio ad absurdum
• 24:35
  of there's one dude
• 24:36
  in his parents' basement
• 24:38
  in his underwear.
• 24:39
  He's mining on his laptop.
• 24:41
  He has 100% market share
• 24:43
  because the year is 2,200.
• 24:48
  All the subsidies gone
• 24:49
  and transaction fees
• 24:50
  are extremely low.
• 24:52
  And so economically,
• 24:54
  the only thing that's justified
• 24:55
  is one guy mining on his laptop.
• 24:58
  and he's clearing the whole world's economy on his laptop
• 25:04
  or providing final settlement services for it.
• 25:07
  And then he's like, okay, well,
• 25:09
  I'm going to start censoring some transactions.
• 25:11
  And so some transactions start getting left in the mempool
• 25:14
  and those transactors, rationally,
• 25:17
  they start bidding up their transaction fee using RBF,
• 25:20
  using Child Pays for Parent.
• 25:21
  And so they're starting to put chips on the table of,
• 25:27
  hey, whoever-
• 25:27
  Create an incentive for somebody else to come along.
• 25:29
  So somebody else whips out their laptop
• 25:31
  and there's nothing that that first person can do
• 25:36
  to stop the second person from coming in
• 25:38
  and taking that money and contributing hash rate.
• 25:41
  And so even though he had 100% market share
• 25:44
  and everybody on Twitter could be saying,
• 25:47
  hey, Bitcoin mining is centralized.
• 25:48
  This is a huge problem.
• 25:50
  Well, it's not really a problem
• 25:52
  until he effectuates and abuses that power.
• 25:55
  and then it's a self-resolving problem
• 25:58
  because there's a market mechanism
• 25:59
  and the transaction fees going up
• 26:01
  that will then cause somebody to enter the set of hashers.
• 26:07
  All right.
• 26:07
  So now I understand the framework that you have there.
• 26:16
  You mentioned before the second function
• 26:19
  being the ordering of transactions.
• 26:21
  Now I want to dive into the hold my beer
• 26:25
  discussion of trying to explain in a way that is digestible, but I do want to get into the
• 26:32
  technicals of how, of what the difficulty target is, what difficulty adjustment is,
• 26:40
  and how it plays into, because if we, if we start from the idea that if I send you a Bitcoin,
• 26:46
  I send somebody else, the network needs to know which one of those happens first.
• 26:50
  and one of them needs to be processed for transaction finality
• 26:56
  or for final settlement and another one invalidated.
• 27:03
  In this scenario you just described for just illustrative purposes
• 27:09
  of discussing the incentives, setting that one aside
• 27:13
  but going to this world of there's a lot of people out there
• 27:15
  running Bitcoin mines all over the world,
• 27:17
  practically speaking every state in the country every country in the world wherever there is
• 27:24
  excess power all of these miners and you mentioned they're all competing with each other
• 27:31
  in your words describe the before we get into the technical of how the difficulty adjustment works
• 27:39
  and the difficulty target what the significance what it is and its significance to the network
• 27:45
  in ordering blocks? So it's really a way of getting the... So I mean, first of all,
• 27:57
  on the subsidy is that you want to prevent seniorage. And so those monopoly profits from
• 28:04
  issuing new currency, you want there to be a cost associated with that. And the only way to have
• 28:11
  that is to have competition over it. And then what form that cost takes for Bitcoin mining is
• 28:18
  essentially the consumption of energy and of semiconductors. On the transaction side,
• 28:26
  you know, the importance there is to make sure that we don't need to have a proof of authority
• 28:33
  of some kind of centralized clock that is saying, OK, whose transactions come in first.
• 28:39
  you know wells fargo very recently they they entered into litigation over this because
• 28:44
  they were reordering people's transactions to maximize overdraft fees oh i remember this
• 28:50
  so whoever controls the clock actually has a lot of control over the user experience of the system
• 28:57
  right um and and that uh conceivably they they would also be in a position to to give themselves
• 29:04
  more Bitcoin. Now, the other part that we have to keep in mind is that when they mine a block,
• 29:17
  they're not imposing it on the network. They're actually proposing it to the network.
• 29:24
  And so all the Bitcoin nodes, when they receive that block, they make sure that not only is the
• 29:31
  proof of work part of it valid, which we'll get into soon, but they also make sure that all the
• 29:36
  transactions are valid and that there's a list of rules of the Bitcoin protocol that are being
• 29:41
  followed, including the new issuance, the block size limit. And this gets into the area where
• 29:54
  there's a lot of debate going on right now about, is Bitcoin just a database or is it money? Well,
• 30:01
  the rules that the Bitcoin nodes
• 30:03
  are filtering
• 30:05
  with,
• 30:07
  to use a lot of terms,
• 30:09
  they're really optimized to create
• 30:11
  a transactional monetary system.
• 30:13
  You know, they're really,
• 30:15
  that's what their intent
• 30:17
  is. It's not
• 30:18
  any other kind of intent And one that also and here where we enter into tension is censorship resistant So for example like the block size limit on one hand you got people
• 30:32
  like Roger Ver who will say, well, you're censoring transactions that could go into making that block
• 30:39
  bigger by having the block size limit. But the counter would be that the block size limit allows
• 30:45
  for one, low cost of operating a Bitcoin node, which you don't want to censor Bitcoin nodes,
• 30:53
  because if you censor Bitcoin nodes, then there's no way to verify the monetary supply. There's no
• 30:57
  way to be self-sovereign. And it also alleviates the problem of censorship of network traffic.
• 31:05
  So a smaller block can fit through the Tor network, for example, more easily than a large block.
• 31:11
  Uh, and so, uh, it's, there, there's always going to be competing claims of this person
• 31:18
  censoring me and the, the retort being, no, I'm trying to make it more censorship resistant.
• 31:24
  Uh, and so, you know, that's, hopefully that triggers everyone.
• 31:28
  Um, where were we?
• 31:29
  I feel like it should have triggered.
• 31:31
  We covered a lot of basis of like triggerable offenses.
• 31:34
  Yeah.
• 31:34
  So now bringing it to, yeah.
• 31:37
  And I think you made an important comment about the blocks being proposed and that the individual blocks have a number of rules that determine validity and that each node checks to see in a block that is proposed by a miner.
• 31:56
  Are all the transactions in it valid?
• 32:00
  Is the amount of Bitcoin being issued or entering circulation?
• 32:04
  Is it consistent with the fixed supply schedule of 21 million?
• 32:07
  but then there's the difficulty target.
• 32:10
  Does this block meet the difficulty target?
• 32:14
  So what is the difficulty target?
• 32:16
  Well, sorry, I wanted to make one additional historical note
• 32:19
  on the blocks being proposed
• 32:21
  because we actually have a case study
• 32:24
  of a block was proposed by a pool
• 32:27
  where they included the fees and the subsidy in the Coinbase,
• 32:32
  but they had a bug in their software
• 32:34
  such that they didn't include any transactions.
• 32:36
  and so for any node verifying it it seemed like they were trying to inflate the supply of bitcoin
• 32:42
  because they were adding the fees without there being transactions to you know to pay for those
• 32:48
  fees uh and the bitcoin nodes all rejected it as invalid okay i remember this block but i did not
• 32:54
  remember i didn't i did not realize that that was the uh that was the issue with it so so so the
• 33:00
  other nodes interpreted because the transactions weren't there but the fees were that all of those
• 33:05
  fees were the
• 33:06
  subsidy, basically. Because they
• 33:09
  weren't actually validating the
• 33:11
  transactions. They were in the block.
• 33:13
  Right. And so
• 33:14
  what it proves is that even though this
• 33:17
  block had the most work,
• 33:18
  and even though you could say,
• 33:21
  oh, you know, if the miners
• 33:23
  are in charge, that block
• 33:24
  would have gotten into the blockchain.
• 33:27
  But the
• 33:29
  reality is that the miners are not in charge.
• 33:31
  So even if they've got the most hash rate, the most
• 33:33
  work, if they have an invalid block,
• 33:35
  then it got discarded and that mining pool received zero revenue.
• 33:40
  Right.
• 33:40
  And part of that, there was that they're competing with each other
• 33:43
  and that they don't have an, because they're distributed
• 33:47
  and they don't know each other, in part,
• 33:48
  they don't have an incentive to let somebody else have
• 33:52
  or get a reward that doesn't have valid work.
• 33:56
  Right.
• 33:58
  Or invalid work.
• 33:59
  Well, it's a bit of a prisoner's dilemma.
• 34:01
  If you don't include a transaction in your block to collect that fee,
• 34:05
  you know, somebody else is going to.
• 34:07
  And so then you have the game theory incentive of including it.
• 34:11
  Right.
• 34:11
  But you made a comment that I think is getting to the heart of it,
• 34:15
  which is, so it's not enough to have the most work
• 34:20
  or the longest proof of work.
• 34:23
  Everything also has to be valid.
• 34:25
  Right.
• 34:26
  But explain the concepts of the most work.
• 34:29
  Right.
• 34:29
  So most work, you could, so let's talk about the difficulty
• 34:34
  because essentially when Satoshi launched Bitcoin,
• 34:41
  he was mining on his Pentium 4.
• 34:43
  And let's say the second person comes online
• 34:47
  and starts mining.
• 34:48
  So when he launched Bitcoin,
• 34:51
  let's say it's, there's,
• 34:53
  and the way that the nodes calibrate things,
• 34:57
  we want to block every 10 minutes.
• 34:59
  And so, let's see, where to start the explanation?
• 35:02
  It's kind of a chicken and egg thing.
• 35:03
  um let's start it with what is uh the the the difficulty it's a number that you're trying to
• 35:11
  get under as the miner because let's think about what mining is first of all hashing
• 35:18
  okay so a hash function you take a piece of data you pass it through a hash function and you
• 35:24
  essentially get a cryptographic fingerprint of that data that's specific to the data you put in
• 35:29
  So if you put the same data in through the same hash function, you will get the same output.
• 35:34
  It's deterministic.
• 35:35
  It could be like, for simplifying terms, so they know it's not this simple, but you put a string of information in and it outputs nine, and then you put a string of information in and it outputs eight.
• 35:47
  You know, it could be, you put a string of numbers in, it could be a hundred.
• 35:49
  Right.
• 35:50
  And in practical reality, it's...
• 35:54
  256 zeros and ones.
• 35:55
  Right.
• 35:56
  Yeah.
• 35:56
  SHA-256 is the hashing function that Bitcoin mining uses.
• 36:00
  So when you say getting under, it's that you're putting the information in and continuing to try to hit a number that's below a certain number.
• 36:08
  Yeah.
• 36:10
  You can think of it like at a Hawaiian luau party, right?
• 36:14
  You're like trying to get under the bar, right?
• 36:17
  I forget what that was.
• 36:18
  If it gets lower and lower, it gets harder and harder.
• 36:20
  Yeah, exactly.
• 36:20
  The difficulty increases.
• 36:23
  And it's probabilistic.
• 36:24
  So if you think about 256 zeros and ones, so a hash function is random.
• 36:30
  So if you put in different data, even if it's different by a tiny bit, then the output is completely randomly different.
• 36:38
  And so when we're doing this with Bitcoin mining, the input data is what's called the block header.
• 36:45
  So the block header is information about that block metadata that includes what's called the Merkle root, which is the sum of all of the transaction hashes.
• 37:00
  And describe that in a way that somebody could interpret.
• 37:04
  Basically, an easy way to summarize all the transactions in the block.
• 37:11
  That's right.
• 37:11
  And in a way that if you change one of the underlying transactions, well, now that hash is going to be completely different.
• 37:20
  And so this is when people talk about Bitcoin being immutable.
• 37:23
  The immutable nature of Bitcoin's blockchain is really enforced by the cryptography of the hash functions.
• 37:31
  When your wallet creates a Bitcoin transaction and broadcasts it out to the network, it'll give you a transaction ID.
• 37:38
  That transaction ID is a hash of the data in that transaction, which includes the inputs that are unlocking the Bitcoin that have your digital signature that you signed with your private key.
• 37:54
  And it has the outputs of where those Bitcoin are going and the addresses that those Bitcoin are going to.
• 38:03
  So we can start with thereof.
• 38:07
  You take that transaction ID, and then when the mining pool is choosing all of the highest fee-paying transactions, it combines all the transaction IDs into one hash.
• 38:22
  And it does this through a cryptographic process called a Merkle tree.
• 38:26
  and so that's why you know the terminology is is the root is because you're you're summarizing
• 38:33
  all of the transactions and you're fingerprinting all of them into a very compact format that way
• 38:40
  the train the the block header is very small relative to the whole block so a whole block
• 38:47
  could be as big as four megabytes the block header is a tiny fraction of that because the the the
• 38:54
  Merkle root is summarizing all the transactions. In that block header, you also have the previous
• 39:02
  block header. So that is where we get into the blockchain, because that's how you chain the
• 39:10
  blocks together is by embedding the previous block header into the next block header. And so that way,
• 39:16
  everything is interrelated, and you can't go back and change Bitcoin's past without also changing
• 39:23
  everything that follows it. And you know, the, and the significance of that is knowing the
• 39:30
  state of ownership at any point in time and how it changes. Yes. So now, you know, what,
• 39:37
  what are we building on top of with this block? We're not building on top of any other block
• 39:44
  header. We're building on top of a specific one, which actually is, is an interesting contrast with
• 39:51
  with proof of stake where you can actually build on top of an infinite number of previous
• 39:59
  blocks, if you want to call them. Anyway, this is going to turn into like a trashing proof of
• 40:06
  stake episode. But so the when, you know, okay, so if we go back to day zero, Satoshi created
• 40:16
  the genesis block um that's actually the only block that was not mined but i want to reassure
• 40:22
  our listeners the 50 bitcoin in the genesis block are unspendable right okay so let's not get into
• 40:31
  a conversation like yeah 19.9 some odd million 50 yeah first block not mined but can't be spent
• 40:40
  Can't be spent.
• 40:41
  Invalid.
• 40:41
  Invalid.
• 40:42
  All right, now, but coming back to, okay, so you got each block having a block header that points to the previous block header.
• 40:51
  It summarizes all the transactions, basically, and what the version of the longest, you know, kind of proof of work up at that point in time.
• 41:00
  But then coming back to, okay, we've got this difficulty target.
• 41:03
  You know, what, you know, what does it, what role does it play?
• 41:08
  Yeah.
• 41:08
  in basically coordinating the network function.
• 41:14
  So the block header has the difficulty target in it as well.
• 41:18
  It's called nbits.
• 41:20
  And this represents,
• 41:22
  and so this can be calculated by any node.
• 41:25
  You don't have to be a miner to calculate
• 41:27
  what the difficulty should be.
• 41:29
  What you do is you look at the past 2016 blocks.
• 41:35
  And so, okay, if we say there's a block every 10 minutes,
• 41:37
  That means there's 144 blocks per day and multiply 144 by 14 days.
• 41:45
  Then you get to that 2016 blocks.
• 41:48
  Okay.
• 41:48
  So that's something really important.
• 41:50
  Yeah.
• 41:51
  Because this would, and we're actually jumping ahead a little bit, but I, but I think there's
• 41:56
  no way to have a linear discussion about this, which is that this was the part in Mastering
• 42:02
  Bitcoin by Antonopoulos where, cause I had this question of like, okay, well, without
• 42:06
  relying on a single source of truth or an oracle to tell you what happens first, how does each
• 42:18
  miner arrive at what the difficulty target was? So you had mentioned difficulty target being a
• 42:24
  number. And when you put in a certain amount of data into this hash function, inputting that data
• 42:30
  set, and it generates a random output, and then you check to see if it's below. And if it's not,
• 42:36
  you keep going but then maybe like talk in a little bit more detail this idea that the miners
• 42:45
  are basically looking at all the blocks proposed seeing the rate at which they're seeing them and
• 42:53
  then how they actually get to the same target yeah so um the on day zero satoshi just picked
• 43:03
  an arbitrary difficulty of what did he reasonably think? Maybe he probably did some testing of like,
• 43:10
  how many hashes can I generate with my Pentium 4 in 10 minutes? And then using that to calibrate
• 43:17
  the initial difficulty level. Then fast forward two weeks later. Now, I say two weeks, let's say
• 43:25
  more hash rate came online. And so instead of 14 days, it was 12 days, still 2016 blocks.
• 43:32
  But because basically, if we think about, okay, what's the probability of finding a number below the threshold of the target?
• 43:44
  The more attempts you have at doing that, the greater the probability that you're going to find a winning hash.
• 43:52
  And the faster you would find it.
• 43:54
  A good metaphor is essentially asking you to flip a coin, heads or tails, and asking you to get heads five times in a row.
• 44:03
  And if you don't get it five times in a row, then you've got to start over, right?
• 44:07
  So if you get it two times in a row and then you do tails, you start over and you've got to get sequential five times in a row.
• 44:15
  If you're really flipping that coin very quickly, you're going to get to a result faster.
• 44:21
  if there's two people flipping a coin
• 44:23
  they're going to get to the
• 44:26
  one of them is going to win more quickly
• 44:28
  than if there's just one person flipping a coin
• 44:30
  and then if I want to
• 44:32
  slow you guys down I'd say alright
• 44:34
  get ahead six times in a row
• 44:35
  I've increased the difficulty
• 44:37
  if I want to speed things up then I'd say alright
• 44:40
  get it four times in a row
• 44:41
  and so it really is
• 44:43
  a question of statistical
• 44:46
  probability
• 44:46
  of how quickly are you going to
• 44:50
  find that winning combination. And that's the intuition behind creating a way to calibrate
• 44:58
  that decentralized clock that is decentralized. So when the nodes look back at the past 2016 blocks,
• 45:06
  they calculate, okay, what is the time at which I saw the most recent block? And what's the time
• 45:14
  that I saw the first block in that set of 2016.
• 45:20
  And so then let's get an average
• 45:22
  of the difference between those,
• 45:26
  and that should be 10 minutes.
• 45:28
  And if it's less than 10 minutes,
• 45:30
  then I going to increase the difficulty such that we get back to an average of 10 minutes And so deterministically now okay So now the both
• 45:42
  Okay.
• 45:43
  So that's really important.
• 45:47
  But explain the significance of the 10 minute going back to like, why, why, why not five
• 45:54
  minutes?
• 45:55
  Why not 15 minutes?
• 45:55
  You could talk about it either in relation to security and orphan blocks or just in terms of the pulling forward of all the issuance of Bitcoin.
• 46:06
  Yeah. So the way that Satoshi set to issue Bitcoin, I think there's two lenses to look at it through.
• 46:18
  One is that it's arbitrary. You've just got to have some kind of schedule that gets the Bitcoin out into circulation in a way that's competitive.
• 46:26
  And so he could have made it so that within the first two weeks, all the 21 million Bitcoin got mined.
• 46:33
  for whatever reason he wanted to spread it out over several decades now half of the bitcoin
• 46:42
  got mined you know in the first four years and so it definitely was front-loaded in terms of
• 46:51
  the issuance and and then it tails off until the year 2140 but really in by the 2030s 99%
• 47:01
  of the Bitcoin will be on the ledger. And so you can see that as like, okay, well, that's arbitrary.
• 47:09
  Maybe the environmentalists would say that he should have front loaded it more,
• 47:13
  because if he had done that, we would actually have less electricity consumption into Bitcoin
• 47:20
  mining today than we do have. And so like what Elizabeth Warren should be advocating for is a
• 47:29
  soft fork to remove the subsidy today.
• 47:33
  And that would actually be feasible.
• 47:36
  And I actually,
• 47:37
  I don't know if I,
• 47:39
  I don't know.
• 47:39
  Well,
• 47:40
  of course not socially feasible,
• 47:41
  changing the rules on all these guys who are running all this power to,
• 47:45
  to,
• 47:45
  to get that subsidy.
• 47:46
  But I don't know if I follow the,
• 47:48
  the idea that it would reduce because it just comes back to how much people
• 47:52
  value.
• 47:53
  No,
• 47:53
  because look like if Bitcoin's not under attack,
• 47:57
  then transaction fees are low
• 48:00
  and if there's no subsidy
• 48:03
  then the top line revenue
• 48:06
  for the mining industry is low
• 48:08
  and so it just justifies a low amount of power consumption
• 48:11
  Right, but then if the nominal amount declined
• 48:15
  and it was just transaction fees
• 48:17
  but 8 billion people started valuing it
• 48:20
  then wouldn't it
• 48:22
  Well, I'm just saying all else equal
• 48:24
  Yeah
• 48:25
  yeah yeah if we had mass adoption then yeah i agree that the transaction fees would go up
• 48:32
  unless you know some brilliant let's say tomorrow some big brilliant product bitcoin protocol
• 48:37
  developer comes up with some cryptographic magic that makes it so that the efficiency of using
• 48:45
  block space increases a million fold and and we get eight billion people on chain but we only need
• 48:53
  like 400 kilobytes per block
• 48:56
  and so transaction fees are
• 48:57
  de minimis, right?
• 48:59
  Like, and then we had no subsidy
• 49:02
  then the power
• 49:04
  consumption of Bitcoin's proof of work
• 49:06
  would fall to
• 49:07
  almost zero. And
• 49:10
  I would argue Bitcoin
• 49:12
  is just as secure
• 49:13
  as it was
• 49:15
  today.
• 49:19
  I'd have to think about that
• 49:21
  because
• 49:22
  ultimately
• 49:24
  like again
• 49:25
  in my framework
• 49:27
  you've got to
• 49:30
  you have to have currency that's valuable
• 49:33
  to
• 49:34
  pay a wide
• 49:37
  number of participants
• 49:38
  and that the more distributed
• 49:40
  and the more decentralized
• 49:41
  it is
• 49:43
  that the more decentralized it is
• 49:46
  the
• 49:48
  more reliable final settlement
• 49:50
  is and
• 49:52
  And the more you can pay people for power in more places, the more distributed it becomes.
• 49:59
  But setting that aside, because we don't want to give Senator Warren any ideas.
• 50:05
  I'm happy to give her lots of ideas.
• 50:06
  I know she has very little follow through.
• 50:09
  Yeah, honestly, it would be good for Bitcoin because it would make more people understand it.
• 50:13
  But coming back to the idea of the target and basically each miner, it's basically like if the blocks are being found faster, then each miner individually looks at the rate in which they saw the last 216 blocks and adjust back to say, I saw the last 2016 in nine minutes.
• 50:41
  I need to make it
• 50:43
  10 divided by 9
• 50:46
  harder to then have the next
• 50:48
  and the significance of that is
• 50:50
  it basically ensures that
• 50:52
  one, there's
• 50:54
  a period of time in which
• 50:58
  that subsidy is distributed
• 51:01
  those decades that you
• 51:02
  mentioned
• 51:02
  and then there also is a security
• 51:06
  component of
• 51:08
  the 10 minute, not in the sense of
• 51:10
  the security budget, but in terms of the ordering of like,
• 51:13
  what's the probability that a block is found at the exact same time
• 51:18
  probabilistically that could create a split in the network.
• 51:23
  Yeah. So there's that.
• 51:24
  And so you want to have enough time for blocks to propagate globally.
• 51:29
  So that also, you know, if you have lots of ad rate in North America,
• 51:33
  if you had a shorter block time,
• 51:36
  then there would be an incentive to co-locate.
• 51:40
  And hash rate would become geographically concentrated due to the block interval.
• 51:46
  But 10 minutes is long enough to get from Texas to Tokyo.
• 51:53
  You know, in fact, obviously, if you look at the speed of fiber optics, like it gives you lots of wiggle room.
• 52:00
  Right.
• 52:00
  But then at any, at any interval, if it was probabilistically 10 minutes, or if it was
• 52:05
  probabilistically nine minutes or 11 minutes, there's some medium where if, if it's probable
• 52:12
  that it's going to take 10 minutes, that, that it's also possible that two, two miners
• 52:18
  find it in very short period of time simultaneously.
• 52:20
  But that happens, right?
• 52:21
  Yeah, it does.
• 52:22
  But, but am I wrong in thinking that, or at least this has been my understanding that,
• 52:26
  that the difficulty, um, and it being set at that, um, it's not that it doesn't happen.
• 52:35
  It's going to happen, but with a certain frequency. And if it was every say two minutes,
• 52:41
  it might happen in greater frequency than if it was ever 10 minutes, just probabilistically.
• 52:45
  Yeah. Um, that's, that's right. Um, I don't, I think I, in my mind, it really is about,
• 52:52
  leveling the playing field from a global block propagation perspective.
• 53:00
  Yeah.
• 53:01
  The other part of it too is on a practical level, we've got hundreds of thousands of blocks.
• 53:09
  From a software perspective, if you had every one minute, you'd have 10 times more blocks.
• 53:16
  And that just, it just adds to the overhead of operating a Bitcoin node.
• 53:22
  If you've got, even if they had like, if you also cut the block size limit by 10, you're
• 53:29
  just increasing the amount of like metadata overhead that the node has to deal with.
• 53:36
  Smaller issue.
• 53:36
  But the only caveat I want to add is that the difficulty is really being set by the Bitcoin nodes.
• 53:48
  And so when a miner proposes a block, it's going to make sure that it's going to calculate the difficulty.
• 53:58
  Use a little bit more specific language where you say the nodes are setting it.
• 54:01
  Well, the nodes are looking at the past 2016 blocks, every 2016 blocks, and that if a miner proposes a block that has an incorrect end bits in it, that it would get rejected as invalid.
• 54:20
  Right. So you could be a non-mining node.
• 54:23
  You're calculating...
• 54:25
  Independently.
• 54:26
  Independently.
• 54:27
  The frequency with which you saw the last 2016 blocks.
• 54:31
  It's actually not what you saw because the timestamp is included in the block.
• 54:38
  I was going to get there.
• 54:39
  No, no, no, no.
• 54:40
  I was going to ask the question because I remember learning this, but then the question is, technically, each miner might see a block at a slightly different point in time.
• 54:54
  So describe how they're able to get to the actual right amount because it makes a difference.
• 54:58
  uh so you're you're um let's say you're you're mining and you you put a time stamp in your block
• 55:05
  header that's like two days in the future uh nodes would reject that as invalid you have to be but if
• 55:13
  you put it five minutes in the future it's fine there's so there's a there's a limit to how far
• 55:18
  into the future you can put your time stamp and so that helps keep things honest uh and then also
• 55:24
  So you can't...
• 55:25
  And so does that mean that...
• 55:30
  Is it five minutes or is it...
• 55:32
  I don't know what the exact threshold is.
• 55:34
  Okay, because I was wondering if someone's hashing
• 55:36
  if their template, if they haven't found a block
• 55:38
  in that interreligious update for the time.
• 55:41
  They have to eventually update the template.
• 55:44
  The other part is that you can't have it be
• 55:48
  so far in the past before the previous block.
• 55:52
  now my understanding is that it can be the timestamp of your block can be slightly before
• 56:02
  the timestamp of the previous block and so you've got negative time between some blocks
• 56:07
  where you know you would where you still like the
• 56:10
  if i my understanding of that is i times i timestamp something um i don't know what time
• 56:21
  it is but i i timestamping two o'clock you timestamp something 202 because you're just
• 56:27
  yeah uh estimating it it can it needs to be in this window you actually solve a block propagate
• 56:36
  it that's timestamp 202 i didn't get really lucky solve a block the next block within the window
• 56:43
  and it's uh it has a timestamp of two o'clock and then like technically my blocks after yours but
• 56:51
  your time still valid even though yeah things are okay but now get back to how if that's the case
• 56:58
  and you know the timestamps aren't precise people are are um ordering yeah 2016 blocks and looking
• 57:07
  backwards to see what was the well because there's limits on how imprecise they can be
• 57:12
  over an average of 2016 blocks that would not have an effect okay but but
• 57:20
  am I right to say that if everyone calculates their own difficulty target,
• 57:30
  do they actually get to the exactly precise same number?
• 57:37
  I'd have to double check that.
• 57:39
  I'd have to double check that.
• 57:40
  My understanding is that they do, but I'm now like trying to,
• 57:44
  that was one thing that was mind boggling.
• 57:47
  But because if you weren't at that exact same number, then a certain number of times somebody would think that they had a probabilistic block that met the target.
• 58:00
  So I want to more confidently say that the end bits is global and is the same for everyone.
• 58:08
  Because the timestamps, even if with the wiggle room we talked about, it's in the block header and it's hashed and it's global as well.
• 58:18
  So everybody's comparing the same timestamps.
• 58:20
  Yeah, so maybe then it might be there's a timestamp of the 2016 block and there's a timestamp of the first one.
• 58:28
  And they might use...
• 58:29
  And then they subtract those and then they divide by 2016 and then they see how many minutes.
• 58:36
  Yeah.
• 58:37
  Okay.
• 58:38
  So now I always try to now make it relatable because you can like start zeroing in.
• 58:46
  but the important thing is there's a mechanism
• 58:49
  to allow everyone within the network
• 58:56
  independently from each other to sequence blocks,
• 59:00
  but then readjust as more power comes online.
• 59:07
  Or less.
• 59:07
  Or less to adjust and synchronize the network
• 59:13
  such that on average, the blocks come through
• 59:16
  every 10 minutes and that it doesn't materially pull forward the issuance of the supply or
• 59:25
  materially slow right um if for a period of time hash rate comes off now
• 59:35
  part of you know in in a in a practical setting the bitcoin hash rate has increased significantly
• 59:42
  over the past, since inception.
• 59:46
  Yeah.
• 59:48
  And so as more work,
• 59:52
  as the difficulty target is going down,
• 59:55
  the amount of work to hit the target is going up.
• 1:00:01
  How do you think about contextualizing
• 1:00:04
  the significance of more work coming online,
• 1:00:09
  not translating to either faster time or more of the currency units?
• 1:00:16
  Yeah.
• 1:00:17
  So because of this exponential increase in hash rate, we're living in the future.
• 1:00:25
  So we've got blocks arriving that if you had just done, I don't even know what the block
• 1:00:32
  height, do we have a block clock in here?
• 1:00:33
  We do have a block clock, but is this working?
• 1:00:35
  It's not on current block height.
• 1:00:38
  It swaps.
• 1:00:39
  All right. Well, you know, we're we're we're
• 1:00:44
  I'm trying to see what this the this Moscow time
• 1:00:48
  right now.
• 1:00:52
  Logan, what block height are we at?
• 1:01:00
  Okay, 912 to 97.
• 1:01:02
  So if you look at,
• 1:01:04
  I think it was January 2nd, 2009
• 1:01:05
  was the first block.
• 1:01:07
  And if you multiplied 912 by 10 minutes,
• 1:01:12
  you know, it's probably several years,
• 1:01:14
  maybe in the,
• 1:01:16
  or at least months in the future.
• 1:01:18
  Um, but, uh, that's why you should not rely on Bitcoin's decentralized clock to, you know,
• 1:01:28
  get on time to your wedding, right?
• 1:01:30
  Like it's, it's, it's not a clock for that.
• 1:01:33
  It's only about making sure that we're not double spending Bitcoin.
• 1:01:38
  Um, and so the, the, when we think about like proof of work, I think that a really important
• 1:01:45
  intuition is that the work is being done by the Bitcoin miners. And so let's say, you know, we're
• 1:01:52
  at one zeta hash per second, right? And so we're generating this astronomical amount of hashes.
• 1:02:02
  Put that into context. Because most people, maybe, I mean, Bitcoin miners would understand maybe what
• 1:02:07
  exahash or zeta hash is
• 1:02:11
  are we at a zeta hash?
• 1:02:12
  But anyways
• 1:02:12
  Almost
• 1:02:13
  That's almost counter to the point
• 1:02:15
  Equate it in power
• 1:02:18
  or
• 1:02:19
  Well, okay
• 1:02:20
  So let's
• 1:02:21
  We can do metric system education
• 1:02:24
  on this podcast, right?
• 1:02:26
  Yeah
• 1:02:26
  So one kilo hash
• 1:02:28
  would be a thousand hashes per second
• 1:02:29
  Right?
• 1:02:31
  A mega hash is a million
• 1:02:32
  A giga hash is
• 1:02:35
  one billion
• 1:02:37
  A tera hash is one trillion
• 1:02:40
  So the latest generation of miners
• 1:02:43
  They might be in the ballpark of like
• 1:02:46
  A hundred to five hundred tera hash per second
• 1:02:48
  Of trillions of hashes per second
• 1:02:51
  Okay, so that would be five hundred trillion
• 1:02:54
  Yes
• 1:02:55
  Let's see, like trillions of the number of the people
• 1:02:57
  Yeah, already like we're talking about the federal budget deficit
• 1:03:01
  Yeah, which no one can conceive of how large that actually is
• 1:03:04
  Yeah, yeah
• 1:03:05
  It's phenomenal
• 1:03:07
  Now, then if we go to one petahash, so multiply that by a thousand, and then...
• 1:03:14
  Okay, so that's quadrillion.
• 1:03:17
  Yeah.
• 1:03:18
  And then you get to exahash, that's quintillion.
• 1:03:23
  And then zeta hash.
• 1:03:25
  So today we're probably at like 800 exahash.
• 1:03:27
  Yeah.
• 1:03:28
  Which means that...
• 1:03:29
  Higher, like 900.
• 1:03:29
  One way to think about it is that there's millions of Bitcoin mining rigs out there that are hashing to get to that.
• 1:03:38
  And each one of those is consuming or requires power to hash.
• 1:03:41
  Yeah. So then if we translate it into energy talk, we're probably around 22 gigawatt hours of like over an hour, 22 gigawatts.
• 1:03:53
  Right, and maybe a way to think about that is that to solve the next, assume that it was, for round numbers, let's assume it's 20 gigawatts, that it would require running, probabilistically, it would require running 20 gigawatts of power for 10 minutes to solve the next block.
• 1:04:23
  I wouldn't, you're right. You're right. You're right. For 10 minutes. Yeah. It doesn't matter. It would require 20 gigawatts to solve the next 10. And that if 10 more gigawatts of power came online for a period of time, the blocks would be found sooner, but then it would adjust to say, Hey, for the next, you know, probabilistically you need to run 30 gigawatts power for, to solve the next block.
• 1:04:47
  Yeah. And then part of that work, like, like, you know, this is a hard way to ask the question, but it's like, you're out there mining. You might know a friend that mines, but you don't know who's, you know, in terms of like identity solving these blocks and the, in each difficulty, the next block and, and the next 10 minutes and the next 10 for the 2016.
• 1:05:17
  but you need a way to know that others are actually doing the same work that
• 1:05:24
  you're doing or that,
• 1:05:25
  or that they're relational to each other.
• 1:05:28
  That basically if like you can't fake power,
• 1:05:33
  I'm trying to find some way to explain that where you might be slightly better
• 1:05:39
  to be able to produce power at a cheaper cost or set up your mind marginally
• 1:05:45
  better but we're all essentially
• 1:05:47
  competing against the same game
• 1:05:49
  that is competed on the margin
• 1:05:51
  such that if you're producing work
• 1:05:53
  I know that
• 1:05:55
  the work is
• 1:05:57
  comparable to the work that I'm doing
• 1:05:59
  but then explain that and the concept
• 1:06:01
  of this coordination
• 1:06:03
  function of we're all doing
• 1:06:05
  work and we're all validating
• 1:06:07
  each other's work to then
• 1:06:09
  sequence blocks
• 1:06:11
  which ultimately add up to
• 1:06:13
  the highest level idea that we're facilitating a currency system and needing to know who
• 1:06:19
  has the money at what time and that people can't spend it. But kind of like just maybe,
• 1:06:25
  you know, kind of going back up a level. We've been deep down in the weeds. We've kind of
• 1:06:29
  figured out how with no central coordination miners have this way to figure out how much work must be done to consider the block valid in addition to all the transactions being valid pointing at the right version of history but then
• 1:06:46
  you know like the connection between okay if you zoom in too close you're like but why are you
• 1:06:50
  consuming all of this power like fit that into um the functioning of the broader network so
• 1:07:01
  they're consuming all this electricity
• 1:07:04
  to generate all of this astronomical number
• 1:07:06
  of hashes and then every 10
• 1:07:08
  minutes they find one hash
• 1:07:10
  that actually has value.
• 1:07:12
  That one hash is what we call
• 1:07:14
  the proof.
• 1:07:16
  All of the work gets
• 1:07:18
  discarded because
• 1:07:19
  it's not valid hashes
• 1:07:22
  but they needed it
• 1:07:23
  to find that one valid hash
• 1:07:26
  the proof and then
• 1:07:28
  when they broadcast that
• 1:07:30
  valid hash, all the Bitcoin nodes can verify that that hash is valid using very little electricity.
• 1:07:40
  So there's this huge asymmetry between you're consuming all this electricity to propose a block
• 1:07:46
  and it takes a tiny minuscule fraction of that to verify that what you did actually is correct.
• 1:07:57
  um and then so that's for that applies everyone in the network can rely upon it as good right
• 1:08:03
  because if you're if you're mining this is and this is the real you know the reason you want to
• 1:08:09
  run a bitcoin node as as a mining pool is that if somebody sends you a block header and a hash
• 1:08:16
  and you don't verify it using your own node and you start building on top of it you could be
• 1:08:23
  building on sand right because it could be an invalid uh hash in which case when you find your
• 1:08:29
  valid hash and you broadcast it out to the network all the nodes are going to say no that's not valid
• 1:08:34
  because everything about it is fine except for the hash that you included in your block header
• 1:08:41
  because ultimately when you were basically looking at the past we're further in the future
• 1:08:48
  If you were building on an invalid block.
• 1:08:51
  Yes.
• 1:08:51
  Yeah.
• 1:08:52
  Yeah.
• 1:08:52
  And so there's a hole in the history that you're trying to build on.
• 1:08:58
  Right.
• 1:08:59
  Contrast that with XRP where they lost the first 30,000 blocks.
• 1:09:06
  That doesn't matter.
• 1:09:08
  It's gone.
• 1:09:08
  Do you know why?
• 1:09:09
  He's creating more of it.
• 1:09:11
  Yeah.
• 1:09:11
  Yeah.
• 1:09:12
  It's fine.
• 1:09:12
  There's no mining going on.
• 1:09:13
  And that comes into this idea that you can't produce Bitcoin without a massive amount of work.
• 1:09:20
  And that, you know, because you said it's basically like you do all this work to, you know, think about it.
• 1:09:26
  It's like 20 gigawatts of power running concurrently.
• 1:09:29
  You know, and obviously I'm using that number to give some order of magnitude, but it's constantly changing.
• 1:09:34
  Power is coming offline.
• 1:09:36
  Power is coming online.
• 1:09:37
  More power is coming.
• 1:09:38
  Machines are getting more efficient.
• 1:09:39
  So we don't actually know how much power.
• 1:09:41
  But let's also contextualize 20 gigawatts.
• 1:09:43
  That's approximately one third of Texas electricity grid.
• 1:09:49
  Yeah.
• 1:09:49
  At its peak.
• 1:09:50
  Yeah.
• 1:09:50
  At its peak.
• 1:09:51
  Yeah.
• 1:09:52
  Yeah.
• 1:09:52
  We're pushing it.
• 1:09:53
  But broadly speaking, on average.
• 1:09:56
  Peaking like the spring or something.
• 1:09:58
  So it's a lot of electricity, but at the same time, it's not like we're using all the electricity on the planet.
• 1:10:05
  No, no.
• 1:10:06
  I don't want to present that.
• 1:10:07
  But that it is a significant amount and that what it does to that process is it clears all of the value, a clearing from the perspective of everyone who sent money and bid enough to be in that next block, which you communicated the significance of that bidding process, that now we can get on to clearing the next financial transactions.
• 1:10:29
  I would argue more importantly, so if 95% of the Bitcoin revenue, mining revenue comes from the subsidy, then what that represents is that we have destroyed the value of the subsidy by consuming it instead of having that be a monopoly profit of seniorage.
• 1:10:53
  and so you know the fact that if we look at like a natural gas power plant they essentially we've
• 1:11:01
  burned natural gas in order to exhaust out the new issuance and not allow any single participant
• 1:11:10
  to you know profit off of that that from a monetary system perspective that's how we maintain
• 1:11:16
  the morality of the system because well now um i'm trying to i'm trying to wrap my head around
• 1:11:22
  it is that effectively,
• 1:11:25
  I'm trying to reconcile that idea.
• 1:11:30
  It's that you're putting some costs to it.
• 1:11:33
  So you're saying exhausting it,
• 1:11:34
  but it's like,
• 1:11:35
  it's ensuring that there's some cost to it
• 1:11:37
  versus there being no cost to it.
• 1:11:38
  Correct.
• 1:11:39
  Because there's one scenario
• 1:11:40
  where there's no cost to it,
• 1:11:42
  which is the senior age,
• 1:11:44
  the central bank,
• 1:11:45
  the trust that third party
• 1:11:46
  can just create the money.
• 1:11:47
  Yeah.
• 1:11:48
  And that we're maybe not destroying
• 1:11:52
  the value of the subsidy, we're unlocking, we're giving it to somebody, but we're ensuring
• 1:11:56
  that they put in work to get it.
• 1:11:59
  Their profit margin is approaching zero.
• 1:12:02
  But there is some profit motive to actually do the operation.
• 1:12:05
  Sure, and some are going to be better than others, right? Some are going to mine at a loss, some are going to
• 1:12:10
  mine at a huge profit. With it, on average, the average
• 1:12:14
  miner should be basically a break-even. The median miner, or well, the marginal
• 1:12:18
  miner really is going to be a break-even, right?
• 1:12:19
  Right.
• 1:12:20
  Yeah.
• 1:12:21
  And that it you know in one way you could create think about as race to bottom on the other side it is as it commoditizes it should become a very stable activity that has a very slim margin but reliable low return
• 1:12:37
  But in the greater context of
• 1:12:40
  how do you coordinate a monetary system
• 1:12:45
  that everyone can participate in permissionlessly
• 1:12:48
  and you need to reconcile
• 1:12:51
  that what transactions happen first,
• 1:12:54
  everyone has to propose that,
• 1:12:56
  but then everyone else has to validate it.
• 1:12:58
  And you need a way to order
• 1:13:02
  and you need a way for,
• 1:13:05
  imagine a thousand people that don't know each other
• 1:13:07
  to look at each other's work
• 1:13:09
  and know that they did something
• 1:13:11
  in relation to the amount of work that was required.
• 1:13:16
  And that's all, I mean, that at its core
• 1:13:18
  is this idea of the difficulty target
• 1:13:22
  and then the adjustment,
• 1:13:24
  the difficulty adjustment being
• 1:13:25
  how do we basically gate or synchronize
• 1:13:31
  as things speed up or slow down
• 1:13:34
  to ensure that we're kind of within a frequency.
• 1:13:38
  That's right.
• 1:13:38
  Without the difficulty adjustment,
• 1:13:40
  if we'd stayed at the initial difficulty,
• 1:13:43
  we'd be finding a block like every second, right?
• 1:13:46
  And more realistically,
• 1:13:47
  what would have happened is that we would have speed run all of the subsidy blocks and uh today
• 1:13:56
  you know the uh and if you and if you're finding a block every second it's more likely that people
• 1:14:03
  are finding blocks concurrently distributing them and the network can't reconcile which one of these
• 1:14:09
  actually did happen and you can't have a block size limit because in order to say we're only
• 1:14:16
  going to have this many megabytes per day you have to have two limits one you have to limit the number
• 1:14:22
  of blocks per day and two you have to limit the size of each block that's a very important point
• 1:14:27
  and so that's those two things go hand in hand as well because it ensures that there's
• 1:14:34
  scarcity of yeah after all of the bitcoin is issued or entered circulation that there's
• 1:14:41
  something to create prioritization or to require that people bid on the space.
• 1:14:50
  Well, so you would still have that in the sense that you'd still have people
• 1:14:56
  using like child pays for parent to keep their order correct.
• 1:15:02
  I know, but if it was like every second, it's like if time wasn't a limiting factor,
• 1:15:08
  then it's functionally infinite.
• 1:15:10
  Right. And so to me, the bigger problem would be that only Google, Amazon and Apple could run a Bitcoin node because of the resource usage.
• 1:15:20
  Yeah. Because you'd just be adding more data.
• 1:15:22
  Right. Yeah.
• 1:15:23
  With greater frequency at which you're changing the state of the network, the faster it is, the more you're compounding.
• 1:15:32
  Yeah. And they'd be the only ones that'd be able to mine too. And then they would decide on the monetary policy.
• 1:15:39
  Okay. And so now last question, and this is going to bring it back up to a high level. A lot of people think about, will describe the difficulty adjustment and, or the difficulty, because like all the times people say difficulty adjustment, but it's like, if you don't have a concept of what the difficulty target is, then how can you have a concept of adjusting it?
• 1:16:03
  so those two things
• 1:16:06
  difficult to target and difficult to adjustment
• 1:16:07
  as being one of the most profound
• 1:16:09
  pieces of the puzzle
• 1:16:12
  that Satoshi
• 1:16:14
  put forward in taking
• 1:16:16
  other puzzle pieces that have pre-existed
• 1:16:18
  over those
• 1:16:19
  30 some odd years where
• 1:16:21
  cryptographers and cypherpunks were trying to figure out
• 1:16:24
  how to make a
• 1:16:25
  digital money system
• 1:16:28
  work
• 1:16:28
  Drew Bonsal
• 1:16:31
  friend,
• 1:16:33
  co-founder of Unchain,
• 1:16:35
  before an idea,
• 1:16:37
  and it's interesting coming from him, because he was
• 1:16:39
  someone that I think has come around to the
• 1:16:41
  monetary side, but
• 1:16:43
  started from the technology side,
• 1:16:46
  has a physics background.
• 1:16:49
  When he did
• 1:16:50
  his deep dive on the history
• 1:16:53
  of Bitcoin and what Satoshi,
• 1:16:55
  what he theorized
• 1:16:56
  Satoshi must have been thinking,
• 1:16:58
  that in Drew's mind,
• 1:17:01
  he thinks that Satoshi must have figured out that he needed a fixed monetary policy
• 1:17:07
  to make all the incentives of the network work.
• 1:17:10
  And then once he would have figured that out,
• 1:17:14
  then he would have needed a way to meter the distribution
• 1:17:19
  and that the difficulty adjustment and target would have logically followed that.
• 1:17:24
  What do you think?
• 1:17:25
  Do you think that...
• 1:17:26
  Well, you could
• 1:17:29
  I think the incentives would still work
• 1:17:32
  even if there was not a limited supply
• 1:17:35
  in the sense that if there was no halvings
• 1:17:37
  I think the incentives would still work
• 1:17:39
  Well
• 1:17:41
  if there were no halvings
• 1:17:44
  you think the incentives would still work?
• 1:17:46
  Yes
• 1:17:46
  It's so
• 1:17:50
  Wait, are you saying if there was no fixed supply
• 1:17:52
  that there would
• 1:17:53
  It's still a
• 1:17:55
  There's no ceiling, but obviously they couldn't create more than 50 Bitcoin per 10 minutes.
• 1:18:02
  No, no.
• 1:18:02
  What I'm saying is like, even if there weren't halvings, but you still had the difficulty adjustment,
• 1:18:09
  you'd still need a way to meter to ensure that it all just didn't...
• 1:18:13
  Yeah you still have a difficulty adjustment But do you think that the fixed supply came first in terms of
• 1:18:23
  in order to make this whole thing where I need to fix supply,
• 1:18:26
  and now I'm just talking about how I distribute it?
• 1:18:28
  Because you could still have the difficulty adjustment.
• 1:18:30
  You could say 50 every minute and adjust difficulty,
• 1:18:34
  or until we get to 21 million.
• 1:18:37
  Yeah.
• 1:18:38
  I would argue that if we look at the white paper,
• 1:18:41
  that he started with,
• 1:18:45
  okay, we've got a centralized time stamping server.
• 1:18:48
  How do we make it decentralized?
• 1:18:52
  Centralized in the sense of like,
• 1:18:54
  this is the way the central bank works.
• 1:18:55
  Yeah, we've got a server, yeah.
• 1:18:56
  You know, how do we get this guy out of here?
• 1:18:59
  He's got the proof of work part from Adam back, right?
• 1:19:03
  And then he's adding in the difficulty adjustment.
• 1:19:08
  And that the issuance...
• 1:19:11
  is he's thinking, okay, well, how do I get the Bitcoin onto the ledger?
• 1:19:15
  He thinks of this clever hack of, oh, I'll add it to the transaction fees.
• 1:19:22
  So that's the way I read the white paper.
• 1:19:25
  And that's where he says, oh, well, the issuance, it can work like gold mining.
• 1:19:31
  And he was the first person to call it mining.
• 1:19:36
  but in that context
• 1:19:40
  in that paragraph
• 1:19:40
  when he talks about
• 1:19:41
  it's like gold mining
• 1:19:42
  he's specifically talking
• 1:19:44
  about the new issuance
• 1:19:45
  he's not talking about
• 1:19:46
  the transaction ordering
• 1:19:48
  but like the white paper
• 1:19:49
  doesn't say anything
• 1:19:50
  about 21 million
• 1:19:51
  doesn't say anything about
• 1:19:53
  I don't think he even
• 1:19:54
  says anything about
• 1:19:55
  the having
• 1:19:55
  so it was only in the code
• 1:19:56
  it's in the code yeah
• 1:19:57
  and then the white paper
• 1:19:59
  like his intent
• 1:20:00
  is to prove that
• 1:20:02
  he solved the double
• 1:20:02
  spending problem
• 1:20:03
  on the transaction side
• 1:20:05
  not that solving the double spending problem was necessary in order to secure a fixed supply
• 1:20:14
  monetary policy one of life's great mysteries i i don't think it's such a mystery he wrote the
• 1:20:20
  white paper right i mean but he but he wrote the code first true but and he didn't he's communicating
• 1:20:25
  to the world of like what has he accomplished to him the value of what he and look i think
• 1:20:31
  we can,
• 1:20:33
  today we can debate whether it was the most important thing or not.
• 1:20:35
  Right.
• 1:20:36
  Because what I was put before was actually,
• 1:20:38
  yeah,
• 1:20:39
  not what we think is the most important,
• 1:20:41
  but what,
• 1:20:41
  you know,
• 1:20:42
  what he thought or,
• 1:20:43
  or the sequence with it,
• 1:20:45
  which he put these things together and that drew,
• 1:20:47
  you know,
• 1:20:48
  in his mind thought that he must've put the,
• 1:20:50
  the 21 million together.
• 1:20:52
  And in an interesting point,
• 1:20:54
  I think Rob Warren put out a thread about this.
• 1:20:56
  So that,
• 1:20:56
  um,
• 1:20:57
  both way die.
• 1:20:58
  And,
• 1:21:00
  um,
• 1:21:00
  I think Nick Szabo, both initially when they saw the 21 million fixed supply, they thought that that was what was going to cause it to never work.
• 1:21:12
  So there could be an idea that he realized that, but also realized that people would discount it.
• 1:21:21
  So maybe he just didn't emphasize that it was critical to the whole thing working in the white paper.
• 1:21:26
  But yeah, I think it's, you know, Bitcoin's working and that's the most important thing.
• 1:21:31
  But I do think it's interesting to think about them in relation to each other of you have the monetary policy and then you have the difficulty target and adjustment that, you know, helps distribute it.
• 1:21:42
  I think that Satoshi's intent was to incentivize early adopters with this bait of the subsidy.
• 1:21:52
  and then he saw a necessity of phasing it out
• 1:21:57
  because at some point either it's taken off or it hasn't.
• 1:22:01
  And so that is really, I think, his thought process
• 1:22:05
  on what he writes in the white paper.
• 1:22:08
  Yeah, and what I think Dhruv would say there is that
• 1:22:11
  the only way to value it, people needed to have some way
• 1:22:19
  to know how many there would be.
• 1:22:21
  And then if you come to that conclusion, then the most logical thing is that...
• 1:22:29
  Hey, people value dollars.
• 1:22:30
  We don't know how many there are.
• 1:22:32
  Yeah, but if he started with the idea that it would always be infinite, then it would be that much harder.
• 1:22:38
  Look, so, I mean, as I mentioned, like on one hand, like we can say, oh, you know, the issuance schedule, it's all arbitrary.
• 1:22:44
  He could have done anything.
• 1:22:45
  I would argue that there's a tremendous amount of wisdom in what he ended up doing.
• 1:22:51
  And so the four years between the halvings, the halvings, the 21 million, that number, I think that the mimetic value of it is tremendous.
• 1:23:07
  yeah well
• 1:23:08
  people might need to listen to this one
• 1:23:12
  a few times like understanding the difficulty
• 1:23:14
  adjustment is not easy
• 1:23:15
  but it is critical to understanding
• 1:23:18
  really how
• 1:23:20
  the network works
• 1:23:21
  and how the miners
• 1:23:24
  are able to work all in unison
• 1:23:26
  and reach a consensus
• 1:23:28
  and so
• 1:23:30
  I appreciate you coming
• 1:23:32
  downtown. Next time we'll talk about the doom spiral
• 1:23:34
  oh yeah we can do an episode on that
• 1:23:36
  Yeah, for sure.
• 1:23:38
  There's always going to be a death spiral.
• 1:23:41
  Yeah.
• 1:23:41
  Every time the rate of issuance gets cut in half.
• 1:23:44
  So if you're willing to do that, yeah, we'll look at the calendar because we should definitely
• 1:23:49
  talk about how we know that there won't be death spirals every four years.
• 1:23:55
  Every four years.
• 1:23:56
  All right.
• 1:23:57
  Peter, thank you.
• 1:23:58
  Thank you.
• 1:23:59
  All right.
• 1:24:06
  Thank you.