Solar Powered Space Pirates: A Threat to Bitcoin?

From Reboil

Solar Powered Space Pirates: A Threat to Bitcoin? was a talk given by Peter Todd at the 2017 Breaking Bitcoin Conference in Paris, France.

Stats

Text

I'm Peter Todd. I do cryptography consulting and have been involved in Bitcoin Core as well. I wanted to do something a little more interesting and tell you a bit about whether or not solar-powered space pirates are a threat to Bitcoin.

You know, if you're sci-fi and space nut like me, you've probably heard of collecting solar power in space. You know, and you put a bunch of solar panels up there and it's nice and shiny, and there's none of those annoying clouds to get in the way, and you beam your power back to Earth and, I don't know, run civilization or something. Or if you're really clever, you'd do Dyson–Harrop satellite all kinds of crazy stuff like I found out about this morning when I actually bothered to do my slides. Apparently that's the sun on the left and all those squiggly lines or something are charged particles getting collected. And long story short, I mean, it's space. There's energy all over the place and at the very least you don't have EPA to worry about, so surely you could run maybe some nuclear reactors and just dump all your waste right off into space. I mean, there's a lot of it, after all, in space.

But, you know, the issue with all this stuff is, well, power transport. I mean, what are we going to go do with all this power we're collecting, you know? What are we going to do with these kilometer square arrays of solar panels? And this is actually off the page I found out about that previous slide, and I really love how they say, you know, the main technical deterrent to this crazy proposal that involves sticking things, collecting ions off next to the sun, or something, is getting the power back to Earth. And I've got to admit, I thought that was a really optimistic statement. You know, the main challenge to building this megastructure in space is just getting the power back, not actually building it or anything. But, then I kind of remembered, you know, we've got this guy, and he's a rocket in more way than one.

So obviously, this is definitely going to happen. Also, I've been watching a lot of The Expanse (TV series). I don't know if you know that TV series. But I mean, it's influencing me. Obviously, everything's possible, right? Aside from having decent labor relations between the poor and whatever, but that's another matter. So Max Keiser of all people, he suggested something really clever. Now admittedly he suggested this in response to Blockstream satellite things. I'm not totally sure he understood what he was talking about there, but he's absolutely right, which is if you do have this solar panel stuff and power in space, well the easiest way to get the power back, or at least get the money that power represents, is in terms of Bitcoin. This is something that actually happens right now, which is that you get things called stranded hydropower, stranded wind power, stranded, I mean any power, where you have some power getting generated in place, but you don't actually have the infrastructure to get it off to the grid where it would actually be used. Apparently this is fairly common in parts of China and other countries around it, where they have a hydro plant built 50 years ago for a mine that was never actually profitable. But now you have this hydro plant in the middle of nowhere, why don't you go load up some Bitcoin miners on a set of trucks and plant them there and get your money out that way? Supposedly this works. Well, Max Keiser is absolutely correct. If you did somehow solve the challenge of solar power in space, adding the Bitcoin miners to it is actually pretty easy in comparison.

And if we're going to start talking about what's the effect of this, I mean, let's start with Blockstream satellite. You may have heard of it, it's being beamed through you right now, which is, you've got a bunch of, I shouldn't say repurposed, but rented, geostationary satellites, usually used for, say, broadcasting TV and tons of other stuff, back to Earth. And they're broadcasting blocks. Long story short is they have ground stations, they send the blocks up to the satellites, and they get bounced back down. And anyone with a relatively cheap, actually, satellite receiver can get this. So, why don't we imagine our simplest use case? I'm a space pirate, and I'm going to go and decide, you know what, I want to go put my solar panels up somewhere within that geostationary range, right? Well, question for me is, is this profitable? And there's your geostationary orbit. Being a space pirate, you know, I'll go do the figures approximately. I mean, it's not like it's rocket science or anything. Well, it is actually rocket science, but it's easy rocket science. And geostationary is, including Earth's diameter, it's like, say, 48 000 kilometers. Speed of light set annoyingly slow, 300,000 kilometers per second. You know, if I go find a block, what I want to have happen is for me to get my block to everyone else before someone else goes and finds a block. Because, after all, if they go find a block and I'm just some little guy with my little space barge without that much hashing power, chances are they're going to find another block after me.

So, you know, this latency figure is very important and latency from pretty much anywhere on earth up to geostationary and back again, as you can see, is about .3 seconds, which is actually pretty fast when you compare it to fiber. The reason is really kind of two things. First of all, in space you can go in a straight line, whereas if you actually look at actual fiber maps of where they actually put the cables in the ocean, well, they go and squiggle all over the place because they've got to go around obstacles like undersea mountains. The other thing is the speed of light in fiber is actually slower than in vacuum. I think the number is like 60 or 70% of the normal speed of light. So this is a bit slower, but it's not as slow as you'd think. It's kind of competitive. And more to the point, I mean, let's do the math. How much money am I going to earn? Relative to a miner with a faster internet connection. Well, I mean, we know that blocks are found as a Poisson process, being with all these French people, I hope I pronounced that right. And if I go out on Wikipedia, I can go and look it up and say, oh, no, I guess I'm looking for when there's no blocks, n equals zero for this time interval out of the block interval, which would be t[fn 1] . So, you know, my time interval for the latency is little t. My time interval for the block is big T. Well, I mean, I don't know, I'm just some space pirate. I don't really know how to do this e thing.[1] I mean, why don't I just go and plug it in WolframAlpha?[fn 2] And there's actually a good reason to do this, which is. Intuitively, if I know that blocks are getting found roughly once every 10 minutes, surely if I take a tiny little slice of that… The amount of blocks I should find in, say, one second versus ten minutes should be kind of linear, right? Because the reason why the Poisson process gets a bit weird is because we're talking about, you know, what's the probability of me not finding a block in, say, 60 minutes? That's based on the probability of me not finding the block for the first ten and so on. So obviously that's kind of a more complex equation. Whereas when we're talking about very short intervals, we should expect it to be linear. And what do you know when I go plot e to the x? Just get an idea of what this e thing does. Well, when the exponent is small, it looks like a straight line. So I'll spare you the details, but long story short, you can approximate this pretty easily.[fn 3] And if you guessed that, you'd actually get pretty much the right answer. And you can check this by hand by feeding in more things to WolframAlpha. But point being there, always think, how can you go approximate it, make it simple? And now that we've made it so simple, and I'll tell you when you do this, you'll find that, you know, in this small interval it's a pretty close result. Well, let's see so 0.3 seconds block interval is 600 seconds. Now this is saying what is the probability of no one finding a block in this time t from, you know, their block getting up to me and then back again. Well ninety nine point nine four percent. All right, so ninety nine point nine four percent of the time if I find a block in that interval I will not be orphaned. All right, so this sounds pretty good right. I mean chances are I already have other expenses that matter more than this 0.06 percent like maybe the fact I'm floating around in space and space food is kind of expensive you know and suits are kind of expensive too plus I need to buy all this oxygen so I can breathe. So, that sounds all good.

But, hang on a second… Why would I restrict myself to geostationary orbit? You know, why would I go put my mining so close to Earth? Like, I looked it up and NASA's planned closest probe[fn 4] that'll get to the sun, that's actually apparently a relatively accurate photo, which is, you know, it's about 10 diameters of the sun away. That's really close to the sun. And if they can put a probe there, chances are I can put solar panels there. And if I can put solar panels there, maybe radiation coming on them, I can go and mine for cheaper than the poor guy who has to put solar panels way out in Earth's orbit. You know, maybe I need to launch more. I don't really know, but it's certainly possible. Well, if that guy's operation is cheaper, what happens with the latency? Well, Earth to the sun is like 150 million kilometers. When I plug this into the speed of light delay, just like I did for geostationary, that's like a thousand seconds. That's a good deal more than the block interval. And, if you go put this curve a little further, Wolfram Alpha is telling me I can no longer use such a simple approximation. Plus, I mean, it's so long that chances are someone's already going to find a block. So, surely my space pirate life baking outside the sun is not going to work out and I'm not going to make any money. But what if, what if I'm like, say, a major, you know, organization? Maybe something I could put enough hashing power in one place that this all flips. Because remember the assumption we made. That equation I started off with assumed that the amount of hashing power I had was negligible. If it's the other way around, if I'm, you know, Weyland-Yutani, and between digging up bio-weapons that go and spit out acid or something, I decide to go build some miners by space, I can probably build a majority of hashing power. And in that circumstance, then it's not me that has to worry about this latency, it's Earth. You know, it's their problem, not my problem. And, my hashing power is all in one place, so I don't really care.

And, of course, the issue with that, too, is, again, like… Even if I'm the space pirate, I want to head off to the Sun after Weyland-Yutani has gone and set up their big station. Well, that orbit is pretty big. You know, even that close to the Sun, it's like 12 million kilometers. Again, let's look at the speed of light delays because chances are Weyland's not going to be too happy if I always position my equipment right next to them. I mean, what if I want to have it on the other side of the orbit? Well, looks like you've got like 80 seconds worth of delay.[fn 5] I mean, I'm not going to bother doing the calculations. I can just think for myself, how likely is it for someone to go get a block in like 80 seconds? I don't know, pretty likely. And I'm already spending all this money on spacesuits. I mean, I don't want to go and spend another 10% of my income there. I mean, this doesn't sound very good. At the very least, I might be incentivized to stick my equipment closer to where they are. But that's also not good for Bitcoin because, I mean, what if there's a solar flare? Do we really want all the Bitcoin hashing power getting knocked out by one event? I mean, it could happen, you know, that's what the numbers are telling me, I'm incentivized. So… We kind of have an interesting conclusion here, which is decentralized consensus only works if it's centralized. And, well, what do we mean by centralized? I mean centralized in terms of the light cone, if you will, which is, are you close enough to your other participants in the so-called decentralized consensus to come to consensus in a reasonable amount of time? I mean, after all, we could go and tweak the parameters and so on. We could make the entirety of our solar system essentially be very close to each other in relation to the block interval, but, you know, if I'm not going to do a hard fork just because I want to do space mining, I mean, maybe we have an issue there.

Equally, you know, if I'm some new altcoin designer and I go say, well, you know, ten minutes is a really long time to wait for a retail transaction, and I turn that knob down to 15 seconds, maybe it's not space miners I'm worried about. Maybe it's people behind the Great Firewall of China. It's kind of a bit of a problem, really. And, you know, as much as I gave my example in terms of space, I mean, the equations don't care how far you're away. All they care about is ratios. So I'm gonna go leave you with that very basic lesson in a bit of latency and let's get some questions.

QUESTION: Hi, I wanted just to ask a question and then make a short statement, just it's probably worth noting that if the majority of mining power is close to the Sun and you said that it's not the Sun's problem, it's the Earth's problem, that means that if still the majority of people live on Earth and the majority of transactions happen on Earth, that means that confirmation times will be huge and the network will be pretty much unusable for the users on Earth. And that means that the economic value of the network will drop and so will the mining rewards.

TODD: Well, so I may just go summarize that, essentially the argument there is because the users are very likely in Earth orbit and the hashing power is off much closer to the Sun, we would have long confirmations and the value of Bitcoin would drop, etc. etc. But, remember what my figure said. From here to the Sun is a thousand seconds, roughly. Well, can we name a cryptocurrency that's maintained its value even though… Blocks were, I mean I should say even the transactions were confirmed in more than say ten minutes? I mean for that matter multiple hours? Why don't we go and, yeah. I mean Bitcoin's distribution of blocks being found is already kind of erratic. There's a good chance if I go send you money it's going to take an hour to arrive. Also, we have another example which is a fork of Bitcoin called Bcash. And Bcash is just fine with multi-hour block intervals. I mean, last I checked, Bcash is still worth quite a bit of money, and if I went and sold Bcash at the dip, I wouldn't have been a very smart person to do that. So, I think right there, what we really got to look at is, well, what are we trying to go prevent? And my argument would be: well, let's go prevent centralization of where the hashing power is, because at least in my rather contrived example of space mining, if one solar flare can take it all out, we're in deep trouble. But you might not have the same design parameters. So really, this gets back to, well, what is your coin supposed to do?

QUESTION: Hi Peter, so I like your approach but you forgot one big thing, so first of all you should construct the miners and this also requires a lot of money and the weight of miners will also be huge, 300 petahashes weight is approximately 150 tons, so if you are using Falcon because you mentioned the Musk, it will cost you 7 000 per kilo, so I don't think it's a good idea. Also transferring a lot of miners, it's not the smallest problem. Cooling and also you should support and maintain all the miners.

TODD: Now can I remind you that Elon Musk is literally a rocket, I mean I agree with you, I think doing this right now would be roughly as crazy as using a home-grown hash function in a two-billion-dollar cryptocurrency.[2][3][4][5]But, look at that guy's face. I mean, maybe it costs $7,000 a kilogram right now, but what do the lower limits due to physics? I mean, maybe it'll cost $7 a kilogram, maybe $70. And the interesting thing is, well, for this particular example, A, I can potentially build things a lot cheaper in space, because once I get my robots going and whatnot, I don't actually necessarily need a structure to it. You know? Space, there's nothing to blow things away. I can just, like, stretch out my miners and solar panels in some big, very, very flimsy role. So, I don't know. Maybe that'll happen. Equally, if you look at embodied energy, if I can make these things last, which, admittedly, is kind of hard, because there's a lot of radiation from the sun, but if I could, the embodied energy of something in orbit, to give you a rough idea, it's as though we're made out of gasoline. And in a world where we have solar space mining, chances are we bootstrap to the point where energy is actually pretty cheap, and in a world where we can go land rockets on their feet, or we're not using them up. So, I mean, I'm not going to say no. And this may very well be an issue. I sure hope so. I really think that'd be cool.

QUESTION: So, assuming we don't just have a single miner, but there's actually competition between different entities, aren't they incentivized to have their hash power spread out from each other because a miner who can attack another miner's hashing power gets more percentage of the network themselves. So even if they're all mining in orbit around the sun, as you showed, they would be spreading out their satellites.

TODD: Well, so if I were Weyland-Yutani, I mean, I'm already willing to go and send people to their death to go and get bio-weapons that could destroy all of humanity. So surely I might not be thinking too much about what I'm actually doing. I might be thinking about the PR implications of it. Can I keep Bitcoin valuable? And it may work out that I could just split up my hashing power into what appears to be multiple independent entities, but it's actually all the same group. If I'm off mining at the sun, I mean, how likely is it that people can actually verify where that hashing power is? I mean, for that matter, I have a hard enough time figuring out where it is, even on Earth, and in theory I can visit most of the places. So, I mean, I think you're kind of right in principle, but there's a lot of second-order and third-order exploits, essentially, against these economic incentives, because the people with this hashing power can go lie about it. They can go lie about who owns it, who has it, and so on, and it's very difficult for that information to then propagate back to the crypto coin and, you know, actually influence the value. Equally, remind me again how much IOTA's dropped, because if even a broken hashing function can't go make things drop in that much value, I mean, maybe these coins don't really have much of a feedback loop in whether or not they actually work. In fact, I should test this. I should try launching a centralized coin where I just go sign all the blocks.

QUESTION: Hey Peter, thanks for the talk. Have you heard there's a scheme called uncles[fn 6] where instead of having blocks that go you have if there's a race between blocks then both of them get paid somehow? Do you know if this scheme is incentive compatible? Does it work?

TODD: It's very complex, and from what I've heard from the authors of this, they're a lot less certain that this actually works than it used to be. You know, because you can still be in situate… After all, you can't allow uncles to go and earn money indefinitely, because at some point there isn't necessarily an incentive to come to consensus. So there's still going to be some form of this, you know, and how much you twist and turn the parameters one way or another, I mean, it's hard to say. I mean, equally, you could imagine a different model where we just push block controls very, very long, but we make confirmations be parallelized over multiple different chains at once, such that you can imagine a system where, if at least, you know, two or three chains show a confirmation, we assume other ones will, and the effect could very well be a shorter block control. Similarly, you've got things such as the weak… I'm not going to remember exactly which term is the latest for this, but essentially the weak blocks idea where you say… While my block may not be fully confirmed, I can partially confirm it by saying I found a block hash that was close to the target, not all the way. And that close to the target means obviously I was actually mining on those transactions. Thus, it's evidence that they will confirm, even if it isn't 100% evidence. So, I mean, there's a lot more to cover there, but I think there's a whole bunch of interesting possibilities there and I'm not going to necessarily rule them all out. But equally, I'm going to be skeptical about all of them.

QUESTION: Hi Peter, thank you for the presentation, it was very thought-provoking. The question I have is, all calculations on profitability for miners is based on the orphan rate, which itself is based on latency and the assumption that it's expensive to verify a block and that it is cheap to propagate a transaction. However, in your model, it becomes expensive to propagate a transaction as well because that thousand-second latency now applies to the people pushing the transactions and it creates a situation where there is a new flurry of attacks that you can perform upon a miner that's very powerful, but very far away from the people that emit the transactions, which is to refuse them some. And since we're talking satellites and, well, the rocket, it's going to be at some point in the future, therefore you could assume that the reward for a miner in a block is going to be overwhelmingly based on transaction rewards, the fees, as opposed to a coinbase, which is going to degrade as we go. So has your model thought about it in this way at all?

TODD: Well, I think your premise there is incorrect, because it's actually very cheap to go and broadcast information through space. I mean, even when you talk about true extremes like contacting Voyager probes, one of the figures I love that came out of a NASA press release once is that per byte, for them to talk to the Voyager probes, which are the furthest man-made objects ever, costs less than your average US cell phone company charges for a text message. Now, maybe that says something bad about cell phone companies, but, you know, that's really not that expensive. When you talk about, say, transactions, which are relatively small, the cost to actually broadcast that information is low. And by the time you get, you know, something like space mining, it's going to get lower and lower because you do things like just fire a laser at your target. The fundamental reasons why information is cheap to broadcast on land don't change when you move things into space. The fixed costs go up because of course it's space and your contractors gouge you. But, the things that actually scale don't change. After all, it's still going to end up being light pointing from one place to another. And whether or not it goes through a fiber or just goes through empty space, it's not going to be that different.

QUESTION: Peter, thank you for your presentation. It makes me wonder in the future when we're constructing, say, colonies all through the solar system, do you see a need for multiple chains with essentially the same features but with hashing power localized to the users and then perhaps a chain with a much longer lock time in order to tie them together?

TODD: This is being brought up even right now for trying to get very short interval cryptocurrencies. The ultimate problem is how do you actually enforce this linkage of users to hashing power? You know, in my example, I mean, the users were still all on Earth, and the hashing power still moved off to wherever it was cheapest. And there is no easy way to enforce that, unless you somehow tie users to hashing power, which maybe could do in some kind of proof-of-stake-ish thing. But even then, you may very well just end up in a situation where people rent out their coins to the hashing power, which is even further away. And now you have the situation where not only does your mining power go out, but also you may lose your coins. I mean, there may not be easy answers to this problem. And frankly, it wouldn't surprise me if there aren't easy answers to this problem. You know, we could be in a situation where the first time we launch our interplanetary probe off to Alpha Centauri, ten years when the signals finally get back, we have a massive block reorg, because they also had Bitcoin. We just don't know. And unfortunately, I don't have good answers to this.

QUESTION: In case they discover subspatial transmission or hyperspace drive, do you think we should do something like prepare softfork now to address that like ASIC Boost?

TODD: You know, I mean, that's a hard one. Like, if I say yes, I'd give it a 99.9% chance I'm going to really embarrass myself. But, if I were creating a currency which had a much shorter interval, I think I would be leaning very hard on the people behind it, and the community in general, to maybe consider a longer interval to be more robust, because while I think Bitcoin does have fairly good resistance to at least earth-borne threats, it's easy to go and make this go away, and if anything, it's kind of surprising how close these numbers are. Once you go add maybe a bit of validation time and so on, and maybe a bit of censorship, then 10 minutes may seem a little dubious. Certainly if it was say 60 seconds or 15 seconds, so that's my advice. Thank you.

History

See also

External links

References

  1. ↑ Baltakatei: 2024-02-11:
  2. ↑ Baltakatei: 2024-02-11: I think this is a reference to a security vulnerability reported on 2017-09-08 in IOTAʼs Curl-P-27 hash function.
  3. ↑ Ethan Heilman; Neha Narula; Garrett Tanzer; James Lovejoy; Michael Colavita; Madars Virza; Tadge Dryja. (2020). “Cryptanalysis of Curl-P and Other Attacks on the IOTA Cryptocurrency”. IACR Transactions on Symmetric Cryptology, 2020(3), 367–391. DOI: https://doi.org/10.13154/tosc.v2020.i3.367-391 .
  4. ↑ Amy Castor. (2017-09-07). “MIT And BU Researchers Uncover Critical Security Flaw In $2B Cryptocurrency IOTA”. Accessed 2024-02-11. Archived from the original on 2017-09-08.
  5. ↑ Neha Narula. (2017-09-07). “Cryptographic vulnerabilities in IOTA”. medium.com. Accessed 2024-02-11. Archived from the original on 2017-09-09. “Last month, Ethan Heilman, Tadge Dryja, Madars Virza, and I took a look at IOTA, currently the 8th largest cryptocurrency with a $1.9B market cap. In its repositories on GitHub, we found a serious vulnerability — the IOTA developers had written their own hash function, Curl, and it produced collisions (when different inputs hash to the same output). Once we developed our attack, we could find collisions using commodity hardware within just a few minutes, and forge signatures on IOTA payments. We informed the IOTA developers, they patched their system, and we wrote a vulnerability report.”

Footnotes

Presentation equations: PDF

  1. ↑ Equation 2
  2. ↑ See link.
  3. ↑ See equation 3 and 4.
  4. ↑ See Parker Solar Probe.
  5. ↑ See Equation 6.
  6. ↑ Baltakatei: 2024-02-11: See “ommer block”. This is Ethereum terminology used in a mechanism in which a block reward could automatically be shared with miners of orphaned blocks. The mechanism was deprecated after Ethereum moved from proof-of-work to proof-of-stake on 2022-09-15 during “The Merge”.
Retrieved from "https://reboil.com/mediawiki/index.php?title=Solar_Powered_Space_Pirates:_A_Threat_to_Bitcoin%3F&oldid=197546"