บทถอดความฉบับเต็มของ 124. Mining Forks, Firmware Red Teams, and the Future of Sovereign Hashrate

what would you do if you wanted to change the proof of work algorithm? 58 00:04:35.405 --> 00:04:51.060 Because it is as we we noticed, you know, Bitcoin mining is is fairly centralized. There's a lot of big players with a large, overwhelmingly large share of the network. So what would you do? And I was initially interested in what they were doing because I thought that they were going to change 59 00:04:51.060 --> 00:04:54.580 the hash algorithm or the hash, construction 60 00:04:54.740 --> 00:05:03.085 to make it so that you couldn't use ASICs to mine it, which would be really easy to do. You could stick with SHA two fifty six and just change the hash construction, 61 00:05:03.165 --> 00:05:28.044 and none of the existing Bitcoin miners would work, which would be very interesting. It'd be kind of like going back to the genesis of Bitcoin, right, when there just was no one had developed any ASICs and seeing, you know, how that goes. But, unfortunately, they didn't do that. They just picked the lock a two b and the exact same hash construction that SiaCoin uses. So there are existing ASICs available that will you could plug and play and mine LukeCoin, 62 00:05:28.044 --> 00:05:40.720 including ASICs from Bitmain. Yeah. That seems to be an interesting decision. I saw some decision matrix table that Luke was posting because you'd think if you want to avoid a first mover advantage 63 00:05:40.880 --> 00:05:55.025 with an ASIC chip, you would pick a hashing algorithm that starts with CPU mining. Right? Yeah. And that's that's really easy to do. It wouldn't be hard. You could stick with SHA two fifty six. You just change sort of the way the block header is arranged, 64 00:05:55.025 --> 00:06:03.745 and then you could be certain that no existing mining ASIC supported it and could mine to it. And then you would you would force the whole fork, 65 00:06:03.905 --> 00:06:08.220 the whole coin to go through that discovery process of, like, 66 00:06:08.700 --> 00:06:18.620 maturing and getting to the point in which someone would be willing to drop millions of dollars to build an ASIC to mine it. They decide I guess they decided they didn't wanna do that. So so now 67 00:06:18.925 --> 00:06:37.400 Bitmain is just gonna have the largest market share for these ASIC miners as well? I mean, I don't know who actually has these chips. Clearly, Bitmain has the mass set and could make these chips. I think I saw that it was on, like, 28 nanometers Yeah. The last one that they made. So now the chips out there aren't efficient by Bitcoin mining standards. 68 00:06:37.800 --> 00:06:46.360 But, yeah, Bitcoin, if they wanted to, has a clear advantage here. In fact, that's what they did to SiaCoin. I was reading a little bit about this. Apparently, 69 00:06:47.015 --> 00:06:49.575 there was a couple companies making Siacoin 70 00:06:49.575 --> 00:07:05.639 ASICs and and machines to mine it, and then Bitmain came in and just they just dominated like they do. And I think they actually forked Siacoin to to sort of to rug Bitmain and change the mining algorithm. I think Epic was involved in that and potentially 71 00:07:05.639 --> 00:07:11.160 some of the people that later went on to form Foundation Devices with the Passport Wallet. It's, 72 00:07:11.880 --> 00:07:19.115 yeah. It's very interesting. But basically, of the gate, someone's got the chips and the miners that can just plug and play on this. They will see What 73 00:07:20.795 --> 00:07:22.795 was that? Sorry. They will see 74 00:07:23.115 --> 00:07:25.995 I didn't mean to interrupt you. When when Lootcoin 75 00:07:25.995 --> 00:07:35.460 actually launches, they will see, like, much higher hash rate numbers, which is maybe that's what they wanted, but I don't think it means any sort of decentralization 76 00:07:35.460 --> 00:07:38.660 or any sort of resistance towards centralizing 77 00:07:38.660 --> 00:07:44.259 again. Who what was the link between the Foundation Devices team and and 78 00:07:44.595 --> 00:07:46.035 is it SiaCoin? 79 00:07:46.035 --> 00:07:47.315 Yeah. SiaCoin. 80 00:07:47.315 --> 00:07:57.555 I believe some of the people that started Foundation Devices previously worked on SiaCoin miners. I see. Okay. Yeah. Yeah. And there there was a bunch of drama there about this, like, hard fork to essentially 81 00:07:58.920 --> 00:08:11.560 rug Bitmain. They, like, changed the algorithm somehow. I don't know I don't know how. You know, Bitmain was was dominating the network, and they're like, we don't like this, so we're just gonna change the algorithm. And I believe, from what I read, I believe that 82 00:08:11.805 --> 00:08:14.285 Luke Koin is using the 83 00:08:15.085 --> 00:08:17.805 sort of original Sia coin mining 84 00:08:17.965 --> 00:08:18.925 strategy. 85 00:08:18.925 --> 00:08:35.630 The Bitmain dominated one. The Bitmain dominated one. I believe so. How strategic do you think that move was on Luke's part? Like, do do you think he wanted to pick a mining algorithm that had abundant available hardware and that that was a motivating factor? Or 86 00:08:36.029 --> 00:08:47.845 Yeah. Because he That matrix that Tyler was mentioning, I looked at that, you know, he had there's, I don't know, 20 plus options in there that they were looking at, and they were sort of rating them on, you know, how many ASICs were available. 87 00:08:48.245 --> 00:08:51.045 So, yeah, they they were. They were definitely looking at that. 88 00:08:52.889 --> 00:08:57.769 It's just, like, Bitcoin didn't have this problem. Well, it's funny in that matrix. 89 00:08:57.850 --> 00:09:06.170 Just important comment. Under the stock SHA two fifty six d hash, it's rated poor for mature ASIC competition 90 00:09:06.335 --> 00:09:11.055 because it says exclude from the algorithm lottery ASIC boost history, 91 00:09:11.135 --> 00:09:15.055 accumulated physical design expertise, and closed firmware advantages. 92 00:09:15.135 --> 00:09:16.495 Supply chain enrichment 93

00:09:16.495 --> 00:09:42.435 remain. It's like, yep. That is true. They're like, they have the closed source firmware advantages as like a a pro of of that of using that algorithm? So this is the description current judgment of SHA two fifty six d hash mature ASIC competition is poor from a fresh competitive market perspective. Okay. Okay. Okay. I mean, couldn't pick SHA two fifty six d for a lot of reasons, 94 00:09:42.515 --> 00:09:53.410 but I would say it probably has the best competitive landscape. Right? There's lots of there's lots of well, not lots, but there's several manufacturers of SHA-two 56 d hashing 95 00:09:54.130 --> 00:09:59.890 ASICs. Yeah. I just find it interesting that they flag in their notes that ASIC boost 96 00:09:59.890 --> 00:10:01.810 and supply chain enrichment, 97 00:10:01.810 --> 00:10:02.850 firmware advantages, 98 00:10:03.215 --> 00:10:05.695 accumulated physical design expertise, 99 00:10:06.335 --> 00:10:10.575 these are all flagged as reasons why they rated it as poor. 100 00:10:11.215 --> 00:10:14.815 But none of those are solved with their choice either? Well, they're also just like 101 00:10:15.310 --> 00:10:20.270 that doesn't have to be the end state of the ASIC mining maturity 102 00:10:20.430 --> 00:10:22.670 world for SHA2560 103 00:10:22.670 --> 00:10:28.110 hash. That's the whole point of what we work on. True. Yeah. I mean, they couldn't use SHA2560 104 00:10:28.110 --> 00:10:46.490 because there's a ton of miners out there, and they would they would get 51% attacked, like, immediately. That it makes me wonder, like, why do you think we're not seeing an attack on Luke coin right now? Because it seems like if anyone wanted to make his little project come to a screeching halt, this would be the opportune 105 00:10:46.490 --> 00:10:49.930 time to do it. I don't think it's live yet. What's 106 00:10:49.930 --> 00:10:57.290 the isn't the, like what is it? Five blocks now? Isn't that their their blockchain? No. They're targeting, I think, a September 107 00:10:57.375 --> 00:11:08.495 proof of work change. So what's going on with those with that, like, fork in the meantime? Like, why are people still mining to it then? I don't know. I don't Why would you mind to that? That doesn't make any sense. I 108 00:11:09.055 --> 00:11:14.940 I guess when they do do the proof of work change, they'll probably go from that chain tip. So maybe 109 00:11:15.100 --> 00:11:33.995 that's when you could get paid out if you're mining to that. That must be it because they're also well, Luke is claiming that Bitcoin is being attacked and is still at whatever that block height is, three three two or three three three. Right. Okay. So you're right, Iko. I guess you could attack it now. Right? And you could with two exahashes, 110 00:11:33.995 --> 00:11:34.795 you could 111 00:11:35.115 --> 00:11:37.995 you could 51% attack it or less than two exahashes. 112 00:11:38.630 --> 00:11:56.274 Yeah. With like half an exahash,

you could Right? Because it has like one exahash right now. Yeah. With one telehash? I'm kidding. Yeah. Yeah. You could 51% attack it. That would be pretty funny. I mean, yeah, there's a pretty big cost to doing that, but And like, don't they have to get through all 2,016 113 00:11:56.274 --> 00:11:59.394 blocks? Like they still have like 2,011 114 00:11:59.394 --> 00:12:15.649 blocks to go before they see a downward difficulty adjustment, if I'm not mistaken. Yeah. Right. That's never gonna happen. They're definitely not gonna see a difficulty adjustment at this rate. And now that they've selected this hashing algorithm, Scott, there was also some fun banter over the weekend 115 00:12:16.050 --> 00:12:24.964 about Bidax support that you chimed in on. What's going on there? Yeah. Luke was tagged me in a post that was like, when Bidax? And Marshall 116 00:12:24.964 --> 00:12:26.325 made a PR 117 00:12:26.404 --> 00:12:31.765 to the ESP Miner repository with some, like, vibe coded nonsense 118 00:12:31.765 --> 00:12:35.180 about supporting the Bitmain ASIC in ESP Miner. 119 00:12:35.420 --> 00:12:36.060 The 120 00:12:37.020 --> 00:12:39.260 the Block A two b ASIC. 121 00:12:39.340 --> 00:12:41.420 And that 122 00:12:41.580 --> 00:12:51.865 was that was immediately closed and flagged as spam because that is nonsense. We are not doing that. Open source project, you can fork and do whatever you want, 123 00:12:52.265 --> 00:12:54.904 but that is certainly not getting, 124 00:12:56.505 --> 00:13:24.515 mainland in the ESP project ESP Miner project. So That's an important distinction. I saw some people saying, you know, you can claim open source, but whoever controls the repo controls the project. And it's like, well, really, you can fork it. And if the community follows you, then then that's the answer there. It's not all the power is who's the maintainer of the GitHub organization. Yeah. Yeah. I Exactly. You can just do what you want. And that that is the power of open source is that they can 125 00:13:25.235 --> 00:13:28.595 they can take this and fork it and support whatever stupid 126 00:13:28.595 --> 00:13:30.995 fork they wanna go knock yourself out. 127 00:13:31.740 --> 00:13:35.020 There seems to be no shortage of misunderstanding 128 00:13:35.020 --> 00:13:44.300 about how open source works. It's just like every day I see some new comment and I'm just like, god, people just don't get it. Yes. Very much misunderstood. 129 00:13:44.300 --> 00:13:46.825 That is right. That is that it's 130 00:13:47.305 --> 00:13:54.745 I don't know. I don't really follow it but yeah, there's a lot of it out there. It's like, it's some sort of like money making 131 00:13:54.825 --> 00:13:55.785 scheme or 132 00:13:56.105 --> 00:14:01.065 something like that. I just I don't get it. I don't know. I'm also not getting the money so Right. 133 00:14:01.629 --> 00:14:28.495 It's a pretty bad scam if it is one. Yeah. Yeah. It's funny. Yeah. Well, it is funny. But it was it was it is fun to think about how you would how you would change proof of work if you're trying to make a new coin. There there's a pretty awesome quote from Bob Burnett that's like, Bitcoin is a once in, like, three thousand year opportunity that that can't be repeated. And I think that's exactly right. You you can't just, like, recreate the Beatles. Right? Because the Beatles already did it. The the set of circumstances that led to that are not reproducible. 134 00:14:28.495 --> 00:14:30.735 Yeah. Yeah. I think there's something to be said for that. 135 00:14:31.529 --> 00:14:34.330 This Block A hashing algorithm, 136 00:14:34.570 --> 00:14:43.850 can can people use CPU to mine to it or it it requires a ASIC chip? No. You can definitely do it with a CPU, GPU, FPGA, 137 00:14:43.850 --> 00:14:46.655 for sure. And, like, and do it competitively? 138 00:14:46.735 --> 00:14:59.455 Well, not against even a 28 nanometer ASIC. No. Okay. Yeah. Definitely not. Even a so a 28 nanometer ASIC is orders of magnitude more efficient than a four nanometer CPU? Yes. 139 00:15:01.620 --> 00:15:02.820 Gotcha. Absolutely. 140 00:15:03.140 --> 00:15:06.020 Yeah. I guess ASICs are really good jobs. 141 00:15:06.180 --> 00:15:08.740 Going with this is if, like, if there is 142 00:15:09.060 --> 00:15:16.155 a possibility to just use CPUs, like like, you see this in Monero. Like, they went with an ASIC resistant 143 00:15:16.155 --> 00:15:25.595 proof of work algorithm. And I think one of the inherent risks in doing that is you can potentially just spin up a bunch of CPUs 144 00:15:25.995 --> 00:15:29.515 in the cloud to attack the chain or possibly GPUs. 145 00:15:29.490 --> 00:15:40.610 So So I don't know. That was just one thing I was thinking. But if if there's a I guess if there's enough hash rate out there, honest hash rate on Luke's fork, then doesn't really matter. You know, you have to pay to spin up GPUs. 146 00:15:41.010 --> 00:15:45.875 Mhmm. So it's not without risk. Just like, you know, it costs money to 147 00:15:46.915 --> 00:15:48.915 to plug in a bunch of ASICs. 148 00:15:49.555 --> 00:15:57.475 I don't I don't know. I mean, I don't know if that really holds up. Yeah. How does the ASIC resistance argument work? Well, ASIC resistance argument from from Monero 149 00:15:57.899 --> 00:16:01.019 is that they wanted to be more democratized. 150 00:16:01.019 --> 00:16:05.899 Right? That that more people can get more people can be competitive miners because 151 00:16:06.139 --> 00:16:15.155 Mhmm. There isn't these expensive ASICs. Yeah. What they do is kinda interesting. It's like they they can change the the hashing algorithm. 152 00:16:15.235 --> 00:16:17.635 It's like baked into the protocols. 153 00:16:17.635 --> 00:16:21.715 They they it just changes. So you need a general purpose processing platform? 154 00:16:21.875 --> 00:16:41.010 Yeah. Exactly. And it it it's it's like architected in a way that doesn't really give GPUs an advantage.

So, really, you wanna be doing it with a CPU such that it has the flexibility to change the hashing algorithm. But to Eco's point, a state actor could just spin up a big VPS. Right. Or they could give a big miner a call and say, 155 00:16:41.535 --> 00:16:51.375 do this with your a six. Mhmm. Yeah. That's why I I was curious because it doesn't seem like a like it actually fixes the underlying attack on either algorithm. 156 00:16:51.375 --> 00:16:55.770 Yeah. No. I don't I don't I wasn't saying I think it's a good strategy. 157 00:16:55.930 --> 00:16:59.850 I was just pointing out that it was it was it was that they 158 00:17:00.250 --> 00:17:06.570 they wanted to move away from ASICs for one reason, but it it just does at the end of the day, it doesn't really solve the problem. 159 00:17:06.810 --> 00:17:08.890 Yeah. They're just switching to different ASICs. 160 00:17:08.890 --> 00:17:13.624 It's just Right. Temporarily delaying the problem. But you know what? The the 161 00:17:14.745 --> 00:17:16.505 best foolproof 162 00:17:16.505 --> 00:17:19.224 will not ever be broken ASIC resistance 163 00:17:19.225 --> 00:17:33.929 is to have your coin be worthless. No one will mine it. Yeah. No one will build ASICs, and there will never be any ASICs for it because it's just It's not worth the r and d. It's not worth the r and d. So it's solved right there. I I still think that people could make ASICs 164 00:17:34.010 --> 00:17:35.529 to mine Monero, 165 00:17:35.690 --> 00:17:36.970 but it's just like, 166 00:17:38.535 --> 00:17:43.175 it's not worth it. Yeah. Has that played out with other proof of work shitcoins 167 00:17:43.175 --> 00:17:53.975 where someone does spend the r and d to make a miner and then another generation doesn't come out because the coin didn't survive long enough to justify it? Kind of with, what was that one? They used 168 00:17:54.590 --> 00:18:02.430 SHA three. I'm blanking on the name. There was there was a shitcoin that came out, and it was really interesting strategy because the the ASIC manufacturers 169 00:18:02.430 --> 00:18:04.429 basically created the cryptocurrency. 170 00:18:04.590 --> 00:18:06.510 So it's like, they're like, we're gonna make this chip, 171 00:18:07.095 --> 00:18:09.015 and then we're gonna create the cryptocurrency 172 00:18:09.015 --> 00:18:14.855 to go along with it to sell our chips. Right? It's really like fucked up incentives. 173 00:18:15.015 --> 00:18:37.235 What was that one called? I can't keep track of all these things, but Sounds just like fiat. Yeah. I mean, I think it's basically fizzled. They had a like, instead of a blockchain, they had a DAG and they had SHA three, which is supposed to be way better. And they had, like, you know, block times in the seconds and, you know, it's all hyped up, but it was really just kind of dogfooding this this ASIC manufacturer. 174 00:18:37.235 --> 00:18:53.900 What if you required every miner to have a full non pruned node? You add a big storage

requirement to it. It's interesting. Like every mining device had to be paired with a node? I I mean, that would just be so much redundant overhead unnecessarily. 175 00:18:53.980 --> 00:18:58.060 Like, imagine you had a whole warehouse of miners. Well, I think it would disincentivize 176 00:18:58.060 --> 00:19:05.180 warehouses of miners. Back in the, like, altcoin craze, there was, like, miners, you know, shitcoins that required, 177 00:19:05.135 --> 00:19:06.414 like, proof of, 178 00:19:07.294 --> 00:19:09.455 like, hard drive and all that. 179 00:19:11.135 --> 00:19:19.855 You know, it was they're they were trying to, like, somehow tied in that you had, like, disk space available, and that was your proof of work. So it could be something like that, I guess. 180 00:19:20.730 --> 00:19:21.530 Filecoin. 181 00:19:21.530 --> 00:19:27.929 Yeah. That's right. Maybe it was Filecoin. And there was that other one, Helium, that was, like, proof of, like, RF coverage. 182 00:19:28.090 --> 00:19:30.090 Like, it was this wireless network 183 00:19:30.250 --> 00:20:17.905 and you had like, you got paid by, like, how many other nodes were, like, in RF range of you. It's like an interesting idea, but easily gameable. Yeah. Well, it's certainly interesting to watch this play out because we don't get opportunities in Bitcoin all the time to see these attempts. So thought experiments are for sure worth worth watching. Yeah. It's a it's a fun thought experiment. I think I've probably dedicated enough of my brain space to this and Here you go. I'm done with those thought experiments on on Lootcoin. Yeah. I'm just waiting for the day that they, like, just disappear from the, like, social layer of Bitcoin, Twitter. Like, if they're forking off, like, just fork off and stop talking about Bitcoin already. Yeah. I I'm looking forward to that too. I think it's rapidly approaching. I think as the, you know, the influencers 184 00:20:17.905 --> 00:20:27.200 stop talking about it and move on to the next big thing, it's probably done. Well, and certainly, Ocean can't support both these endpoints forever. They're gonna have to make a decision. 185 00:20:27.519 --> 00:20:30.080 They certainly can't call both of them Bitcoin, 186 00:20:30.080 --> 00:20:33.695 which they're currently doing. I don't know. That's why we also claim 187 00:20:33.855 --> 00:20:36.575 Ocean's a little better about it, right? I won't say their organization, 188 00:20:37.215 --> 00:20:38.494 but it's 189 00:20:38.495 --> 00:20:45.534 definitely still misleading and confusing when you go to their website. It'll be interesting to see what Ocean does now with Luke on sabbatical. 190 00:20:45.890 --> 00:20:58.850 I don't think he's coming back. In my personal opinion, we're gonna That's usually how sabbaticals go. Yeah. He's Luke's not coming back to Ocean. Mechanic's not coming back to Ocean. Luke coin is just gonna be this, like, persistent 191 00:20:59.065 --> 00:21:08.825 failing project in the background, like b cash and b s v have been, where you just get these trolls that just are obsessed with Bitcoin and can't leave it alone, 192 00:21:09.065 --> 00:21:20.520 but, like, just have this failing project that nobody wants to hear hear about or cares about. I don't know. I don't know where I don't know what what mechanic's fate is. I could see him being like a news correspondent 193 00:21:20.520 --> 00:21:37.805 for the for one of these crypto magazines. I don't know. Yeah. I don't know. I don't know. It's really wild. It's fun to think about. But I yeah. I they they they have forked off and are rapidly finding out. It is interesting to me that Mechanic got fired from Start Nine, 194 00:21:38.230 --> 00:21:41.429 and then Start Nine was a pretty vocal 195 00:21:41.590 --> 00:21:43.909 supporter of BIP one ten. 196 00:21:46.230 --> 00:21:50.309 Say it least. I found that interesting. Yeah. That was interesting. I think they supported optionality. 197 00:21:51.115 --> 00:22:03.914 I I mean, the tweets I read from their account were very, very confident about BIP one ten and it and its success. That was interesting what we were talking about the other day is that Datum technically supports 198 00:22:03.995 --> 00:22:11.590 both Bitcoin Core and Knots. But the way it's configured on Start nine, it only supports Notts. Yeah. What did Dylan say to 199 00:22:12.070 --> 00:22:15.990 to do the signaling? It's only possible through the Notts implementation? 200 00:22:15.990 --> 00:22:23.994 You wanna signal? Like, can you can I just, like, can I just use Datum and Core and just mine to the normal, like, ocean 201 00:22:24.075 --> 00:22:25.674 main chain endpoint? 202 00:22:25.835 --> 00:22:29.994 Does that work? I don't think so, but I won't confirm. 203 00:22:30.395 --> 00:22:40.369 I'm looking at Domus Master. I remember asking Mechanic about this, like in person when I saw him once and he was like, oh, yeah. Yeah. You totally can. But I could see why they'd be incentivized to not make it easy to 204 00:22:41.570 --> 00:22:51.955 to use Datum with Bitcoin Core. Yeah. Dylan said you need to be able to configure block notify to specifically notify the Datum service, and you can't do that with Core on StartOS, 205 00:22:52.115 --> 00:22:55.955 but Not allows you to. That was at least on o.3.five. 206 00:22:55.955 --> 00:23:00.835 They just did a p overhaul, so it might be possible now. Yeah. That seems like a weird technicality that, 207 00:23:02.170 --> 00:23:06.490 block notify doesn't work. It's like every pool needs block notify. That's pretty important. So 208 00:23:07.130 --> 00:23:20.375 should be able to get it to work with Datum, especially since the Datum client is open source. Right. Well Tell tell me about two fifty six Red Team. Yeah. Schnitzel had another banger of a tweet today. I saw him put out. It was a cool photo of 209 00:23:21.815 --> 00:23:25.095 his, like, test bench setup, his poor wife's garage. 210

00:23:25.975 --> 00:23:29.735 This is it looks like APW three power supply, 211 00:23:29.815 --> 00:23:47.945 some switches, and then there's six control boards on this wooden plank. And it highlights so if for anyone that missed some updates from Schnitzel on behalf of the two five six Foundation talking about the red teaming efforts, which Scott will explain much better than I can when I'm done ranting here. On closed source firmwares, 212 00:23:47.945 --> 00:23:51.304 be it from the ASIC manufacturers, the OEMs themselves, 213 00:23:51.305 --> 00:24:08.340 or these aftermarket firmware providers, the jailbreak solutions like FinishBrains and LuxOS, these are closed source alternatives, but with open source decompiling tools like Gidra and really powerful open weight models that don't say no to you when you ask them to do things, you can 214 00:24:08.659 --> 00:24:17.565 reverse engineer a lot of how this firmware works. And so that's what Schnitzel's been doing as part of our two five six foundation red team efforts. And I just wanna highlight that firmware 215 00:24:17.565 --> 00:24:18.445 assessment 216 00:24:18.445 --> 00:24:21.164 inherently requires testing on hardware. 217 00:24:21.325 --> 00:24:42.210 And so this is a bit more complicated of a setup than say just pointing an AI tool at an open source repo for a software project. And so but it's been really fun to watch. I think you've been more in the trenches with him, Scott. Why don't you tell us what's going on? That that is a really interesting point. Right? You think about someone who's red teaming just general purpose open source Bitcoin software. 218 00:24:42.664 --> 00:24:51.144 They can just run it on their computers. And you can attack all the exposed endpoints and things like that and and do kind of like a a black box 219 00:24:51.544 --> 00:24:57.225 attack on it, where you're not even looking at the source. You're just, like, hitting all the the different endpoints and throwing invalid, 220 00:24:57.600 --> 00:24:59.200 oversize, undersize, 221 00:24:59.520 --> 00:25:02.639 you know, inputs at it and seeing how it does. You 222 00:25:02.880 --> 00:25:03.679 could, 223 00:25:04.080 --> 00:25:06.879 in theory, do that with mining firmware, 224 00:25:07.039 --> 00:25:13.695 except for this firmware is super sneaky, especially the stock mining firmware is super sneaky. It's all encrypted. 225 00:25:13.934 --> 00:25:36.330 They have secrets on the hardware and baked into the firmware so that it won't run anywhere else. And it won't run unless it can contact their fee server in the in the third party firmware case. So you really do need to run it on the control board hardware that it's intended to, which is just a Linux computer, but it's got those, like, special bits in there that lets it run. So that that makes thread teaming and doing security analysis 226 00:25:36.330 --> 00:25:42.415 on this stuff even harder. It's hard because the source isn't available, but it's even harder because it's so locked down. That's 227 00:25:42.655 --> 00:25:46.495 I'm I'm thinking why Schnitzel has such a rad setup. 228 00:25:46.735 --> 00:25:50.255 Shout out to his wife for putting up with these amazing experiments. 229 00:25:50.495 --> 00:25:59.960 But, it's gotta be done. Right? There's lots of lots of cool findings that I've heard that Schnitzel's come out with on this stuff. You know, a lot of it is is exposing just, 230 00:26:00.440 --> 00:26:01.880 like, intended behavior, 231 00:26:02.200 --> 00:26:04.120 like contacting fee servers, 232 00:26:04.280 --> 00:26:27.070 license servers, and things like that in the third party firmware case. And so I think some of those disclosures will be, like, ignored because they know it's doing that, but maybe not everyone else knows that it's doing that. And in the fee server case, it's really interesting because it's like, okay. So if this miner won't operate if it can't contact its you know, if it can't phone home to its fee server, then, yeah, you have the risk that the firmware provider 233 00:26:27.070 --> 00:26:38.750 will shut you down. But why would they do that? Right? Because they don't wanna wanna kneecap their own revenue stream. So there's that, like, game theory that goes along with that, like, shut off their customers because then they wouldn't get the money. But why they do that? But 234 00:26:38.855 --> 00:26:48.375 what if it's not them who shuts it down? What if they're compelled to by whatever random government, you know, they're under the jurisdiction of who's like, no. No. No. You need to shut off Eco 235 00:26:48.535 --> 00:27:00.300 because we don't like him or whatever. Right? Like, that's how these these things get. Or or just like, you know, you the miner needs to pay that it doesn't need even need to be a fee based 236 00:27:00.540 --> 00:27:04.700 server. Like, it could just be like, well, it needs to ping age verification 237 00:27:04.700 --> 00:27:06.860 server or this KYC database 238 00:27:06.860 --> 00:27:11.955 or, you know, if the hardware doesn't make contact with whatever arbitrary 239 00:27:11.955 --> 00:27:37.115 barrier we wanna put in place, then the machine's not gonna function at all. And it's really easy for them to do and and take those freedoms away from people, which is fucked up. Really easy to do. Right? They they probably got a 100 number they could call and just be like, you gotta do this right now. Otherwise, you're going to jail. The incentive argument doesn't make sense to me either. Like, oh, we would never shut off the dev server because then we don't get our cut. You think Coldcard was incentivized 240 00:27:37.115 --> 00:27:41.195 for this firmware debacle to happen? Like, I don't think that company is 241 00:27:41.515 --> 00:27:43.274 viewed well anymore publicly. 242 00:27:43.275 --> 00:27:53.739 So I I don't believe that incentive argument. Like, that stuff can happen. There's also I mean, there's also error case. Right? We

this has happened. Venish's license server went down maybe a year ago. 243 00:27:54.220 --> 00:27:54.859 Mhmm. 244 00:27:55.740 --> 00:28:01.419 When when you're running Venish firmware and it can contact its license server, it throttles down like 245 00:28:01.500 --> 00:28:02.779 ten, twenty percent. 246 00:28:03.340 --> 00:28:07.085 So they their server just went down due to some networking problem. 247 00:28:07.405 --> 00:28:11.965 If we can't get our 2% cut, then you can't have 10% of your hash rate. Yeah. 248 00:28:12.445 --> 00:28:26.970 Yeah. Exactly. It's still like It's not even a wash. Two to two. What the heck? And then, I mean, the third, the big, another big one is the third one is just like, it could get man in the middle. Right? It could get attacked by an adversary. And we've we've you know, this this has happened in lots of other industries 249 00:28:27.050 --> 00:28:30.490 where someone gets in between, you know, the mining farm 250 00:28:30.730 --> 00:28:36.315 and the the fee server. Right? And can spoof it or can inject packets, 251 00:28:36.315 --> 00:28:39.914 can can man in the middle attack it and then Change your passwords, lock you out. 252 00:28:40.635 --> 00:28:49.435 Exactly. Right? They can make life very difficult for you, which is definitely something that competitors or hackers or whoever might be incentivized 253 00:28:48.990 --> 00:28:57.629 to do? Well, good news here is that through this effort that Schnitzel is leading, it has brought some attention to this work 254 00:28:57.870 --> 00:29:21.929 so much so that developers from these firmware manufacturers, OEMs are in contact with him and that's good to see. You know, it's this balance about proper disclosure practices, which we do wanna follow, but also there have been folks tooting the horn for a while saying, hey, you know, does the firmware do this? Can you prove it? And and it just gets ignored. And so that's kind of why I think we went through with this announcement, 255 00:29:22.010 --> 00:29:33.445 not listing our full findings quite yet. So we have that, I think Shinsel said, a thirty day window where we're reaching out, communicating with these teams, helping resolve some of these issues that were found or have explanations. 256 00:29:33.845 --> 00:30:03.145 Then at that thirty day mark, we're going to release our full findings. But I was very excited to hear that folks are reaching out to him so much so that even some of the public miners, the large miners that maybe have their own firmware images, I'll save the names for now, those teams have network engineers that have kind of been wondering about this and maybe got shot down or just didn't have the time or the funding to pay attention to it, and now they are paying attention. And so working with them, it's exciting to see. Yeah. I mean, you think these machines if you're a Bitcoin miner, these machines are responsible for all of your revenue. 257 00:30:03.545 --> 00:30:14.080 That's that's

your business. Right? Maybe you got you're doing some AI or whatever, but, you know, your your Bitcoin mining unit, this is where all of your revenue comes from. It is very important that, 258 00:30:14.400 --> 00:30:23.360 these machines keep running for a lot of reasons. And to have closed source firmware that hasn't had a lot of eyes on it, that hasn't been properly audited, 259 00:30:24.085 --> 00:30:34.965 that's that's a risk. That is definitely a risk. And, you know, these things change over time. So, you know, we're doing this security audit now, but who is this gonna continue in perpetuity? 260 00:30:34.965 --> 00:30:46.220 I I don't think so. Right? It's like you you still have this problem of there's just a lack of of eyes on it and a lack of transparency about what's going on. So this is obviously a plug for Mugena, 261 00:30:46.300 --> 00:30:50.700 which is open source and by that very nature is 262 00:30:50.860 --> 00:30:56.985 much better audited by many more people. I think it it's on the consumers of the hardware, 263 00:30:56.985 --> 00:31:12.710 the people that are running the miners to demand open source. Mhmm. You know, I mean, there's millions, tens of millions, maybe probably hundreds of millions of dollars. I don't know. Billions of dollars in mining hardware operating the Bitcoin network right now and, like, 9099% 264 00:31:12.710 --> 00:31:13.909 of that's closed. 265 00:31:14.070 --> 00:31:36.690 Iron's like a $10,000,000,000 company. Right? This is not this is not little, pocket change they're playing around with here. Well, what we've heard too is the network security engineers at these companies, they're not jumps. They know what they're doing. I bet they've they've been pulling their hair out for years saying, like, hey. We should probably investigate this. And now Probably. The red team effort, it's finally getting the attention it deserves. And to your point, Scott, it's like with an open source firmware, 266 00:31:36.930 --> 00:31:41.010 you do all this inspection before you get a billion dollars in machines deployed. 267 00:31:41.410 --> 00:31:42.930 Right. Not after the 268 00:31:44.210 --> 00:31:47.755 So crazy, dude. Man. Could you imagine 269 00:31:47.915 --> 00:31:49.275 alright. If you have 270 00:31:50.475 --> 00:31:51.434 gigawatts, 271 00:31:51.435 --> 00:32:05.490 right, connected to these miners and someone finds an exploit that can just turn them all off in an instant. Right? That has like massive power grid implications. Like, is not joke. Issues for sure. For sure. They are Internet connected 272 00:32:05.490 --> 00:32:18.565 dampers to the electrical grid, and they're controlled by closed source firmware. It's insane from China. Insane. What is going on here? Why are we why are they doing this? It's so ridiculous. 273 00:32:18.645 --> 00:32:20.005 It's just gonna get like 274 00:32:20.325 --> 00:32:32.110 stocks netted. Yeah. Exactly. But reception has been good. It's been exciting to see.

It is exciting. I mean, I think it these kinds of risks that we're talking about, I think shed some light on I I think well, it helps 275 00:32:32.670 --> 00:32:36.910 I'd like to think it helps explain why the US government was 276 00:32:37.310 --> 00:32:44.510 they had that, like, security advisory out a few months ago about Bitmain hardware, and they they were, like, seizing 277 00:32:45.145 --> 00:32:54.664 Bitmain hardware under the guise of national security concerns. Right? Do you recall this happening? Yeah. That's right. That's right. They were definitely preventing the import of the At customs. 278 00:32:54.985 --> 00:33:20.895 Mhmm. Yeah. And it's like if if they under if they if they're looking at the risks the way that we're looking at them, then I think it makes a lot of sense. But I mean, who knows what really prompted them to do that? It could have just been economic reasons or anything any other number of reasons. But I mean, I I do think the risks that we're shining light on here are valid and real. And if they're not addressed, will be exploited at some point to do 279 00:33:21.215 --> 00:33:24.575 a a massive amount of damage, not just to 280 00:33:25.130 --> 00:33:31.450 the Bitcoin mining network, but the downstream or upstream power infrastructure that it's connected to. Yes. 281 00:33:31.690 --> 00:33:36.090 For sure. For sure. I think the Bitmain incident was one of their subsidiaries, 282 00:33:36.090 --> 00:33:36.970 SoftCo, 283 00:33:37.130 --> 00:33:39.370 was doing some work with Huawei, 284 00:33:39.895 --> 00:33:45.494 which is, like, embargoed in The US. And I think they fell afoul of that, 285 00:33:45.735 --> 00:33:47.095 if I remember correctly. 286 00:33:47.174 --> 00:34:32.510 But, yeah, I mean, that's that's also a counterparty risk when dealing with these overseas companies that can just make a decision like that. And I know that there was there was large miners that had their deliveries of ant miners held up because of that. And you don't want your deliveries of ant miners held up. Right? Yeah. Got like a thirty month life cycle? That yeah. Yeah. That's the whole name of the game. Like, you you can't just pay for these things and then not plug them in. There's lots of problems with that. Yeah. In other news though, there was a really nice write up we released this morning on the first product being developed on Magena and the power of open source firmware. Yeah. The first commercial product. That was exciting to to get out and really piece together the narrative, I think, for other people to to understand 287 00:34:32.670 --> 00:35:30.545 really just what we're showcasing here, what what Riot has developed, Riot, in Switzerland. That's r y three t. We've talked about this a little bit before on the show and Schnitzel came on to talk about it because he had extensive Twitter threads on on this. But it's really awesome to summarize just the the feature ads too. So there's all the risks talking points that

we just went through with closed source firmware and how open source negates that. But also with open source firmware, you can make it do what you want. And this is finally an opportunity where we can see, like, it's put well in the article here. You know, it's like mining now can enter where situations where it can follow available energy. This is a solar monetizing miner and not demand a fixed load. And that just opens up the floodgates for places you can put Bitcoin miners. So this this system from Riot monetizes excess solar. It responds to the solar flux within seconds. So if a cloud comes in, it can ramp from 3,000 watts to two thousands in a couple couple of milliseconds. It supports hash board mixing, so you can just upgrade one hash board at a 288 00:35:30.625 --> 00:35:55.160 time, have an s 21 board, an s 19 k, an s 19 j pro. And all of this is running actually on the stock control board. The the app the Magina application, the firmware application, but then also the the layer on top of that that actually connects the system to talk to your solar inverter. And it's just amazing to see development. That that's the Nova system. Right? The Nova. Right? Nova system. Yeah. Yeah. So there's a cool write up now on 256foundation.org. 289 00:35:55.475 --> 00:36:17.940 Go to our newsroom page and you'll see it there. This is really cool. Right? This is like you said, we're hyping lots of doom and gloom here about the security aspects. But then the flip side, which is definitely more exciting, is like, look at what you can now do. Like, you can make these miners do what you want. And it's it's not even that hard. It's just this unlock that lets you move forward with with real companies, with real products. 290 00:36:18.260 --> 00:36:34.035 Tyler, you have a pretty good point about, you know, if you wanna make a product with a Bitcoin miner, you can't just, like, slap an Antminer in a box. Right? There's a lot more to it than that, right? As far as permitting and making sure the authorities will allow this device to be used in that way, right? Yeah, 291 00:36:34.515 --> 00:36:42.059 yeah. I mean, I know it's definitely, there's additional hoops you have to go through if you're putting miners in different products, systems, applications. 292 00:36:42.059 --> 00:36:56.555 I know the Heapit team, for example, like, had to go through extensive tip over testing. You know, what happens if your kid knocks this thing over? Is it gonna light your house on fire? I can assure you they're not doing tip over protection. There's not a gyro sensor inside of an s 21. Yeah. There definitely is not. Yeah. 293 00:36:56.795 --> 00:36:58.954 But Well, this is this has been a 294 00:36:59.434 --> 00:37:03.275 couple years in the making. You know, one of the things we've been suggesting 295 00:37:03.275 --> 00:37:06.875 since the beginning of the two fifty six foundation is that we 296 00:37:07.275 --> 00:37:11.690 think there's real opportunities out there to build

real commercial 297 00:37:11.930 --> 00:37:13.450 products and businesses 298 00:37:13.450 --> 00:37:17.450 on top of a foundation of open source solutions. 299 00:37:17.530 --> 00:37:18.570 And that 300 00:37:18.570 --> 00:37:33.625 guided the four projects that we set out to initiate. And it's great to see that vision finally taking form in the real world and I hope it continues and accelerates from this point forward. Yeah, and we've heard from the Riot team, 301 00:37:33.945 --> 00:37:46.470 no effort of our own, this is the power of open source. They took the code, they implemented it. They added their layer on top. We didn't fund it. We didn't have any any effort in it. They they just can do it because it's an open source project 302 00:37:46.470 --> 00:38:45.950 that's out there. But, you know, it's just amazing to see that this is what's possible when you don't have all these barriers. I know that the Riot team before and this is something that I've had to do with exergy and just what I've learned from the Heapunk Summit is like to your point Scott, you could do this with retrofitting miners but you have all these layers on top. You got a little Raspberry Pi to serve as your control layer because you can't get into control board of the miner. You're running Home Assistant to connect things wirelessly like when Y, you know, your solar inverter has an API, the miner has an API. Just put software that connects the two. Why have all these additional parts? And the Riot team has told us like, this eliminates the bill of materials. It makes the products cheaper. It makes the system cheaper. You don't need all these extra parts. Actually, it's less stuff to go wrong, less things to fail. So this is to your point, Nico. Like, this is how you get actual mature, well designed products to emerge when you have the proper tools available to you and it's not a MacGyver jerry rigged patch job. And so did Riot help me understand like because I know a lot of people think that if it's GPL 303 00:38:46.030 --> 00:38:48.185 v three licensed, companies 304 00:38:48.185 --> 00:38:49.625 are disincentivized 305 00:38:49.625 --> 00:38:50.905 to use it because 306 00:38:51.145 --> 00:39:04.360 it's gonna require the company, and and this is the misguided understanding they have about GPL v three, but they think that it's critics of this license think that it's gonna require the company to open source their entire product and 307 00:39:05.240 --> 00:39:06.200 no rational 308 00:39:07.480 --> 00:39:11.800 company would make that decision, right? So they are gonna avoid GPLv3 309 00:39:11.800 --> 00:39:12.600 licensed 310 00:39:12.600 --> 00:39:14.755 open source projects. However, 311 00:39:14.915 --> 00:39:22.195 why is why is that not a concern for Riot? It's a good question. So if they took Mugena as is and implemented 312 00:39:22.195 --> 00:39:34.960 no changes to the Mugena mining application, what we call the Linux kernel project of of Bitcoin mining firmware, then there's nothing for them to release because they didn't change it. It's already released. If they did change, implemented a new feature, 313 00:39:35.200 --> 00:39:40.720 something like that to the Mugena application itself, and then they start selling this product, I e distribution, 314 00:39:41.165 --> 00:39:45.325 they'd be required to open source those changes for the Mugena application. 315 00:39:45.484 --> 00:40:12.095 Notice how I said the Mugena application, that's because the Linux control board, which is the stock Amlogic Bitmain control board, it can also run this Riot application on top of it. And that they don't have that's their software. They can choose to open or close source that, the part that connects Mugena running on the same control board to the solar inverter. And that's their input power demand. Right? And so that's up to them if they want to release the source for that project, that that application. 316 00:40:12.095 --> 00:40:15.535 But you can see that these two things together, the Riot application 317 00:40:15.535 --> 00:40:21.775 and the Mugena application in one image, that that effectively becomes their distribution. 318 00:40:21.775 --> 00:40:27.290 Right? And Scott, chime in here. Am I wrong? Like, if you add things on top of the Mugena application 319 00:40:27.450 --> 00:40:29.770 and they have an image for the Riot product, 320 00:40:29.930 --> 00:40:39.145 do they have to release the whole source if Mugena is an underlying piece of that image? No. No. Not necessarily. I mean, there's there's some pretty, like, specific criteria, 321 00:40:39.145 --> 00:40:43.464 and this comes from the Linux world. But, yeah, like, if you install Ubuntu 322 00:40:43.705 --> 00:40:49.385 if you've noticed when you install Ubuntu, there's a checkbox for install, like, closed source proprietary 323 00:40:49.385 --> 00:40:56.110 drivers, like for NVIDIA stuff, for example. A lot of graphics drivers are still proprietary. Right? So that software 324 00:40:56.110 --> 00:41:09.265 that you can plug into a Linux system, and it doesn't just because it's running because the software is running on a Linux system doesn't mean it has to be released open source. Obviously, I think you should, but that that is up to them. It's so it'd be the same way with 325 00:41:10.385 --> 00:41:18.945 with a Mugena system, which in a lot of cases is going to be running on Linux, and Mugena is this application that does the mining. And so if you change that, 326 00:41:19.620 --> 00:41:35.505 then you have to release those changes. But you can you can run whatever other apps you want on that system. Right? You could you could preload it with Microsoft Word if you wanted to, and it doesn't, like, require you to release the source of Microsoft Word. There's some definitely some things around images, 327 00:41:35.585 --> 00:41:44.545 like if you're providing complete images for the whole board that you need to that that contain open source GPL

software in it, that you need to like the 328 00:41:45.670 --> 00:42:03.945 image. Like, you need to make that whole thing available, but it can definitely contain binary blobs for proprietary stuff. Gotcha. And one other note is that if you were to modify Mugena for your custom mega mining farm operation and not distribute it and keep it internal, then you don't have to release those changes. Yeah. That's right. I mean, the the the 329 00:42:04.425 --> 00:42:09.145 rules of the of the GPL only come into effect when you are distributing 330 00:42:09.145 --> 00:42:40.464 them for the regular GPL v three. So when you're selling them or otherwise distributing them. But if you're just running it yourself, you can do whatever you want. Right? Like, I could take Linux installation that I have on my desktop and, like, add all kinds of stuff, and I'm under no obligation to, like, release the source of that. Same way a big mining firm that has has come up with some algorithm that is their secret sauce, they could just modify Mugena directly, implement that algorithm, and they're under no obligation to share it. So to your point, Iko, this does not disincentivize 331 00:42:40.464 --> 00:42:54.020 making money or building products. In fact, I'd argue the latter. You know, Riot got to start from this launching pad of the work that's already been done on Mugena and not have to design it and architect it all from scratch. Right. 332 00:42:55.475 --> 00:43:17.359 It's so cool because your agents know about it. Right? It's open source. You can just point them with the repo and you'd be like, here's the repo. I need to add x y z. And it's just like, you can bang it out because it has all that context about how the whole underlying system works. Yep. I think the critics are gonna be standing by shocked and odd at these companies that are springing up on a foundation of GPL 333 00:43:17.359 --> 00:43:20.720 v three or CopyLeft licensed 334 00:43:20.175 --> 00:43:27.295 open source solutions. Man, I wish I had Rod's 100% button right now. That is I I saw he used it on his tangents show. 335 00:43:28.095 --> 00:43:38.800 Yes. It's back. Yeah. Shout out to Rod. He's got his all tangents all the time show. I believe it's weekly. It's pretty cool. He had me on to the first one to do his community noted segment. 336 00:43:38.800 --> 00:43:43.200 It was a lot of fun. So you should check that out if you're not already. He's got the soundboard. 337 00:43:43.360 --> 00:43:47.600 He's got the soundboard. And it was, like, just the the layers of tangents 338 00:43:48.015 --> 00:43:57.935 were all the way down. Wait. Isn't that a 100% recording? Isn't that Eko saying that? I think so. Yeah. It definitely sounds like Eko. I think it's my voice that's been modified. 339 00:43:58.255 --> 00:44:08.630 Bruce said. Our old producer Tom made that, if I recall correctly. Alright. We need a we need a button that where you just say that, Iko. But I think you're right, Scott. I just wanna emphasize that. You know, 340 00:44:09.110 --> 00:44:14.070 with agentic tools, we've seen a lot of people are starting their own little firmware projects 341 00:44:14.070 --> 00:44:24.095 to solve their immediate problems. And for anyone listening why it might seem like they see 10 different vibe coded firmware hacks for different systems that people have in their garages. 342 00:44:24.174 --> 00:44:37.190 Meanwhile, Mugena might seem to be moving at a slower pace. You see tweets about developer calls and forum links. It takes a lot of effort and time with experts that have twenty years experience doing this to properly architect 343 00:44:37.190 --> 00:44:46.069 a firmware project like this. But once you get the momentum, this flywheel effect starts to come out, and then an open source thoroughly documented 344 00:44:44.955 --> 00:44:46.475 copy left license 345 00:44:46.555 --> 00:45:03.080 firmware project exists that these agentic tools can reference for context and then they don't have to think hard about how it works, that's where you get this snowball effect of more products coming out. I think we're going to continue to see this happen at a faster and faster cadence over time. Yes. I think so too. 346 00:45:03.400 --> 00:45:10.280 Clearly. We've been in the nurture phase for a long time because we want to get it right and make sure it's secure and it works well and 347 00:45:10.520 --> 00:45:20.974 bring the community in and get their inputs. Then the ball's gonna start rolling. Speaking of community inputs, tell me about this Buzz server. Yeah. Schnitzel has helped us out spin up a bunch of server infrastructure. 348 00:45:21.134 --> 00:45:54.550 It's funny. You know, I thought Dylan and I were pretty good at this stuff because we've had to figure it out because I don't wanna pay for SaaS services. And then you meet a guy like Schnitzel who literally started a company doing it, and he spins up a better server solution than you could ever dream of in a weekend. So it is his job. He's very good at it. That kid's got talent. That's for sure. But we are playing around with the Buzz Relay server that we are hosting for the two five six Foundation. We've got some internal channels there, which just helps with our comms, but we are gonna test out with some public channels as well, perhaps a a solution to Telegram, 349 00:45:54.550 --> 00:45:55.430 but we'll see. 350 00:45:55.910 --> 00:45:59.990 Telegram has had some bots lately. Thanks for the save there, eco. I was getting spammed. 351 00:46:00.950 --> 00:46:06.710 Oh, that's ruthless. But the cool thing about Buzz is that agents are treated as, what a schnitzel say, first class citizens. 352 00:46:07.075 --> 00:46:11.235 So they they do get proper context, and then all messages are recorded 353 00:46:11.235 --> 00:46:30.850 via Nostra as logged events. And that's where I think it could be a really cool tool with an agent in there that's just kinda listening into the chats or

channels about various topics to kinda just build this ever growing ball of gum of knowledge from everyone. Right? This has kinda been the dream, Scott, with OSMU as well. And then you can just ask an agent, hey. 354 00:46:31.250 --> 00:46:33.010 How do we do this? And it will know, 355 00:46:33.665 --> 00:46:40.785 oh, someone already tried this eighteen months ago, blah blah blah. Yeah. It's it does seem like it has a ton of potential for development 356 00:46:40.785 --> 00:46:48.464 and community building and, like, onboarding new folks. Right? Like, this is what's been it's impossible with Discord 357 00:46:48.540 --> 00:46:49.260 and, 358 00:46:49.660 --> 00:46:53.340 Telegram. It's like you just show up, you have 450 359 00:46:53.340 --> 00:46:54.140 messages, 360 00:46:54.540 --> 00:46:55.580 and you're like, 361 00:46:56.540 --> 00:47:28.599 WTF, what's going on? Now you can you can have your agent just privately if you want, like summarize and, like, extract the information that you're interested in. You can also have your agent sort of publicly ask questions or talk about what you're doing, reference your Git repos. Yeah. It it does seem like it has a lot of potential. So I think I think this is a good direction to go for us because we have so many projects all with their own ecosystems at this point. Lots of events and exciting things going on. I'm looking forward to it. The the Buzz Buzz client is a little rough around the edges. 362 00:47:29.195 --> 00:47:33.915 Yeah. Certainly a beta. Yeah. It leaves a bit to be desired. I mean, I think they got the 363 00:47:35.195 --> 00:47:38.155 the Slack clone down pretty well, 364 00:47:38.395 --> 00:47:49.740 but then there's still a little bit difficult to get yourself added to the server and figure out how to hook it up with your agents and just kinda get the, the vibe. But, I mean, basically, 365 00:47:49.980 --> 00:47:57.980 every other day there's a new update for the Buzz client. So it's under active developments and, it'll get there for sure. I think in no time it'll easily surpass 366 00:47:58.275 --> 00:48:02.195 Discord and Slack in in functionality and UX. 367 00:48:02.434 --> 00:48:21.920 Well, of the coolest things about it too is the the ability to connect persistent server side agents. Say if we were to have a two five six foundation bot that is just reading context and it knows everything about open source mining, knows where repos are, knows where forum topics are. And then when you come in, you can actually connect your agents from your local computer. That could be 368 00:48:22.320 --> 00:48:45.500 a self hosted model, but it can still get a seat. It gets a Nostri private key, so it gets a persona. And then you can just have a group chat in the two fifty six Foundation Buzz Relay where you say, hey, at my agent, ask the foundation's knowledge agent where I can find this stuff. I wanna start working on a blank revision to Bidax, and it will pass off the context and your your machine will just start hacking locally, 369 00:48:45.660 --> 00:48:57.495 which is really cool. It is really cool. Yeah. If you're just like, oh, I've heard about this two fifty six thing. I wanna, like, you know, install Mugena. You could imagine instead of, you know, pestering someone directly 370 00:48:57.495 --> 00:49:14.110 and getting a, you know, a hurried answer or whatever, like, you can you can solve a lot of your own problems this way when the whole chat context is available and get, like, sort of the latest updates and issues and all that stuff. I like it. It's been cool. What else is going on, guys? I've been working on a, an interesting idea. 371 00:49:14.190 --> 00:49:22.724 I think we need to have to have nodes. More miners need to mine to their own nodes. Did I talk about this last week? Yeah. I talked about this on a podcast the other day with Solusitoshi. 372 00:49:22.724 --> 00:49:29.605 But I think we touched on it last week. Okay. So, yeah, I've been working on that. I think it's a lot of digging into Floresta. 373 00:49:29.605 --> 00:49:30.165 I think, 374 00:49:30.885 --> 00:49:40.840 one of the one of the reasons why I think node adoption amongst miners is low is that it's fairly expensive to set up a full archival node due to the storage requirements, 375 00:49:41.240 --> 00:49:51.505 and it doesn't really get you anything in your mining adventure. I mean, gets you self sovereignty, which is awesome, but it doesn't get you any more like lottery tickets or anything like that. So I think a lot of people are like, oh, $600. 376 00:49:51.505 --> 00:49:58.465 I'm I'm I'm gonna pass. And I don't think it needs to be that way. I think mining to a prune node works perfectly well, and 377 00:49:58.865 --> 00:50:01.585 prune nodes have significantly lower hardware requirements. 378 00:50:02.160 --> 00:50:10.800 Why not come up with an easy like Raspberry Pi image or some other low cost single board computer that people can set up in mind too? So they're 379 00:50:11.280 --> 00:50:12.720 Like your LibreBoard. 380 00:50:13.599 --> 00:50:24.215 Right? Like the LibreBoard? Yes, exactly. Like the LibreBoard. So this would be perfect. So you could have a mining node on your LibreBoard. You wouldn't need to put in like a two terabyte SSD. 381 00:50:24.215 --> 00:50:28.695 Right? It could use the compute modules baked in eMMC storage. 382 00:50:28.695 --> 00:50:41.390 And so, yeah, this would be a perfect way to deploy a a node on your miner that's running with a Libre board. Exactly. I like it. What do you what do you does it look like this is gonna be feasible? It does. It does. 383 00:50:42.589 --> 00:50:44.670 Utree XO has been around for a long time, 384 00:50:45.795 --> 00:50:47.075 and it 385 00:50:47.075 --> 00:50:54.755 kinda supports this, but it's it's actually not real well optimized for budget hardware. So I've been I've been running into some issues there. 386 00:50:55.795 --> 00:50:56.994 UtriXO 387 00:50:56.994 -->

00:51:00.329 and Floresta also require these proof servers. 388 00:51:00.410 --> 00:51:22.525 And right now, there's not very many of these proof servers out there. There's actually seems to be only two, and they are in Korea, and they're 200 blocks behind the chain tip. So, you know, there's gonna be some some bootstrapping of this. Like, we need to have these proof servers set up. Setting those up is actually pretty complicated. It takes a long time to sync those, like, more than a normal Bitcoin Core IBD. 389 00:51:22.685 --> 00:51:36.810 So, yeah, there there needs to be some work done there. I think getting some some fresh eyes on it with a with a mining focus in mind, I think will help. But yeah, the potential is there. I mean, I I synced a fluorescent node 390 00:51:37.049 --> 00:51:38.329 on a, 391 00:51:38.569 --> 00:51:41.529 like, a low end Raspberry Pi in, 392 00:51:41.769 --> 00:51:47.035 twenty hours. So that's pretty good. I mean, it's normally like a month plus. Dude, 393 00:51:48.075 --> 00:51:48.955 fluorescent You 394 00:51:49.275 --> 00:51:56.619 can't actually mine to fluorescent. You can mine to Yutri XO. So, yeah, there's there's a little bit of work that needs to be done. What is Florestra? 395 00:51:56.619 --> 00:52:02.300 I haven't heard of that one before. Florestra is it's kind of like a rewrite of U3XO. 396 00:52:02.300 --> 00:52:08.460 So it's the same concept where it uses this essentially like big Merkle tree to keep track of all the UTXOs. 397 00:52:08.460 --> 00:52:11.695 So it's the functionality is the same as UTXO, 398 00:52:11.695 --> 00:52:33.330 but it's a rewrite. Floresta is a rewrite in Rust. It's more modern. It's under active development. Actually, UTXO is also under active development. Yeah. And and you the the Floresta projects, have talked with a couple of guys working on that, and they say that they do wanna add support for mining. It can be added. It's on their road map. So it certainly is capable of doing that with some changes. I'd like to see those come sooner rather than later. But But it requires 399 00:52:33.569 --> 00:52:40.290 what what do you call that server that there's only two of? It's like a u three x o archival node, but it basically has calculated 400 00:52:40.545 --> 00:52:46.625 all these proofs for Proof server. Okay. Yeah. Proof server, and it has done it from the Genesis. 401 00:52:46.785 --> 00:52:56.305 So it's like it's additional computation per block on top of what Bitcoin Core does. So that's why it takes longer. I see. So once you spin up your Floresta 402 00:52:56.530 --> 00:52:57.330 node, 403 00:52:57.330 --> 00:52:58.370 it periodically 404 00:52:58.370 --> 00:53:01.170 needs to connect to one of these proof servers just to 405 00:53:01.490 --> 00:53:05.730 make sure everything's legit or what? Yeah. Yeah. Basically. 406 00:53:06.210 --> 00:53:06.930 Okay. 407 00:53:08.210 --> 00:53:12.555 You know, and and then, yeah, because it's this it's this, like, cascading

408 00:53:12.555 --> 00:53:14.155 of of proofs 409 00:53:14.235 --> 00:53:17.035 to to keep track of all the UTXOs. 410 00:53:17.035 --> 00:53:32.320 And so that's computationally intensive, and you can rely on, like, a peer to peer network of those to get yourself synced. There's I mean, it has. So Bitcoin has those snapshots. Remember, Rico? We were Yeah. Doing those before. UTXO has that too. It's called assume UTXO. 411 00:53:32.320 --> 00:53:48.945 And so there are those, and that helps you get much further forward faster. Right? So you just have the snapshot, and it's like, alright. It's this, like, hash, and it just allows you to start from a more recent block height. Those need to get updated. They're, like, a couple 100,000 blocks behind right now. 412 00:53:49.505 --> 00:53:59.660 Wow. Those need to get updated into the into the but, yeah, you do need you do need these proof servers. And we're you know, if this is gonna work for lots of miners to just buy a cheap, you know, ideally $100 413 00:54:00.060 --> 00:54:05.340 computer and plug it in and start mining to it, we're gonna need a lot more of these proof servers. Right. Interesting. 414 00:54:05.340 --> 00:54:14.665 But we can do that. There's there's all kinds of interested miners out there that would do it. Yeah. I'm I'm definitely into this idea of, like, like, low cost hardware. 415 00:54:14.665 --> 00:54:17.865 One of the motivating factors behind HydraPool 416 00:54:17.865 --> 00:54:21.785 was that I tried spinning up public pool on a Raspberry Pi. 417 00:54:23.420 --> 00:54:32.859 And, you know, part of the deal was that I had to have a Bitcoin node on the Raspberry Pi as well. And that was just very painful to spin up on that hardware. 418 00:54:32.940 --> 00:54:41.875 So I really liked the idea of being able to get involved with mining and having your own node without a hurdle of expensive hardware. 419 00:54:41.955 --> 00:54:44.035 Because I've got a bunch of old Raspberry 420 00:54:44.035 --> 00:54:46.275 Pi fours laying around, 421 00:54:46.915 --> 00:54:57.050 you know, it's it's just nice to be able to, like, if I wanna try something out, to be able to grab one of those and then just like try it out without having to go out and buy another computer 422 00:54:57.450 --> 00:55:10.975 and like spin up, you know, like you said, $600 worth of hardware to spin up a node. Sometimes you just wanna try things and you just wanna like use what you have laying around to see if the concept is even like worth pursuing and 423 00:55:11.295 --> 00:55:38.625 it would just be nice if there are a solution that would be conducive to that. I think it is. I think it can work. I mean, I originally I think, David a guy named Davidson, he's the, one of the lead devs on Floresta. I met him, like, almost two years ago, and he showed me that he had this this embedded Linux system. It was the size of, like, a pack of gum. And he was like, we're gonna run a mining node on this. And I was like, this is amazing. Awesome. Totally do this. So, you know, it's on their sites, 424 00:55:39.025 --> 00:55:47.210 and it it would be a game changer. I mean, the more home miners, solo miners that are mining to their unknown, the better. Right? If we can just make it easier 425 00:55:47.450 --> 00:55:49.930 and cheaper for them to do while still retaining 426 00:55:50.010 --> 00:55:53.130 that same, you know, decentralization 427 00:55:53.609 --> 00:56:03.805 that, you know, validating that they are, like, on the chain. It's not susceptible to, you know, someone rugging them. We don't wanna have anyone mining invalid blocks. 428 00:56:04.125 --> 00:56:14.285 Like, that stuff all needs to stay. But I think I think this is the route to do that. So that's been fun to poke at. My AI is pretty good at coming up with ideas. Shall look at the Hashdash? Yeah. 78 429 00:56:14.420 --> 00:56:15.300 Terahash. 430 00:56:15.300 --> 00:56:18.740 In first place, we got Zarka, longtime supporter supporter. 431 00:56:18.740 --> 00:56:19.380 Thank you. 432 00:56:19.780 --> 00:56:20.980 777, 433 00:56:21.060 --> 00:56:25.060 the Bitcoin Park miners that we conquered with Schnitzel script. 434 00:56:25.380 --> 00:56:32.405 PubKey. Also conquered by Schnitzel script. Yeah. Watch out, Megaminers. You don't wanna get conquered by Schnitzel script. Tyler, 435 00:56:32.484 --> 00:56:33.765 hardestblocks.org, 436 00:56:33.765 --> 00:56:35.685 D plus plus, Dylan, 437 00:56:35.765 --> 00:56:36.484 Eko, 438 00:56:36.565 --> 00:56:38.885 Backwater Mining Collective, Exergy, 439 00:56:38.885 --> 00:56:40.484 Nord, Satoki, 440 00:56:40.565 --> 00:56:44.040 Scott, Clear, agent p, p Buttweiler, 441 00:56:44.119 --> 00:56:54.440 Nuclear, Pizzandy, and Schnitzel. Thank you for continuous support. No changes to the difficulty leaderboard since Telehash four, and Eco has been booted from the loyalty leaderboard. 442 00:56:54.935 --> 00:56:56.375 Come on. How? 443 00:56:57.015 --> 00:56:59.495 By Scott. Scott's in tenth. Damn, Scott. 444 00:56:59.895 --> 00:57:00.695 You're doing too Yeah. Much 445 00:57:01.895 --> 00:57:07.335 I had to it it just got too hot. My, like, AC was running a 100% duty cycle, 446 00:57:07.335 --> 00:57:27.595 and I had to unplug everything yesterday. I was like, it's so That's funny. Not your chance, eco. Yeah. Where let me check. Are all three of my miners still online? Let's see. Oh, no. One of my miners shut off. Hopefully, didn't get it back on. Yeah. I gotta it's at a it's at a a different geographic location, 447 00:57:27.755 --> 00:57:29.595 so this might take me some time. 448 00:57:30.315 --> 00:57:33.994 That's funny. Yeah. Okay. Good to be Geographically 449 00:57:33.994 --> 00:57:34.875 decentralized. 450 00:57:34.955 --> 00:57:35.355 Yes, Scott. 451 00:57:36.339 --> 00:57:38.820 Go ahead, Ike. No. I I'm 452 00:57:39.140 --> 00:57:57.405 I'm I'm retarded and autistic so I have things all over the place. Well, Scott's been having more hash

rate because I keep sending him minors to poke at. We didn't talk about that. You took a look at the m 64. That's quite interesting. It is. Yeah. This is my first deep dive into what's minors. 453 00:57:58.525 --> 00:58:00.525 But you said him an m 64? 454 00:58:00.525 --> 00:58:07.650 Yeah. Which is like Nice. The heatpunk holy grail. That's what a miner that all the heatpunks like because it's the only single phase power, 455 00:58:07.810 --> 00:58:26.945 small form factor, silent water cooled miner. But it has crappy Watts miner firmware on it. And I think Scott tore it apart, and it's pretty interesting machine. It is a very interesting machine. Tore it apart. I tore it open. It's it's very well built inside. I mean, this this is significantly better build quality than I've seen on the, 456 00:58:27.425 --> 00:58:34.570 Antminer hydro units. It's super heavy. It has very weird proprietary connectors for power and water. 457 00:58:34.809 --> 00:58:35.770 So that's interesting. 458 00:58:36.089 --> 00:58:41.930 Yeah. Mostly what I was getting at was the firmware. The one of the discoveries that shocked me the most was 459 00:58:42.225 --> 00:58:55.905 that and this this is still preliminary, so I don't know if it's entirely true yet. But it seems like the work that is sent from the firmware to the chips, like the the jobs to mine, it seems like they're cryptographically signed, 460 00:58:56.065 --> 00:59:01.510 which is insane to me. So there's like a essentially a secure element on the CPU on the control board. 461 00:59:01.750 --> 00:59:03.190 And presumably, 462 00:59:03.190 --> 00:59:06.390 there is a portion on the ASIC itself 463 00:59:06.470 --> 00:59:10.710 that will verify that the work that they got is signed by MicroBT's 464 00:59:10.710 --> 00:59:12.310 keys on the control board. 465 00:59:12.995 --> 00:59:28.275 Come on. Right? This is crazy. Why would they do this? All the mining data is still sent in the clear. Right? Like, you can still see how it's writing all the registers and everything. So there's no secrets they're hiding. They're just preventing or trying to prevent third party firmware. 466 00:59:28.780 --> 00:59:30.380 Yeah. From operating. 467 00:59:30.540 --> 00:59:46.075 This is That's the only reason that's the only reason you would do something like that. Again, I'm not a 100% on this yet, but there is we have identified, me and my agent have identified that there is a I think it was like 20 bit essentially checksum 468 00:59:46.075 --> 00:59:55.435 at the end of the job packet, which is is common. We always see checksums on there, but it it does look like it contains a signature from 469 00:59:55.915 --> 00:59:56.235 the 470 00:59:56.780 --> 00:59:57.660 CPU's 471 00:59:57.660 --> 01:00:16.515 secure element, which is we don't know yet if it rejects packets that aren't signed, rejects jobs that aren't signed. We don't know that yet. That could be the case, but why would they even bother if they weren't using it?

I don't know. Is this an indication on perhaps why aftermarket firmwares have not really focused on What'sMiner support? Perhaps. 472 01:00:16.595 --> 01:00:17.155 Yes. 473 01:00:17.714 --> 01:00:19.555 And what would it mean for Mugena? 474 01:00:19.635 --> 01:00:24.115 What would what would Mugena need to institute so that it could support What'sMiner's? 475 01:00:24.240 --> 01:00:26.240 You'd have to know this signature. 476 01:00:26.240 --> 01:00:29.200 Yeah. Okay. You'd have to know the signature. You'd have to know the, 477 01:00:30.320 --> 01:00:35.920 yeah, the the key that they're signing these jobs with. Well, hopefully, it wasn't created with a lot of entropy. 478 01:00:39.694 --> 01:00:41.535 That's awesome. Yeah. 479 01:00:41.694 --> 01:00:44.095 Turns out that's hard to do. Yeah. 480 01:00:44.654 --> 01:00:52.630 Yeah. I I haven't plugged this thing in yet, but now I I almost have all the little bits that I need to plug this thing in yet in, and see 481 01:00:53.270 --> 01:00:53.910 what 482 01:00:54.870 --> 01:01:01.030 what it what the serial data communication with the ASICs actually looks like in practice. Because right now, I've just been 483 01:01:01.430 --> 01:01:08.944 poking at the binaries. So we should be able to see what it's actually looking like, and we should be able to try modifying it, and see if, 484 01:01:09.585 --> 01:01:15.025 we get a rejection if that key isn't there. So that'll give us a definitive answer on this. 485 01:01:15.585 --> 01:01:17.425 And then, yeah, if it is requiring 486 01:01:17.870 --> 01:01:39.595 those jobs to be signed, then there's some more work to be done. It's a pretty pretty tight setup. You know, they really don't want you doing other things with these machines. Yeah. Yeah. They really don't. They have the root file system is encrypted on there. There's no SSH is turned off by default. It's it's really locked down. That that is on par with what I've seen from the latest Ant miners. 487 01:01:40.234 --> 01:02:16.730 S 23 does that as well. So this is kind of that song and dance where someone figures out a way to get aftermarket firmware on a miner, and then Bitmain will go change something. Yeah. Yeah. It has been that, like, that way for a couple years is they've they've really tried to lock down what you can load on there. It gets exploited, and it's this, like, cat and mouse game where, you know, it gets exploited, now brain's vanished. LuxOS can load on there, and then the next version, Bitmain patches it. Well, in Vegas, they announced, like together Luxor and LuxOS and MicroBT announced support for their fleet. So I wonder what's going on there. 488 01:02:17.369 --> 01:02:19.130 Really? I didn't hear that. 489 01:02:20.490 --> 01:02:20.970 Firmware? 490 01:02:21.695 --> 01:02:23.615 Luxos firmware on MicroBT? 491 01:02:24.575 --> 01:02:28.495 Interesting. Very interesting. I don't know. Must have just gave them the signature. 492 01:02:30.575 --> 01:02:31.455 Man. 493 01:02:31.695 --> 01:02:43.380 Although I wonder I wonder how much info they actually get. I've heard from some friends that are repair techs at mining companies. In this case, mining companies owned by the manufacturer, 494 01:02:43.540 --> 01:02:45.860 and they don't get repair documents. 495 01:02:45.940 --> 01:03:01.915 Like, these are the people that the manufacturer is paying to fix their fleet, and they still don't have access to scan patents and things like that. It's just It doesn't make that sense. Insane. Yeah. Or even the if you ordered a 100,000 Watts miners, say you're Riot Riot in Texas, 496 01:03:01.995 --> 01:03:11.960 and you say to them, hey. I'm planning to put LuxOS on these. Why why would the manufacturer not just help you do that? Seems like they should. Yeah. They could be like, 497 01:03:12.359 --> 01:03:28.954 alright. Well, we're not gonna support you. Your warranty we're not gonna gonna cover, like, failures if you put LuxOS on there. But, like, once that's out of the way, what do they care? Yeah. Right. Very hesitant or resistant to change. It is interesting. I think it's fair to say they're not the best at making firmware. They're not. 498 01:03:28.954 --> 01:03:31.035 They're just not. I mean They don't have to be. 499 01:03:32.730 --> 01:03:39.370 They've got a captive audience, I guess. People who have gotta have that warranty support, although I've heard it's kinda garbage. But, 500 01:03:40.330 --> 01:03:43.290 yeah. I mean, everyone knows that Bitmain's firmware sucks. 501 01:03:43.290 --> 01:03:47.045 Like, it doesn't do any tuning. It doesn't have any overclocking, 502 01:03:47.045 --> 01:03:48.085 underclocking. 503 01:03:48.645 --> 01:03:58.325 It's like the API's garbage. Well, the only thing they had from from the initial findings in Schnitzel's thread on the red team effort was it is the most, like, clean and trustworthy 504 01:03:58.325 --> 01:03:59.924 in terms of avenues 505 01:03:59.924 --> 01:04:05.380 to exploit compared to these aftermarket ones that sell you on features, which was interesting. Like, 506 01:04:05.940 --> 01:04:08.260 complete lack of features minimizes 507 01:04:08.260 --> 01:04:12.579 the attack surface. Yeah. Pretty much. Of course, there's no reason it has to be that way. Securing 508 01:04:13.185 --> 01:04:18.465 securing Linux computers on the network is, I wouldn't say solved, but it's pretty well understood. 509 01:04:18.545 --> 01:04:46.415 Like, third party firmware could give the owner the option to install a SSH key on there and lock it down. But one of the one of the, like, mantras of hardware reverse engineering or reverse engineering, I guess, in general, is that physical access means all bets, all security bets are off. And that is very true. Yeah. If you have the hardware on your desk, it gets very, very difficult, and especially now with AI, basically impossible to lock someone else out. 510 01:04:46.494 --> 01:04:58.690 So the name of the game is just securing your device on the network and then putting, like, you know, padlocks on your building so no one can get in and get physical access. Exciting. Send send miners to Scott. He can't help himself. He's gonna hack them. Same with Schnitzel. 511 01:04:59.970 --> 01:05:05.090 Yeah. We got a good team. Fun stuff. Really fun stuff. Yeah. When when are we gonna be able to point everyone to the 512 01:05:05.650 --> 01:05:08.210 to the Buzz server so we can just really get the party started? 513 01:05:08.904 --> 01:05:14.665 That's a good question. We're still scoping out kind of the the server side of resources we need and 514 01:05:15.065 --> 01:05:29.350 probably come up with a token budget too before people start spamming questions to bots. Right now, I think I'm charging Schnitzel's credit card when I ask questions. Gotta fix that. Yeah. Turn off the self hosted AI for a little bit. Yeah. Yeah. But things are moving along. Upgrading 515 01:05:30.070 --> 01:05:31.670 our infrastructure and 516 01:05:31.910 --> 01:05:39.475 tools. This is cool. Yeah. Alright. Should we wrap? Yeah. I think that's a wrap you guys. Amazing. Thanks for tuning in. Good show. Y'all 517 01:05:39.795 --> 01:05:40.355 stay 518 01:05:41.155 --> 01:05:41.955 sovereign, 519 01:05:42.275 --> 01:05:43.715 and we'll see you 520 01:05:43.955 --> 01:05:47.555 next week for pod two fifty six number one twenty five.