Latest / The Starting Block / DoubleZero infra to supercharge Kalshi - Ft. Austin Federa

Transcript

DoubleZero infra to supercharge Kalshi - Ft. Austin Federa · The Starting Block

0:06
Welcome back to The Starting Block,
0:07
a happy Friday to you, wherever
0:09
you're joining us today.
0:10
It's the end of the week, you made
0:11
it.
0:12
The markets aren't looking too bad,
0:14
so plenty to smile about.
0:16
I'm broadcasting live from the
0:17
Netherlands. It's very hot today, so
0:19
if you're across Europe and you're
0:21
experiencing this heat wave, I hope
0:23
you survive, I'm doing my best to.
0:25
If this is your first time joining
0:26
us on the show, welcome to The
0:27
Startling Block.
0:28
This is The Block's daily live
0:30
show. We've been running for just
0:31
under two months.
0:33
Crazy to think, two months ago we
0:34
were in Abu Dhabi.
0:35
Sitting with CZ Live
0:37
to launch this fantastic
0:39
show. And it's been a fantastic
0:40
journey since, and we keep going
0:43
strong. Today, we'll be
0:44
joined a little bit later by Austin
0:46
Federa. He's the co-founder of
0:48
Double Zero. And of course, many
0:50
of you out there who are big fans
0:52
or users of Solana will know him
0:53
from his time at the Solana
0:55
Foundation as well.
0:56
So we've got Austin coming on the
0:58
show to talk about a big partnership
1:00
slash integration between
1:02
Double Zero and Kalshi.
1:03
Which will be a really interesting
1:04
conversation coming up in the hot
1:06
seat in a few minutes time.
1:08
But if you've never joined us
1:09
before, we always start the show
1:10
with the Block Briefing, which is
1:12
our rundown of the top headlines of
1:14
the last 24 hours curated
1:16
from the Block's fantastic editorial
1:18
team. So it's a great pleasure for
1:20
me to highlight some of the stories
1:22
that our news team have been working
1:23
on throughout the week and
1:25
over the last twenty-four hours.
1:26
So without further ado, let's jump
1:28
straight into our first headline of
1:30
the day. And I think probably one of
1:31
the biggest.
1:33
Pieces of news concerning stable
1:35
coins and Tether in general.
1:37
In the last 10 years, Tether has
1:40
cleared a milestone critics have
1:41
been chasing for years.
1:42
The stable coin giant says KPMG
1:44
US has issued an unqualified
1:47
that's a keen opinion on the
1:48
full 2025 financial statements
1:50
of Tether International, which is
1:52
the entity behind USDT,
1:55
the world's largest stable coin.
1:57
CEO Paolo Arduino told the
1:59
block yesterday it's the company's
2:00
first full independent financial
2:02
statement audit and that it covered
2:04
all of Tether's financials.
2:07
KPMG separately confirmed the work
2:09
for the world's largest stablecoin
2:11
issuer long dogged by questions
2:12
about its reserve.
2:14
A big four sign off is a
2:16
significant credibility marker
2:18
and just a few more anecdotes on
2:19
this story.
2:20
I heard that KPM actually
2:22
went into the bunker where Tether
2:24
is holding all of gold reserves and
2:26
they counted bar by bar.
2:27
The gold that's backing their
2:30
tokenized gold offerings that
2:32
they have.
2:33
I am actually going to be speaking
2:34
to Paolo Arduino a little bit later
2:35
this afternoon. We'll be sitting
2:36
down for a pre-recorded interview,
2:38
so keep an eye out.
2:40
On the block socials, we'll be
2:41
releasing some snippets of that
2:43
interview as we have it later
2:45
today. Unfortunately, he couldn't
2:46
join us on the show, but we
2:48
will be looking forward to chatting
2:49
to Austin Federer in a couple of
2:51
minutes' time.
2:52
Moving on to our next story,
2:53
Gemini's first earnings as a public
2:55
company. Sent shares sliding more
2:57
than 7% after hours Thursday.
2:59
The Winklevoss twins exchange
3:01
posted $107 million
3:04
net loss for the second quarter,
3:06
even as their revenue grew
3:08
by 37% to
3:10
45 and a half million.
3:12
The bright spots were outside crypto
3:14
trading, credit card revenue surged
3:15
231% to
3:17
$16 million, and
3:19
prediction market volume jumped 93%
3:22
quarter over quarter with cumulative
3:24
contracts topping.
3:25
$225 million.
3:27
Management struck a candid note
3:29
saying the company still has, quote,
3:32
work to do.
3:33
So an interesting one.
3:34
And hopefully we can get the
3:36
Winklevoss twins on the show in
3:38
the next couple of weeks for a catch
3:39
up. I met and interviewed them last
3:41
year at Bitcoin Amsterdam,
3:44
where they were launching
3:45
Cypherpunk, their Zcash
3:48
treasury company.
3:49
So we'll go knocking on their door,
3:50
and maybe we can dig a little bit
3:52
more into that story.
3:53
Our next headline, Peter Thiel
3:54
backed exchange bullish is still
3:57
bleeding red, but less of it.
3:58
The company reported a $280 million
4:00
net loss for the second quarter
4:02
narrowing sharply from $604
4:05
million in
4:07
losses in the first quarter.
4:09
It's adjusted revenue rose 62%
4:11
year over year to $92 million.
4:14
The pain point here, digital asset
4:16
sales fell 44%
4:18
dropping to 32.6
4:21
billion. From 58.6
4:23
billion.
4:24
It's a mixed scorecard for the newly
4:26
public exchange, top
4:28
line growth and a slimmer loss set
4:29
against a steep slide in
4:32
trading volumes. And I think, you
4:33
know, this financial report really
4:36
indicating the current state
4:38
of things in the crypto industry.
4:41
Then regulators drew a hard line
4:42
around Kalshi in Washington state.
4:44
A court issued a final order
4:46
requiring the prediction market to
4:47
stop offering contracts tied to
4:49
sports. Elections, politics,
4:51
entertainment, culture, technology,
4:53
and science.
4:54
Culture must geofenge
4:56
Washington, uses out of
4:58
those markets by the 2nd of
4:59
September, but it's not a total
5:01
shutdown. The platform can still run
5:03
commodities, climate,
5:05
economics, and finance markets in
5:06
the state, but it is another
5:08
regulatory blow for the event
5:09
contract space, which is fighting
5:11
battles on multiple fronts in
5:13
the United States.
5:14
And staying on that vein, the
5:16
prediction market legal parlor.
5:17
Continues to grow, Baltimore,
5:19
Maryland's largest city
5:22
has sued Kalshi and Polymarket,
5:24
alleging their sports events
5:25
contracts amount to illegal sports
5:27
betting.
5:28
It's not stopping at the operators,
5:30
Mayor Brandon Scott and the city
5:32
council roped in distribution
5:33
partners, Coinbase, Robinhood
5:35
and Weeble as defendants too.
5:38
The city is seeking penalties,
5:39
restitution and a court order
5:41
blocking what it calls unauthorized
5:43
sports betting for an industry
5:44
already under regulatory fire having
5:46
a major city name.
5:49
Mainstream brokerages raises the
5:51
stakes here. And just a really
5:52
interesting kind of situation
5:54
playing out in the United States
5:55
where prediction markets and
5:57
gambling kind of see some
5:59
blurred lines.
6:00
So we've been talking about this
6:02
plenty on the show in the last
6:03
couple of weeks.
6:04
Our final story of the day, a costly
6:05
reminder that the threat often
6:07
starts at the top of a
6:09
search page.
6:10
A hyperliquid user appears to have
6:11
lost roughly $550,000
6:14
in USDC.
6:15
After falling for a phishing scam
6:17
promoted through a Google search ad,
6:20
according to the co-founder of
6:21
crypto firm Flash Rescue.
6:23
If it's a pattern, security
6:24
researchers have been warning about
6:26
back in April, the non-profit
6:28
Security Alliance will seal,
6:30
said it had blocked 356 malicious
6:32
Google ad URLs in just a
6:34
few weeks.
6:35
The lesson, double check that
6:37
link before you connect a
6:39
wallet. And that really goes without
6:40
saying, just seeing the spate of
6:42
crypto crime in the last
6:45
year.
6:46
And recent issues with
6:48
lots of AI assisted
6:50
hacks and exploits going on
6:52
in the industry.
6:53
So take care of what you're doing
6:54
online and definitely double check
6:56
what you are clicking.
6:58
That's it for the block briefing
6:59
today. We're gonna move straight
7:01
into the hot seat this afternoon.
7:03
It's a great pleasure to welcome
7:04
back to, well, actually for the
7:06
first time to the show,
7:08
Austin Federer is of course the
7:10
co-founder of Double Zero.
7:11
And was previously with the Solana
7:13
Foundation. Austin, it's great to
7:14
have you on the show for the first
7:16
time. I keep saying that because
7:18
I've spoken to you so many different
7:19
times at my previous employer,
7:21
but this is the first you're on the
7:23
starting block.
7:24
So first things first, welcome to
7:25
the show. It's good to catch up.
7:26
How are you doing?
7:27
I'm good, congratulations on the new
7:28
show.
7:29
Thank you very much.
7:30
We've been working hard and it's
7:32
been a great two months so
7:34
far and just on the
7:35
up and up.
7:36
And once again, it's always good to
7:38
catch up with you and have you on
7:39
the show. So we appreciate your
7:40
time. Let's talk about the big
7:41
headline. We'll jump straight into
7:43
it.
7:44
You've just announced a big
7:46
partnership slash integration with
7:48
Kalshi.
7:49
They're coming to essentially use
7:51
Double Zero's infrastructure,
7:53
fiber infrastructure.
7:55
You can give us the TLDR a lot
7:56
better than I can. Give us, give us.
7:59
The long and short of it.
8:00
Yeah, certainly.
8:01
So what we're doing with Caoshi and
8:03
Caoshi research is basically
8:04
enabling users to subscribe
8:07
to every single
8:09
event contract and every single
8:10
state change with full market depth
8:12
delivered over double zero and
8:14
multicast. And there's a lot of
8:15
words to kind of unpack there.
8:17
But at its core, what this is is a
8:19
much faster way to receive
8:21
a complete view of all of
8:23
the data of what's going on on
8:24
Caoshi event contracts, including
8:26
liquidity depth.
8:27
So this is called L2 data in
8:29
the financial world.
8:30
And what this really provides is a
8:32
few different things.
8:33
So today, most people who consume
8:36
any type of data from this whole
8:38
category of new finance, whether
8:39
it's crypto and hyperliquid
8:41
and finance and Coinbase,
8:43
or whether it is prediction markets
8:45
like Cal-She, they're basically
8:47
using the same way that Slack
8:49
receives updates from your
8:51
coworkers.
8:52
It's something called a web socket
8:53
connection and they never really
8:55
were built for this type of thing.
8:57
In the traditional financial space,
8:59
you actually see this technology
9:01
called multicast as something that's
9:02
been an underpinning financial
9:04
technology for 20-plus
9:07
years at this point.
9:08
And the reason for that is it
9:10
delivers all of the data to traders
9:12
as quickly as possible, and it
9:14
does it in a much more equitable way
9:17
because all that data is delivered
9:19
at the same moment to all
9:20
the people in the same location, as
9:22
opposed to what traditionally
9:24
happens is you send one piece of
9:26
data out and another piece of out.
9:28
It's much more akin to like radio
9:30
and streaming TV.
9:31
And so in working with Cauchi on
9:33
this, what this means is
9:34
sophisticated traders, whether those
9:36
are large, high-frequency trading
9:37
firms, which believe it or not, are
9:39
actually very active on prediction
9:41
markets, or if it's just like a few
9:43
guys running a trading strategy
9:44
powered by AI agents,
9:46
we now allow them to consume
9:48
much more detailed information
9:50
and faster information about what's
9:53
going on on Cauchi.
9:54
And that's a big deal in these
9:56
markets, right, because literally
9:58
milliseconds will make the
10:00
difference between trades going
10:02
through or you being beaten to
10:04
an arbitrage or a specific trade.
10:07
Do you think this is kind of
10:08
indicative of how quickly the
10:10
prediction market space is maturing.
10:12
Absolutely. I mean, we have seen
10:14
prediction markets go from sort of
10:16
two or three years ago, like a weird
10:17
thing that a bunch of political
10:18
nerds did in their spare time
10:20
to like core financial
10:22
infrastructure.
10:23
You know, we, we have folks
10:25
in discords that are trying to
10:27
solve the exact same problems and
10:29
questions that an entire team
10:31
of experts at
10:33
a large Wall Street trading firm are
10:34
also trying to solve.
10:35
What is inflation going to be?
10:37
What is CPI numbers going to be?
10:38
What are the jobs numbers going to
10:40
be. And often times prediction
10:41
markets are getting it.
10:42
More accurate than the
10:45
traditional Wall Street firms are,
10:46
which is fascinating.
10:48
I think the argument that this stuff
10:49
is somehow not
10:51
part of core financial
10:52
infrastructure is really missing
10:54
the point.
10:55
This is taking something that
10:57
every single large financial actor
10:59
has had access to, which is the
11:00
ability to hedge exposure to
11:03
anything that is not by their own
11:05
design.
11:06
You've had hotels effectively
11:08
shorting the weather.
11:10
Or running prediction markets
11:12
for the weather through complex
11:13
insurance products.
11:14
What is crop insurance, except a
11:17
prediction market outcome on yield
11:19
harvests for a specific piece of
11:20
field?
11:21
And now, suddenly, a mom and pop ice
11:23
cream store can also hedge the
11:25
weather, because, of course, you
11:26
sell much more ice cream when it's
11:28
beautiful and sunny out than when it
11:29
is cold and rainy.
11:31
That's a really good anecdote,
11:32
actually.
11:33
I also wanted to just pick your
11:35
brain a little bit on how much
11:36
learning you've done about
11:38
prediction markets, obviously, in
11:39
the buildup to this partnership and
11:40
this integration, because there has
11:42
been sort of
11:44
the naive assumption
11:47
or belief that prediction
11:49
markets are just a dressed up
11:51
gambling, but that's not quite the
11:53
pace.
11:55
It's not true, it's not accurate.
11:57
Maybe you can give us your kind of
11:59
take on prediction markets and why
12:00
you think they are so powerful and
12:03
they are actually quite different
12:04
from just conventional gambling.
12:07
I mean, prediction markets are,
12:09
at its core, just a new type
12:11
of financial primitive that's now
12:13
available at a much lower
12:14
cost of access.
12:16
Again, we've had these type of
12:17
complex instruments for a while.
12:20
You could always go to Lloyd's of
12:22
London and get an exotic policy on
12:24
something extremely expensive to
12:25
hedge risk. They famously insured
12:27
the space shuttle.
12:29
And these sorts of things are
12:31
available at certain level of
12:33
sophistication and on Terry Premium.
12:35
That just has not been available to
12:37
individuals before or even small
12:39
businesses.
12:40
And so, from my perspective here,
12:42
you have to ask yourself, is it a
12:44
game of chance or is it
12:46
prediction based on skill?
12:48
I think it's very hard to look at a
12:50
political market or an oil market
12:52
or a crypto futures contract
12:54
and say, there's not skill involved
12:56
in trying to predict what the
12:57
outcome was.
12:58
If there was no skill involved and
13:00
it was purely luck, you would not
13:01
have people employed at many Wall
13:03
Street firms. Whose job it is to try
13:05
and predict exactly these things,
13:06
let alone people at intelligence
13:08
agencies who are trying to predict
13:09
the exact same thing.
13:11
And so for me, that's kind of a
13:13
pretty sharp line that separates
13:15
things that may be considered
13:16
something along the lines of
13:17
gambling from something that is
13:19
simply aggregating
13:21
human intelligence to try and
13:23
produce accurate outcomes for
13:24
financial gain.
13:25
That's the name of the game.
13:26
That's what capitalism is all about.
13:28
Yeah, and it's it's fascinating that
13:30
really that's what it comes down to
13:31
is that the the likelihood
13:33
of an outcome happening is
13:35
Directly correlated to
13:37
the financial positions that people
13:39
are putting in so literally the skin
13:41
in the game indicates how
13:43
likely something is to happen and it
13:44
it's incredibly powerful because
13:46
No one's gonna put money on
13:48
on a certain outcome happening just
13:50
just by chance You know other than
13:52
if they're willing to just gamble,
13:54
you know, like I used to work
13:56
in the sports industry and
13:57
I had a betting agency that
13:59
sponsored my trip to the Rugby World
14:01
Cup. And before I went to the rugby
14:02
world cup, I was like, I think South
14:04
Africa are gonna win this rugby
14:05
world Cup. So I'm gonna put a
14:06
hundred dollars on that.
14:08
That was a gamble.
14:09
I was right and it happened, but it
14:11
was a complete gamble.
14:12
And prediction markets, that's not
14:14
quite the same thing at all.
14:16
And I really loved the notion of,
14:18
yeah, the amount of money being put
14:20
on the table for a
14:22
certain outcome is very interesting.
14:24
And I think you need look no further
14:25
than even the US elections a couple
14:27
of years ago, and you know,
14:29
Polymarket and kind of Carl Shee,
14:31
you know indicating to us that
14:33
Donald Trump was going to win when
14:34
in fact, people tracking the polls
14:36
and conventional news websites, etc.
14:38
Would be telling you a different
14:39
story.
14:40
And if that doesn't tell you
14:41
everything you need to know,
14:43
then I don't know what does.
14:47
I do want to keep just talking about
14:48
prediction markets for a little bit
14:49
longer here.
14:50
When it comes to the infrastructure
14:52
that you have built, maybe you
14:53
can just give our audience that
14:55
don't know Double Zero a quick TLDR
14:57
on what you do and
15:00
why and how this kind of
15:02
prediction market angle came out to
15:03
you. Because as you were saying,
15:04
Orfe, it's very interesting to move
15:06
away from the crypto use case that
15:07
you imagined when you found
15:09
the Double Zero to what
15:12
kind of primitives are now being
15:13
built on top of that infrastructure.
15:15
Certainly. So under the hood, double
15:17
zero is a blockchain protocol.
15:19
It the sort of core network
15:22
of the double zero network
15:23
is contributed by
15:26
17 different individual companies
15:28
at this point, saying the same way
15:29
that there are multiple entities
15:30
that run validators for
15:32
Solano or Ethereum or hyper liquid.
15:34
There are multiple companies and
15:35
entities out there that run fiber
15:37
for the double-zero network.
15:39
And so what we build and what the
15:41
folks at the labs company build is
15:43
really the software system that
15:44
allows these individual pieces of
15:46
fiber to mesh together and
15:47
talk to one another.
15:48
And the whole purpose of this is to
15:50
go faster than the public internet
15:52
allows data to move and to
15:53
use technologies like multicast that
15:55
are simply not available on the
15:57
public Internet. And so this whole
15:59
market data feed, which is called
16:00
Double Zero Edge, is just one
16:02
of many different products and
16:04
services that are built on top of
16:05
the Double Zero Network the same way
16:07
that a DEX or a Borrow-Len protocol
16:09
might just be one thing built on a
16:11
Ethereum.
16:12
And so for us, this
16:14
high-speed data movement for
16:16
market data, it can carry all types
16:17
of data, whether that is equities
16:19
data or commodities data,
16:21
securities data.
16:23
Anything that is a
16:25
data feed that people would like to
16:27
consume rapidly for any
16:29
reason at all can be transmitted
16:30
over the double zero network.
16:32
And so you hear a lot
16:34
on now is like, oh, are
16:36
these event contracts commodities?
16:38
Are they securities? Are they
16:39
something else?
16:40
It really is agnostic to the double
16:42
zero network because we
16:44
run smart contracts under the hood
16:45
to provision access.
16:47
And so any type of granular access
16:49
control that a data feed needs,
16:51
we can program that into the base
16:53
layer. And you might assume that
16:54
like, you can do this for feeds from
16:56
traditional finance and Wall Street,
16:58
and we can, but none of the existing
17:00
providers for this type of thing
17:02
actually have that level of smart,
17:04
intelligent routing and control.
17:05
Most of these are still.
17:07
Physical cross-connects and paper
17:08
contracts and these sorts of things.
17:10
And so really, the Double Zero
17:12
Edge product is all about expanding
17:14
access. And a lot of this is
17:15
targeted at AI agents, that a
17:17
huge number of people are trading
17:19
prediction markets, as well as other
17:20
types of digital assets or digitally
17:22
accessible markets, through
17:24
AI agents now.
17:25
And so you can go to Double Zero.
17:27
You can buy access to calcium data
17:28
or Solana data or other data
17:30
coming soon.
17:31
And your agent can now ingest
17:33
that feed and set up trading
17:35
strategies.
17:35
And it's an entirely new way
17:37
both to think about trading and
17:39
to think about the data that powers
17:41
that trade.
17:43
I mean, I don't mean to make this a
17:44
competition, but it kind of sounds
17:46
like you're edging into competition
17:47
with, you know, players like
17:49
Chainlink. Is it kind similar,
17:51
some of the offering and the data
17:52
that you're putting out there now?
17:53
So there's a pretty good distinction
17:55
between these two things.
17:56
And so if you look at like what
17:58
chain link or pith or an oracle out
17:59
there is doing, they're trying to
18:01
look at sort of the aggregate market
18:02
and tell you what a good price
18:04
is, right?
18:05
Like you see this price on
18:07
this deck or this website.
18:09
How does that relate to the average
18:11
price of Bitcoin on the market?
18:12
What double zeros focused on is
18:14
quite different. It is what is the
18:16
price at this specific venue so
18:18
you can make your own decision so.
18:20
In a mature end state of double
18:21
zero, you could imagine we would
18:23
have feeds from multiple centralized
18:24
exchanges.
18:25
Whereas what Chainlink and Pith do
18:27
is they aggregate that data together
18:29
and they give you an editorial
18:31
perspective.
18:32
We think the price of Bitcoin today
18:34
is $65,283.34
18:37
with a certain amount of confidence
18:39
interval.
18:40
What double zero instead will give
18:41
you is this is what Coinbase thinks
18:43
the price is. This is what Binance
18:44
thinks the prices.
18:45
And for a lot of traders.
18:47
What they're looking for is actually
18:49
a better purview about what's
18:51
going on across multiple venues.
18:53
And arbitrage is the name of the
18:54
game, oftentimes, in
18:56
that. And so as we start to
18:58
see hip four markets and hyperliquid
19:00
take off, as we started to see this
19:02
sort of diversification of
19:04
the prediction market, you might
19:06
find, hey, I have a high conviction
19:08
about this thing.
19:09
And you know what? Polymarket
19:10
actually, other traders
19:12
have the same conviction as me about
19:13
it. Maybe that's not the case on
19:15
calcium. So maybe you have an
19:16
opportunity there to say.
19:18
Okay, I've reached confidence
19:19
on a prediction that I would like to
19:21
make where actually is the best
19:23
market for me to express
19:25
that outcome potential, whether
19:27
it's in terms of liquidity or
19:28
whether in terms of how far I think
19:30
the market is off from the truth.
19:33
I want to talk to you a little bit
19:33
about AI agents and
19:35
that convergence now.
19:38
Anyone who's spun up Claude
19:40
and is using frontier models,
19:42
I think, would be looking for
19:43
different trading strategies,
19:45
especially on prediction markets,
19:47
because you can do a lot of
19:48
interesting things there.
19:49
And we all read these outrageous
19:50
headlines about crazy trades
19:52
and insights and money that people
19:54
have made on prediction markets.
19:57
And the block also
19:59
has Prediction Markets News, our
20:00
recently launched podcast that's
20:02
dedicated to the sector.
20:03
So we're watching this stuff kind of
20:05
very closely.
20:06
But when it comes to the AI
20:07
integrations, how important is
20:09
infrastructure like
20:12
Double Zero Edge in
20:14
giving people and their agents an
20:16
edge over other people?
20:18
I feel like it could
20:20
make or break some trading
20:21
strategies.
20:22
Yeah, it's hugely important.
20:24
I think, you know, if you'll allow
20:25
me a digression for a second, one of
20:26
the things that we have seen with
20:28
the rise of these frontier AI
20:30
companies is.
20:32
For the first time in probably 20
20:33
years, the majority of people who
20:35
work in tech are starting to
20:36
understand that the servers and the
20:38
infrastructure truly matters.
20:40
One of the beautiful things about
20:41
the cloud is that it abstracted all
20:43
of that away from you now.
20:44
But you can't abstract away
20:46
a frontier model.
20:47
You need access to GPUs
20:49
in very high performance data
20:51
centers. And the physical location
20:53
of those data centers matters
20:54
because the speed of light is
20:56
very fast, but the speed light is
20:58
nowhere near fast enough.
20:59
And so the type of people now who
21:01
are trading with AI agents
21:03
on prediction markets or traditional
21:05
markets, they're thinking about
21:07
where physically is my agent
21:09
running. Is it physically close to
21:10
the matching engine for Kalshi?
21:12
Is it physical close to matching
21:13
engine for a NAZI or NASDAQ?
21:15
Or is it running in an abstract
21:17
environment somewhere that's pretty
21:19
far away?
21:20
And so with double-Z, we're going to have to look at that.