Mr. Andrew Miller is a PhD Candidate at UMD. This is our first stab at producing this Podcast, it was intended to be a dry run. However, Andrew is so interesting to talk to, I figured I’d share this with you all. Here’s Andrew explaining how Bitcoin is actually based on a very unproven hypothesis. In that Bitcoin makes the underlying assumption that Miners will act altruistically. Andrew makes some excellent points about Rational vs Altruistic behaviour and what it means for the long term survivability of the Bitcoin Experiment. Really interesting listen, I’ll break down the full interview into a few more Podcasts as time goes on. Full article with transcript available here.
Paul Rausch: So if you could highlight of all of the risks that target Bitcoin, which do you feel is the number one risk that might cause a disruption to Bitcoin?
Andrew Miller: Well the obvious risk is that it’s a self-crumbling fallacy. Ideally, it should be a self-fulfilling fallacy, meaning that if everything goes well, if everyone keeps using it. But it has no value other than people keep using it. The basic risk is that the whole thing just doesn’t work. There’s no reason to think that the whole thing works. It could be fatally flawed, and we have no reason to think that we’ve found all the fatal flaws. It’s not going to be from cracking the signatures of the cryptography but Bitcoin uses completely untested moon-level crazy cryptography. It’s not the cryptography that’s magic about bitcoin, it’s the distributive algorithms. It does things that people think were impossible in computer science. It’s a highly experimental future research water that it’s traveling on. It seems to be doing a great job.
Paul Rausch: What do you mean by it being the opposite of a self-fulfilling prophecy?
Andrew Miller: So, Bitcoin has an incentive system, right? Part of the game is that people get rewarded for participation. That’s the whole idea behind mining, but everyone agrees that incentives drive behaviour right? But the Bitcoin paper, as opposed to the Bitcoin community mythos, the folklore, the dogma isn’t about an incentive Model. There’s nothing in the paper that describes why the incentive model has any sort of desired effect. In the paper, the paper describes the system in not a rational model, but in a majority-honest model, an altruistic model, so there’s a lot of the original paper and some degree the way that thee community acts, bit coin is driven by altruism. There’s some sense that people who are Bitcoin miners are doing a great public service by doing so, and the condition by which the network succeeds is that half the people involved have this altruistic behaviour.
But what happens if not everyone is altruistic? Nobody really expects that miners are doing it to be altruistic. People are miners because they want to make a bunch of money out of it. So there’s this extra layer that people think its’ a good idea to stay in Bitcoin and more and more people will stay in Bitcoin, and it’s possible that this system has that desired effect, but that’s the part that’s ground-breaking, experimental research level, crazy level idea that Bitcoin is.
Nobody has any proof that it works. Nobody has a clear description that it works. The fact that the community is split between sides, some people follow the paper and think that it’s about minds being altruistic, and other people just think that the incentive thing works correctly, but they really don’t have any justification of that.
The biggest risk to Bitcoin is that the altruistic model isn’t realistic, people aren’t mining because they’re altruistic, they are doing it for money. So then the risk becomes that the incentives are misaligned and that people will begin cheating each other just out of normally predictable, economically rational behaviour, so that’s what I mean by systematically self-destructing, self-crumbling fallacy. If you can make a lot of Bitcoin by harming the Bitcoin network, then you can expect that pretty soon, someone will figure out how to do so.
Paul Rausch: That’s really interesting. I never thought about that. Because you hear people in the Bitcoin community chanting this ethos that once we run out of coins to mine, everybody’s going to make money off of transactions, and that’s the magic of Bitcoin. Bitcoin is cryptocurrency, its magic, but you make an excellent point. Right now, the incentive is aligned with altruism. That might not necessarily always be true.
Andrew Miller: Right. There’s very little theory that says that it’s true, but if that can be true that’s everything you would want in a well-designed system. If it’s still possible that Bitcoin’s incentives are aligned correctly. I think that they aren’t aligned correctly, only for some very subtle reasons, but those reasons can be fixed. Once Bitcoin gets the proper kind of scrutiny, researchers start working on it, and the Bitcoin community shows a little more imagination in terms of tolerating these changes. If you can make a system whereby honest behaviour is the same as rationally optimal behaviour, then it just means that it’s going to get better and better and better, the more that people figure how to game the system, the stronger it gets.
Andrew Miller: That’s exactly what you would want, and if it turns out that Bitcoin shows how to make that obtainable…that’s why it’s such a revolutionary invention. That’d be amazing. That’s definitely not the case with any type of other structure that we have in society, whether banking systems, very large companies, or even very large organizations, you get perverse incentives. The opposite is perverse incentives, right? when you have a system with perverse systems, then it means that the more proficient people get at it, the worse the system functions, and that’s system that isn’t going to last very long or will fail spectacularly when it does.
So, the hope of Bitcoin, the unstated assumption of Bitcoin is that the incentives are aligned correctly, and that should be the goal.
Paul Rausch: Now, is it because the incentives are so simple that it’s much easier to align them properly?
Andrew Miller: I think that’s a huge part of it. Bitcoin’s incentives actually work like a lottery, like the more you play, the more you win. It’s not a good example because… I’m going on a tangent right now on how Bitcoin is a lottery.
Paul Rausch: That’s for another day. It’s very easy to define if somebody’s desire is altruistic or not.
It’s a very unique system in that someone is very easily define if someone is in one of these two camps: altruistic and not altruistic. It’s very difficult to bend these rules…
Andrew Miller: No, I disagree that’s the case. It’s possible that’s the case now, but that’s not a part of how Bitcoin works. For example, so it’s really simple to mine for Bitcoins. You can just download a Bitcoin miner, get some mining equipment like a GPU, and you can be mining Bitcoins, and that’s really great, but people also join in pools. You can join a mining pool and get a lower variance, like playing the penny spot machines versus the biggest jackpot ones, if you join a mining pool. One of the decisions you can make is to join a mining pool and how you can do it. But suppose there are different choices you can make in terms of mining software. Say that you have an altruistic mining pool that give you some percent. You can also join an ethical mining pool that didn’t take shortcuts, if known, but you would make less money from doing so.
It’s not the case right now that there are any large mining pools that differentiates themselves by calling themselves rational optimal mining pools — we’re unethical and give the savings to you — but you can easily imagine that that can happen. If there’s any way that you can make more Bitcoin by being irrational. Or by not following the protocol exactly. This is where scalability issues come in. One of the requirements of behaving in a way consistent with altruism is being altruistic, you have to perfectly validate every transaction. (9:51) But if it turns out that it is possible at some point to save money by not validating transactions, then now, you don’t technically fall into the altruistic camp, because now it means that you have some chance of behaving other than the specified protocol.
It’s very possible because of scalability problems that it will become much cheaper not to validate Bitcoin transitions, and at that point, we’ll definitely need to realize that we’ll have to look at a rational model rather than an altruistic model, because if it starts being too expensive to be altruistic, then nobody’s going to be.
Paul Rausch: What would a rational model look like?
Andrew Miller: I worked out a really specific example. That I can probably explain. In the event that this ends up being a really exciting story, I still wouldn’t want to air it, because it’s too close to being an attacked, but also probably not really, so I’ll just talk to myself into explaining it.
Okay, e.g., the rule of Bitcoin altruism is really simple: the thing that defines correct functioning of Bitcoin is really simple and objective. The rule of mining is that you should always be mining on the longest, valid block chain. Or more precisely, the valid block chain that has the most amount of proof of work. That’s the rule of mining, the only rule of mining: it has to be the longest, and it has to valid. The simplest way to not violate that rule is: what happens if you mine on something other than the longest block chain? That is, it is valid, but requires more work. What’s an example of when it would be in your best interest to do that?
Well, the best example is that, suppose that here’s a very large transition fee, something like a 1000 Bitcoin. This happens when somebody has a glitched client, and accidentally makes a transaction that has no transaction outputs. Then the total balance of the transaction would be a reward to a miner. That means that if you’re the winner of this block, then you get this 1000 Bitcoin reward from that awesome transaction that just gave you this huge fee. Normally a Bitcoin reward is just 25 Bitcoin per block. So if you are the lucky winner of this particular block, you get an unusually huge amount. If someone else wins the block (your chance of winning any particular block is pretty small, right?), then they get that enormous fee not you, so if you follow the first rule of mining, as soon as someone finds a winning answer, you say “oh they have the lower block, they win the fee. I’ll just build on theirs” that’s what altruistic behaviour would be. But in this case, rational behaviour would be: Screw them. I’m going to win this block for myself. I’m not going to mine on the longest block chain. I’m going to mine on the one before that. Now I have a chance to win this large transaction. And if you’re get really lucky, and you’re able to somehow win, e.g. the next two blocks in a row, or you’re somehow able to circulate your version of this block faster than the other guy’s able to convince everyone that it is his, then you would potentially be able to have that extra chance of getting that huge fee, deviating noticeably form the protocol. And it’s plausible to me that if there’s rational mining software – there isn’t any rational Bitcoin mining client that you can download right now – but I can just imagine that it’s plausible that someone will make one of those at some point especially if Bitcoin catches on, and more people with money care about Bitcoin, and someone will build an optimized nonaltruistic Bitcoin client, then this nonaltruistic Bitcoin client would deviate in a systematic way , which is that if there’ s a really large bit reward, then all the rational clients will reveal themselves, because they will stubbornly keep fighting over that same block.
Paul Rausch: So by rational, you mean self-interested?
Andrew Miller: Self-interested, yeah, wants to make the most bitcoins for itself
Paul Rausch: So how would you fix that or avoid that? what could be built into the Bitcoin protocol to prevent that?
Yeah, so here’s something pretty cool, having thought about this I actually think that Bitcoin kind of has a built in solution to this problem. And here is what it is, so if everyone is fighting amongst themselves, that’s also not a terribly good solution, because even if you happen to be a lucky winner one of these blocks, it’s not going to be very helpful anyway if everyone else is fighting over it. So you might as well just turn off your miner since all of the blocks around that time are going to be contested anyway. So if you do win the block, there is a different strategy which is that as soon as you take a fee for yourself, what I think the rational optimal behaviour is take a portion of the fee for yourself, and then send the rest back to everyone else
Paul Rausch: Pay off the hoard
Andrew Miller: Pay off the hoard! Exactly! You take a tiny bite of this thing for yourself, but you can’t hold on to it, you can’t play king of the hill, you can’t keep the whole carcass from the pack of hungry dogs so you take a bite off while you have the opportunity and throw the rest back to everyone else. But they won’t fight over you any more so you can enjoy your bite in peace. Right, so that’s what I think is the rationally optimal behaviour for Bitcoin. In that case it’s actually sort of self stabilising in that it’s possible that someone could fight for more than that but it has this property, that the best way to play is to take the proportional amount of the award anyway. So the best strategy is exactly the same as the altruistic strategy.
Paul Rausch: Well that’s interesting, so what you’re saying is that even if you had a self interested group of miners trying to destabilise the currency for themselves, it would still bring them back full circle to an almost altruistic behaviour because slowly everyone would move to rational mining and it would self stabilise again
Andrew Miller: Exactly, and this is the kind of analysis that I don’t think anyone in the world is doing right now. I think is the kind of thing that needs to be understood about Bitcoin, forget about the 51% honest what’s going to happen when everyone switches to rational mining mode and what impact does that have. That’s one ideal about what might happen, but I’m not confident about what rational behaviour would be.
Paul Rausch: Absolutely, we’re just two people theorising here.