Latest / The Starting Block / Breaking down the Coldcard exploit & ramifications for self-custody

Transcript

Breaking down the Coldcard exploit & ramifications for self-custody · The Starting Block

0:06
Welcome back to The Starting Block.
0:08
Happy Monday to you wherever you are
0:10
around the world.
0:12
It's week number five, I think we're
0:13
number six. No, I we're week number
0:15
six, so we're a
0:17
good six weeks into the start of the
0:18
show. It's been an absolute pleasure
0:20
broadcasting every day.
0:22
This is The Block's new daily show,
0:24
and if you're joining us for the
0:25
first time, my name is Gareth
0:26
Jenkinson. I'm your host and head of
0:28
multimedia here at The Block.
0:30
If you wanna catch up on some of the
0:32
content we've been putting out over
0:33
the last few weeks, you can go and
0:34
check out The Block YouTube channel.
0:36
Give us a subscribe
0:38
if you're there and watching us
0:40
today. And if you watching us live
0:42
on X this afternoon or this
0:44
morning, if you in the United
0:45
States, please like and read
0:46
broadcast so we can get as many
0:47
people joining us today for what
0:49
promises to be a jam-packed
0:51
show. As you can see by the name of
0:52
the show, we are gonna be dedicating
0:54
a lot of time today to talk about
0:56
the cold card exploit.
0:58
And if don't know what that is, I
0:59
will give you a full rundown once
1:01
we have our guests on the show
1:03
a little bit later today.
1:05
If you are here for the first time,
1:07
the show is very simple.
1:08
We start off with a news bulletin
1:09
that's called the Block Briefing.
1:11
We'll run down six or
1:13
seven of the top stories
1:15
in the news over the last 24 hours.
1:17
Obviously it's a Monday, so we'll
1:18
look at some of the stuff from the
1:19
weekend and then we'll jump into
1:21
the hot seat today.
1:22
We won't have a market pulse because
1:24
we're gonna be deep diving into that
1:25
cold exploit and what it means for
1:27
Bitcoin self-custody going
1:29
forward. Really, really excited to
1:31
have Zach Herbert from Foundation
1:32
and Jameson Lopp.
1:34
From Casa to talk
1:36
about Bitcoin self-custody and
1:38
the cold card exploit.
1:40
But before we do any of that,
1:41
I'm gonna give you a rundown of the
1:43
top stories you need to know this
1:46
Monday.
1:47
Starting off with the big headline,
1:48
and this news dropped less
1:50
than 15 minutes ago,
1:53
Michael Saylor's strategy continues
1:55
to sell Bitcoin.
1:56
A fresh SEC filing shows the
1:58
firm offloaded 1,638
2:02
Bitcoin last week.
2:03
For around $105 million at
2:06
an average price near $64,000.
2:09
That trims its total stack to
2:11
roughly 842,000
2:13
Bitcoin, still around 4%
2:15
of all the Bitcoin that will ever
2:17
exist, but now carrying close
2:19
to an $11 billion paper
2:21
loss at current prices.
2:24
My producer Jordan was teasing
2:25
us all fair a little bit earlier
2:27
once he saw this news come out, but
2:29
just to put this in perspective,
2:31
this is about 0.
2:32
1% of their total Bitcoin
2:34
stack that they've sold and
2:36
strategy says the proceeds went
2:38
towards preferred preferred
2:40
stock Distributions and buying
2:42
back its STR shares a
2:44
clear shift away from sailors long
2:46
running the buy only Playbook
2:48
and if you go and look at X
2:49
and strategies announcement, there's
2:51
plenty of negative comments there
2:53
But I am seeing a few Bitcoin
2:55
is that I follow quite closely
2:56
giving a dose of perspective here
2:58
and kind of calling
3:00
out strategies,
3:02
building up of their credit standing
3:05
as they build up a much
3:07
bigger US dollar reserve
3:09
to give themselves more than
3:11
a two year runway to meet
3:13
dividend obligations.
3:14
Moving on to our next story,
3:17
a big on-chain move raised eyebrows
3:19
while it's tied to Trump Media, the
3:21
company behind True Social shifted
3:23
2,628 Bitcoin
3:25
with around $165 million
3:27
to crypto.com over the weekend.
3:30
Spread across two transactions,
3:32
but the company says don't read
3:34
too much into it.
3:35
A Trump media spokesperson told the
3:37
block the Bitcoin was transferred
3:39
and not sold, the same explanation
3:41
they gave back in May.
3:43
The tag wallet still hold roughly
3:45
4,261 Bitcoin,
3:47
about $260 million worth
3:50
closely matching the collateral
3:51
backing its convertible notes as
3:54
of the first quarter filing.
3:56
Over in Washington, a regulatory
3:58
turf war is heating up.
4:00
The SEC has hit pause on
4:02
NASDAQ's PHLX's
4:04
plans to list cash settled Bitcoin
4:06
index options after CME
4:08
group petitioned the commissioners
4:10
to overturn a staff approval.
4:13
CME's argument here, these
4:15
contracts are commodity option swaps
4:17
that belong squarely under the
4:19
CFTC's oversight,
4:21
not the SEC.
4:22
The commission granted CMEs petition
4:24
for review and kept the
4:26
approval state.
4:27
Written statements are due on the
4:29
24th of August, later this month.
4:31
It's a jurisdictional
4:33
tug of war that could help shape
4:35
who regulates the next wave of
4:37
crypto derivative products in
4:39
the United States.
4:41
Then a sobering story out of Hong
4:42
Kong. Police say a local insurance
4:44
agent lost more than $3.3 million
4:46
after being lured into
4:48
a fake cryptocurrency investment
4:50
scheme through a romance scam.
4:52
It wasn't isolated.
4:53
It was the single largest case among
4:55
25 romance scams reported
4:57
in just one week, from July 24th
4:59
to the 30th, adding up to nearly
5:01
$9 million in total losses.
5:04
It's a stark reminder as crypto goes
5:05
mainstream, so do the scams
5:07
targeting everyday people.
5:10
Then prediction markets continue on
5:12
a tear.
5:13
Combined trading volume across
5:14
Polymarket, Polymarket U.S.
5:16
And Kalshi hit a record.
5:18
$50 billion in volume
5:20
was up 7.7%.
5:23
From June, according to the block's
5:25
data dashboard.
5:26
Continued momentum around the World
5:27
Cup helped drive that, but there's
5:29
a twist under the hood.
5:30
Polymarket US's volume surged
5:32
54 percent, while the main
5:34
polymarket platform's volume
5:36
actually fell 26 percent
5:38
compared to June, a sign that
5:40
regulated US-facing prediction
5:41
markets are pulling in more of
5:43
the action as the sector keeps
5:45
gaining steam.
5:46
And finally, South Korea,
5:49
but the country's second largest
5:50
crypto exchange.
5:52
Has officially confirmed it's
5:53
targeting a public listing in
5:55
2028.
5:57
In a notice posted on its
5:58
website, the exchange laid out a
6:00
multi-year preparation plan,
6:02
including a request for a
6:03
preliminary listing review in 2027
6:06
before pushing for the IPO the
6:08
following year.
6:09
It's a notable step for a major
6:11
Asian exchange moving towards public
6:13
markets, and another sign that
6:14
South Korea's crypto industry is
6:16
maturing fast.
6:18
We'll be watching how that roadmap
6:19
unfolds.
6:20
Over the next couple of
6:22
years. And that's it for the block
6:24
briefing.
6:25
I think we are gonna take a very
6:26
short break before we bring Zach
6:28
Herbert on and Jameson Locke a
6:30
little bit later on the show.
6:32
Stay with us as we talk about the
6:34
cold card exploit and everything you
6:35
need to know about what it means
6:37
for Bitcoin self custody going
6:39
forward. Don't go away.
6:40
We'll be right back.
9:35
Welcome back to the starting block.
9:37
We're jumping into the hot seat and
9:38
we're gonna dedicate two interviews
9:39
today to the hotseat segment.
9:42
It's a great pleasure to have Zach
9:43
Herbert, who's the co-founder and
9:45
CEO of Foundation, joining
9:47
us today to talk about the Cold Card
9:49
exploit.
9:50
And if you're joining us for the
9:51
first time and you don't know about
9:52
the cold card exploit, Cold Card is
9:54
of course a Bitcoin-only
9:57
hardware manufacturer that
9:58
was hit last week with an
10:00
exploit that...
10:02
As far as we would know right now,
10:04
could impact all of their devices.
10:06
And we've seen more
10:08
than a thousand Bitcoin drained
10:10
from certain models
10:12
of cold card wallets
10:14
in the past 36
10:16
hours or more.
10:18
It's been a concerted effort across
10:20
the Bitcoin ecosystem to,
10:21
first of all, make sure that
10:23
everyone knows that this is
10:24
happening, especially for cold card
10:26
users that have not been online the
10:28
last few days.
10:29
They might not even know, and I've
10:30
seen some harrowing stories as well
10:32
that. Some people went away on
10:33
holiday for a couple of days, saw
10:35
the news, they weren't able to get
10:37
to their devices. And when they got
10:38
home to their device or got someone
10:39
to actually open
10:42
up the devices and check their
10:44
balances, Bitcoin had
10:46
been drained. So a really dire
10:48
situation.
10:49
And we're gonna be talking about
10:50
that at length today with
10:52
Zach and Jameson Lopp, who's joining
10:54
us a little bit later on the show.
10:55
Zach, first things first, thank you
10:56
so much for taking the time to chat
10:58
to us today.
10:59
It's the first time we've spoken in
11:01
person, virtually.
11:02
So it's good to meet you.
11:04
I can imagine it's been a rough few
11:05
days, how are you doing?
11:06
I'm doing okay a
11:08
little tired and some of our
11:10
team is a little tire as well
11:12
But I mean compared to those folks
11:14
who lost money right and it's just
11:16
yeah and comparable and it has been
11:18
a really difficult day for
11:20
Bitcoin
11:22
Bitcoiners and Most hardcore
11:24
Bitcoiners right that I've been told
11:26
self-custody and that have helped
11:27
other self- custody So it's
11:29
definitely everyone's reeling from
11:31
the events of the last few days.
11:33
I don't think it's gonna
11:35
uh, stop either anytime soon,
11:37
we're going to be feeling the
11:38
effects of this potentially for
11:40
months, maybe even years.
11:42
And so, um, yeah, thank you for
11:44
having me on to, uh, talk about it.
11:47
Yeah, sure. And I think just for
11:49
our viewers out there that actually
11:50
might not be too familiar
11:52
with the story, maybe you can give
11:54
us your kind of TLDR
11:56
of what happened here,
11:59
a very high level overview before
12:01
perhaps we get into some of the
12:02
specifics.
12:03
For sure. And I'll try to keep it
12:05
brief so that you can ask me
12:06
questions, you know, because I don't
12:07
know how to in the weeds, right?
12:09
We want to get here.
12:10
But essentially a cold card,
12:12
right, is a, is a popular Bitcoin
12:14
only hardware wallet made by a
12:16
company called coin kite.
12:17
That's MDK or Rodolfo Novak's
12:19
company. You know, he's a pretty
12:21
huge name in the, in the
12:23
Bitcoin space.
12:24
And on Thursday afternoon reports
12:26
start to come out that funds were
12:28
being stolen.
12:29
It seemed like from folks
12:31
with cold card devices.
12:33
And so you can imagine the
12:35
entire industry scrambled to
12:37
figure out what was happening, what
12:38
was going on.
12:39
And the root cause that was
12:41
first disclosed, I think probably
12:43
blocked it, the best job, getting
12:45
on top of this and disclosing it
12:47
fast and writing up a report.
12:48
So many of us were kind of figuring
12:50
out in real time and
12:52
obviously checking that for all the
12:53
other hardware wallet companies,
12:55
checking that our customers were
12:56
also not affected.
12:57
But there was a bug introduced many
12:59
years ago into the cold card code
13:01
base in early 2021
13:04
that effectively reduced
13:07
by a significant factor
13:09
the amount of randomness
13:11
or we call entropy that goes into
13:13
the seed generation process.
13:15
And so what that meant is that
13:18
cold card seeds starting from early
13:20
2021 and even basically
13:22
through today or I should say maybe
13:24
through over the weekend, because
13:25
they released an update a
13:27
day or two ago,
13:29
which you could install and would
13:31
fix that ongoing on new devices.
13:33
But from 2021, basically,
13:36
until now,
13:38
all seeds generated on cold card
13:40
devices are insecure.
13:42
Now the level of
13:46
the security for the seed varies by
13:48
model, right?
13:49
So in this case, the mark three
13:51
model, which was the model that was
13:54
introduced around that time
13:56
a little bit earlier is the one
13:57
that's most affected.
13:58
And the thought is that
14:00
most of the funds have already
14:02
been taken from
14:04
those wallets and devices.
14:06
When we say taken, we don't
14:08
actually mean that you need any kind
14:09
of physical access to the hardware
14:11
wallet, right. And so that's what
14:12
makes this really difficult,
14:14
really acute.
14:15
It means that an attacker can
14:17
sit down at the computer, they can
14:19
write a script that can run on,
14:21
very often it could run on your
14:22
laptop, it could run on Nvidia GPUs,
14:25
they could be renting, you know,
14:26
compute somewhere, and they're just
14:28
cracking seeds.
14:29
And so they're going through seed by
14:31
seed, they're generating potential
14:32
seeds, and then they have scripts to
14:34
check, is there is
14:36
there money at this Bitcoin address,
14:38
right, that's that they're able to
14:40
derive that based on the seed.
14:41
And so.
14:42
Our best guess now is that there's
14:44
multiple attackers doing this right
14:46
this copycat once you hear about it
14:47
there's other bad actors that want
14:48
to jump on And um,
14:51
I know the counts the counts of our
14:52
said over a thousand bitcoin
14:54
Uh stolen I believe it's
14:56
now something like over 1700 from
14:58
what I can tell Um, it's very
15:00
hard to know exactly though.
15:02
So we're only going to figure it out
15:04
in the coming weeks and months
15:06
And even though it's not a huge
15:08
number of Bitcoin compared to some
15:09
of these mega exchange collapses
15:15
So I'll stop there, yeah.
15:17
Yeah, sorry, someone's ringing my
15:19
doorbell. I'm sure you can hear that
15:20
in the background.
15:22
Someone will get that downstairs.
15:24
Yeah, so I think let's take it from
15:26
there. Obviously,
15:28
once news of this had come out,
15:30
there was a concerted effort across
15:32
the industry to try and figure out
15:33
what this happened.
15:34
One of the questions I had
15:36
posed to me online, just in
15:38
discourse with other people was,
15:39
how much do you think AI tools
15:41
played in the eventual.
15:44
Figuring out of this
15:46
bug.
15:47
Do you think these frontier models
15:49
that we've seen come out in recent
15:50
months might have been the
15:53
key that has allowed some attackers
15:55
to figure out this
15:57
exploitable attack vector and go
15:59
through it and then obviously now go
16:00
through the process of generating
16:03
seeds and then running them through
16:04
all these addresses to see if they
16:05
can steal this Bitcoin basically?
16:09
My quick answer is yes.
16:10
Absolutely.
16:12
Some folks think that
16:14
maybe the reason that this happened
16:16
on Thursday was because
16:19
the latest Kimi model had just come
16:20
out some number of days before.
16:22
I think it was just a few days
16:23
before and these
16:26
models that are open weight models
16:28
mean that
16:30
there's no cyber security safeguards
16:32
and so anyone is
16:34
allowed to use them and they
16:36
will do a pretty good job
16:38
finding bugs and code.
16:41
We as foundation have access
16:43
to like GPT's cyber
16:45
program, right?
16:46
And so we're also able to run the
16:48
frontier models on our own code
16:50
or on other code, but.
16:53
As you know, new models are coming
16:55
out every month or every two months.
16:58
And in our experience,
17:00
up until recently, the Frontier
17:02
models have not been
17:04
the best at embedded engineering
17:06
work, meaning like firmware for
17:07
physical devices.
17:09
We've been running Frontier Models
17:10
that are our own code for months
17:12
now, and there's been so many
17:14
false positives.
17:16
If there's, a misunderstanding
17:18
of how our code bases work,
17:20
because typically you have
17:22
this firmware and it's calling on
17:24
other libraries.
17:24
And then sometimes the bugs exist in
17:26
the interaction between the hardware
17:27
and the software on the phone
17:29
or computer.
17:31
And AI still has a really hard time
17:32
holding all of that in context.
17:35
So I think we may have crossed the
17:36
threshold in the last few weeks
17:38
with some of these new models,
17:40
because I pointed GPT
17:42
5.6 sole, which is
17:44
the latest model from OpenAI.
17:46
At the cold card code base
17:48
that afternoon on Thursday.
17:50
And it was able to find the bug
17:52
without any issue.
17:54
Well, there was some
17:56
inflection point that took place
17:59
and there's a few lessons learned
18:00
but I do not think the takeaway
18:02
is that open source is bad.
18:05
My takeaway is actually that free
18:07
and open source software is
18:08
necessary because
18:11
If this had been a different
18:13
project in the space, like let's say
18:14
there was a common library
18:17
that means like some block of code
18:19
that multiple companies are
18:21
using and building off of.
18:22
And there's a lot of examples in
18:23
Bitcoin, like Block has Bitcoin
18:25
Development Kit, EDK.
18:26
There's like Rust Bitcoin libraries.
18:28
There's all these Bitcoin libraries
18:29
that are used by multiple companies.
18:31
They're free and open source
18:32
software.
18:33
I think that the likelihood of
18:35
someone catching that, if it's in a
18:37
commonly used library, is much
18:38
higher because of all these real
18:39
companies that are using the code
18:41
and sharing on it and building on
18:42
it.
18:43
In this case, this was a cold card
18:45
library that was
18:47
a set of code that was
18:49
publicly available.
18:50
So we say source available,
18:51
but not fully open source.
18:53
And my understanding, though I can't
18:55
say with certainty, but my
18:56
understanding is they were the only
18:57
company that was
18:59
using this specific library.
19:03
Now, obviously, this was just a bug
19:05
that was introduced back then.
19:06
There was no, it was not
19:07
intentional, right?
19:08
It's, there's a lot of
19:10
bugs that exist in any code base.
19:12
Every time we turn a frontier model
19:14
out of our code, it finds bugs.
19:16
The question is, like, what level of
19:17
criticality, right, is that bug?
19:19
Is it like a critical bug?
19:20
Is it a high, medium, low, and so
19:22
on? I do think AI is
19:24
part of the reason why this
19:26
happened. I also think it's part of
19:27
a solution.
19:28
And so it's just like any new
19:29
technology, right.
19:30
You have the good effects, you have
19:32
the bad effects.
19:33
If we try to create a thought
19:35
experiment about what happened if
19:36
the cold card code base was closed
19:38
source and this exact same thing
19:40
happened, which is totally possible,
19:42
by the way, his attackers, like they
19:44
reverse engineer binaries, they hook
19:46
stuff up in the lab.
19:47
You could have like a bad actor
19:49
within the company or at a different
19:50
company that has or even at
19:52
Microsoft who has access to all the
19:53
GitHub private repos.
19:54
So like, let's just say is the
19:56
thought experiment. If this was
19:57
closed sourced, we would not
19:59
have seen this immediate response
20:00
from the Bitcoin community.
20:02
Figuring out exactly what happened
20:03
and helping thousands of people get
20:05
to safety, we probably would have
20:07
seen like a fog of war for days or
20:08
weeks because no one would know what
20:10
happened or why it happened or what
20:11
the bug was.
20:12
So I'm still a huge open source
20:14
advocate, but this was definitely,
20:16
I think,
20:18
definitely had something to do with
20:19
the latest AI models.
20:23
Obviously, Foundation does
20:25
hardware wallets as well,
20:27
and we've seen lots of reactions
20:28
from different self-custodial
20:32
hardware wallet service providers
20:33
coming out into port,
20:35
but I can imagine internally, there
20:37
was a huge fear initially that,
20:39
like, holy shit, what happens if
20:41
this is something that is affecting
20:42
all devices?
20:44
How quick was it, or how hard was it
20:45
to rule it out and figure out that
20:47
this only the
20:49
specific bug that that was
20:51
impacting. Mainly the MK3s.
20:53
And how important is
20:55
it for people in the industry and
20:58
wider crypto natives
21:00
to understand that this is not
21:02
impacting all hardware wallets?
21:03
Because I've been seeing a lot of
21:05
people saying online that
21:06
self-custody is dead.
21:07
And I fundamentally don't believe
21:09
that, and I wanna push back against
21:11
that narrative myself.
21:12
Yeah, I'll answer that.
21:13
But one thing you just said, I want
21:14
to make really clear,
21:16
um, MK threes have been,
21:18
are, are most affected, but
21:20
all MK foreign leaders are still at
21:22
risk. They still have some better
21:23
entropy. So just like, if you have a
21:25
cold card, if it's MK three
21:27
or newer, I would, I will get your
21:29
funds out of there as fast as
21:30
possible. So that's, that's the
21:31
first thing I want to address, um
21:32
with self custody.
21:35
It's really, it's really dangerous
21:36
to say self custody is dead because
21:38
of this. I mean, I, I completely
21:40
disagree.
21:42
However, it's scary.
21:46
Obviously, every hardware
21:47
wallet maker, probably every wallet
21:49
maker in the last few days across
21:50
the entire industry has been, I
21:52
hope, turning AI loose
21:54
on their code base and reviewing
21:56
some of the most critical parts of
21:57
the code.
21:58
There's been some amazing Bitcoiners
21:59
that have been helping with that.
22:02
Like Rob Hamilton at Anchor Watch
22:03
has been using, I believe
22:05
Kimmy to submit bug
22:07
reports to virtually every hardware
22:09
wallets, software wallet.
22:10
We've been talking with him.
22:12
He submitted stuff to us,
22:14
there's one that scared us because
22:17
Kimi thought that we had a potential
22:19
bug. It was nothing like as
22:20
bad as the cold card bug,
22:22
but we had
22:24
to really dive into it with human
22:26
review and realize that it was
22:27
wrong. And so there's another aspect
22:29
here where there's
22:31
going to be a lot of panic and
22:32
uncertainty in the short term
22:34
because customers are thinking, is
22:35
my hardware wallet safe?
22:37
They're running AI models now
22:38
against their hardware wallets that
22:40
they're using, and sometimes AI
22:41
models are saying, oh, there's some
22:42
critical bug, but the bug doesn't
22:44
actually exist because the AI models
22:46
are still not at that level
22:48
and probably will never be at that
22:50
levels of perfection, right?
22:51
Nor are humans.
22:53
And so there's a lot of complexity
22:54
happening. So yeah, I mean,
22:55
obviously, the first thing I did
22:57
Thursday evening was
22:59
run the
23:01
Frontier models against our code
23:03
base, and our CTO did the same.
23:05
We breathed, you know, a sigh of
23:06
relief, right? That there was no
23:07
entropy bug. And I'm sure that exact
23:09
same thing happened at all the other
23:11
wallets.
23:11
Um, yeah.
23:15
Look, I think the
23:17
next kind of question that comes
23:19
into my mind now is the
23:22
notion of trust in Bitcoin, because
23:24
the old saying goes
23:26
very hard, don't trust, verify, and
23:28
a lot of people trusted here.
23:29
And I think that's the great tragedy
23:30
of this, is a lot people trusted
23:33
CoinKite, the
23:35
company, and NVK, and
23:37
I'm not here to bash them, I'm just
23:39
not that person. And no one could
23:40
have known, and it was very hard to
23:42
find this bike.
23:43
And no one found it for a very long
23:45
period of time, right?
23:46
But there is that element of
23:48
trust that most people don't even
23:50
realize is still there when
23:52
they're using a hardware wallet like
23:54
this. And it does make
23:55
me kind of wonder like how does the
23:57
industry address that?
23:58
Like this element of trusts because
24:00
it was supposed to be trustless,
24:02
right.
24:03
Like here's a hardware wallets,
24:05
now you can be the master
24:06
of your own coins.
24:08
And in this instance, that failed.
24:11
How do you feel about that?
24:13
Yeah, I think it's funny.
24:14
I'm running a blog post right now
24:16
that has to do with don't trust
24:17
verify to talk about all this
24:19
because I think what
24:21
happened is that um
24:23
The whole Bitcoin media and
24:25
podcast. I don't want to say the
24:26
whole but a majority of the Bitcoin
24:27
media podcasting community
24:30
has
24:32
Trusted
24:34
You know, I don't want to stay
24:36
blindly trusted this company I think
24:37
I think you know, they have good
24:39
intentions I do not think that
24:41
this bug was intentional.
24:43
MBK has been a Bitcoiner for a long
24:45
time, but
24:47
this don't trust verify is really
24:49
important.
24:50
And I feel like the Bitcoin
24:51
community should be leaning more on,
24:53
if you're going to trust a wide
24:55
array of technical, you
24:57
know, folks in the Bitcoin
24:58
Community, it shouldn't be
25:00
based around who
25:03
sponsors what podcast,
25:05
who's friends with whom, it
25:07
should be much more about
25:09
trusting technical leaders within
25:11
the community.
25:12
And I do think that the
25:14
media has to be careful, both within
25:16
Bitcoin and within all of crypto,
25:18
with recommending specific hardware
25:21
wallets or software wallets, which
25:22
is tough, right, because you want to
25:23
be able to take sponsorship money.
25:24
I mean, we've sponsored stuff before
25:26
and sometimes still do, but
25:28
it creates a really difficult
25:30
cascading effect.
25:31
And I know so many amazing.
25:35
Know, Bitcoiners who have large
25:37
audiences who are just torn up about
25:39
this completely, right, because
25:41
you recommend to so
25:43
many people a
25:45
specific device, you tell them to
25:47
self-custody, not
25:49
your keys, not your coins, don't
25:50
trust verify, you help so many
25:52
people, thousands of people, tons of
25:54
thousands of people on board to self
25:55
custody, and then this happens.
25:56
And so I do think that
25:58
there needs to be some introspection
26:00
within the Bitcoin community.
26:02
I don't know if that's a good enough
26:03
answer.
26:04
Like I wish I had a better answer to
26:06
your question.
26:07
I think the principles are still
26:09
there.
26:09
I just think that we as a community
26:11
need to really figure out what that
26:13
means. One thing I would say is that
26:14
like, what
26:16
are we looking for in terms of
26:17
verification?
26:19
Things like security audits are
26:21
really important, engaging with
26:23
the security research community.
26:25
Like we'll get bug reports all the
26:26
time. It's really important when you
26:28
engage with security researchers
26:30
that you do so with a great amount
26:31
of respect.
26:32
It can sometimes be an antagonistic
26:34
relationship, right?
26:35
Someone reaches out to us, they say
26:37
they found a critical bug.
26:38
Sometimes your first reaction as a
26:39
wallet company is to get upset about
26:41
that. And I think it needs to be the
26:43
opposite.
26:44
And then also just free and open
26:45
source software, collaborating with
26:47
each other, building on shared
26:49
libraries.
26:51
All those things I think are really
26:52
helpful. And when we say don't
26:54
trust, verify.
26:55
I think that's the kind of
26:56
verification that the space should
26:57
be moving towards, you know, less
26:59
about marketing
27:03
a specific company, more
27:05
about giving
27:07
the technical leaders and the
27:08
security experts a voice as much as
27:10
possible, especially in the coming
27:12
weeks.
27:13
Yeah Zack we've got
27:15
a lot backstage.
27:16
So I'll ask my final question to
27:18
you I mean just just in terms of
27:20
the the hardware wallet space, you
27:22
know We've seen kind of newcomers
27:24
come in in recent years I think you
27:26
know block streams Jade bit
27:28
key from block a few
27:30
newcomers that have definitely added
27:32
More hardware wallets to the
27:34
industry. I'm kind of curious.
27:35
Do you think that this situation
27:37
does call for for more?
27:40
Options in the industry and will
27:42
that kind of help drive things
27:44
forward? And from a usability
27:46
perspective, it does kind of beg the
27:48
question. It's like, is it still too
27:50
complicated? Does it need to be
27:51
easier than it is?
27:53
Because a lot of cold card users
27:55
were told like,
27:57
this will be safe.
27:58
You're fine.
28:00
But if you really want to go the
28:01
extra step, you need to do the
28:05
the coin toss dice
28:07
roll.
28:08
Option to go and generate a seed,
28:10
that's hugely complicated for the
28:12
layman who's looking to self-custody
28:13
their Bitcoin.
28:15
So A, do we need more
28:17
variety of hardware wallet
28:18
manufacturers in the industry?
28:20
And do we have a user experience
28:22
problem?
28:25
I think that more is not
28:27
necessarily the answer, though I'm
28:28
conflicted, obviously.
28:31
I think more free and open source
28:34
software and hardware options
28:36
is part of the
28:38
answer.
28:39
Some of those newcomers you
28:40
mentioned are open source,
28:43
BitKey is open source.
28:45
It was not fully open source at
28:46
first.
28:47
Many fiercely criticized that,
28:49
myself included, and they have now
28:51
open sourced everything.
28:52
Jade is open source, there's
28:54
many that are open source.
28:55
And so I think the open
28:57
source aspect is very important
28:59
here. And I think only
29:01
newcomers to the hardware wallet
29:03
world should be taken seriously if
29:04
their stuff is open source, is a
29:06
big takeaway that I want to
29:08
push as part of this moment
29:10
in conversation.
29:12
With regard to the UX, absolutely a
29:14
problem.
29:15
I mean, 100%.
29:16
And I've actually been trying to
29:17
think about how to react to that,
29:18
because I've been in a lot of these
29:19
Twitter spaces.
29:21
Trying to help some folks over the
29:22
last few days.
29:23
Um, there's been some incredible
29:26
Twitter spaces, uh,
29:29
one, one guy who goes by wicked
29:30
has been hosting one basically for
29:32
the last like four days straight is
29:34
extraordinary that he's been able to
29:35
do that, but everyone's coming
29:37
up on stage asking about dice rolls
29:39
and coin flips, and I don't think
29:40
that should take away here.
29:41
I think it makes sense.
29:42
If you're a cold card customer and
29:44
you, and you can't switch to another
29:46
device and you're like, how do I
29:47
generate a seed securely now?
29:50
What could I have done differently?
29:53
But for every wallet that we know
29:54
of, every mainstream hardware wallet
29:56
besides cold card, like the
29:57
C generation code has been
29:59
okay, right?
30:00
It's been okay.
30:02
And I'm sure Jameson who's gonna
30:04
be up next will talk about
30:05
multi-vendor, multi-SIG, which I
30:07
think is a solution.
30:08
There's some good solutions
30:10
that offer a good user experience as
30:12
part of that.
30:13
From my perspective though,
30:16
I think it really does come down to
30:18
making sure that you're
30:20
code is bulletproof for the C
30:22
generation as a hardware wallet
30:23
vendor, which is not that much good
30:25
to, you know, inspect and
30:27
trying to create a really great user
30:28
experience, both for onboarding and
30:30
backups.
30:31
You know, we use like NFC key
30:33
cards that we call them, where you
30:34
can tap to the device as an option
30:35
right now.
30:37
BitKey does some really interesting
30:38
stuff.
30:39
We're going to be, I think, further
30:41
improving our onboarding and
30:43
backup process because you want to
30:45
make it as easy as possible for
30:46
folks. And so...
30:48
I don't think the takeaway should
30:49
be do not trust all
30:52
random number generation after this.
30:54
I think that's the takeaway I've
30:55
been seeing on, you
30:57
know, these spaces and everything
30:58
because people are scared.
31:00
But I do think that this was more
31:02
about a horrible bug
31:04
with a specific vendor.
31:06
But if you are advanced enough and
31:08
you do want to generate your own
31:09
seed words, You can pick seabirds
31:11
out of a hat, for example.
31:13
Uh, you can enter them into the
31:14
device, a device like passport can
31:16
auto fill the last word for you,
31:18
like the 11th, the 12th or the 24th
31:19
word, where there's a bunch of other
31:21
ways that folks have been covering,
31:22
like you should do that.
31:23
Right. Cause it's empowering to be
31:24
able to do that yourself and not
31:26
rely on anyone else.
31:27
So I think that would be my major
31:29
takeaways, but if I get just the one
31:30
big takeaway, right.
31:32
Free and open source software I
31:33
think is, is our, uh,
31:35
longterm solution.
31:38
Zach, thanks so much for taking the
31:39
time to chat to us today.
31:41
And for anyone watching out there,
31:42
you can follow Zach on X and go and
31:44
check out Foundation as well if
31:45
you're looking for alternative
31:46
options. It's at Z Herbert on
31:48
X and at Foundation
31:51
HQ I believe is the X handle.
31:53
Zach, thank you very much.
31:54
We're gonna take a very short break
31:55
just to check Lop's
31:57
audio and visuals.
31:59
We'll be right back. Don't go away.
34:28
Welcome back to the starting block.
34:29
We are continuing our hot seat
34:31
conversation. We've got Jameson
34:32
Lopp, who's of course a CTO at
34:34
Casa, very well-known name in the
34:36
Bitcoin industry.
34:37
Jameson, it's great to have you on
34:39
the show.
34:40
First things first, I'd love to hear
34:41
your kind of recollection of events
34:43
last week, when you first saw
34:45
the cold card exploit
34:47
news and what your immediate
34:49
thoughts were and what you did in
34:50
the hours after that.
34:53
Well, actually and
34:55
oddly enough, I believe I
34:57
was one of the first people to
35:00
have a hint of it.
35:02
And that's because I had someone
35:04
email me privately through my
35:07
website in
35:10
the wee early hours of
35:12
the morning, because they were
35:13
actually over in Australia in
35:15
that area.
35:16
And they detailed that
35:18
they were a cold card user and they
35:20
had seen a partial loss
35:22
of funds and they described their
35:24
entire setup and their best
35:25
practices to me.
35:27
I was pretty baffled for several
35:29
reasons.
35:30
First of all, because it was only a
35:31
partial loss of fund and usually you
35:33
see full loss of funds when a seed
35:35
is compromised.
35:37
And at the moment,
35:39
I had just gotten back to them
35:41
and said, hey, we
35:43
need more information.
35:46
Need to better understand if
35:48
you have done something that could
35:50
have caused this.
35:52
And then of course, as the day went
35:54
on, we started seeing more and more
35:55
reports and the
35:58
only real commonality was that
36:00
these were all cold card users.
36:02
And so that's when the
36:04
security community really
36:07
woke up and said, okay, well,
36:09
we still don't know what it is,
36:11
but... Signs are pointing to
36:13
cold card and that's when a lot of
36:14
people started going and inspecting
36:17
the firmware more closely and
36:19
I think it was by the end of the day
36:20
that it was determined that
36:22
there was essentially
36:24
an entropy problem
36:27
you know with the random number
36:28
generator.
36:30
Once that was discovered then
36:33
really the goal was to
36:35
understand what the severity of it
36:37
was and you know
36:39
how Basically,
36:41
how long... People had before
36:43
they really could expect to get
36:45
popped, depending upon a number of
36:47
different variables of how they had
36:49
generated their seed phrase.
36:52
I mean, it's pretty crazy
36:54
how much Bitcoin has been stolen
36:56
since, but it's
36:58
kind of still ongoing, right?
36:59
And I saw a few anecdotes over the
37:00
weekend of people being
37:02
on holiday away from the cold card
37:04
devices, knowing that this has
37:05
played out and they can't do
37:06
anything, they can get to the
37:08
devices and move their funds
37:10
off.
37:12
Yeah, I guess my first kind
37:13
of question to you
37:16
is when you kind of saw that
37:18
the issue was with the random number
37:20
generator and the entropy involved
37:21
in mainly the MK3 devices,
37:24
but perhaps more of the cold card
37:25
devices.
37:28
You said to me, Orfea, that this is
37:29
something that one of the
37:31
sort of value propositions of
37:32
something like CASA is
37:35
ruling out this kind of
37:37
single point of failure.
37:41
I mean, I don't want to say, do you
37:42
think that this was kind of long
37:43
time coming that we would see an
37:45
issue like this with a hardware
37:47
wallet manufacturer?
37:48
But- Is it quite in your
37:50
mind rare that something like this
37:51
happened?
37:52
And do you think it sets a stake in
37:54
the ground where hopefully we don't
37:56
have this issue ever again with
37:58
any hardware wallet manufacturers?
38:02
It has happened, I want to say,
38:04
a dozen times before, and I
38:06
fully expect it to happen again
38:08
now.
38:12
The good part
38:14
about having critical
38:16
security failures like
38:18
this is that
38:21
the result is that
38:23
there is a sort of shared
38:24
learning.
38:26
And that it will result in
38:29
increased protections and
38:31
better practices taken,
38:34
but not by everyone.
38:36
And so, you know, the.
38:38
Various classes of security
38:41
failures tend to happen over
38:43
and over and over again because
38:46
even though our collective
38:48
understanding and consciousness
38:50
absorbs them and learns from
38:52
them, there will always be
38:54
new entrants in the future who come
38:56
in and don't learn the
38:58
hard lessons of history and end up
39:00
repeating them.
39:02
So, there's trade-offs
39:04
to all of this, good and bad.
39:06
And
39:08
I'm somewhat hopeful
39:10
that at least the
39:13
ecosystem-wide impact will
39:15
be less severe than what we've
39:17
seen from centralized exchange
39:20
providers that have gone bust
39:22
for various reasons.
39:24
I think
39:25
one nice
39:27
thing about this is that it is
39:29
somewhat distributed.
39:32
I don't know exactly the
39:35
size of Cold Cards
39:37
business and how many people bought
39:38
from them, but I think that
39:40
they are a lot smaller than
39:42
like Trezor, Ledger, whatever.
39:44
So on orders of magnitude impact,
39:47
it would have been far worse.
39:48
I think if this had occurred in
39:50
Trezr or Ledger.
39:51
But, and this is something that
39:52
we'll probably end up talking about
39:54
a lot in a variety of
39:56
ways, but
39:58
the size of the
40:00
organization and the project is
40:02
also somewhat
40:05
correlated to the security
40:07
of the project.
40:08
And so I think there's a strong
40:10
argument to be made that one
40:12
of the reasons why this
40:14
happened and went undetected for
40:16
so long was because
40:18
CoinKite is one of the smaller
40:20
hardware vendors.
40:21
And so, you know, didn't have the
40:23
same level of.
40:25
Eyeballs and other resources
40:27
focused on
40:30
auditing the code.
40:31
Now, we
40:33
know that people have looked
40:35
at the code over the years and
40:38
this particular bug was
40:40
fairly obscure.
40:42
It was not obvious I think just from
40:44
looking at the line
40:46
of code that was calling the random
40:47
number generator. And that's why it
40:48
was missed for so long.
40:51
And yes, I think AI
40:54
advancements have led
40:56
to this being found when
40:58
it was found.
40:59
Could it have been found earlier?
41:01
Sure.
41:02
You could probably play
41:04
coulda, shoulda, woulda
41:06
all day long,
41:08
but ultimately there is the
41:10
fact that advancements
41:12
in large language models are
41:14
drastically changing the security
41:16
landscape of everything.
41:19
This is something that
41:21
has been concerning.
41:23
To those of us in the security
41:25
community, really for the past
41:26
year, because we've been
41:28
seeing obscure
41:31
hidden vulnerabilities
41:34
in a variety of different popular
41:36
code bases, getting found,
41:38
some of which had been there for
41:39
five, 10,
41:41
even 15 or 20 years,
41:44
simply because...
41:46
You know, being open source is
41:48
not panacea.
41:49
It is not a guarantee that all
41:51
bugs will be found.
41:53
It's a combination
41:55
of being open-source
41:57
plus the level of scrutiny
42:00
and resources that are applied
42:02
upon that code.
42:04
And it's just that the LLM stuff
42:06
is it's
42:08
reducing the resource requirements
42:10
of applying level of screening
42:12
by orders of magnitude.
42:13
And, you know,
42:15
the the models that are
42:17
coming out on a monthly basis are
42:20
improving so
42:22
much that
42:24
it's
42:27
hard for the defenders to keep up
42:29
with the attackers because the
42:31
tool sets are improving for
42:33
both of them, but I think the
42:35
attackers are more motivated to
42:37
take that first mover advantage.
42:41
Yeah, I mean, it's been interesting
42:42
to see like cold kites,
42:44
I think, in one of the earlier sort
42:46
of announcements suggested
42:48
that it might be, you know,
42:51
AI related, obviously, we've
42:53
got these frontier models and
42:54
they're so powerful now that they
42:55
can go and search hard for these
42:57
bugs and then suggest ways to
42:59
exploit them.
43:00
But no one has really come out and
43:02
said like, this is probably how
43:04
it happened and this is who's doing
43:05
it. Our industry has obviously
43:07
been plagued by DPRK.
43:10
Hackers in the last few years,
43:12
would you expect sometime in the
43:14
next few months for us to actually
43:16
have a decent post-mortem that does
43:17
point a finger at someone with
43:19
the on-chain analysis that
43:21
we have today.
43:25
It's complicated because
43:27
obviously even though there was one
43:29
actor who was the first,
43:32
there's probably at least half a
43:34
dozen and like the number of
43:35
people who will be following
43:38
their footsteps and trying to
43:40
iteratively improve upon the
43:41
sophistication of the,
43:43
we'll call it search for
43:46
these keys,
43:48
that's ongoing.
43:49
And it
43:52
does sound like
43:54
the first attacker was
43:57
probably pretty naive
43:59
and didn't
44:02
fully understand the
44:03
intricacies of the protocol.
44:06
And we're probably doing all of
44:07
this by just like chaining LLM
44:10
prompts together.
44:13
And so that was actually, as
44:15
I mentioned at the very beginning,
44:16
one of the things that was baffling
44:17
to me was that the first person who
44:19
contacted me only had a partial
44:21
loss.
44:22
And it appears that the
44:24
reason for that is not because
44:26
the attacker
44:28
had only compromised some of the
44:30
keys, but rather the attacker was
44:32
unsophisticated in actually
44:35
doing their wallet
44:37
derivation and address
44:39
and funds search on Shane.
44:42
So they probably naively
44:44
had some LLM do
44:46
a fairly shallow search
44:48
for funds related to these
44:50
keys, and they were hitting various.
44:53
Roadblocks that were causing them to
44:55
stop searching the key
44:57
space of a given wallet, even after
44:59
all of the keys had theoretically
45:00
been compromised, you still have to
45:02
go generate
45:04
and do a search basically
45:07
in the UTXO set to figure out which
45:08
ones are available.
45:11
So yeah, I mean, that first one
45:13
or possibly few may have been more
45:14
naive and left enough of
45:17
a bread come trail that they might
45:19
get caught, but I expect that.
45:22
The later attackers are
45:24
just going to be more and more
45:25
sophisticated.
45:28
Lop, obviously, your website's a
45:30
wealth of information that I often
45:31
go to for reading and
45:33
recommendations on different things,
45:34
and obviously, you've got a whole
45:36
section dedicated to wallet
45:38
basics,
45:40
different hardware, manufacturers
45:42
that you would recommend,
45:44
et cetera.
45:45
The industry has always said that
45:47
self-custody is how you should do
45:48
Bitcoin, right? You know, you're
45:50
supposed to be banking yourself.
45:53
After this incident has happened,
45:54
you've been around a long time, so
45:55
you've seen worse.
45:56
I mean, I would say you've see worse
45:57
than this, but nothing like this has
45:59
ever really happened up until now.
46:02
How do you feel about, you know,
46:03
like self custody as it stands right
46:05
now, having seen this play out?
46:06
And do you think that we're getting
46:08
to a point where self custody really
46:10
isn't for anyone?
46:11
I don't wanna put any words in your
46:12
mouth. I just wanna get a sense of
46:14
where your head is at, just
46:15
considering the severity of this
46:18
exploits and what it might mean for
46:19
self custody going forward.
46:23
So although I don't think we've
46:25
had random number generator
46:26
vulnerabilities in hardware devices,
46:29
there have been many vulnerabilities
46:31
found in hardware and devices over
46:34
the years.
46:36
And there have many software
46:38
wallets and other,
46:41
I would say even more convenient
46:43
self custody
46:45
products that have
46:47
had random number generators
46:48
failures.
46:51
Self-custody is
46:54
for anyone
46:56
who is willing to
46:58
take on the responsibility that
47:00
comes with it.
47:02
So, you know, with great
47:04
power comes great responsibility.
47:06
This is true for self custody.
47:08
It greatly empowers the
47:10
end user.
47:11
And I think that
47:13
it's...
47:16
It can be a bit tricky,
47:18
like it may be too
47:20
easy for people to get set up
47:22
in self custody.
47:23
I know a lot of people are saying,
47:24
oh, self custody is hard.
47:25
Well, in order to
47:27
have a robust self custody with
47:29
no single points of failure,
47:31
yes, that does take some work.
47:32
And that's what I've spent over
47:35
a decade doing myself
47:36
is trying to make that easier for
47:39
people.
47:40
But...
47:41
It's deceptively simple to just
47:43
download a Bitcoin wallet,
47:45
a software wallet, and run it and
47:47
send money to it, and really have no
47:49
idea of all of the implications
47:51
behind that.
47:53
Um,
47:55
I think when one thing
47:57
that might.
48:01
Need further conversation
48:02
is, and
48:05
you were saying it earlier and I
48:06
think it's gonna be coming up a lot
48:07
in these discussions,
48:09
is the concept of don't trust,
48:11
verify.
48:15
I mean, obviously I think that
48:16
that's a good mantra and
48:19
that one should do
48:21
that whenever one can,
48:22
but you have to understand that
48:25
verification of
48:27
complex software and hardware
48:29
is simply not feasible
48:31
for 99.9%
48:34
of the population.
48:35
And I've had this idea floating
48:37
around in my head for a few years, I
48:39
actually have like a
48:41
draft blog post I've been working on
48:42
regarding it where,
48:45
ultimately what happens is because
48:47
it's just not feasible to
48:49
verify a lot of these things,
48:52
what we end up doing is
48:54
we end trusting.
48:56
Someone that that we suspect
48:58
has verified it, right?
48:59
And so you do end up with
49:02
various layers of trust,
49:04
which is understandable.
49:07
This is also in
49:10
a way how human civilization
49:12
works, like if you think of like
49:14
specialization of tasks, right.
49:16
So, you know, I think there
49:18
are a lot of people out there who
49:20
trust the various security experts
49:22
in the space.
49:23
Or just influencers in the space,
49:25
many of whom are gonna be getting a
49:27
lot of backlash for recommending
49:29
products that
49:32
resulted in people losing money.
49:37
But.
49:37
Perhaps it will be
49:39
easier if we just stick to
49:41
the don't trust aspect
49:43
and understand that you probably
49:45
can't verify.
49:47
And once again, and you know, this
49:49
is obviously shilling my own
49:51
product, but like the entire point
49:53
of CASA is to not trust
49:55
any one thing.
49:57
Not trust any one hardware vendor,
50:00
not trust anyone piece of software,
50:02
not even trust CASA ourselves.
50:03
And the main point that
50:05
I have here is that...
50:08
I understand there's
50:10
going to be a lot of hate and
50:12
vitriol and sadness
50:14
and other emotions and stuff over
50:17
this mistake.
50:21
I don't believe that
50:23
there's any evidence to show that it
50:25
was intentional.
50:26
I do believe that this
50:28
was an honest mistake.
50:31
And there will be good points
50:33
of like, well, this was their one
50:35
job and they failed at their one job
50:36
and that's completely fair.
50:38
But from an
50:40
end user perspective,
50:42
I believe it's
50:44
only fair to understand
50:47
that everyone is fallible.
50:50
Every company out there
50:52
is fallable.
50:54
Yes, like CoinKite has
50:57
been a smaller company with less
50:58
resources and so maybe more prone
51:00
to fallibility than others,
51:02
but ultimately there is
51:05
no such thing as
51:07
perfect software.
51:09
Perfect hardware,
51:11
perfect project.
51:13
Even Bitcoin Core, which is arguably
51:16
like one of the
51:17
most well engineered,
51:21
strongest, robust
51:23
open source projects
51:24
out there has had
51:26
critical vulnerabilities.
51:30
There have been critical
51:31
vulnerabilities over
51:33
the past year found in
51:36
low-level libraries like SSL
51:38
and SSH, like
51:40
code that runs on
51:42
billions of devices around
51:45
the world, and you could argue
51:47
is responsible for
51:49
the security of
51:51
many, many, or orders of
51:53
magnitude more time than
51:55
the entire Bitcoin ecosystem.
51:59
Open source is not panacea.
52:00
Like there, and there is no such
52:02
thing as perfection.
52:03
And so.
52:05
This is why I believe
52:06
that instead of
52:09
saying, OK, we're going to
52:11
trust any one given piece of
52:13
software or hardware, because we're
52:14
going to just verify the hell out of
52:16
it, whenever possible,
52:18
you should instead be
52:21
distributing your trust
52:23
across multiple different
52:25
diverse things.
52:28
Diversity is strength is one
52:30
of our mantras at CASA.
52:33
Because we expect that everything
52:36
can fail.
52:37
And so you want to put yourself in a
52:39
position where you
52:41
expect failure and you
52:43
are able to
52:45
recover from failure.
52:47
And so there's a million different
52:49
paths we could go down here.
52:51
And I was hearing and you were
52:53
talking about dice rolling a bit
52:54
earlier.
52:56
And yes, it's interesting that in
52:58
this particular case the
52:59
people who followed the
53:01
super high paranoid roll
53:04
your dice stuff were actually safe.
53:06
And, and I think it's,
53:08
it's somewhat ironic and unfortunate
53:10
that, um,
53:13
you know, I myself have
53:15
recommended rolling dice.
53:16
I have a blog post from three or
53:18
four years ago that I
53:20
entitled like how to generate
53:22
a secure seed phrase and,
53:25
and this is meant to be normie
53:27
friendly. And I said, basically
53:29
you want...
53:30
You want real-world entropy,
53:32
such as via playing
53:35
cards or throwing dice or whatever.
53:37
And in my guide,
53:40
I specifically say that the
53:42
easiest way to get a
53:44
super high security,
53:46
high entropy seed
53:48
phrase is to buy
53:50
casino-grade dice and buy a cold
53:52
card and use the built-in dice
53:54
rolling functionality.
53:55
And, as
53:57
far as we know, that claim
53:59
still ...
54:00
Holds true.
54:02
But I would note that,
54:04
you know, even that is
54:06
not a guarantee.
54:08
It's not panacea.
54:09
The reason for that is that
54:12
it would
54:14
have been and still is
54:16
entirely possible that there
54:18
is a bug in the
54:20
dice rolling algorithm
54:22
in the cold card.
54:23
Now, we...
54:26
As far as I know, that's not the
54:27
case. And people have actually been
54:29
looking at that over the past few
54:30
days.
54:31
And that code seems to be fine.
54:33
My point is, once
54:35
again, there is no guarantee that
54:36
that code is 100%.
54:38
There's no guarantee that any code
54:40
is a 100% perfect.
54:42
It just so happens that in this
54:43
case, that's the way to go.
54:44
Now,
54:47
going forward,
54:48
like if you wanted, for example,
54:51
to take kind of my perspective
54:53
of how you diversify your
54:55
trust since you can't really verify
54:57
any of the stuff, I
54:59
think the best way
55:01
to go forward would be to have
55:02
multiple different
55:05
means of
55:08
rolling the dice and
55:10
inputting that input and
55:12
and getting output from completely
55:14
different software and hardware.
55:16
That's how you verify that
55:19
it's unlikely to have bugs.
55:22
If you're doing the same thing
55:23
against multiple different software
55:24
and hardware, the
55:26
likelihood that they both
55:27
implemented the exact same bug
55:29
is very, very low.
55:31
And that's the same reason why
55:33
we are
55:36
highly supportive of
55:38
this idea of multi-vendor,
55:39
multi-SIG.
55:40
It's just because you expect
55:42
that anything can and will have
55:44
bugs, But.
55:45
As you diversify
55:47
your trust across many
55:49
different organizations,
55:51
then it's just a
55:53
sort of hedging game.
55:56
Yeah.
55:57
Talking about the hedging game,
55:58
obviously you've been at CUSA
56:00
a long time, you've been in the
56:02
security game a long, do you
56:04
think there needs to be a message
56:05
out there now considering everything
56:07
that has happened about A, best
56:09
practice for self custody, does
56:10
it need to be multi-seg,
56:13
where's your head at with regards to
56:15
that?
56:17
I mean, it's gonna be the same
56:19
thing that I've been preaching for
56:21
a decade.
56:24
It's very difficult to talk about
56:26
self-custody generally
56:28
because there are so many variables.
56:32
You know, I'm not gonna tell people
56:34
everybody needs to put
56:36
their Bitcoin into a
56:38
geographically distributed
56:39
multi-vendor multisig.
56:41
I believe that only makes
56:43
sense when you get to the point that
56:45
you have life
56:47
saving. Things.
56:49
When you have enough
56:51
Bitcoin that losing it
56:53
would be a sufficiently
56:56
large financial catastrophe
56:58
that you would never recover from
57:00
it, that's when you
57:02
should be getting to
57:04
the level of
57:05
paranoia where you're putting in
57:07
enough resources to ensure
57:10
you don't have single points
57:12
of failure, where if one thing goes
57:13
wrong, you have catastrophic loss.
57:17
But, you know, if you're just
57:20
playing around with Bitcoin, if you
57:21
only have a few hundred dollars
57:22
worth of Bitcoin, then yeah, leave
57:24
it on an exchange, put it
57:26
in a hot wallet.
57:29
If that blows up, then
57:31
it's not going to, you know,
57:33
negatively impact the rest of your
57:34
life.
57:35
If you have,
57:37
I generally say, if you have
57:39
thousands to tens of thousands
57:41
of dollars, That's
57:43
when it starts making sense to
57:45
invest a hundred or a few hundred
57:46
dollars into a hardware device
57:49
to get
57:51
your keys off the internet.
57:53
That protects you from
57:55
99% of
57:57
common loss
58:00
vectors.
58:02
But it's when you get beyond
58:04
that is when you should start
58:06
investing more.
58:07
I mean, generally just think of
58:09
it as.
58:10
And this is true for all types
58:12
of security.
58:14
The amount of
58:16
time, effort, and resources that
58:18
you invest into securing
58:20
a given asset should
58:23
be correlated to
58:25
the value of that asset.
58:28
So whether that's,
58:30
you're like insuring a house
58:32
or a car or
58:35
protecting the
58:37
physical security of your...
58:40
Family and loved ones,
58:42
you know, you should be spending
58:44
an amount on that that
58:46
is relative to how much you
58:48
care about it, how valuable it is,
58:50
and how irreplaceable it is.
58:54
I'm gonna quickly finish
58:56
off with a final question to you.
58:58
I mean, I know I think a lot of
58:59
people out there will naturally
59:01
gravitate towards some sort of
59:03
a custodian
59:05
or a custodeal solution.
59:07
CASA is still self-custodial,
59:10
right? Do you ever get people
59:11
knocking on your door and they're
59:12
like, I don't wanna use CASA because
59:13
I wanna be in full control and maybe
59:14
they don't understand that is
59:16
what it is.
59:19
And yeah, I don't know
59:21
if you want to just comment on the
59:23
other hardware wallets out there.
59:25
And I guess the coming together
59:27
of the industry, because it's been,
59:28
the last few weeks in Bitcoin has
59:30
been hugely divisive around, you
59:31
know, Bip 10, Bering in the
59:33
background. And this kind of seems
59:34
like everyone suddenly forgot about
59:36
that and got together and tried
59:38
to support each other as best as
59:39
possible as Bitcoin is first
59:41
and foremost.
59:43
Yeah. So, you know, the
59:45
way that I've framed
59:47
it is that,
59:49
um,
59:51
you
1:00:00
Diversity of
1:00:02
security properties and that
1:00:04
collectively added all together
1:00:06
uses this
1:00:08
really magic of
1:00:11
additive security properties that
1:00:13
puts you in a much more secure
1:00:15
position than you could ever be if
1:00:17
you only have one key of
1:00:19
any type.
1:00:22
But we're not reinventing
1:00:25
the wheel.
1:00:26
We're not doing anything novel from
1:00:28
a protocol point of view.
1:00:30
So it has always and will
1:00:32
always be possible for someone to
1:00:35
create their own multi-vendor
1:00:36
multi-SIG using nothing but
1:00:38
free open source software
1:00:41
that there's no subscription or
1:00:43
anything like that.
1:00:44
I think the reason you would
1:00:46
want to pay for
1:00:48
CASA Is
1:00:50
your your You're paying for
1:00:51
consulting, you're paying for
1:00:54
being given a
1:00:57
really a guided
1:00:58
software experience that
1:01:01
we
1:01:03
expressly do not support a lot
1:01:05
of advanced things that Bitcoin
1:01:06
supports because we think those are
1:01:08
foot guns that are too
1:01:09
easy to backfire.
1:01:11
So really, we're guiding
1:01:14
you through what we consider to be
1:01:15
the best practices.
1:01:16
And saving people a lot of time
1:01:19
from having to spend days,
1:01:21
weeks, months learning about all
1:01:23
of these best practices and
1:01:25
employing them.
1:01:26
And you're also, you're paying for
1:01:28
the CASA recovery key
1:01:31
that we safeguard and
1:01:33
build additional authorization
1:01:35
parameters around.
1:01:37
And there's plenty of other
1:01:39
functionality as well that
1:01:40
theoretically you could build
1:01:42
yourself, but we've been.
1:01:43
Building for eight years.
1:01:44
And so we've probably been able to
1:01:46
put a lot more resources into it
1:01:47
than any one person.
1:01:50
So optionality is great.
1:01:52
Just be aware that when
1:01:55
you are, if you're architecting
1:01:57
your self custody on your own,
1:02:02
it's problematic for
1:02:04
the same reason that these
1:02:06
other vulnerabilities have been
1:02:08
found, which like, as I
1:02:10
said, open source is not Patasia.
1:02:12
And it really comes down, I
1:02:14
think, to the principle of various
1:02:15
and numerous, which is strength
1:02:17
and numbers.
1:02:19
And when you're using CASA,
1:02:21
you're a system that a lot
1:02:23
of people are using that
1:02:25
we've had a lot
1:02:27
of engineering talent look at that
1:02:28
we get regular
1:02:31
security audits on.
1:02:33
And actually at this point, we have
1:02:35
our own AI pin test that
1:02:37
we're running all of the time.
1:02:39
So...
1:02:39
You're basically paying for
1:02:42
the expertise of a
1:02:44
security focused organization.
1:02:46
Whereas if you're doing it yourself,
1:02:48
it's highly likely that simply
1:02:50
because you're the only one looking
1:02:52
at it, you're just going to overlook
1:02:53
some potential vulnerabilities
1:02:55
and something over a long enough
1:02:57
time frame could go wrong.
1:03:01
Yeah, my final question to
1:03:03
you, I find this funny that,
1:03:05
you know, there was some influences
1:03:07
and stuff online being like, you
1:03:08
know, Bitcoin self custody is dead.
1:03:10
You're better off just, you know,
1:03:12
buying the Bitcoin ETFs.
1:03:14
Like my general take on that is
1:03:16
like, I fundamentally believe
1:03:17
Bitcoin is money and it will be
1:03:19
money that we all use
1:03:21
in the future.
1:03:22
And I want to own it and I want to
1:03:23
have total control of it.
1:03:25
I fundamentally believed that.
1:03:26
And I pushed back, you know, on
1:03:28
those assertions online, but Do
1:03:30
you think it's funny that people are
1:03:31
saying that online and is that just
1:03:33
completely misguided and we should
1:03:35
really still take Bitcoin
1:03:37
self-custody as the first prize and
1:03:40
really just be in control of
1:03:41
our own keys?
1:03:45
I mean the freedom of the protocol
1:03:47
is that people
1:03:49
can put
1:03:51
themselves into what security model
1:03:53
they want. They can choose whether
1:03:55
they prefer convenience
1:03:57
or security. That's really what this
1:03:58
all comes down to is that
1:04:01
if you want to
1:04:03
be empowered, if want
1:04:05
to actually
1:04:07
take full advantage of.
1:04:10
What I would say Bitcoin's
1:04:12
fundamental premise of not trusting
1:04:14
third parties is,
1:04:16
you gotta put in some work to do so.
1:04:18
If all you want is financial
1:04:21
exposure to Bitcoin as a
1:04:22
volatile asset, and
1:04:24
you don't
1:04:26
care too much about
1:04:28
the risks of
1:04:30
custodial failure, despite the
1:04:32
history of that happening in
1:04:34
the space, then you're
1:04:37
free to do that as well.
1:04:38
The the only thing that worries me
1:04:40
and the reason why I've spent the
1:04:42
past decade focused on self-custody
1:04:45
is That I do believe that
1:04:47
if we don't continue to iterate
1:04:49
on it and make it
1:04:51
easier for people make
1:04:53
people Believe
1:04:55
that they are capable
1:04:57
of taking on this Responsibility
1:04:59
then it will end
1:05:01
up being a niche thing that
1:05:03
only like super nerds
1:05:05
and paranoid folks are
1:05:07
going to end up bothering to do
1:05:09
and ultimately that is
1:05:12
arguably not great for
1:05:15
the fragility
1:05:18
of the overall ecosystem because
1:05:20
it would result in massive
1:05:22
amounts of Bitcoin being centralized
1:05:24
into a few trusted third parties and
1:05:25
then we're just recreating the old
1:05:27
system all over again.
1:05:28
Yep.
1:05:29
Completely agree with you.
1:05:31
I don't think I could have said it
1:05:32
better than that.
1:05:33
Jamison, thank you so much for
1:05:34
taking the time to chat to us.
1:05:35
I know you're probably busy on this
1:05:36
Monday with the amount of people
1:05:37
that are just looking to migrate
1:05:39
funds or cold card,
1:05:42
but salute you and every other
1:05:43
Bitcoin out there that's doing what
1:05:45
they can to help people in these
1:05:46
pretty tough times.
1:05:48
So Jamison we'll say goodbye to you.
1:05:49
Feel free to drop off the stream.
1:05:52
Thank you so for watching the
1:05:54
starting block today.
1:05:55
We broadcast live every day,
1:05:57
8.30 a.m. Eastern Time, 2.30
1:05:59
p.m., Central European Time, if
1:06:01
you're in Europe like I
1:06:03
am.
1:06:04
Please come back tomorrow.
1:06:05
We have Ophelia Snyder, who is
1:06:07
one of the co-founders of 21 shares.
1:06:09
She exited the company last year.
1:06:11
We're going to be chatting about the
1:06:12
institutionalization of Bitcoin and
1:06:14
many other cryptocurrencies.
1:06:15
So we'd love to have you join us
1:06:17
on the show tomorrow,
1:06:20
wherever you are around the world.
1:06:21
Look after yourself.
1:06:22
Look after your Bitcoin.
1:06:24
We'll be back tomorrow Until then,
1:06:25
bye-bye.