전사 🔗
WEBVTT
NOTE
Transcription provided by Podhome.fm
Created: 8/10/2026 11:49:40 PM
Duration: 4061.571
Channels: 1
1
00:00:00.000 --> 00:00:05.120
I think yours could be a little higher, Scott. Really? Yeah. Alright. It looks alright.
2
00:00:05.440 --> 00:00:16.915
Alright. I'll You just choke it. Whoever from last week said I was too loud. I'm sorry. Yeah. We're trying to get our audio levels figured out here. You just have to speak confidently into the microphone. Microphone. There we go.
3
00:00:17.475 --> 00:00:37.470
Thanks for the thanks for the comments from our our many, many listeners about the audio quality of the last issue. We will make sure to address that loud and clear. No stone will go unturned on pod two fifty six, episode one twenty three, August tenth. Welcome. We're doing it. What's up? It's been a crazy couple of weeks.
4
00:00:37.950 --> 00:00:40.030
You know, the cold card exploit,
5
00:00:40.030 --> 00:00:41.550
just to recap quickly,
6
00:00:42.350 --> 00:00:44.350
there was a latent bug found
7
00:00:44.845 --> 00:00:48.765
in the way the cold card generated its entropy,
8
00:00:48.925 --> 00:00:57.165
which made the seeds very weak, roughly 30 to 40 bits of entropy, where they should have at least a 128,
9
00:00:57.165 --> 00:00:58.685
preferably 256
10
00:00:58.685 --> 00:01:00.480
bits of entropy. So
11
00:01:00.719 --> 00:01:03.600
all in all, roughly $100,000,000
12
00:01:03.600 --> 00:01:18.204
worth of Bitcoin got drained from people's cold cards. Coldcard released a hotfix to get an updated firmware out there so people could spin up a new wallet with proper entropy and move their funds. And they ended up introducing
13
00:01:18.365 --> 00:01:23.005
a critical vulnerability that ended up breaking some cold cards. And then
14
00:01:23.325 --> 00:01:26.100
there was some indicators
15
00:01:26.180 --> 00:01:30.820
that one of the cold card team members, the CTO, Peter Gray,
16
00:01:31.060 --> 00:01:33.540
was the developer behind
17
00:01:33.540 --> 00:01:35.299
the libngu
18
00:01:35.299 --> 00:02:00.250
library where the vulnerability stemmed from. And that's just not a very good look. So, yeah, that was a complete fucking train wreck. Sorry for anyone who lost their funds. Yeah. Yeah. And then BTC pay server had an issue found as well. And, yeah, and there was some stuff going on with Lightning. Like there were Lightning channels getting drained apparently. I think Zach from Foundation said that. Yeah, I think it was like a Lightning
19
00:02:00.490 --> 00:02:02.170
channel associated
20
00:02:02.170 --> 00:02:02.729
with
21
00:02:02.970 --> 00:02:11.255
BTC pay server was vulnerable, and a few people got that drained. The only one I heard of was Zach from Foundation Devices,
22
00:02:11.255 --> 00:02:27.640
the makers of the Passport Wallet, which have obviously been in the news a bunch. He said, you know, luckily he said it wasn't much, but we do we ever hear what the what the actual problem was? I know it was like, there was a patch that was issued that allowed you to redo something called a macaroon,
23
00:02:27.720 --> 00:02:50.905
and then there was the advisory of like, just turn it off for now. Yeah, I'm not really sure what exactly was the cause of that issue. I think if if I were like more into like the lightning social circles, I would probably know, but I'm just like, I've had very little confidence in lightning for many years and I don't really pay that much attention to it. So I think I'm just kind of out of the know.
24
00:02:52.610 --> 00:02:53.970
Professional
25
00:02:53.970 --> 00:02:55.170
optimist. Yeah.
26
00:02:55.570 --> 00:02:56.050
So
27
00:02:56.690 --> 00:03:24.270
do your own research. Be careful about BTC Pay Server right now. I mean, from what I know, the people behind that project are pretty rad and I'm sure they're working hard on it, fixing that up. It's a proper open source project. Well, and this is part of the, I think the Bitcoin Red team, the emergent Red team, which is a group of people that seem to be pointing LLM credits, tokens, frontier open source models, open weight models that basically don't say no and
28
00:03:24.590 --> 00:03:32.830
talk down to you, like the closed proprietary models from many US leading LLM companies. But yeah, there's a large effort to review,
29
00:03:32.910 --> 00:03:33.710
audit, inspect
30
00:03:34.245 --> 00:03:36.485
many, many, many Bitcoin repos,
31
00:03:36.724 --> 00:03:37.525
sensitive
32
00:03:37.525 --> 00:03:38.325
infrastructure
33
00:03:38.325 --> 00:03:39.045
tools,
34
00:03:39.125 --> 00:03:57.500
BTC Pay Server being one of them. So, yeah, I don't think there's any malice as it relates to this story, more so just these AI tools and this emergent effort to inspect the shit out of them has led to finding some interesting findings. Yeah. The adversaries are clearly doing it. So it makes sense to have
35
00:03:58.780 --> 00:04:02.540
white hat hackers doing it as well. So it sounds like people are using,
36
00:04:03.115 --> 00:04:04.715
what like OpenAI
37
00:04:04.715 --> 00:04:13.355
and Claude to try and fortify software by finding bugs and properly disclosing them, doing responsible disclosures.
38
00:04:13.355 --> 00:04:20.590
But their agents are telling them, no, you can't reverse engineer this because that's exactly what a black
39
00:04:20.830 --> 00:04:49.730
hat hacker would do. Right? But then, so they're turning to like Kimi three, these Chinese models that are based on open source weights, whatever that means. And the Kimi three is telling them, yeah, you can do that. Yeah. Rob Hamilton's had some good posts on it lately, but my understanding is I'm looking at some of his screenshots here. My understanding is you can just be much more blunt about the goal with these open weight models. Like, Hey, I am trying to maliciously attack this software
40
00:04:49.889 --> 00:04:53.170
because I want to figure out flaws that we can fix. It's
41
00:04:53.650
Load full transcript
--> 00:04:57.490
more messed up than that. I've definitely run into this myself
42
00:04:57.570 --> 00:04:58.850
with OpenAI
43
00:04:58.850 --> 00:05:05.465
models, which, you know, are the closed weights one that you're talking about. If you if you even mention security
44
00:05:05.465 --> 00:05:07.065
or exploit,
45
00:05:07.305 --> 00:05:17.920
it will just flat out refuse to do it. So this can this can legitimately come up when you're working on your own project. Right? You're like, I should have AI do a review of my own project. And it refuse
46
00:05:18.320 --> 00:05:21.919
if you talk about network security exploits,
47
00:05:21.919 --> 00:05:23.840
anything like that. But furthermore,
48
00:05:24.000 --> 00:05:33.035
you know, you can kind of get around it, right? If you ask it a very targeted question like- Which is a total waste of time. Check the network security of this. I want to look into,
49
00:05:33.035 --> 00:05:34.395
like, overflowing
50
00:05:34.715 --> 00:05:46.680
a buffer on this API, and it'll start doing it. But I've had it where it'll start doing that, and then it'll convince itself that it's working on network security and stop. That's bullshit. It's such bullshit.
51
00:05:46.919 --> 00:06:01.785
Rob's screenshots here say primary model was rejected by provider. OpenAI triggered cyber policy classifier and fell back. So it might just boot you down to a dumber model and say, oh, yeah. You can't have worst out flat out refused with OpenAI.
52
00:06:01.945 --> 00:06:05.865
But they do they do say, like, oh, you should sign up for our cybersecurity
53
00:06:05.945 --> 00:06:06.825
program.
54
00:06:06.905 --> 00:06:22.800
And I clicked on that link, it was, like, fully KYC yourself, and then just wait for approval. But apparently, Rob Hamilton has done that and gotten that approval, and he's still getting blocked when doing some of this red team work. I saw that tweet from him, and that's extra fucked up.
55
00:06:23.040 --> 00:06:42.815
Extra fucked up. Yeah. So he submitted his KYC information. They know exactly who he is, what he's doing, and they're still telling him, no, you can't. The software is capable of doing it, but we're just not letting it because we don't trust you, even though we now have a honeypot of all your information. So meanwhile, you can take these open weight models
56
00:06:43.110 --> 00:06:46.949
from China, like KimiK three, and you can
57
00:06:47.189 --> 00:06:50.470
run that. And it doesn't have any reservations
58
00:06:50.710 --> 00:06:56.710
about doing security stuff. You know, you could straight up tell it, I wanna do a, you know, malicious
59
00:06:57.164 --> 00:07:09.910
hack. And it's like, okay. Yeah. Like, it'll just do it. One of the problems is it's very expensive. I mean, you know, they they must know what they've got. Right? And they wanna charge appropriate for it. But, yeah, they KimiK3
60
00:07:09.910 --> 00:07:13.110
is from a Chinese company called Moonshot AI,
61
00:07:13.590 --> 00:07:18.390
and they don't offer subscriptions. So you have to pay for the tokens through their API directly.
62
00:07:18.550 --> 00:07:20.870
And it gets expensive fast.
63
00:07:21.270 --> 00:07:23.750
Some people that we've talked to are working on this. You know,
64
00:07:24.485 --> 00:07:26.165
you can rack up $30
65
00:07:26.165 --> 00:07:32.245
in tokens from, a single prompt, and, yeah, it adds up really fast. Whereas with OpenAI,
66
00:07:32.245 --> 00:07:43.540
you can get a subscription. You could pay a $100 a month and get, you know, enough tokens to to get the job done if they'll let you. What's the alternative here? Like, can people just
67
00:07:43.780 --> 00:07:52.900
self host these things instead of paying someone else for tokens? Well, mere mortals cannot self host this. It's Why? It the the hardware,
68
00:07:53.060 --> 00:07:54.419
like the GPU required
69
00:07:54.675 --> 00:08:01.155
to run these models is like it it's something like a million dollars in hardware to run this.
70
00:08:01.555 --> 00:08:05.315
Come on. You need a lot of VRAM. You need a lot of Yeah.
71
00:08:06.354 --> 00:08:09.430
You need a lot of RAM and fast PCI
72
00:08:09.430 --> 00:08:11.350
and a badass processor.
73
00:08:11.350 --> 00:08:12.310
And so it's
74
00:08:13.270 --> 00:08:19.430
not practical for anyone to run this. Of course, it's gonna be super expensive in electricity to run this. But I have seen, there's
75
00:08:20.230 --> 00:08:33.795
a lot of work being done to optimize these models and make them so that you can run them, you know, on, you know, just tens of thousands of dollars worth of hardware. But it's it's still not there yet. And, you know, you wanna like, if you're doing a
76
00:08:34.355 --> 00:08:36.115
a security audit, right,
77
00:08:37.030 --> 00:08:52.245
you wanna make sure that you get everything. And so trusting it to a lesser model just adds risk, right? I mean, CoinKite famously said that they ran AI security models on their software many times before, and it never revealed this bug. Hopefully
78
00:08:52.325 --> 00:08:55.524
other providers though that have compute,
79
00:08:55.764 --> 00:08:58.645
if the model is open source and open weight,
80
00:08:58.805 --> 00:09:42.660
then you could run that on various competitors could run that on their infrastructure. I think of our friend Mark's at Maple AI. That's effectively what they do. I don't know how much horsepower they have. I saw that they are looking into offering Kimi k three. I saw a post yesterday about that. So that's that's exciting. Yeah. I mean, I from what I understand is this this model is open source, so you can just get it and run it if you have the hardware available. And so I assume if there aren't already, there will be several places where you can just buy tokens from their API for their API and and do it. And hopefully, that will, you know, provide a little competition maybe and bring the price down. Why are they using
tokens instead of Bitcoin? Well, tokens are the, like, the actual mechanism by which you communicate with the
81
00:09:42.980 --> 00:09:57.565
with the the the model. So I think it'll always be it's not like, you know, like, shitcoin tokens like we think about in that regard. So you could have a service very easily where you pay for those tokens with Lightning. I know OpenRouster
82
00:09:57.020 --> 00:09:59.660
Rouster is. Rouster. Rouster?
83
00:09:59.660 --> 00:10:04.540
Yeah. It's it accepts Lightning and e cash, and you there's no accounts, no KYC,
84
00:10:04.540 --> 00:10:12.845
and it's just Bitcoin, e cash, or Lightning directly for tokens. Nice. That's the way to do it. That's the way to do it. I was looking at OpenRouter
85
00:10:12.845 --> 00:10:26.765
and they I was excited because they had like a pay with crypto button, but there's definitely no Bitcoin option. Yeah. My understanding a token is vaguely an approximation of a certain amount of words or letters. So like a thousand words is 750
86
00:10:26.690 --> 00:10:45.885
tokens or maybe it's vice versa. And then you're charged on the tokens in because you have to feed words from your prompt to the model. Then it gives you an output in words in human language. And it charges you more for that typically because it's already done the thinking. That's how it's metered, what you feed it and then what it feeds you as a result. Crazy. This is like the billing units.
87
00:10:46.205 --> 00:10:55.245
Yeah. Not the billing units, but the compute units. If you were charging for kilowatt hours or something like that, it's kind of along that same path. What do you run at the space?
88
00:10:55.870 --> 00:10:58.990
Don't you self host some IoT? Yeah, we've got
89
00:10:59.470 --> 00:11:03.550
a cluster of four by Nvidia A6000s,
90
00:11:03.550 --> 00:11:10.095
I wanna say. So it's a single 2U blade with four GPUs in it. And I think each one has 128
91
00:11:10.095 --> 00:11:16.815
gigs of RAM. So we're looking at around 500 gigs of RAM, five twelve. And we were running MiniMax 2.7.
92
00:11:16.815 --> 00:11:23.135
I don't think we've run Kimi. It's been used as a development platform. We test a bunch of things. That's pretty cool. But even that is
93
00:11:23.550 --> 00:11:40.395
two forty volts. It's really loud. It makes a lot of heat. I don't even know what the GPUs cost. I didn't buy them, but I think it's out of reach for the average person. It's definitely an order of magnitude more than your gaming desktop with a GPU in it. I saw this interesting post from NVIDIA.
94
00:11:40.475 --> 00:11:44.075
I saw it this morning, and it was, I guess right now,
95
00:11:44.555 --> 00:11:56.190
to switch models sort of like mid project, you kinda have to like start over. And they made some advancements by which you can actually like shift the context from one model to another.
96
00:11:56.430 --> 00:12:13.565
So I could imagine, you know, like, you're working with a a local or or or, you know,
some other model that's way cheaper, and you kinda, like, get things set up with that. And then if necessary, you like shift the context over to a bigger model and just do the like hardcore compute that you need.
97
00:12:13.964 --> 00:12:19.885
I'm I'm just waving my hands here, that that sounds like it could be really neat, especially if you wanna save some money,
98
00:12:20.365 --> 00:12:37.390
you could do a low, you know, self hosted maybe even lesser model and like kind of get things set up and get to where you need to do the heavy work and then start paying the big bucks for that. Well, is interesting how it relies on video memory, VRAM as it's called in the GPU space,
99
00:12:37.824 --> 00:12:48.785
But then the unique architecture of Apple Silicon where they have shared RAM between the GPU and the CPU, the system on a chip, allowed them to take advantage of that for local models.
100
00:12:48.945 --> 00:12:51.824
Whereas if you have a Windows laptop or even Linux,
101
00:12:52.940 --> 00:13:14.645
the RAM that your motherboard has access to for random access memory for the processing is not what the AI uses. It needs explicit video RAM from the graphics card. So if you don't have a discrete graphics card. And then I saw, speaking of announcements from NVIDIA, effectively some Windows machines and Linux machines pivoting to this integrated shared memory
102
00:13:14.725 --> 00:13:16.325
platform architecture
103
00:13:16.325 --> 00:13:32.204
specifically for local compute for LLMs. It's fascinating how it affects the hardware. It seems like it makes a lot of sense to have it all be the same, especially with RAM prices so high. I I guess you gotta still keep all your regular OS memory in there, so you're not gaining
104
00:13:32.204 --> 00:14:01.675
any free RAM, but I don't know, shuffle things around. Well, one of the exciting things from this Bitcoin Red Team development was, okay, these tools are so powerful. You can point them at repos, they find bugs, they pretty much act unrestrained, they do what you ask, they don't give you pushback, they don't virtue signal you. And it makes us wonder, you know, if the tools are so powerful, what happens if you point them at, say, projects that aren't open source, where there's not a public GitHub repo,
105
00:14:01.755 --> 00:14:05.195
rather you have to do some extra lift, some extra digging.
106
00:14:05.435 --> 00:14:21.519
And over the weekend, I learned from the smart guys hacking in the open source mining space what Ghidra is. Why don't you tell us what that is, Scott? I Oh, think it's pronounced Ghidra, but Ghidra? Okay. Ghidra. So Ghidra is a tool that's been around for a long time, and it's a disassembler
107
00:14:21.519 --> 00:14:23.360
and a decompiler.
108
00:14:23.360 --> 00:14:25.920
So it can take machine code, which is,
109
00:14:26.774 --> 00:14:46.550
you know, when you take source code and you compile it and then you assemble it into machine code, you get a binary, right? And that's an compiled application that runs directly on your CPU. And so it's it it it it is the program, but it's very hard to interpret what it's actually doing because it's just it's literally manipulating
110
00:14:46.550 --> 00:14:47.430
instructions.
111
00:14:47.430 --> 00:14:57.125
It's a list of instructions directly to your CPU. And so backing out what the actual program is doing from that directly can be very hard for a for a mere human.
112
00:14:57.285 --> 00:15:07.449
But AI is pretty dang smart. And so you can use this tool, Ghidra, which was developed by the NSA, which goes backwards. And so it will disassemble
113
00:15:07.690 --> 00:15:09.050
and decompile,
114
00:15:09.050 --> 00:15:13.290
and what you'll get on the output is C or Rust code
115
00:15:13.449 --> 00:15:27.404
that was created by running the process in reverse. It's still pretty hard to interpret when you're looking at this decompiled code because you don't get like function names, right? When you're writing C code, you might create a function that's called like generate entropy.
116
00:15:27.725 --> 00:15:28.524
But
117
00:15:28.925 --> 00:15:37.230
the compiler strips away all of those, you know, human friendly words. So what you get out of the decompiler is just kinda like function numbers.
118
00:15:37.470 --> 00:15:52.605
But it will give you an idea of what the structure of the program is. You can see the arguments to the functions and things like that and, like, where the code is jumping and, like, loops and things. But, yeah, it's still very hard to to understand. But AI is wicked smart. Right? So you can plug it in this disassembled
119
00:15:52.605 --> 00:16:05.010
decompiled code into AI and be like, alright. What's it doing? And that's where it gets really powerful is you can you you can just make some sense out of this madness, way more so than you ever could before
120
00:16:05.329 --> 00:16:09.410
with with tools like Idra. There's a there's a there's a really popular
121
00:16:09.490 --> 00:16:14.850
tool called IDA Pro, which is, you know, licenses cost like $50,000,
122
00:16:14.850 --> 00:16:27.335
and it's just traditionally only used because you have a huge financial incentive to figure out what this program is doing. And it provided a lot of tools for humans to kinda make these connections between this, like, totally obfuscated
123
00:16:27.335 --> 00:16:28.135
c code.
124
00:16:28.560 --> 00:16:43.360
But now you can kinda just do that with AI. So Gidra's free open source and it's it's very the program itself is very scriptable, so it works great with AI. You can just point it at a binary and be like, go. Tell me what's going on here. It still does make a lot of mistakes
125
00:16:44.075 --> 00:16:52.715
because it's so ambiguous as to what this code might be doing. But now more than ever before, we can start to figure it out. So is it nondeterministic
126
00:16:52.715 --> 00:17:04.200
in the sense that you could write a bunch of
different functions that output the same goal, the thing you're trying to result. And so when you try to reverse from a result or machine instructions
127
00:17:04.280 --> 00:17:12.214
to the actual human readable code that does that, you can kind of reverse engineer source code, but it might not be a 100%
128
00:17:12.215 --> 00:17:18.214
what they wrote. It might just result in the same thing. Is that fair to say? Yeah. Yeah, that's correct. Right. The process
129
00:17:18.375 --> 00:17:19.654
of compiling
130
00:17:19.654 --> 00:17:32.010
code with a compiler, like a lot of information is lost. All a lot of that information is what's useful for humans to understand it. You know, compilers are very good at generating the right binary
131
00:17:32.010 --> 00:17:44.234
from the code. And so going backwards, it's it's pretty good. But, yeah, there's a lot of context about what it's actually doing that's stripped out when you come to To me, this just says like, if anybody wants something to be open sourced,
132
00:17:44.475 --> 00:17:47.434
it's about to be. Yeah. Yeah. I mean, wouldn't want
133
00:17:47.995 --> 00:17:59.769
to like try and modify a project. Like let's say you use git drone binary and you do compiled it right now. You're like, I want to modify it. It gets really hard to go back to a,
134
00:18:00.250 --> 00:18:07.565
to change things and go back to a working program from there. Because it's like, yeah, there's just no context. You
135
00:18:08.685 --> 00:18:10.845
know the architecture of the program,
136
00:18:11.085 --> 00:18:18.205
which is the architecture of the program is by the original programmers who kind of set up things like objects
137
00:18:18.420 --> 00:18:30.900
and they name them and they structure how these objects all work so that it like makes sense and it is clear on how to modify it. I think a lot of that gets lost when you decompose. But if you wanted to, for example, inspect
138
00:18:31.305 --> 00:18:43.385
what kind of communications are happening, what's going on with network traffic, it's a good tool for that. Yeah, it's a great tool for that. And the other tool, especially for network traffic, is Wireshark,
139
00:18:43.385 --> 00:18:48.070
which is just for sniffing packets on the network. Also very complicated,
140
00:18:48.070 --> 00:18:53.429
but AI can make a lot of sense out of it too. So you could you could have AI look at the at the binary,
141
00:18:53.669 --> 00:19:02.804
be it, you know, the the decompiled binary. You could have AI look at the actual network traffic from the running binary and start to put those things together, and you can get a real good picture
142
00:19:03.125 --> 00:19:13.470
of what's going on. Wild. It is wild. It's wild. It's really neat. So it it was just like, I've tried to do this before, you know, before the AI revolution,
143
00:19:13.470 --> 00:19:26.174
and it's just like it's so it's such madness, you know. It's like you just, like, can't keep all these things in your brain as you're trying to go through and figure out what this doing. But you can now just be like, hey. Just write me a markdown file of what's going on. And it's just like,
144
00:19:26.735 --> 00:19:30.335
you go. And you're like, wow. Alright. No kung fu. So,
145
00:19:30.575 --> 00:19:38.780
I mean, we're kinda dancing around the the root of the subject here. What's been done with Ghidra and decompiling
146
00:19:38.780 --> 00:19:43.180
over the last few days? Close-up. So our friend of the show and
147
00:19:43.420 --> 00:19:46.540
core maintainer of the LibreBoard project, Schnitzel,
148
00:19:46.620 --> 00:19:47.660
took KimiK3
149
00:19:47.660 --> 00:19:55.585
and a tool like Ghidra and pointed it at a plethora of different mining devices that are running with various
150
00:19:55.585 --> 00:19:58.945
firmwares and started sniffing. The first of which was,
151
00:19:59.425 --> 00:20:03.505
I think he's still testing others, but the first of which was with Venish firmware,
152
00:20:03.585 --> 00:20:11.280
which I actually haven't tested myself. My understanding is he just kind of asked, like Scott said, give me a report AI on, hey, I have
153
00:20:11.520 --> 00:20:50.865
these risks that I'm curious about. Is there a license fee connection that may kill my hash rate if the dev license server goes offline from the aftermarket firmware provider? How is network traffic encrypted? How are my shares accounted? Things of that nature. Are there remote kill switches and back doors? And so he's working on writing that up and really getting to the bottom of how it works, because as Scott just articulated that you do kind of have to poke around and guess how some of this is working because you don't have the full context from the manufacturer. But there were certainly some things that were brought up, I think Scott can speak to better, where it's like, why are they doing that? Do tell. We I mean, we've we've known for a while that
154
00:20:51.025 --> 00:20:52.544
Venish has
155
00:20:52.625 --> 00:21:04.030
it's they take a dev fee. So that's they take a portion of your hash rate and send it to to them, and that's how they extract the fee. And we we have known for a while that if you block
156
00:21:04.270 --> 00:21:07.950
that connection from your miner to their dev fee server,
157
00:21:08.030 --> 00:21:11.230
that the miner will go into a reduced mining state.
158
00:21:11.310 --> 00:21:40.559
They do that as to sort of penalize you for doing that. Right? You're essentially saying, no, I'm not paying you your dev fee. So they're like, okay, well, we're gonna essentially make your miner not work very well. Right? We're gonna reduce your hash rate by 10%, which in many mining cases means it's just not profitable. And you're like, fuck, I don't wanna do that. But what we haven't seen before is exactly how they do that. And so I think that's that is really cool to get into look a look into with with what Schnitzel's been digging
into is how exactly
159
00:21:40.559 --> 00:21:44.959
do they implement that and what, you know, what are the avenues for it to be misused?
160
00:21:44.960 --> 00:21:47.134
And we found out that
161
00:21:48.095 --> 00:21:54.575
the Venetius connection to their DevFee server is encrypted and authenticated over TLS,
162
00:21:54.575 --> 00:21:56.335
whereas the mining necessarily
163
00:21:56.335 --> 00:22:01.455
isn't. Right? Because TLS mining over Stratum is not really very well supported.
164
00:22:01.610 --> 00:22:06.330
That's So kinda funny. But, you know, there is an attack vector here where
165
00:22:06.410 --> 00:22:08.250
someone could man a middle connection
166
00:22:08.490 --> 00:22:15.690
and and and block that that connection from your firmware to the Venish license server effectively, like, killing your mining operation.
167
00:22:15.770 --> 00:22:18.615
Right? I think you can get in there and do that without you necessarily noticing.
168
00:22:18.855 --> 00:22:23.095
And I think he was looking at some, you know, possibilities
169
00:22:23.095 --> 00:22:37.060
for, like, remote code execution, meaning if you have, like, a fake license server set up, could you send back some information to the server, to the miner that would get it to do something else outside of its intended operation?
170
00:22:37.060 --> 00:22:39.139
That's where it starts to get really interesting.
171
00:22:39.300 --> 00:22:39.940
That's
172
00:22:40.580 --> 00:22:45.780
a big vulnerability for all network connected software is like how well are
173
00:22:46.015 --> 00:22:46.815
they
174
00:22:47.215 --> 00:23:06.750
what do they call it? Sanitizing their inputs, I think is what they call it. Is this what they call fuzz testing where you just kinda put random inputs to things and see how it responds? Yeah. I mean, I think fuzz testing is normally like at the the source code level, like how functions and and Oh, okay. And components communicate with one another. But yeah, I mean, I guess you could you could fuzz test a network connection too.
175
00:23:07.150 --> 00:23:09.070
This is sort of like educated
176
00:23:09.070 --> 00:23:17.264
fuzz testing where fuzz testing is like, I'm just gonna throw everything at it as much as I possibly can come up with and see what breaks. If you can get a look at the firmware,
177
00:23:17.265 --> 00:23:29.519
now you can make some very educated guesses about what might break it when you throw something at it. Right. And the concern there is, well, if sending this command from their license server gets your hash rate to dip 10%
178
00:23:29.600 --> 00:23:41.875
because they don't want you to block the license server connection for the dev fee split, is there something else they could send as a command to my entire fleet and shut it off? Yeah. That's the big one, right? That's
179
00:23:41.875 --> 00:23:45.634
the famous ant bleed exploit that Bitmain ran in 2017
180
00:23:45.634 --> 00:23:46.434
where
181
00:23:46.434 -->
00:23:47.074
they
182
00:23:47.394 --> 00:23:50.674
you know, they had a connection from stock firmware
183
00:23:50.835 --> 00:23:51.955
to their server,
184
00:23:52.610 --> 00:23:55.170
And depending on how that server responded,
185
00:23:55.170 --> 00:24:03.650
they could just totally stop your miner from mining. So that's a really big one. There's also All these miners have APIs.
186
00:24:04.050 --> 00:24:09.315
Tyler, I know you love using those, right? Where you can say, hey, like ramp down the hash rate, change
187
00:24:09.955 --> 00:24:12.195
my mining pool information.
188
00:24:12.275 --> 00:24:19.555
And so getting a look at those APIs, seeing how well they're sanitizing their inputs, seeing how well those are secured. I mean, they're not secured at all.
189
00:24:20.630 --> 00:24:21.190
Potential Those are
190
00:24:22.230 --> 00:24:27.910
attack vectors, or they are attack vectors. They're the potential for very mischievous.
191
00:24:28.470 --> 00:24:41.204
But look at what we were able to find out with, you know, $40 and Kimi K3 credits and some directed prompting from people who know what they're looking for in a tool like Kidra. Who knows what we'll find? There's many
192
00:24:41.684 --> 00:24:46.820
firmwares to point this exercise at. Yeah. Yeah, there really is. And I mean,
193
00:24:48.100 --> 00:24:58.019
I feel like a lot of this stuff has been out there and just sort of not important to the mining community. Like, they just don't seem to care about it. But
194
00:24:57.375 --> 00:25:06.895
I think exposing all these things with very specific details, I think will be very useful as we try and as we enter this glorious new age of AI
195
00:25:06.895 --> 00:25:08.095
and knowing
196
00:25:08.335 --> 00:25:15.610
that your adversaries are doing this. Particularly too, if you have large mining sites, and the same goes for AI HPC
197
00:25:15.690 --> 00:25:20.090
that if they were to suddenly shut off could substantially
198
00:25:20.090 --> 00:25:24.330
shock or damage a very critical piece of energy infrastructure.
199
00:25:24.330 --> 00:25:43.380
That's the thing here is like, you know, it's not just a couple machines plugged into your outlet. It's like, oh, there's two gigawatts in Texas where if they if they shut this off remotely, that could damage and break a lot of things. And and I just feel like Bitcoin mining has been so lax about this stuff. I just you know, they're they're probably because it's it's all proprietary,
200
00:25:43.860 --> 00:25:51.380
firmware, and you just don't have a lot of eyes on it, I I I think it's way too lax. There's gonna be insane gaping holes
201
00:25:51.685 --> 00:25:57.205
that, will really affect Bitcoin mining. I mean, Bitcoin itself is is much more
202
00:25:57.765 --> 00:26:10.539
reviewed and scrutinized because it's open source, and, you know, people are constantly thinking adversarial about it, and there's a huge financial incentive to do that. Yeah. I think Bitcoin mining is is gonna be in for some,
203
00:26:11.419 --> 00:26:17.339
rude awakening when when a lot of this stuff gets out because it's just been it's been so overlooked.
204
00:26:17.340 --> 00:26:20.620
And I really think that that's a big benefit of
205
00:26:21.025 --> 00:26:36.259
open source, like, we're trying to do with Mugena is that, like, now it's out there and you can look at it. Both white hats and black hats and the developers and and the the people who are running this can look at it and be sure for themselves. Speaking of rude awakenings,
206
00:26:36.260 --> 00:26:56.755
sorry, did you have a question, Tyler? Go ahead. Well, I was gonna kinda see if we could shift gears a little bit and talk about BIP 110. Okay, let me just ask one closing comment. With these open source, open weight, powerful frontier models like Kimi K3 and a tool like Gidra or the wire you're sniffing one for network traffic, Wireshark.
207
00:26:56.995 --> 00:26:58.355
Does it effectively
208
00:26:58.435 --> 00:27:02.515
tell us all of the special sauce that these firmware manufacturers
209
00:27:02.515 --> 00:27:16.470
have been trying to keep close to the chest? My question being like, if this is available for anyone with $50 and an open router account, what are they why are they so adamant about staying close source then? Or is there something we can't extract that they can still hold?
210
00:27:17.185 --> 00:27:38.139
I don't know, man. It's like, why are they doing this? I mean, maybe it was successful before, but it's not gonna be anymore. It's not gonna go. We've already looked, gotten such a deep dive into this stuff, and I think we're just scratching the surface. Okay. So things like their whatever performance tuning of the chips that gives them 7% more hash rate, we could also figure that out Absolutely. With enough
211
00:27:39.419 --> 00:27:41.820
Okay. Absolutely. There's
212
00:27:41.820 --> 00:27:45.099
several people working on this now in the open source miners community.
213
00:27:45.259 --> 00:28:07.380
And yeah, it's you know, if you have hardware on your desk that's doing this, now more than ever, you can figure out what it's doing. Yeah. I think they're just they're afraid of, like, losing money and their business is shutting down. That's why they're so resistant to open source. They think that they're closed source solutions.
214
00:28:07.700 --> 00:28:11.299
They think that they've protected their profit
215
00:28:11.860 --> 00:28:12.980
by
216
00:28:13.460 --> 00:28:37.340
enclosing it behind these closed source walls. But as those walls come down, it's it's gonna get exposed. And, you know, there's already solutions like Mugena out there that don't require a dev fee. Like, the day is coming where anyone can switch to a non dev fee firmware for their miner very quickly and easily. So yeah, I mean, if you've built a company, I guess I can sympathize
217
00:28:37.340 --> 00:28:39.980
with what they're doing and why they're doing it.
218
00:28:40.380 --> 00:29:08.159
Build a company, your profit model is you get a percentage of other people's hash rate who are running your firmware, and they are incentivized to run your firmware because you have special optimizations that make their miner work better. And that being open source means it's less likely that you're gonna be able to protect your profits. Right? Like Yeah. Yeah. I mean, it's it's rent seeking. Right? Yeah. They're Yeah. It is. They're providing
219
00:29:08.559 --> 00:29:12.799
a value in this this, you know, additional tuning capability
220
00:29:12.799 --> 00:29:21.164
and easier to use software and all that stuff. But, you know, the the the sort of locking you in so you have to continually pay for this software
221
00:29:21.245 --> 00:29:30.605
is what is especially bad and Mhmm. Which is is going away. It's absolutely going away. I think software companies need to revert back to,
222
00:29:31.399 --> 00:29:37.399
you know, making money, selling things that work good. Right. TurboTax 2,009
223
00:29:37.480 --> 00:29:47.544
on a CD ROM. Yes. That's so bullshit. Like, you have to buy it every year. It's just I don't know. I haven't done this yet, but it's gonna be so good at doing your taxes. Like,
224
00:29:48.505 --> 00:29:49.144
jeez.
225
00:29:49.625 --> 00:30:02.360
Yeah. It's gonna be What happened? A lot of things. What happened, Iko? Some some bib thing happened? Well, I don't know. There were these for, like, the last two years, anyone who used Bitcoin and didn't support Luke Dass Jr. Being called a pedophile.
226
00:30:02.360 --> 00:30:03.560
Oh, boy. And
227
00:30:04.040 --> 00:30:27.629
and then supposedly there was some event where Luke was supposed to become the the the pope of Bitcoin. And they tried, but only got two blocks in, and the rest of the network left them behind. What I don't get is why did they activate already? Right? Wasn't the whole point of the signaling thing is to gauge the interest of the network in adopting this? And they only had, like, a few percent,
228
00:30:27.950 --> 00:30:28.990
but they activated
229
00:30:28.990 --> 00:30:57.860
it anyways. I don't get it. What wasn't the whole point of signaling to be like, yeah, I'm into this. I'll do it. I think it was, like, set on a like, at that block height. They're like, at this block height, it's gonna activate or the like the counting's gonna start. Right? And so there's like the two thousand sixteen block period where they're counting to see if there's enough signaling support for it to activate. I think the activation was scheduled for September like roughly September 1. So I think this, like, 2,016
230
00:30:57.860 --> 00:31:02.580
block epoch that we're in right now, like, this is the the signaling period,
231
00:31:02.740 --> 00:31:04.179
if I understand correctly.
232
00:31:04.660 --> 00:31:40.374
But there's Or am I get am I getting that wrong? I I don't I I my mind's kinda blown here. So but the I guess they wanted to, like, be double sure that if you're signaling, you're into this, and that's what just happened. I don't know. I just it seemed like it was like, okay. Start signaling, and then not very many people did, or not many miners, not many nodes did. And then they went ahead and started enforcing the rules anyways. Yeah. So I think, like, people were free if I here's the way I understand it. Like, any miner was free to, like, signal their support for BIP one ten, like, prior to block height
233
00:31:40.534 --> 00:31:43.734
nine six one six three one. Yep.
234
00:31:44.375 --> 00:31:48.054
And so that is kind of, like, a way to roughly gauge,
235
00:31:48.135 --> 00:31:51.495
like, how much support there is among miners for this proposal.
236
00:31:51.970 --> 00:31:57.730
Okay. But then, like, the way the the rules work is once it activates
237
00:31:57.970 --> 00:32:05.170
or once once the signaling period begins, which is what block high nine six one six three one.
238
00:32:05.835 --> 00:32:15.354
Saturn. Yeah, that was like the beginning of the signaling period. So it's like- The mandatory signaling. Yeah, so it's like once that block hit, then there was,
239
00:32:15.755 --> 00:32:17.914
know, now we're in this difficulty
240
00:32:17.914 --> 00:32:26.790
epoch number four seventy seven. So we have twenty sixteen blocks for all the miners who wanted to signal for BIP 110
241
00:32:27.110 --> 00:32:27.830
to
242
00:32:28.870 --> 00:32:34.630
show up and show that. And none of the, you know, that's not happening. There's been zero
243
00:32:34.790 --> 00:32:35.590
blocks
244
00:32:35.804 --> 00:32:37.004
on mainnet
245
00:32:37.004 --> 00:32:41.964
that are signaling for BIP 110 support in this epoch. And so then the
246
00:32:42.924 --> 00:32:43.484
the
247
00:32:43.965 --> 00:32:47.164
fork happened because they were rejecting
248
00:32:47.164 --> 00:32:47.965
Bitcoin
249
00:32:48.280 --> 00:32:49.159
blocks
250
00:32:49.240 --> 00:32:52.200
from main net that had what they considered
251
00:32:52.280 --> 00:33:00.039
spam. And so those blocks went one way, but then nobody wanted to build on those blocks. And so the rest of the network just left them behind.
252
00:33:00.600 --> 00:33:04.485
And so we still have this signaling period that's continuing,
253
00:33:04.645 --> 00:33:06.085
but it's
254
00:33:06.085 --> 00:33:10.164
statistically impossible for BIP 110 to overtake
255
00:33:10.325 --> 00:33:24.710
the two seventy five block lead now. I feel like the way I imagined it working, which sounds like this wasn't quite correct, but it's like you go into this signaling period where the rules are not enforced. And if in any one epoch,
256
00:33:24.790 --> 00:33:29.669
you know, over a threshold, like 55% or probably ideally higher are signaling,
257
00:33:29.990 --> 00:33:35.005
then you start enforcing the rules, right? Cause you've gauged the interest.
258
00:33:35.325 --> 00:34:02.059
And you could put that into the into the
software, I guess, to look at what percentage of the blocks are signaling. But then, you know, then you don't get in this situation, this, like, embarrassing situation that they were in on Saturday where, like, no one, no one wanted to enforce the rules. I see what you're saying. Yeah. It's like they're, they've activated at the beginning of the signaling period instead of waiting till the end of the signaling period to activate. Yeah. Yeah. I'm not sure, dude. Those guys are fucking crazy.
259
00:34:02.684 --> 00:34:04.764
Well, now what's the alternative?
260
00:34:04.924 --> 00:34:23.170
Hard fork, change the proof of work algorithm? Yeah. And Luke's been talking about that for years. So, I mean, the guy's just really into forking off. He doesn't, he's not into Bitcoin. He's into something else. Like he's created some totally different thing. And you know, what's crazy to me is that like in 2021,
261
00:34:23.490 --> 00:34:38.115
when Diverter wrote Mining for the Streets, and I wrote my mining guide, and mining at home started like putting an industrial sized miner in a home started to become a thing that people were interested in. Luke was very vocal about
262
00:34:38.275 --> 00:34:55.060
people should not be mining at home. Really? Yes. And I I've I've saved a lot of the screenshots, and I can share them with you. But he's like, I've had arguments with him about why mining at home was a good idea, and he was one of the largest critics against
263
00:34:55.220 --> 00:35:05.705
people mining at home and these small scale miners. And That's wild. Did did you remember any of the reasons he gave? Like, what So it was things like it's it's not profitable
264
00:35:05.705 --> 00:35:12.825
to mine at home. You shouldn't be mining at home because, like, your hash rate's never gonna make a difference. And just
265
00:35:13.770 --> 00:35:18.970
I'll pull up some of the screenshots and send them to you. And they're
266
00:35:18.970 --> 00:35:34.325
just stupid reasons. And I would argue with him on Twitter about why mining at home was a good idea, why people should do it. And he would say that they shouldn't. And so what's evolved over the last several years is like, you know, people identifying themselves as plebs
267
00:35:34.325 --> 00:35:36.005
and being pleb miners.
268
00:35:36.404 --> 00:35:42.110
And it's interesting to me that Luke ended up starting Ocean.
269
00:35:42.110 --> 00:35:43.470
And their
270
00:35:43.470 --> 00:35:44.510
entire
271
00:35:44.590 --> 00:35:45.710
target
272
00:35:47.310 --> 00:35:48.670
market was
273
00:35:48.670 --> 00:35:50.990
this pleb sized
274
00:35:51.230 --> 00:36:14.340
miner or these pleb miners, because those are the ones whom I've seen supporting BIP 110. And it's like, BIP 110 is for the plebs. We have a voice. We're going to make a difference. Bitcoin is for monetary transactions only, not for spam. And like, it's just really interesting to me that the guy who was like telling people not to
275
00:36:14.820 --> 00:36:19.700
become small scale miners ended up building a business and a movement
276
00:36:20.020 --> 00:36:21.460
to fork Bitcoin
277
00:36:21.865 --> 00:36:30.425
that depended on the support of these small miners that he was initially encouraging not to mine at all. It's just looking at it at this long,
278
00:36:31.225 --> 00:36:32.025
arcing
279
00:36:32.025 --> 00:36:56.095
timeframe that I've been watching and interacting with Luke and paying attention to mining and how mining has evolved. It's just really interesting to me that this guy built ocean and then built the BIP 110 movement on the backs of people whom he was telling not to get into mining at all initially. And now he's reliant on those people supporting him, and no one who is,
280
00:36:56.495 --> 00:37:25.695
like, a serious player in this game gives a flying fuck about bip one ten, and that's evident now for anyone to see with the dead fork that he's created. It's plainly there for anyone to see. No one gives a fuck about BIP one ten. It's a retarded project that should have never made it this far or sucked so much air out of the room for so long. Amy, you know Well, it certainly was over quick. Yeah. I'm glad we've got something done with. And they had to fucking cheat to try and even make it work.
281
00:37:25.935 --> 00:37:28.655
They had to cheat and change the default endpoint
282
00:37:28.655 --> 00:37:41.560
of Ocean to point to a BIP one ten node, and not everybody got the message. They were effectively stealing people's hash rate is what it comes down to, and they still lost. These guys are scammers. I saw there was like
283
00:37:41.800 --> 00:37:50.165
this there was an article about this growing movement of people demanding leadership change at Ocean because of that whole ordeal of changing the default endpoint.
284
00:37:50.244 --> 00:37:59.365
And it just struck me as so funny as, like, leadership change isn't gonna isn't the problem here. Right? The leadership there isn't the problem. It's the centralization.
285
00:37:59.605 --> 00:38:05.680
Like right? We we now have the Bitcoin 10 chain, which is like the most centralized
286
00:38:05.760 --> 00:38:06.560
chain
287
00:38:06.880 --> 00:38:10.640
Right. Ever. Right? I mean, it's only, you know, Roughnecks.
288
00:38:11.120 --> 00:38:16.865
It's, yeah, it's it's only roughnecks. It's only one pool. Right? Like, they could 51%
289
00:38:16.865 --> 00:38:20.385
attack that with ease, which they're going to do, right? They're gonna
290
00:38:20.625 --> 00:38:21.505
change the
291
00:38:22.305 --> 00:38:33.040
The proof of work algorithm. The proof of work algorithm. It's just, it is annoying. Yeah, it's like, why even waste your breath on trying to get Ocean to change its leadership when you can just change the configuration
292
00:38:33.520 --> 00:39:02.930
of your miner and point your hash rate somewhere else? Yeah. Yeah. I mean, I don't exactly know where you would change it to. We're not
necessarily in the best position yet on that, but we're working on it. Yeah. I mean, full pay per share is total bullshit. I mean, we all know that, but I mean, like, to to me, it seems like you're pointing your hash rate to Ocean, who's effectively stealing it from you and using it to try and support the Luke Dash Junior's failed fork project,
293
00:39:03.170 --> 00:39:16.715
that's worse than any shenanigans that have come out of full pay per share, in my opinion. I don't know. I don't know. I think the, like, domination of amp pool and foundry is probably worse due to its potential,
294
00:39:16.955 --> 00:39:17.835
effectiveness
295
00:39:17.835 --> 00:39:34.350
in centralizing the potential for shenanigans, I guess. Yeah, but they I mean, to date, they haven't done anything. And now we have, like, actions that have been taken by Ocean's staff and Luke to take something from people. I'm sure it was in their terms and conditions. Probably.
296
00:39:34.750 --> 00:39:38.830
I mean, I'm sure. But it's still just like, that doesn't make it right.
297
00:39:39.310 --> 00:39:39.790
And
298
00:39:40.110 --> 00:39:43.470
yeah, I mean, I'm not saying that full pay per share
299
00:39:44.265 --> 00:39:46.425
a great alternative. It's like those
300
00:39:46.905 --> 00:39:49.225
are two very shitty choices in my opinion,
301
00:39:49.944 --> 00:40:08.710
but they're not the only choices. Now we've got Parasite Pool. We've made spinning up your own pool very easy with Hydro Pool. I would love to see somebody take that and gain some traction and have several miners using it. Easy to deploy PPL and S model. I think there's gonna be more we have grid
302
00:40:09.030 --> 00:40:18.865
pool. We had agent p on the show talking about grid pool. So you know, and I think there's we have P2P pool on the horizon. I know that's not available today, but
303
00:40:19.265 --> 00:40:21.265
I think the trajectory is clear.
304
00:40:22.145 --> 00:40:23.025
Even if it
305
00:40:23.985 --> 00:40:29.250
was only temporary, like switching your mining operation back to a full pay per share pool,
306
00:40:29.410 --> 00:40:30.530
just to get
307
00:40:30.530 --> 00:40:32.369
away from Ocean and
308
00:40:32.450 --> 00:40:33.090
the
309
00:40:33.410 --> 00:40:35.090
scamming that they've done
310
00:40:35.569 --> 00:41:09.930
until you can figure out a better solution, whether that's one of these other options that's available now, or just holding tight until p to pool comes out. I don't know. But I think that trajectory is clear, and it's coming. And, like, if I had a large mining operation and I had been pointing at Ocean and they stole my hashrate to support Luke's failed fork, and then they were, like, issuing me a refund right now. I wouldn't I would and, you know, this is just me. My moniker's burn the bridge. But, like, I would cut that relationship, and I would never look back. Also, how do they so you saw that they're gonna refund point 3 Bitcoin
311
00:41:10.010 --> 00:41:17.155
for eighteen hours? Like That's what I saw. How does that add up? We could work out the math on how much hash rate that equals. Yeah.
312
00:41:17.635 --> 00:41:21.955
But It doesn't it doesn't seem to add up to me, but maybe they're just addressing some specific
313
00:41:22.035 --> 00:41:34.740
claim from a Right. From a miner. Must sure I'm sure there's others that just didn't say anything, so they're not getting anything. For people that want to construct their own block templates, but opt into
314
00:41:35.060 --> 00:41:39.620
a tool like Datum that lets you team up with other people and get more consistent
315
00:41:39.805 --> 00:41:47.485
regular cadence payouts. Is Brains the only one that offers this now with SV2? What other options are there? I mean, there's I domain
316
00:41:48.525 --> 00:41:57.460
don't think you can make your own templates with Brains. Haven't they been marketing SV2? They have, but it doesn't support the job declaration protocol, right? It's just the encrypted
317
00:41:57.460 --> 00:41:59.300
comms and the binary.
318
00:41:59.300 --> 00:42:02.420
Who cares about that? Yeah. No one. Okay.
319
00:42:02.580 --> 00:42:06.740
Well, there's still lots of work to do. There's lots of work to do. It's exciting.
320
00:42:09.425 --> 00:42:10.945
SoloCKPool.
321
00:42:11.025 --> 00:42:25.030
They've implemented SV two now. Yes. That's right. But it's not it's solo mining. So Right. I don't think we have a shared pool SV two option. Oh, Blitz. No. That was solo mining too. I guess. What about demand? Demand.
322
00:42:25.030 --> 00:42:36.070
Yeah. I don't know if that's open access yet. I don't think it is. Oh, you still gotta like sign up and- Yeah. I think you actually have to KYC yourself. That was the last I heard. Maybe they've changed it. Well,
323
00:42:36.414 --> 00:42:38.175
yeah, the options are
324
00:42:38.335 --> 00:42:40.095
pretty sparse. But dude,
325
00:42:40.494 --> 00:42:42.255
I'd be fucking pissed
326
00:42:42.255 --> 00:42:47.535
if I had been mining to Ocean. I am pissed, and I wasn't even mining to Ocean. We can tell.
327
00:42:47.775 --> 00:42:49.135
So they're refunding people.
328
00:42:49.535 --> 00:42:51.694
The really frustrating part, right, is Ocean
329
00:42:51.900 --> 00:42:53.980
promoted themselves as so
330
00:42:54.380 --> 00:42:55.820
dang decentralized,
331
00:42:55.980 --> 00:43:05.915
and it's not. Right? Not at all. They could just unilaterally make these decisions Yep. And it affected people. I think that's the really shitty part. Like, I like to see I've
332
00:43:05.995 --> 00:43:08.075
always said we need more pools
333
00:43:08.235 --> 00:43:12.235
and ocean is another pool and it's good in that respect,
334
00:43:12.395 --> 00:43:14.635
but it doesn't solve the decentralization
335
00:43:14.635 --> 00:43:38.145
issue at all as we've seen. Well, in credit where credit's due, it's like the only popular and viable way to pool your hash rate, but still
construct your own block templates right now. Yeah. Yeah. Shout out to them for doing that. It's just it's a bummer that it got so wrapped up in the the Bitcoin 10 drama. Right? I mean, Yeah. But whose fault is that? Luke's. It's Luke's fault. This is 100%
336
00:43:38.145 --> 00:43:59.080
squarely on Luke Dash junior's shoulders. Yeah. Yeah. I mean, okay. So devil's advocate. Ocean, forget about BitPoint 10 for a second. I guess you could technically mine with them and not not touch the BIP one ten nonsense, although maybe you're supporting it indirectly. But let's just say you weren't. Like, it is probably the most the most not fully decentralized, but the most decentralized
337
00:43:59.160 --> 00:44:05.255
public shared mining pool. Because they let miners submit their own templates. Yeah. Which
338
00:44:05.575 --> 00:44:11.975
I mean, okay. If, like, we wanna look at it as just like a very, like, narrow
339
00:44:12.055 --> 00:44:12.935
definition,
340
00:44:13.174 --> 00:44:18.055
yeah. I don't know of any other pool that is sharing mining payouts
341
00:44:18.360 --> 00:44:32.040
and allowing miners to submit their own templates. Which is a big accomplishment. I mean, what problem are they aiming to solve? Compared to a lot of the other mining pools, this is the realm of SOC two compliance and bullshit
342
00:44:32.040 --> 00:44:40.305
legal teams. So, know, I don't know what it's like running a company of that caliber. So I do want to give them credit in that regard that that is an achievement
343
00:44:40.305 --> 00:44:42.145
for sure. Yeah. It is
344
00:44:42.545 --> 00:44:45.265
so hard to bootstrap a pool, right? I mean,
345
00:44:46.065 --> 00:44:53.940
you're SOL until you have, like, this momentum of significant hash rate of, you know, multiple exahash.
346
00:44:53.940 --> 00:45:03.485
You can't do anything until you have that. So it's really hard to do, but they just shot themselves in the foot with the the BIP one ten association, the NAATS association,
347
00:45:03.645 --> 00:45:04.924
and then overstating
348
00:45:04.924 --> 00:45:06.285
their decentralization
349
00:45:06.285 --> 00:45:11.964
claim. Well, definitely see ECO's frustration with if they're on the high horse of decentralization
350
00:45:11.964 --> 00:45:26.420
and then they flip a bit, a default bit, and have you pointing towards BIP one ten by default. And then they say, sorry, we have to pay you back. That is frustrating. That is frustrating. I mean, again, in their defense, this is
351
00:45:26.740 --> 00:45:30.660
I'm weakly supporting this, but, like, they only do that for their stratum v one endpoints.
352
00:45:31.125 --> 00:45:37.205
So it didn't apply if you had data. They bootstrapped their pool on the backs
353
00:45:37.445 --> 00:45:38.805
of plebs,
354
00:45:39.045 --> 00:46:17.110
whom are the exact group of people Luke was discouraging from even trying to mine at home in the first place. And that's one of the things that really pisses me
off about them, those guys. Yeah, the plebs are big supporters of them. So that's a shame to hear. I hope he doesn't still think that. Yeah, I hope his opinion has changed, which people can do. Well, yeah, he'll encourage people to mine to his pool because it it supports him, and it that's how he lines his pockets. But before he had his pool, he was telling people not to mine. Genius. Man, we need to get p two pool v two, grid pool. These projects need, we need them now. We need them yesterday.
355
00:46:17.110 --> 00:46:17.990
Yesterday.
356
00:46:18.630 --> 00:46:24.150
Yeah. This this seems like an imperative for the Bitcoin mining community is like,
357
00:46:24.390 --> 00:46:25.110
we need
358
00:46:25.510 --> 00:46:27.605
to get these shenanigans
359
00:46:27.605 --> 00:46:43.050
out of the hands of these centralized players. Like, you just you can't you can't trust. Right? Because opinions change. Incentives change. Like, that's the whole point of Bitcoin is you, you know, you don't have to trust, and you don't have to continue to trust. Pool protocol,
360
00:46:43.050 --> 00:46:47.930
certainly the future. So maybe we should just fork it and fix it for good, right? Yeah.
361
00:46:48.090 --> 00:46:58.625
Or everyone could solo mine. I think everyone should solo mine, but I guess when you got a little bit more money on the line. Mara's doing it. Yeah. Come on, be like Mara in that regard. Yeah.
362
00:46:59.665 --> 00:47:04.785
It'll be interesting to see what follows. Actually, I'm not that interested, to be honest. I like
363
00:47:05.185 --> 00:47:09.745
to see these issues come to the forefront of the current news cycle,
364
00:47:10.145 --> 00:47:20.080
some of these issues that we're championing. People are thinking about it. Yeah. Yeah, it's good. The more people learn about mining, I think, the better. I do think that's a very underappreciated,
365
00:47:20.160 --> 00:47:21.120
under understood
366
00:47:21.760 --> 00:47:40.839
domain of Bitcoin. There's been a lot of shit talking on the Internet about, you know, leading up to this, like, who controls the network? Is it the miners or is it the nodes? It's both. Right? Mhmm. Iko, is it is it it's both. Is it? Is that right? I mean Well, I I think the lesson in 2017
367
00:47:40.839 --> 00:47:44.440
was that the nodes control the network. Okay. And
368
00:47:44.760 --> 00:48:05.165
I think there's a good argument there for that. Because the miners will just they're just processing the transaction. I mean so I mean okay. So then is it they run? Is it that what's the argument for miners control the network? Is it that they are the ones selecting the transactions to include in the block? Well, they're choosing what nodes to point to, I guess. True.
369
00:48:05.325 --> 00:48:20.460
I mean, they're the ones generating the blocks. So Like, let's say let's say you had all you know, an overwhelming majority of the nodes that are supporting some change, but none of the miners are. Like, Bitcoin still comes to a standstill if the miners aren't adding new blocks to the blockchain. True.
370
00:48:22.155 --> 00:48:23.195
So, yeah,
371
00:48:23.915 --> 00:48:30.955
it's the nodes, but it's the nodes that the miners have selected. Yeah. I mean, I guess like me as an individual
372
00:48:30.955 --> 00:48:32.555
running a node at home,
373
00:48:32.955 --> 00:48:38.220
I wouldn't really argue that I'm controlling the network. I'm controlling my local mempool.
374
00:48:38.940 --> 00:48:40.060
I'm controlling
375
00:48:40.140 --> 00:48:41.580
broadcasting transactions
376
00:48:41.580 --> 00:49:00.335
to the network, but me at home just running a node and choosing to which transactions I'm gonna propagate or not, I wouldn't really argue I have any control over the network. And I guess that was kind of like the BIP 110 argument was that, right, if I'm understanding them correctly, it's like, if I'm running a node that's
377
00:49:00.335 --> 00:49:02.415
BIP 110 compatible,
378
00:49:03.055 --> 00:49:05.530
then I'm only gonna propagate
379
00:49:05.770 --> 00:49:08.730
these strictly monetary Bitcoin transactions,
380
00:49:08.810 --> 00:49:12.570
and I'm gonna filter out the spam transactions,
381
00:49:12.890 --> 00:49:13.770
and therefore
382
00:49:14.010 --> 00:49:30.625
I have control over the network. But if that is their logic, I'm not sure I believe, I don't think that's, I don't think it's true. Yeah, just seems you can't have one without the other. No wonder they used to be the same thing. The true detections are gonna get through anyway. I said no wonder the nodes used to also be miners.
383
00:49:31.140 --> 00:49:38.020
Yeah. That's right. Yeah. I mean, can't mine without a node, but you also can't, like, process any transactions without the miners. True.
384
00:49:38.180 --> 00:49:51.675
Issuance and settlement. Yeah. You need both. Yeah. It's both. I'll go with both. Let's go with both. You heard it here. Solution found. What else is on the docket this week, gents? How's all your hacking going on, Scott? Oh, man. It's going good. The
385
00:49:51.914 --> 00:49:55.994
Bidax Proto with the four proto chips is working very well.
386
00:49:57.320 --> 00:49:59.720
Very close. I think this this week, I'll,
387
00:50:00.200 --> 00:50:17.105
get some boards made with the new PCB changes put in there. And Johnny has done a bunch of work on the firmware for the Bonanza chips that is also working. He released. It was lost in the in the other news, but I think on Saturday,
388
00:50:17.345 --> 00:50:35.620
he released a version of ESP Miner that supports, the BITX Bonanza chips, and that's that's working really well. So as far as the BITX Bonanza goes, there's a few hardware changes that'll be needed there, but then we can release that. So I think basically both BidX Proto and BidX Bonanza will be released at the same time.
389
00:50:36.660 --> 00:50:37.540
Awesome.
390
00:50:37.700 --> 00:50:41.875
Yeah. That's really
cool. I like to see that. Two American
391
00:50:41.875 --> 00:50:56.320
ASIC chips. American ASIC chips made at TSMC like the best of them. Yeah. You know? No. I mean, it's been a huge goal of mine for a long time to get off of just bitumen chips. And so I'm really stoked to see that this is coming together.
392
00:50:56.960 --> 00:51:02.560
You know, you just look at the if you just look at sort of the nameplate specs on these new miners,
393
00:51:02.560 --> 00:51:14.215
they don't they don't compare all that well with the with the bitumen chips, especially the the new s 23 chips that have just come out. So, you know, it it's a little bit lackluster as far as just, like, headline, you know, specs,
394
00:51:15.015 --> 00:51:28.720
there's so much more potential here with these. I really like focusing on those, get off of just a sole supplier that isn't actually a supplier and doesn't want to supply us anything. Yeah. If you're not paying attention to -.256f.org,
395
00:51:28.799 --> 00:51:44.285
where we have our gamified dashboard for our hydro pool, that is where you get the inside scoop on all the miners getting developed. I think I saw some from Schnitzel earlier of all the different firmwares that he's attacking viciously with Kimi K3 and Ghidra.
396
00:51:44.525 --> 00:51:52.910
So I've seen Proto Bitaxe on there. Yeah, there's a bunch of cool stuff. Adi, shout out to Adi. He's had some cool stuff from Kanan on there. Multiple fronts.
397
00:51:53.150 --> 00:51:58.270
Multiple fronts. He's made, I think he's just about to release it, but Mugena on
398
00:51:59.710 --> 00:52:01.390
Kanan Avalon Nano S3,
399
00:52:01.985 --> 00:52:03.105
which is Yeah.
400
00:52:03.665 --> 00:52:17.900
Great. We'd like to see the community porting Magina or adding Magina support for other ASICs and other boxes. We need like a you could almost have there's the live hashers, users and workers, but you could almost have like a Skunk Works category.
401
00:52:17.900 --> 00:52:18.700
Oh yeah,
402
00:52:18.860 --> 00:52:21.420
that would be cool. That would be really cool.
403
00:52:21.820 --> 00:52:26.700
Of course you could fake it, but you know, it'd be a neat way to like show off what you've got working to our
404
00:52:28.380 --> 00:52:36.455
don't see anything wild on there right now. Yeah, mine are not plugged in at this very moment. They're currently disassembled. I should have organized that a little bit better, but
405
00:52:36.775 --> 00:52:38.535
see Andy's filament dryer.
406
00:52:39.575 --> 00:52:46.215
We'll check it out there. It doesn't look like there's been any changes in our difficulty leaderboard either, nor the loyalty leaderboard.
407
00:52:46.380 --> 00:52:55.660
Iko, you're still holding out. Am I number 10 still? Yeah. Awesome. Hanging on. Yeah. That's fantastic to see. You can check it out @dash.256f.org.
408
00:52:55.660 --> 00:53:03.515
Great way to support the show by chancing a solo mind block with all proceeds going to fund
the developers working on these projects. I,
409
00:53:03.995 --> 00:53:13.115
went down rabbit hole last night and stayed up way too late, but we need to have a cheap mining node. I really think that
410
00:53:13.619 --> 00:53:33.135
we need more home miners, more solo miners mining to their own node. Right? This is good on a lot of reason for a lot of reasons. Right? You're making your own templates when you're mining to your own node. You are in sole control whether, you know, you're allowed to mine to the to the Bitcoin network. Right? You're owning your your whole stack.
411
00:53:33.135 --> 00:53:46.000
You're in charge of your own security, everything like that. So I think we need a ideally $100 box that you can plug in and start pointing all your home miners at that box on premises
412
00:53:46.000 --> 00:53:46.800
ASAP.
413
00:53:46.800 --> 00:53:48.800
It gets tough because
414
00:53:48.800 --> 00:53:49.760
traditionally
415
00:53:50.080 --> 00:53:51.600
nodes have required,
416
00:53:51.600 --> 00:53:55.760
you know, a one or these days, maybe one and a half, two terabyte
417
00:53:55.760 --> 00:53:56.640
NVMe
418
00:53:56.640 --> 00:53:57.280
fast
419
00:53:57.705 --> 00:53:58.505
SSD.
420
00:53:58.505 --> 00:53:59.785
Right? And those are expensive.
421
00:53:59.945 --> 00:54:20.960
So are Raspberry Pis nowadays, Scott? What's that? Raspberry Pi fives are a $100 nowadays too. Yeah. Right. And that's you don't even have the the two terabyte drive to go with it at that price or even the way to plug it in. But we have a solved software solution for the storage requirements on mining nodes, and that's pruned nodes. So pruned nodes are full
422
00:54:21.040 --> 00:54:28.925
validating nodes. They're first class citizens in every way, but they just do not store old blocks. They only store like the
423
00:54:29.005 --> 00:54:31.325
the latest in the UTXO set.
424
00:54:31.645 --> 00:54:33.725
And that is a 100%
425
00:54:33.805 --> 00:54:38.605
perfectly good node configuration to mine to. We have confirmation from
426
00:54:39.279 --> 00:54:40.559
doctor Konkolovas
427
00:54:40.559 --> 00:54:52.055
that that's that's good. Eric Foscol from Libbitcoin was also like, that works. I think it it sort of pained him to suggest a, Bitcoin Core project that could work, but he did, admit,
428
00:54:52.055 --> 00:54:53.815
yes, it can work. So,
429
00:54:53.894 --> 00:54:56.855
you know, I think we need to normalize HomeMiner's
430
00:54:56.855 --> 00:54:58.615
mining to your own pruned nodes
431
00:54:58.775 --> 00:55:02.375
and set up a sort of ready to go node image
432
00:55:02.375 --> 00:55:18.930
that people with this cheaper hardware can set up in mind too. There's been some there's been some cool projects that could really streamline this and reduce the, you know, the requirements of the box that you're running this on. One that you can you can do right away is assume UTXO.
433
00:55:19.315 --> 00:55:36.079
So there's these snapshots that you can download, and the the checksum of these snapshots is baked into Bitcoin Core, and that allows you to just instead of syncing it with a network, you can just download these snapshots and run them on your pruned node and get going, much quicker. Current way that that works is you can download these snapshots.
434
00:55:36.079 --> 00:55:45.200
PiconCore will validate the checksum of it. Now you have this UTXO set, and you can start mining right away. But in the background, it actually does the whole block download and verifies everything.
435
00:55:45.635 --> 00:55:50.515
That could take a long time, but you're kinda up and running in the meantime. There's a really cool project called,
436
00:55:51.155 --> 00:55:52.035
UTXO
437
00:55:52.035 --> 00:56:00.050
and a node software called Floresta that runs on top of that. And they have this, like, tree structure of UTXOs
438
00:56:00.050 --> 00:56:08.770
that I think it has, like, this Merkle root. Right? So you can confirm by this, like, running hash that all the UTXOs that you have are valid. That
439
00:56:09.330 --> 00:56:11.545
that could really make small
440
00:56:11.625 --> 00:56:12.505
machines
441
00:56:12.505 --> 00:56:20.665
work as mining nodes. The only problem is it doesn't have the ability to build block templates in it. So it won't currently work for my it
442
00:56:21.225 --> 00:56:37.359
will build empty blocks if you want. If you if you wanna just do, you know, Coinbase only mining and mine empty blocks, you can do that with Floresta today, but obviously that's not the ideal. So there's little bit of work needs to be done on that to building block templates as well as submitting blocks to the network.
443
00:56:39.405 --> 00:56:43.005
I'm really looking forward to lib Bitcoin getting popular. I
444
00:56:44.125 --> 00:56:44.925
am
445
00:56:44.925 --> 00:56:49.645
really a fan of Eric Voskol's work. And if you're not following him on Twitter, you should be. Yeah.
446
00:56:49.885 --> 00:56:59.340
He's been working on this for a long time, but he goes hard. He's had some amazing stats lately about full initial block download that's in less than an hour. Yeah, it's insane. It's insane.
447
00:56:59.580 --> 00:57:03.100
I think he's running those on pretty heavy duty machines, but still that's
448
00:57:03.100 --> 00:57:07.740
a big win. Right. Yeah, no, I think he's got a really good approach and
449
00:57:08.964 --> 00:57:15.685
I'd like to I'd really like to see that project gain more in popularity. Me too. Me too. We need more node implementations,
450
00:57:15.925 --> 00:57:17.205
and it seems like,
451
00:57:17.685 --> 00:57:25.310
you know, he's he's been working on this for a long time and has the potential to be a real, you know, contender
452
00:57:25.310 --> 00:57:30.270
or a competitor, I guess, for Bitcoin Core. I do think there's something to be said for this UTXO
453
00:57:30.270 --> 00:57:33.710
idea too. Because I tried it I mean, not UTXO,
454
00:57:33.710
--> 00:57:49.445
but Jamison Lobb had that snapshot. Oh, assume UTXO. Yeah. Yeah. And I I tried that on an initial block download and it helped for sure. That's cool. And it was on a Raspberry Pi. Yeah. I think that can help a lot. I mean, it gets it up and running much faster.
455
00:57:50.965 --> 00:57:54.010
There's a lot of, like, network time saved.
456
00:57:54.250 --> 00:57:58.089
And I think more importantly, a lot of, like, essentially CPU
457
00:57:58.089 --> 00:58:04.890
time saved in in validating all those things, like, to get up and running. Yeah. It's intensive for sure. And,
458
00:58:05.130 --> 00:58:11.635
you any Bitcoin node can generate a snapshot at any block height you want. So there is the possibility of generating
459
00:58:12.115 --> 00:58:13.875
your own snapshots
460
00:58:13.875 --> 00:58:24.070
from a node that you run and trust to kinda get yourself more of a head start. We want more of them out there. That's for sure. Really need more people mining to their own node. So if you're solo mining,
461
00:58:25.589 --> 00:58:42.515
that just makes a ton of sense. I think the the problem is it doesn't gain you any more lottery tickets. Right? So a lot of people don't get fired up on doing it. Whereas they might, you know, have no problem spending a $100 for a bid x. They're like, oh, am I gonna buy this node and and gain, you know, no more chances of winning the lottery? So,
462
00:58:42.755 --> 00:58:53.089
yeah, if we can bring the cost of of mining nodes down, I think more people might go for it. What else do you guys wanna hit on this week? Gosh. I feel like my brain's been melted by this stuff going on, like, the last couple days.
463
00:58:53.329 --> 00:59:03.744
Know. Even before that. I mean, do you see the LibreBoard helping? That's kinda like the idea we designed the LibreBoard for, was that you could have your node and your miner in one box. Or is that still too
464
00:59:03.984 --> 00:59:10.865
much? Because I noticed you said you did say something about like NVMe is a non starter. A two terabyte NVMe is a non starter.
465
00:59:11.025 --> 00:59:32.945
Think that you know, if you have something like the LibreBoard that has an m two NVMe slot in it, then you could get a much more affordable, like, 32 gig or 64 gig drive and plug it in there. And then, yeah, if we have this this software setup, that would absolutely run on a Libre board. Mhmm. And then you could have a node right in your miner, which I think is really cool.
466
00:59:33.905 --> 00:59:37.665
Because two two terabyte SSDs, man, they're they're expensive and,
467
00:59:38.810 --> 00:59:45.930
it just takes a long time to get to the point where you can start mining to your own node. Because
468
00:59:45.930 --> 00:59:51.290
of the initial block download, you mean? Mhmm. Yeah. Yeah. Yeah. I see
469
00:59:51.690 --> 00:59:52.250
your vision.
470
00:59:53.445 --> 01:00:03.525
I think it would be useful. Because it's like you open the box, you plug
it in, and you're up in mining, right? And it's like generating your templates there in the device.
471
01:00:03.845 --> 01:00:06.470
Yeah, yeah, right. And we
472
01:00:07.190 --> 01:00:13.110
have to be very careful that we're not introducing a lot of trust in this, where you're sort of trusting someone to set up your
473
01:00:13.349 --> 01:00:15.750
node properly and seed it properly.
474
01:00:15.910 --> 01:00:16.470
But
475
01:00:16.710 --> 01:00:17.430
there
476
01:00:17.510 --> 01:00:22.735
are these pieces that exist in the Bitcoin ecosystem, I think, to do it right. And so you know, it's
477
01:00:22.895 --> 01:00:26.895
probably more like plug it in and, like, within the hour, you're ready to go. Which
478
01:00:27.295 --> 01:00:44.050
is typical of any new hardware. Right? Like computer hardware. Yeah. When I buy a new laptop, like, I have to set it up and it needs to update and all that. Yeah. I think that'd be really cool. That'd be really, really cool. I mean, we have Futurebit that has a node, you know, with the miner in there. Yeah.
479
01:00:44.290 --> 01:00:44.770
But I thought
480
01:00:45.490 --> 01:00:52.145
Those take a long time. Right? A long time to they're expensive and they take a long time to get synchronized.
481
01:00:52.145 --> 01:00:56.785
I did think Stephanopoulos had some good comments in that thread too.
482
01:00:57.025 --> 01:00:59.665
Just about like, well, about like design
483
01:00:59.745 --> 01:01:01.985
choices that they've made. But,
484
01:01:02.145 --> 01:01:03.710
you know, I think his
485
01:01:04.430 --> 01:01:08.990
standpoint is like the full archival node. Right? So
486
01:01:09.630 --> 01:01:20.895
Yeah. Mean, we we don't want we don't want less full archival nodes. Those are very important to the Bitcoin network. Right. But it's almost a different goal than and and it's kind of separate from miners
487
01:01:20.895 --> 01:01:30.015
building their own templates on their own node. Mhmm. Yeah. Because the miner is just interested in, like, generating valid blocks. Right? And for those purposes,
488
01:01:30.470 --> 01:01:41.670
does it really need the full archival node? I would argue it doesn't. No, it doesn't. It definitely does not need the full archival node. And, you know, people should People who want to and who are running
489
01:01:41.910 --> 01:01:49.655
full nodes, full archival nodes, that's great. More people should be doing that too. There's a lot of great solutions for that, right? There's
490
01:01:49.655 --> 01:01:51.815
Futurebit, Umbral, Start9,
491
01:01:52.055 --> 01:02:00.140
those are great, and you can do a lot of cool self hosting with Umbral and Start9, right? You can self host a lot of different apps on there, and that's fantastic,
492
01:02:00.380 --> 01:02:02.700
but there's a big barrier to entry because those
493
01:02:02.940 --> 01:02:17.405
boxes are pretty spendy. I guess the concern is that if the miner's building templates without the full archival history, there's If I understood what John was trying to say, or what If I understand what John was saying correctly,
494
01:02:17.565 --> 01:02:22.125
there's a risk that the miner could include a UTXO
495
01:02:22.125 --> 01:02:38.050
that has already been spent, and therefore, it would, like, jeopardize the validity of the block and there that miner's work would get wasted if it submitted an invalid block. Is that Yeah. That yeah. That was what I I how I understood his comment too. My
496
01:02:38.130 --> 01:02:45.765
point there is that that's not a problem that's unique to pruned nodes. Right? That that can still happen with a
497
01:02:46.244 --> 01:02:48.725
Even with a full archival node? Okay. Yeah.
498
01:02:49.125 --> 01:02:56.820
Right? If you have some sort of disk corruption that's still, like, the data on your disk gets corrupted, but yet somehow it's still a valid UTXO,
499
01:02:56.820 --> 01:03:04.020
that would be a problem if if that transaction got included in a block template and then you mined it, because then you would have mined an invalid block.
500
01:03:04.180 --> 01:03:09.315
Mhmm. I think it's kind of rare that you would have disc corruption that still is a valid UTXO,
501
01:03:09.315 --> 01:03:19.315
but I guess it could happen. But that can happen on a full archival node just as easily, I think, as a pruned node. I don't think there's any difference there. I see. Okay. I'd have to agree with you.
502
01:03:20.570 --> 01:03:28.570
Think he was I think John was was thinking that I was talking about we're gonna run these on SD cards, which I do not think is a good idea. Right? SD cards are notoriously
503
01:03:28.570 --> 01:03:30.890
flaky. They get corrupted all the time.
504
01:03:31.290 --> 01:03:46.645
That that is not that is not at all what I'm proposing. I have suffered a corrupted SD card many a time, and I wouldn't wanna rely on that. But we can rely on I think, I hope, we can we can get good quality NVMe SSDs
505
01:03:46.740 --> 01:03:50.980
that are just smaller and therefore cheaper. So they have the same reliability.
506
01:03:50.980 --> 01:04:09.385
They're just smaller. Right. Right. I think you could do a 32 gig NVMe, which might even be hard to find. Right? So maybe could do 64. Right? And those are I don't know. I think those are gonna be, like, in the the $20 range. I got $20. Alright. I got $20. Let's go. Let's do this. I got 20 on it. Tyler, you in? I'll find some in the couch cushion. Okay.
507
01:04:10.825 --> 01:04:22.920
Yeah. Let's do it. We love hacking. We love building. Yeah. The fluorescent and project has really got a lot of potential here. I think that that would be a really neat space to explore, but it's just
508
01:04:23.480 --> 01:04:27.880
not quite there yet for mining. Well, I gotta run here at the thirty minute mark, guys, but
509
01:04:28.200 --> 01:04:29.560
what should we hit on before we wrap?
510
01:04:30.275 --> 01:04:35.555
So I guess the moral of the story of
this show is that AI is accelerating
511
01:04:35.555 --> 01:04:38.675
quickly and breaking all things Bitcoin related.
512
01:04:39.155 --> 01:04:51.820
Especially your closed source projects that you thought no one would ever figure out how they work. Yes. Yeah. So you gotta stay vigilant, number one. And number two, it's, you know, the other side of that coin is that these tools are being used
513
01:04:51.980 --> 01:05:07.135
to also find vulnerabilities before someone else does and fix them. So that's good. August 31, mining Monday mining Monday at Bitcoin Park. Ryan's gonna come down and we're gonna do a a cool mining workshop, dig into
514
01:05:07.535 --> 01:05:13.940
setting up setting up Mugena, running it against a Bidax, how that whole stack works. He That's gonna be good.
515
01:05:14.260 --> 01:05:15.620
Are you guys recording it?
516
01:05:16.660 --> 01:05:20.180
Do you record those? I don't think we record those. Maybe we could record those.
517
01:05:20.820 --> 01:05:33.595
Yeah. Ask Andrew. Yeah. We have to talk to Andrew about that. That could be really neat. I know this workshop, Ryan did it in Nairobi as a part of Bitcoin plus plus there, and it was very well received. So I'm
518
01:05:33.755 --> 01:05:38.555
excited to see it myself. It's really cool. He's got an amazing setup where
519
01:05:40.315 --> 01:05:42.770
he set up test net network,
520
01:05:42.850 --> 01:05:44.210
you know, locally,
521
01:05:44.290 --> 01:05:47.170
and then everyone gets Mugena set up and HydroPool.
522
01:05:47.170 --> 01:05:52.930
You can mine to that from Mugena from your from your bid acts and like see how basically everything works.
523
01:05:53.565 --> 01:05:54.845
Yeah. And he's got it
524
01:05:55.165 --> 01:06:37.685
set up to do the fractional difficulty. So you can even, like, just do it with your laptop. You can mine to the your little local pool there at the workshop with your laptop. That's cool. Yeah. That's really cool. I mean, just kinda thinking about like, why what is fractional difficulty? Why do we even need that, you know, in the context of CPU mining? I think these are really neat things that are can be really complicated for people to look into. So excited to expose more folks to like, you know, what that means. Like, what what what's the difference between CPU mining and and ASIC mining? Right. So it should be pretty awesome. Come check it out. It's free. Bitcoin Park, be there. Heck yeah. And, get on the wait list for the Heatpunk Summit. We've already
525
01:06:38.240 --> 01:06:47.760
We've already had probably over one third of the seats filled out, so you don't want to miss that. We're going to focus this year's third summit heavily on workshops
526
01:06:47.760 --> 01:06:54.215
revolving around the two fifty six Foundation projects because I view them as critical building infrastructure,
527
01:06:54.215 --> 01:07:00.615
builder infrastructure for all the systems you may want to build for home mining, heating systems, solar, whatever.
528
01:07:01.335 --> 01:07:03.255
It'll be fun. Those
529
01:07:03.255 --> 01:07:08.579
lead engineers will be there in the flesh, it's a great opportunity to come chat with them. You should definitely
530
01:07:08.579 --> 01:07:09.460
be there.
531
01:07:10.019 --> 01:07:20.500
The Heatpunk Summit is the best show in Bitcoin. I don't know. It's your opportunity to go shred on a snowboard with Scott, so don't miss it. It's got it all. Good people, rad spot,
532
01:07:20.579 --> 01:07:25.135
super fun, awesome topics. Definitely not gonna wanna miss this one. Hepunks.org,
533
01:07:25.135 --> 01:07:37.165
you'll see the tickets on there. Awesome, guys. Well, good episode. Thank you to all of our hashers. I didn't go through the full list, but I see a lot of the same names we know and love you. Thanks for your support. Keep it up. Thank you. And we'll
534
01:07:37.485 --> 01:07:40.525
catch you next time. Right, Eco? You wanna close this out? Thanks everybody.
535
01:07:40.525 --> 01:07:41.325
Bye bye. Bye.