0:10 You know, by way of introducing Sam, for those of you who don't know Sam, you know,
0:16 he was recently elected to be the chairman of the board at the Sovereign Foundation.
0:23 So congratulations, Sam. In looking at Sam's background and reading, you know,
0:29 some of the of his past works. I was just in awe and I felt very lucky, you know, by
0:34 way of another past guest of the podcast, Phil Winley for introducing him to us. You
0:42 know, he comes, you know, with the background as both an entrepreneur, a
0:46 consultant, you know, very prolific in his writing, looks likes to spend time at the
0:53 intersection of AI, blockchain, and decentralized identity. So, with that, I'm
0:59 going to turn it over to Sam. And Sam, thank you very much for joining us from
1:02 whereabouts in Utah are you from? – A little town called Linden, which is a sub,
1:10 you know, in the greater metropolitan area for Salt Lake City, Utah. You know,
1:17 everybody in Utah lives in a three-county area. The rest of the state is mostly
1:21 empty. – Well, I'm going to, you know, turn it over to you. Thank you so much.
1:27 And take it away. All right. Thank you. So I'm going to talk about the economics of
1:35 its and bits. And that's going to cover a little bit about digital identity,
1:40 freedom, privacy, control, and security. And I left security for last because it's
1:45 actually the most important part of this conversation. And those that know me
1:50 probably aren't surprised when I say that. So when I talk about economics, I'm really
1:55 talking about value, creation and capture. And so in the environment of digital
2:01 identity, what do we mean by that? Well, we talk about value extraction exchange
2:06 and of course exploitation. And in order to have economics, somebody to benefit
2:14 from economics has to have control over value. That's the key thing. And if you
2:19 want control, then you need security. And so this cuts at the root of everything
2:25 that we do in digital identity. If we don't have security, we don't have control
2:29 over value and that leaves us free for exploitation by others. So, how do we get
2:35 control over value? Well, this is where the its and the bits come in. When I say
2:39 its, I mean atoms, physical things. So, a person is made of atoms. So, a person is a
2:47 physical thing. Buildings, cars, planes, ore, minerals, all of those things are
2:54 physical things. And here's the thing. It's easy to have security over physical
2:58 things because physical things are not easily replicable. You have to create more
3:06 atoms or you have to build or extract or get more atoms. And it costs money. Atoms
3:11 are relatively scarce. Some atoms are very scarce. And you can maintain control over
3:17 those atoms with what's called physical security. Just keep the atoms in place
3:22 where somebody can't get to them. Now, when we talk about bits, we have a
3:27 problem. Bits is digital information. The problem with bits is that it's easy to
3:34 copy them. We actually use subatomic particles to transfer and store
3:39 information, but those subatomic particles, electrons and photons are, you
3:45 know, virtually inexhaustible. So, from a cost perspective, information wants to be
3:50 free, easily copied, very difficult difficult to secure, especially if we want
3:56 to control it because if somebody copies it, we lose control of it. So the question
4:00 is in ITS, we have certain entities that have control over other entities or other
4:09 physical bits. In information, what fills the role of a legal entity like a person
4:15 or an organization or a corporation or a government? What fills that legal role
4:20 that actually gives the same degree of informational security that I have in
4:24 physical security. So I'm not going to answer that question right now, but I'm
4:28 going to pose that question because I wanted to motivate the reason why that's
4:33 so important. So if we look at 60 years of its and bits, I looked at the Fortune 20
4:39 in 1960. You can look at the companies and the Fortune 500, they always in the old
4:47 days, they did it in terms of revenue. That's how they rank them. You see that
4:50 all but one company on this list gets its value from capturing its, not bits. Only
4:58 AT&T is primarily gets its value from capturing, storing, transmitting bits of
5:04 information. If we go to 1980, well, we got IBM in there. You could maybe say that
5:10 it got it from bits. It was a equipment manufacturer, but it was equipment to
5:17 capture value from bits. When we get to 2000, There's a few more in there. We've
5:24 got Southern Bell, Hewlett-Packard, Compact Computer, and AT&T. But still, the
5:31 majority of value capture, creation capture is in the it's, not the bits.
5:39 2020, things start to change. We see now that for the first time, we've got Amazon,
5:44 Apple, you know, they've moved up into the top spots, Alphabet, Verizon, But this is
5:54 revenue. And if we look at something much more telling, it's not revenue, but market
6:01 cap. This is people saying who is going to capture value in the future? What
6:06 companies are going to capture the value? And this is a scary list. Now, in terms of
6:13 market cap in 2020, we've got Microsoft, Apple, Amazon.com, Alphabet, Facebook, and
6:19 we've got
6:24 five, well, three, $2 trillion companies. Alphabet was, Amazon was trillion dollars
6:30 for a little bit in 2019 and Facebook and then Verizon, AT&T are still there. So,
6:38 the top value captures in the world are the ones that capture value from
6:42 aggregating bits. They have figured it out. They have figured out how to capture
6:48 value. And they're doing it through something called network effects. A
6:53 network scaling law basically describes how network value scales with the number
6:58 of participants. And so if we have a multi -sided network like this, then Metcalfe's
7:04 law may apply. And when Metcalfe's law applies, we get this very powerful
7:08 scaling. So this equation says that the value, the average value to a participant
7:13 on the network V is proportional to the number of participants on the network. And
7:20 the total value of the network is the average value per participant times number
7:24 of participants. So, it's proportional to the number of participants squared. And
7:30 any square loss scaling is going to dominate when the network gets large
7:35 enough. And this can be the main value that these companies are able to aggregate
7:42 is these network effects. And it's the network effects from bits of information
7:46 that they're aggregating. And the primary value is reach. If you have reach, like
7:52 you can reach people in Facebook ads or Google ads, or you can reach people in
7:58 your Apple app on the App Store, those contribute to value that scales
8:05 tremendously. And if you extract even a little bit of value from a square law
8:09 network effect, you're going to be an extremely valuable company. And Met-
8:14 calfe's law has been extensively validated. I wrote a paper on this some
8:19 these are some of the papers from the quote. You can go look at the paper, look
8:22 at them if you're not familiar with it. So the question is how do we recapture the
8:28 value in our data? Because the value in our data is being captured with good
8:32 effect by these super aggregators. There's two ways that we can capture it. We can
8:38 leverage cooperative network effects and we can retake control of our data. And
8:43 when I say retake control, I mean really take control of our data, not sort of take
8:47 control of our data. So what do I mean by cooperative network effects? Well, what
8:53 happens to value when two smaller networks combine? And understand, this is how we
8:58 fight people that are the largest, most profitable, most powerful corporations in
9:04 the world, right? How do we fight them? Because if we can't fight them, we'll
9:08 always lose. Because they have more power. They are now at the state size where they
9:15 are nation states in and of themselves. themselves, they can change how nation
9:21 states treat them. Well, Metcalfe's law comes to the rescue. If you look at
9:27 Metcalfe's square law effect and I take two networks, N1 and N2, and I combine
9:32 them and I look at the square law value, I get this bigger rectangle. And so N1 by
9:39 itself has this smaller rectangle. N2 by itself has a smaller rectangle. But
9:44 combined, each of them gets an an added rectangle that is in N2's cases, N1's case
9:52 is bigger than N1 was to start with. And so without going through the math, what
9:57 you get is a bonus of an N1 times N2 proportional factor for any two networks
10:03 that combine. This is essentially free just for combining, just by combining
10:10 participants from two networks. So why wouldn't any two networks combine? That's
10:16 a good question. Right? And this is actually the regimes in which combination
10:24 makes sense. When two networks are value symmetric, that means they're of the same
10:29 size value, then it's always more profitable for both to combine. And in
10:33 this paper, positive demand externalities is an operational research code word for
10:40 network effects. When two networks are value asymmetric, it's always more
10:45 profitable for smaller networks to combine. This is lifetime value of the
10:48 combination. Now, when I have two networks where one is much larger, it's not always
10:53 more profitable for the larger one to combine. That's the problem. If you're
10:57 large enough, you don't want to combine. But if the network size is below a
11:02 threshold, then it's still more profitable for the larger one to combine. And that
11:08 threshold is based on a feedback adoption growth rate curve, which tends to follow
11:13 sigmoidal adoption curve. And when you're down here close to the bottom of this
11:18 curve, even if you're a much larger network, it's still to your advantage to
11:23 always combine with smaller networks. If you're up here in a saturation regime,
11:28 there's no advantage to combination. So, what is this? What do competing small
11:34 networks do? Well, let's look at what happens if we had several competing small
11:38 networks. Each of them has a value proportional to their side. They're all
11:43 the same size, right? And if they're square law effect. If I now make a super
11:50 network of these cooperating small networks, the value of each sub network is
11:56 10 times its value before they combine when the size of the total network is 10
12:02 times the size. So, I get a 10 times increase in value for free just by
12:07 cooperating to the extent that cooperation is free. I mean, there's always some cost
12:14 to cooperation if you don't have interoperable systems. But if you have
12:17 interoperable systems, then that cost is pretty small. The total value of the super
12:22 network is 100 times the value of the individual networks combined. This is
12:27 huge. This allows a bunch of small networks to become super aggregators
12:32 merely by combining. And when I say super aggregators, I mean benefit have a network
12:40 whose network value benefits from the network effect. So that is a powerful
12:45 force that we can use to fight the
12:52 extraction and exploitation of our data value, how we can fight back. And so this
12:57 is an equation that says the proportion will value for any network to join a
13:02 larger network is the ratio of the size of the larger network to the size of the
13:07 smaller network. So the network effect from a subnetwork joining and cooperating
13:11 as a super network is the subnetwork's values increased by the ratio of the
13:14 subnetwork size to the supernetwork subnetwork size. This is the advantage of
13:19 cooperation. It's a huge advantage. So the small network strategy is to one strategy
13:26 is to create what's called a cooperative network cascade. What that means is I've
13:30 got a small network and I'm competing with the larger network. If I convince the
13:35 larger network to cooperate, then the combined network can now outcompete the
13:40 next larger network and so on. And so you can cascade to where you can force even
13:47 the largest existing non-cooperating network to realize that it's more
13:52 profitable for them to cooperate than to compete because if you're benefiting from
13:57 Metcalf's network scaling effects. So, how do we remove the primary barrier to
14:03 cooperation? Even with this knowledge, even with this knowledge that cooperation
14:07 is more beneficial than competition when you're in the right regime, which happens
14:13 a lot, but people still don't cooperate. because they're competitors. So how do we
14:17 remove that barrier? Well, one way to remove that barrier is to look at
14:20 different value contexts. Look at networks that aren't directly competitive. If we
14:27 can find value that's transferable between contexts, then we have a way of gaining a
14:35 cooperative network effect without ever forcing competitors to cooperate because
14:41 they're not competitors in the first place. They just have to cooperate. And I
14:44 call this transcontential value creation and capture. So if we can have value that
14:49 is transitive or transferable between contexts, and context in this case would
14:53 be products and services in the same market segment, then we have an
14:59 opportunity to prime or bootstrap cooperative network effects. So we use
15:05 transcontential value creation and capture to fuel cooperative network effects. And
15:10 then that starts the cascade. So the final thing we want to do is since the question
15:16 was how do we regain control of our data value is that we want these cooperative
15:23 networks to be participant controlled. We want to extract the value from it, not
15:28 some administrator of a network. Even a cooperative network administrator can
15:32 extract value. We want the participants to be able to extract appropriate levels of
15:36 value. So we want participant controlled networks and we can get participant
15:40 controlled networks if we use the right tech. And what that means is that
15:44 participants can then create a virtual participant-centric network. And this
15:50 enables participants to amplify their own value across multiple contexts, which
15:54 maximizes adoption pool and creates the fastest spin-in of cooperative network
15:58 effects. That's the economics of cooperation. And another benefit of
16:04 cooperation is what I call the cooperative long tail network effect. If you think
16:07 about the long tail, and that's the area under the the long tail in this curve,
16:12 this is a power log distribution for a market. The long tail is essentially a
16:18 bunch of different contexts because they're all too small and disparate and
16:23 segmented to have enough individual value to be here where people are targeting. And
16:29 so if we recognize that the long tail is different context, we can create a
16:34 cooperative network that aggregates the value in that long tail and and get a
16:41 network effect just from the long tail. So, where do we find transcontextual
16:44 value? Well, we find it in transaction costs. The theory of the firm, the theory
16:52 of why any organization has value is that it reduces transaction costs to its
16:59 customers. There are three ways we can group transaction costs into three
17:04 buckets. There's lots of ways to do it, but this is a very useful grouping. First
17:10 set are triangulation costs. These are about connection, fine filtering and
17:14 matching participants. The second is transfer, facilitation, transport,
17:18 delivering payment. We all know business models that are focused on one or both of
17:23 these. The third one, the one that is hidden that most of us don't realize or
17:27 see is trust transaction costs, security, competency, reliability, privacy, and
17:34 liability. These trust costs are often hidden costs and they're often not well
17:40 optimized. They're not reduced. And in many cases, people mistake the value of a
17:45 network effect from their triangulation transfer and ignore the fact that they
17:50 actually solved a trust transaction cost. And this is where digital identity and
17:54 decentralized identity come to play because all of the technical stuff, trust
17:59 over IP, right? It's about reducing trust transaction costs. So, platforms and
18:05 networks sell reductions in transaction costs. To a consumer, all costs look like
18:10 transaction costs. It doesn't matter if it's its or bits. All costs look like
18:13 transaction costs. The principal super aggregated pull in many cases is reduced
18:18 trust action costs. Your Facebook login account. You can log into Facebook.
18:23 They're going to manage all of your trust for you, all of your security so that you
18:27 can interact with other participants on the platform and you don't have to worry
18:31 about it. So, but the thing is trust may be highly transferable. If you think about
18:37 it, reliability, credibility, accessibility, those things, an entity or
18:43 an individual tend to be transferable. Those sorts of behaviors tend to be
18:50 transferable. Security tends to be transferable. Digital identity is
18:55 transferable. Login, onboarding, all of those things are transferable. So, a
19:02 reduction in trust transaction cost is a primary network effect value from
19:05 cooperation. We can leverage that. So, what we want is we want to have transitive
19:11 value virtual network scaling law. What does that look like? Well, if we have a
19:16 set of transcontextual cooperating networks, that means there's value that
19:20 can transfer between, but they're not direct competitors, and there's some
19:25 transitivity factor, average transitivity factor, then we can rewrite Metcoff
19:29 scaling law this way and don't read the math, but it basically gives us down at
19:35 the bottom a square law scale effect just from the value that is extractable, that
19:41 is transitive. And even if T is really, really, really small because it's a square
19:46 law, we get huge value just from cooperation. So, that's how we fight. And
19:53 you can read the details in this paper below. Just look up meta platforms and
19:57 cooperative networking network effects. So, the second strategy was retake control
20:03 of our data. How do we retake control of our data? data. We have certain toolkits
20:08 for controlling our data. Now, you've heard this thing that says if all you have
20:13 is a hammer, everything looks like a nail. Well, the truth of the matter is if you
20:18 have nails, there's different types of nails and different types of hammers work
20:24 better. So, that's okay. But in this context, usually people use it to say,
20:28 well, you need to get more tools in your toolbox and not just have hammers. The
20:32 problem is that an information we only have one set of tools. We can do things
20:38 with screws and we can use screwdrivers, but hammer is on an option. In fact, there
20:45 aren't very many options. We have one set of tools for truly secure data control.
20:49 That's it. And this gets us back to answering the question because the hard
20:56 question for data is secure control, right? Information security. So, we want
21:02 control over value. you. Physical security, that's a solvable problem.
21:07 Information security, we continue to have hacks of computing infrastructure. We
21:14 continue to have the network is not secure. We continue to have, in fact, many
21:19 of the talks today were about how to how better to achieve informational security
21:23 in your systems. And it's not a and it appears to not be a well-solved problem.
21:30 Well, the solution, the set of tools comes from digital uniqueness. The reason
21:35 physical security is a solvable problem because we have physical uniqueness of
21:39 atoms. We do not have physical uniqueness of bits. We need to somehow bootstrap to
21:46 something that looks like that. That bootstrap, there's only one. It's called
21:51 informational entropy. It's the only tool we have. It's like the screws in the
21:57 picture before and the drivers are entropy -derived tools. These are cryptographic
22:04 one-way functions. You know them as hashes, scalar multiplication, digital
22:09 signatures, EKPs. That's it. That's all we've got. If we don't know how to use
22:13 them well, we don't have informational security. If we use them poorly or we use
22:18 them inadequately or we use them in not best practices, we fail over and over and
22:24 over again. We have to do this right if we ever want to actually gain control over
22:29 our data. It's like we're in la-la land if we think otherwise. And unfortunately,
22:36 many people imagine that there's some magical way to get informational security
22:40 that doesn't use these tools. So information uniqueness comes from captured
22:46 entropy. But guess what? Anybody can capture entropy. That makes captured
22:51 entropy an inherently decentralized tool. So we individuals, participants can take
22:58 control of our data, but we first must retake control of our identifiers because
23:02 the identifier control is what the tools give us. We can control our identifiers.
23:09 And the type of identifier that allows us to do that are called self-certifying
23:14 pseudonymous identifiers. And there's a white paper I wrote that talks about the
23:19 theory of self-certifying pseudonymous identifiers and why they enable us to have
23:23 true decentralized roots of trust to control over our data. But obviously it's
23:27 not, it's and you can go look at the paper, but it's beyond the scope of this
23:31 talk. What that simplifies to are the four A's of secure data control. That's the
23:37 author, authentic, authorized, and authoritative. Each data item can be
23:45 imbued with these four A's. So, I call that A4 data control. That establishes
23:51 that control via self-certifying pseudonymous identifiers of the data and
23:57 the author and the authorization consent to all of those can be can be put in the
24:02 data. And so I coined a new term called sapored data. So the term sapored means the
24:08 quality and substance that affects the sense of taste, savor, flavor. So if we
24:12 want data that is controllable, we need to sapored the data. And what that means is we
24:18 can securely provenance the data to its authors. And if we can't securely
24:23 provenance the data to the authors, we can't securely extract control. That
24:27 attribution is necessary. And that means that we we build sapored data supply chains
24:32 that enable consumer pull in addition to regulatory push so that we become
24:37 conscious consumers of sapored data. So we go back to our diagram. We have the its that
24:44 control the bits. And now I've got rid of the question mark here because this is
24:50 cryptographic pseudonymous identifiers. We can control those. We can control those in
24:54 a decentralized secure way and there's no there are no exploits other than you can
25:01 protect it with purely key management. That means that we can have control and
25:06 attribution. So we can control our data and we can get attribution back, which
25:10 means we can extract value of our own data. And what we want to prevent is
25:16 correlation between these identifiers and the its. That's the thing that we need to
25:22 do to get privacy. But if we throw away all of this in order to get privacy, we
25:27 throw away our ability to extract value and to control our own data. And that's
25:32 the problem with the regulatory regime. Most regulations are throwing away all of
25:38 the bits in order to prevent the correlation. They say, oh, you don't want
25:43 correlation? Get rid of the bits. Then you won't have correlation because there's
25:46 nothing to correlate. But then you get rid of the value as well. So it's a
25:50 diseconomic incentive. So what does that mean? Privacy. You have to choose. You can
25:55 have this kind of privacy. You're a lone person on an island. Don't interact with
25:59 anybody or you can have privacy where you interact with people. But if you're alone
26:05 on the island, it can be strong because you're not disclosing data to anyone.
26:10 You're not extracting value from your data. You're just living on your own. If
26:14 you want to interact, have commercial economic interactions, then the only
26:19 privacy you can have is weak. And that sounds depressing, but it's not. So what
26:24 do I mean by strong privacy? It means uncorrelated interactions over unbound in
26:29 time and space. And I get in these discussions with people who want strong
26:33 privacy and it's not possible because super aggregators and state actors have
26:39 effectively unlimited storage and compute capacity. That means eventually all
26:42 disclosed data, that means not just the third parties, the second parties will be
26:46 at least statistically correlatable at some time. So all we can have is weak
26:51 privacy. What does that mean? We can have uncorrelated interactive actions over
26:55 bounded time and space. That's what we get. That's what we you have to work with.
27:00 And when is it weak? When the cost of correlation exceeds the value of
27:04 correlation of the data, then it will be uncorrelated. Then we'll have weak
27:08 privacy. So it's an economic privacy. So let's look at the economics of
27:13 correlation. On this graph, I have a data life cycle. So now is data in the present.
27:21 And when I'm looking at predictive value of that data, that's the primary value
27:24 that aggregators get from aggregating our data is predicting. So, if I look at the
27:30 blue line, this is the value of correlating my data. Notice that if I can
27:35 correlate data in the future, like predicted value, it's very valuable. But
27:39 the older data is, the less predictive it is, the less value it has. The brown graph
27:47 is about the correlation cost. It's harder to correlate data that's present because
27:53 it hasn't been shared much. You can keep it close. You can keep it relatively
27:59 private. You can only share it, disclose it with certain people. Over time, it gets
28:03 cheaper to correlate because data is correlatable. Now, maybe, you know, some
28:08 data may just be pretty hard to correlate, but eventually it might become zero. I
28:14 don't know what this curve looks like. What that means is that where these curve
28:19 crosses right here tells you something. That tells you where the extractable value
28:24 is. This is the value that somebody's going to go after. So if you can move this
28:30 curve by making it a little bit harder to extract, you can reduce the extractable
28:35 value that increases your privacy because it just won't be extracted because it's
28:40 not economic. And if you can get more value out of your data, then that changes
28:48 things too. So let's look at the economics of the decorrelator. That's you. You want
28:53 your data to be decorrelated, right? So you have a decorrelation value. This is
28:57 the value of your data when it's decorrelated because then you can extract
29:00 value from it. But here is your cost to prevent people from extracting value. And
29:06 guess what? Your costs go up exponentially the older the data is because now you're
29:10 fighting against the super aggregators and the nation states and they're going to
29:14 beat you every time. So the only place you can extract value is here. It's in the
29:20 immediate past and in the future. But this can be a lot of value. And if we do things
29:25 right, we can move this red curve to the left so that we can extract more value.
29:31 But we have to control. We have to be able to control our data or we can't extract
29:35 our own value. And it depends on the operating regime. Those curves are
29:40 operating dependent. And this is one of the problems is that when we talk about
29:44 privacy and correlation, if we're talking about legitimate actors, legitimate
29:49 economic actors and legitimate political actors, then regulation is a useful tool.
29:56 Contracts are a useful tool. But as soon as we're talking about illegitimate
30:00 actors, illegitimate political actors, neither of those tools work. By
30:04 definition, they don't keep the rules and illegitimate economic actors don't keep
30:09 the rules. So if we're in an illegitimate political regime with illegitimate
30:12 economic actors, the only way we can protect our data is to hide and bribe. We
30:17 just never disclose our data. So we can't ever get value from it unless we bribe
30:22 somebody, which means we're not getting very much value. If we're in a legitimate
30:28 political regime, but dealing with illegitimate economic actors, we might be
30:32 able to hide or bribe. And if we're in a, I'm not sure what happens in a with
30:37 legitimate actors and illegitimate political regime. I guess if they're just
30:41 honest, then we can do business with them with contracts. If they're dishonest, then
30:46 they can get away with it because they can, they're an illegitimate political
30:50 regime, but doesn't enforce the rules or enforces rules that we don't have. If we
30:55 don't, and the thing is, is if we build privacy preserving mechanisms that aren't
31:02 recognizing the regime we're operating in, we'll build systems that won't work well.
31:06 And often things that work well down here are horrible up here because we can't get
31:12 value because we're just we're throwing away our value in order to work in this
31:16 regime. Because in this regime, pretty much the only thing you can do is to hide.
31:19 But in this regime, you don't want to hide. You want to extract your value. You
31:23 just want to extract it securely under your control and protect yourself through
31:27 contracts and regulation. So when we talk about freedom, freedom is a balancing act.
31:33 We have freedom from and freedom to. We have freedom from exploitation,
31:37 intimidation, censorship. We also need the freedom to extract value of our own to
31:42 build relationships to build community. The possibility of ratio of our data
31:46 increases the possibility of censorship. Anonymity is the loss of value from
31:52 attribution. Fairness requires data attribution. So there's trade-offs between
31:58 these values of freedom and privacy. And we have to… Regulations and technology has
32:06 to recognize those trade-offs. So what is a possible solution to those trade-offs?
32:11 Well, class-based A4 pseudonymity. It has the ability to have both high degrees of
32:18 hiding and non-disclosure while still maintaining control and value extraction
32:25 based on the class. And then we can separate our identifier trust bases so
32:31 that if we have to hide, we can erase without being commingled with other trust
32:39 bases that prevent erasure. And KERI allows us to do both of those. And KERI is
32:44 a future technology that doesn't exist today, but is something that we're
32:48 building for the future. So, that's the end of my talk.
32:54 And I'm sorry I went three minutes over.
33:00 I can't hear you, Steve.
33:06 Steve, I can't I can't hear you. Steve, you're on mute for everyone. Nobody can
33:09 hear you. Yeah, sorry. Sorry. Yeah, my double mute is protecting me from myself.
33:16 Sam, I have to say that the that talk was epic and I've early feedback. from people
33:22 on instant message and LinkedIn suggests that our minds are blown and we maybe
33:28 should have you gone earlier in the day. But for everybody who likes podcasts, I
33:35 assure you, Sam has promised me he'll come back and record a podcast with me because
33:43 for me, that's something I need to play like over again and over again and over
33:47 again and play it slowly and then play it fast. and take a transcript and it needs a
33:53 little bit of repetitiveness for it to sink in. Yeah. And I'll post the slide
33:57 someplace. I'm not, I guess. That was another question. Yep. Yeah. You just tell
34:00 me where to just send me the link and I'll post them there. You may have already done
34:03 that, but I just. If you could, you know, just email me a link or, you know, the
34:08 file, I'll make sure. Yeah. I'll put them up on my GitHub page and then I'll send
34:13 you a link to them. Awesome. I appreciate it. And I will be following up with you
34:18 because, you know, Phil has been on a podcast. We've had Drummond, we've had
34:22 Anne, quite a few folks. Good luck in your new role at Sovereign. I'm going to be
34:29 watching that and looking forward to chatting again with you soon.
34:44 We'll see you next time.