October 9, 2024 • 43 mins
Nate and Maria give a degenerate gambler's guide to the election, covering everything from Nate’s $100k gamble on Florida to listener-inspired prop bets. Maria explains why poker bots are shaking up the world of online poker. Plus, Nate and Maria lament the spread of AI “slop” online, and re-evaluate their p(doom) in light of recent developments at OpenAI.
Further Reading:
Should Kamala Harris gamble on a Blue Florida? from the Silver Bulletin
The Russian Bot Army That Conquered Online Poker from Bloomberg
A Further Investigation into the Existence of the BotFarm Corporation from Poker Pro
Poker-Bots: The Call Is Now Coming From Inside The House from Legal Poker Sites
For more from Nate and Maria, subscribe to their newsletters:
The Leap from Maria Konnikova
Silver Bulletin from Nate Silver
See omnystudio.com/listener for privacy information.
Mark as Played
Transcript
Episode Transcript
Available transcripts are automatically generated. Complete accuracy is not guaranteed.
Speaker 1 (00:15):
Pushkin. Welcome back to Risky Business, our show about making
better decisions. I'm Maria Kanakova.
Speaker 2 (00:29):
And I'm Nate Silver. Today on the show will be
giving you a degenerate gambler's guide to the election, including
a big proposed election bet which has not been reciprocated yet.
Speaker 1 (00:39):
After that, we're going to talk about the threat of
pots and automation to one of the things that Nate
and I hold dear to our hearts, poker and online
poker specifically, and then.
Speaker 2 (00:51):
We'll re evaluate p doom, the probability of severe misalignment
from artificial intelligence in light of recent news at Open AI.
This is a good episode, Maria. I feel like we
were kind of hitting all the basic angles today.
Speaker 1 (01:10):
I feel like we are too, so so let's uh,
let's get it started. Nate, I couldn't help but notice
a recent exchange on Twitter that you that you engaged in,
As so often happens, your views on Twitter were challenged.
I know this is this comes as a shock to
everyone every time I hear it. You know, people don't
(01:32):
always agree with you, and this particular challenge was about
Florida right and about the likelihood what percentage by what
percent Trump was going to carry Florida. So do you
want to do you want to talk us through a
little bit what happened and why we're why we're starting
our show with this particular Twitter exchange.
Speaker 2 (01:53):
Yeah. So we published an article at Silver Bulletin late
last week pointing out that that Florida's kind of close
in the polls, Harris is down by about three and
a half points, similar to Trump's margin of victory in
twenty twenty, and that maybe the Harris campaign, you have
infinite money, it's worth actually investing in the Sunshine State.
(02:15):
The American investor Keith Erboy, He's like, you're an idiot.
Trump's gonna win Florida by This is not the literal
verbatim version, right, You're an idiot. Trump's gonna win Florida
by at least eight points, if not ten or twelve.
And I'm like, okay, bro, how much are you willing to, like,
actually wager on this? And he says one hundred K,
which seems like a seemed like a good amount to me. Right,
(02:37):
it's a serious it's a serious adult bet. And by
the way, there's a chance that Trump will win Florida
by eight points or more if you look at polymarket,
about twenty five percent. But I'm willing to take one
hundred k bet where I win seventy factions of the time.
And so I said, okay, you got to send me
a contract though, because you've been sketchy lately, and I
don't really you know, he's not like a known I mean, ironically,
(02:59):
like known gamblers are like more reliable for payout stuff, right,
But I don't want him saying like, oh, the election
was stolen. I don't want him saying, oh haha, I
was just joking. So need some type of third party
to step in and create a contract or at least
create an obligation expectation. So far that has not happened.
We had a little bit of negotiations by back channel email.
(03:22):
The mutual friend said I'm willing to do this and
then and mister Riboy has not replied yet, so I'm
not sure kind of how much more time I want
to give my I don't like giving people free options, right.
You know, if there's a poll that comes out this
week showing Trump way up in Florida, then all of
a sudden, his bet becomes not plus TV but better,
but Yeah, So anyway, I think this is an honorable
(03:45):
way to resolve differences, right that if you actually have
to pay a tax for your bullshit on Twitter, right,
seventy five percent chance of losing one hundred K, then
that to me seems like you know, I think the
world will be better off if people spent more time
betting and less time trolling on Twitter.
Speaker 1 (04:02):
Absolutely, I think that there should be a bullshit tax,
for sure. And just so that the the audience is
up to date, this is not the first time you've
been challenged. To date, not a single person has actually
taken you up on your Nobody ever, actually, nobody has
goes through Yeah, nobody has Heath ever done.
Speaker 2 (04:24):
To a further step than most people do, right, So
give them credit for that. But like, but yeah, they
almost always back down because if I'm making a bet,
they're typically carefully. I'm not trying to eke out you know,
one or two percent ev bets right, in part because
I have a lot on the line anyway as far
as the election goes, right, So I'm not trying to like,
but if you're gonna give me a seventy five twenty
(04:46):
five coin flip for one hundred K then I feel
like I would be negligent to pass up on that.
Speaker 1 (04:52):
Yeah, absolutely, So you know, tune in and uh and
we'll we'll see how that goes. But the news for
people who do want to make bets like this not
on Twitter with Nate Silver, but in general on elections.
We talked a few weeks ago about Calshi, which is
(05:13):
a prediction market that actually operates within the United States
real money bets and Calshie had been trying to for
years basically make their way through the US court system
to make election betting legal in the United States, to
make those contracts legal here, and they had prevailed, and
(05:35):
then the order was frozen, right, so their options on
the election went live for a few hours and then
they were no longer allowed to continue while the Court
of Appeals reviewed this, and they have now ruled in
favor of Calshie, and so people in the United States
(05:56):
are once again allowed to bet on election outcomes. I
actually got a very funny email from Interactive Brokers, which
is an online trading platform on which I happened to
have my retirement account, and they sent an email to
everyone saying, hey, guys, like you can now bet on
(06:16):
all of these election things we're going live, So it's
something that immediately it seems like a lot of platforms
were ready for this and are really excited to start
taking these sorts of bets. But we actually had some
listeners who were interested in our takes on a few
different bet ideas. So one listener, David Paul, wrote in
and asked, would we bet that twenty twenty four voter
(06:36):
turnout will be higher or lower than in twenty twenty
and why? And I thought that that was actually kind
of an interesting bet proposition because you and I have
discussed this a little bit in terms of how close
an election is perceived to be and what the turnout
is as a result of that, and also obviously how
(06:56):
passionate people are and how engaged people are, right, and
different different subsets of the population are more likely to
actually go out and vote with their feet at the
polls as opposed to vote on Twitter talk about how
you know one candidate or the other, but then they
don't actually go and cast a ballot. So what do
we think about this? What do you think about this?
Speaker 2 (07:18):
Well, let me ask an important question. Are we talking
about the raw number of voters or the share of
the eligible voting population, I don't know.
Speaker 1 (07:27):
Let's let's let's discuss because because I actually don't know
what has happened with voter registration. Has that gone up
with you know, Taitay's endorsement and all of that. Are
we actually seeing a much larger number of voter registrations
than we saw four years ago?
Speaker 2 (07:43):
Look, so, first of all, I would probably bet on
turn up being a bit lower than in twenty twenty,
especially with you know, if you are accounting for population growth,
then it might be kind of a tass up. But
if you're looking at the share of the population in
part because like twenty twenty was a bit of an outlier,
sixty six percent of eligible voters cast ballots. That was
up from sixty percent in twenty sixteen and fifty nine
(08:06):
percent in twenty twelve. And this is data from the
University of floor or as Election Lab website. But we
have had an increase, is you know? So turnout in
the twenty twenty two midterm was forty six percent, which
is high for a midterm, but down from twenty eighteen. Right,
you know, when it was Biden versus Trump, people were
distinctly unenthusiastic about this selection. With the events of the
(08:28):
summer with Harris taking over for Biden, then there's more
enthusiasm now. But but you know, twenty twenty turnout was
really high, in part because it's a pandemic. There were
more ways to vote than ever before. And I would
vote on I know, sixty three percent of the voting
eligible population instead of sixty.
Speaker 1 (08:42):
Six I suppose, so you would put your money on
on lower turnout. And to be clear, we're not actually
making a bet, We're not actually putting money on the
line because this is not a topic that either one
of us has probably a lot of conviction. And do
you have to.
Speaker 2 (08:58):
Actually have a model. So the model predicts, oh, there
we go, one hundred and fifty six million, six hundred
and eighty five thousand, six hundred and eighty people exactly
will vote. Okay, I obviously it's fake precision. Let me see, uh,
let me see what the numbers were in twenty twenty.
They say one hundred and fifty eight uh point four
(09:23):
million people voted, and I'm predicting our mouths one fifty
six point seven, so that would be off by just
a million or two.
Speaker 1 (09:32):
Okay, but lower but lower. So so your your gut,
your gut was correct, and now the silver bulletin model
has validated your gut at least as of now. And
these numbers might change right as you get more data.
Speaker 2 (09:46):
Yeah, no, I mean, so what we look at is
polls asking people how enthusiastic they are about voting in
the election, and people actually say they're pretty enthusiastic. However,
turn out with just really high in twenty twenty, So
until we see more confirmation of that than we're expecting
some reversion to the mean.
Speaker 1 (10:04):
Great, well, if we're if we're going down that kind
of Djen rabbit hole. I actually think it's in interesting
because it does kind of touch on different pressure points
of the election. So some other ideas, what's our over under,
for one, major news outlets will actually call a winner.
I think that's an interesting one because I don't think
either one of us particularly feels like the election is
(10:24):
going to be over on November fifth, given how close
it is and given the the way elections have gone
in recent years.
Speaker 2 (10:32):
Yeah, look, I mean the over under might be you know,
November fifth is a Tuesday, maybe Saturday, excuse me, Thursday morning.
The seventh. I mean keep in mind that, like the
networks send to be really conservative about making these calls,
except for Fox News, which weirdly called Arizona for Trump
when it was great premature to do so. But leave
that aside for a moment. Yeah, I remember doing in
(10:53):
twenty twenty. I was kind of like like in the
boiler room, it's during COVID doing like these are PEATV
hits over and over for ABC News, and like the networks,
it was obvious. I thought by like midday Wednesday that
Biden was going to have enough past to victory, but
the networks didn't call it until Saturday morning. But I
don't mean to make light of this. I mean, like,
(11:14):
you know, another disputed election outcome where Trump is sowing
doubt about the results, that would not be great for
the US. I think people are not quite pricing in, uh,
what's going to feel like if you have it, if
you have an unclear, ambiguous, disputed election result.
Speaker 1 (11:33):
Yeah. I mean, I'm actually hoping that we do get
a winner announced on the fifth or you know, on
early morning of the sixth. I hope that we do
have a very clear cut election, because I think that
that is something that we need because we have you know,
we've had some close calls and we've had a number
of disputed elections in recent memory. So so that's not
(11:54):
necessarily a great place to be when you're looking at
kind of when you're looking at democratic out outcomes and
peaceful transitions of power. No.
Speaker 2 (12:03):
I mean, look, you don't need that much of a
polling er to have a pretty clear all right, if
Trump beats his polls by three points, then he wins
all the swing states. Likewise, for Harris, you could even
win Florida, which was a source of dispute with Keith
or boy. So, yeah, you know, on what other prep
bets do we have? Yeah?
Speaker 1 (12:24):
So, so there were a few others. One was in
which of the swing states will polling to be turned
out to be the most accurate and in which will
it be the least accurate? So do you have as
ours our pole guru, do you have a sense of
how the polls are doing this year?
Speaker 2 (12:41):
I mean, maybe Georgia for most accurate. The polls have
tended to be pretty accurate in Georgia in the past.
And also like like in Georgia, like you kind of
know everyone's going to vote, right, you have like the
evangelical whites, the African American population, and like you have
the you know, suburban so it's not like it's just
a matter of kind of racing to see who can
get more turnout, and that tends to make the polling
(13:02):
a little bit easier versus states where you have like
a lot of swing voters. But I'll go with Georgia
as the most accurate polling state. Do you have an opinion, Maria, No, I.
Speaker 1 (13:12):
Don't because I don't know. You know, I don't follow
the polls as the way that you do. I just
look at your outcomes, and I tend to trust you
obviously when it comes to pulling red when it comes
to reading the polls, so I just kind of look
to see what the prevailing wisdom is. But I don't know,
you know, I think that it's one thing to say, oh,
(13:34):
you know, these are high quality polls, these are worse
quality polls, you know, et cetera, et cetera. But I
think it takes expertise that I, frankly do not have
to be able to say how accurate is polling within
a state.
Speaker 2 (13:46):
I'm so torn on deferring to X, you know, because
I want to how your friends like Keith r who
clearly is outside his depth in this particular enterprise. A
lot of the time the experts are full of shit
and the non experts if I'm more full of shit? Right,
Life is hard. It's hard to make predictions about things.
Speaker 1 (14:03):
It is hard to make predictions. And let's do a
shout out to the work of Phil Tetlock, who was
one of the first people to show how hard it
is to make predictions and how you know experts are
also often wrong. And but if we can't actually I'm
curious to hear the answer to this. If we can't
defer to experts, then you know what, who do we
(14:23):
defer to? I mean, there's obviously, you know, going back
to what we started off with, you know, with prediction
markets and places like cal shan As. We know you're
paid advisor for polymarket. I just I hate having to
throw that disclaimer out every single time, but but I
think we do have to throw it at every single time.
(14:44):
Our prediction markets a better place, Like what do we
defer to? Then if we can't, if we say, like
you know, you can't really defer to experts, how do
you figure out you know, how do I form an
opinion on this.
Speaker 2 (14:55):
Yeah, look, obviously my fan of prediction markets. I think
I think one of if you read Phil Telllock's book
that you reference, We're Expert Political judgment. I mean, you
know that actually forced experts to make probabilistic predictions, right,
Like what's the chance that the USSR will break up
in the next ten years might have been a question
that he asked. For example, when you encourage people to
do that, then you actually create more accountability, You create
(15:17):
a paper trail because the most part like experts don't
actually have incentives to make accurate predictions, you know what
I mean, they have incentives to be ambiguous so they
can't be proven wrong. And so you know, stuff like
what we do at Silver Bulletin is actually quite unusual,
and if experts were forced to come up with actual
answers then they would do they would potentially be a
(15:39):
little bit better, although I think we'd also expose people
who are not very vigorous in their thinking.
Speaker 1 (15:43):
Yeah, so then maybe we would be able to have
kind of a hierarchy of expertise, which I think is
something that is very would be very very helpful, at
least to someone like me the other thing that happened
in the news, kind of putting our little prop bets aside.
Obviously we had the VP debates since you and I
last spoke. Did that make any difference whatsoever? I know
(16:05):
you've said over and over and over that you know,
the VP debates don't really matter, piece don't usually matter.
Have we seen them matter at all in the last
week or no change.
Speaker 2 (16:15):
There's been oddly little polling of any kind. I think
the polsearch just all decided to take the weekend off.
I don't know, man, you can let me say something
that will get clipped onto the Twitter, right, Like, I'm
not into Tim Walls really, I mean he's fine, but like, uh,
you know, first of all, there's all this discussion about like, oh,
(16:37):
you don't want to the term sain washing. It's like, oh,
you don't want to make Trump Advance seem insane, and
he's like he's doing this whole nice guy thing. And
I think Vance is just objectively like a better debater
than Tim Walls. And Tim Walls was obviously extremely nervous
for the first fifteen minutes of the debate, which is
a part that people often remember, like it won't matter
I don't think very much of the end. The polls
(16:57):
showed that people thought the debate was a tie basically,
but I don't know, you know, it's kind of you
have your one job thing and you should really nail
that debate. And he I think it was tactically not
very good at going after JD. Vance and I'm I'm
just not into it. I don't really understand, you know what.
(17:17):
There's like feeling of euphoria after Harris replaced Biden, and
I think people therefore overrated everything she did in that
period after that. And if it were me, I would
rather have, say, the extremely popular governor Pennsylvania, which is
the most important tipping point of the state.
Speaker 1 (17:35):
Yeah, well, let's see what ends up happening, and after
the break, we'll talk about betting on something else, which
is betting on cards and poker bots. Nate, if I
(18:02):
remember correctly, way back when you made a living professionally
playing poker online playing limit hold him? Is that correct?
Speaker 2 (18:13):
That's correct? For preire to two for three years, I
guess three years. Yeah, my main source of income was
playing fricking party poker.
Speaker 1 (18:24):
And when was this let's put it on a timeline.
Speaker 2 (18:27):
It's two thousand and four to two thousand and six,
and then then I had a year where let's not
tell them, but I actually lost money in two thousand
and seven after the game's cut. So the US Congress
passed a bill called the U I G E A
EUG U G. Yeah, and once that happened, the game
started to dry up and Party Poker closed. And but anyway, yeah,
(18:47):
I'm actually I was able to change my whole career
by playing online poker basically.
Speaker 1 (18:53):
Okay, So the reason that I asked about this is
this past week, we've had several blockbuster investigation pieces come
out about bots in online poker and bots as kind
of an existential three to the game. So the first
of these came out on September twentieth, and it was
(19:14):
a piece by kit Chilo, who has been doing this
deep dive investigation into a massive, massive Russian bought. He
calls it an army. So usually in the online poker
world we talk about bot farms, but when you read
this piece, you realize that, holy shit, army might actually
be a legitimate way to look at this. So he
(19:38):
had kind of found a few breadcrumbs that then led
him to poker forms to people who actually were involved
in this bot operation. And it turns out that this group,
originally this group of students, then it grew much bigger,
is working, you know, in the depths of Siberia, and
(20:00):
these are stats nerds, college students who need extra income
and they want to get better at poker, at online poker,
and so they start kind of playing around with basically
what you know today we'd call solver software. Then they
start developing little bots to try to implement some of
these ideas. Fast forward to kind of the capabilities that
(20:21):
these bots can actually be deployed onto poker platforms, and
these bots start being used and start winning money on
online poker sites in great numbers. And then it turns
out that some online poker sites might even be deploying
bots themselves in order to generate artificially larger player pools,
(20:43):
because if you're an online poker site, you don't particularly
care if the person playing the rag is a little
bot or a human as long as you get the money.
And so then after this Bloomberg piece came out, a
second piece came out by Jonathan Robb and he investigated
further and actually named names of some online poker companies. Anyway,
(21:03):
this is a very long introduction, so we'll talk more
about this later, but it seems like we see quite
the existential threat to online poker and to players and
it like yourself who look at online poker as a
way of making money. So what what do you think?
You know, have you been following this poker BT explosion
at all and have you given any thought to kind
(21:24):
of the state of online poker play.
Speaker 2 (21:27):
I mean, first of all, I don't look at online
poker as a way to make money. I look at
it as a form of recreation, I suppose, because like
to me, it seems like a somewhat unsolvable problem that
you have solvers now that can that can give you
(21:48):
a game theory optimal solution to a poker hand, never
mind anything more advanced about like doing data mining particular opponents.
You know, to me, unless you're interested when you're actually
having somebody like physically on a camera or something. Then
you know, combined with the fact that, like as it
says the Bloomberg article, it's kind of a Darwinian struggle
(22:11):
to begin with online poker, right, people are already playing
pretty close to optimal that I just don't. I wouldn't
want to have to live my life now trying to
make money playing poker online. And and there are home
games that people say, but I just think the risk
is pretty is pretty high.
Speaker 1 (22:29):
Yeah, And when we when we look at kind of
these further investigations. So this second article that I alluded to,
the Jonathan rob piece in Poker Pro, he actually found
kind of this one site, jack Poker that seemed to
be almost entirely creating their playerpool out of bots, which
(22:50):
is kind of crazy, right, And he you know, signed
up as a real person with real money, and then
he noticed these crazy patterns in their tournaments, et cetera.
Went down that rabbit hole and actually saw that they
were connected to their user interface was connected to this
corporation that Chile had wrote about, the Bot Farm Corporation BFC,
(23:13):
and what kind of they're deep play software, which is
kind of these bots that they were deploying. And if
it's true that sites like that are actually kind of
embracing bots in some ways like that to me is
quite worrying. By the way, time for my own disclaimer,
as people know, but I should say I'm an ambassador
for Poker Stars, which is also an online poker site.
(23:35):
Poker Stars was not implicated in any of this, but
you know, I should, I should throw that out there.
Speaker 2 (23:42):
I mean, like there are incentives to not have perceived
or actual cheating. I suppose it just that, you know,
I mean when you can like, I don't know. I
mean it's I tried to play poker against chatchipt once.
Speaker 1 (23:55):
Really, how did that go?
Speaker 2 (23:57):
It's very nitty. No matter what options you give it, it
always just checks the pot down, even if it has
like the nuts. So you know, maybe we still have
some time for humans to be better than Chetchapet at poker.
But like, but you know, it's an asymmetric arms race.
It's just hard for me to imagine that someone who
is connected to a computer can't find efficient ways to cheat. Basically, yeah,
(24:22):
I mean.
Speaker 1 (24:22):
This is clearly a real problem, and there are I
do think there are solutions. By the way, you can
usually figure it out when you're looking at deep analysis.
So Nate, as a statistician and someone who does this
kind of analysis, you can probably appreciate that if you
have a database of you know, hundreds and thousands of hands,
you can start seeing is this a person or is
(24:44):
this a computer that has some sort of regular type
of output and you know, flagging and getting rid of
those things immediately is a is a big one the
other one, And I don't think anyone wants to do this,
but I think it could. I think it's potentially kind
of the future of a safe poker environment is what
(25:05):
if we play on camera? Right? Like, what if you
have those types of checks where you actually try to
monitor who's playing, how they're playing, et cetera. And even
that's not perfect obviously, but there are certain things that
you can start doing. But it takes a lot of resources,
and the incentives, unfortunately, are often with the players. And
as you said, it is an arms race, like we
(25:26):
can learn to try to figure out, oh, what's this
kind of bot look like? What does that kind of
bot look like? And then someone comes up with something better,
someone who deviates or you know, something that looks more
human like like the Turing test. Right, every single year
becomes more and more likely that we'll actually have someone
be able to beat the Turing test. I think all
of our listeners probably know what the Turing test is,
(25:48):
but do we want to pause for a second to
explain what that is for people who might not know.
Speaker 2 (25:56):
Yeah, I think probably most people know. But the Turing
test is basically, can you program I mean it's actually
anticipates chat GPT pretty well, right, but can you tell
if you're having a conversation with a human being versus
a computer? And yeah, no, I mean like CHUTCHIPT, I
mean it has signature for example, frequently says as a
(26:18):
large language model, which is something that like. But the
thing even that though, like, you know, so if you
do play online poker and you're playing a lot of
hours and sometimes you get a prompt saying prove that
you're a human, right, do some task now that proves
that you're a human and not just a bot. But
the thing is, though, like you could intentionally introduce some
degree of like randomization. I mean, one thing about cheaters
(26:41):
in general is that most cheaters are really greedy, or
at least the ones that get caught right, and so
they cheat too much. When if you were a smart
cheater and poker could just kind of get away with
the edge cases where you magically happen to know in
the spot where you're a different what your opponent has
on the river, And instead people get very greedy, but
I just think it's kind of asymmetric warfare. In the end,
(27:02):
life poker is good, though, yes, more life poker.
Speaker 1 (27:06):
I agree. I think life poker is certainly a for know,
though not altogether safe. You know, we've we've had live
cheating scandals as well.
Speaker 2 (27:14):
And this has always bugging me at like poker is
quite forgiving. I think of cheating other bad behavior and
ways like yeah, it should just be banned for many
casino for lifetime or for five years and for things
like that, and instead they're kind of like slap on
the wrist at the time. Yeah, you know, Isaac Heckson,
I talked to you for the book, one of the
best players in the world, and told me that his
(27:37):
ROI and online poker is about ten percent, which is
not actually very good, or said under ten percent in
the single digits, right, So forever, forever, thousand dollars he
invests in playing a tournament, that's one of the best
players in the world, you know, maybe he has an
expected return of fifty or sixty dollars from that, Like
that's actually, given the variants of poker, not a very
(27:58):
good deal, right, And if he one of the ten
to twelve best players in the world is only making
that much, then I I don't think very many people
are really beating online poker sites without cheating in the
long run.
Speaker 1 (28:09):
Yeah, I think that that's that's very sobering. I did
not I did not realize that that I could communicated
those numbers to you if if one of the best
players in the world says that, but I will caveat
that was saying. Mike also plays in a lot of
much harder games, right, He's probably not playing in some
of the lower buying tournaments that you know, you and
(28:29):
I would would play in. He's probably playing the high rollers.
Speaker 2 (28:33):
But still, it's pretty it's pretty frightening.
Speaker 1 (28:35):
Yeah, for sure, for sure. So I'm very curious to see,
you know, what the fallout is going to be of
all of this, whether sites are going to address it,
whether they're you know, what the what the changes are
going to be. There was you know, for instance, I
the last poker live Poker I played was European Poker
Tory PT Barcelona, and at that stop that happened right
(28:57):
after there were some cheating allegations. We we we don't
have to go into into details on those, but of
live poker and of certain methods of being able to
kind of see cards live at the table. And so
what Poker Stars announced at the stop was some new rules,
(29:21):
including no phones on the tables, no phones on the rails.
And I think that, you know, things like that actually
show that, Okay, we're we're trying to take some of
these threats seriously as they're evolving. And I went to
step further and I said, you know what, like we
should ban phones at the table period, We should ban
(29:41):
sunglasses at the table. You know, there are certain things
that like we just know that there's cheating technology that
can be used in those instances. And I love having
my phone, but you know, I'm willing to put it
on the floor and to not touch it for an hour,
Like I'm okay with that, And I actually think it
would be very good for everyone's mental health to not
(30:02):
have your phone on you.
Speaker 2 (30:03):
Twenty four se day, Day twenty six of fifty two
with the Worldter's of Poker, You're playing in some eight
hundred dollars freeze out ten handed with the guy from Buffalo,
who I mean is taking fifteen seconds to fold. I mean,
you want your fucking phone, don't you?
Speaker 1 (30:20):
I do want my fucking phone, but for the greater
good of the things.
Speaker 2 (30:24):
Why I'm making fun of Buffalo by the way, we
love Buffalo.
Speaker 1 (30:29):
Oh, Buffalo has taken some flack from us in the past.
I just remembered because you were.
Speaker 2 (30:33):
Forgot the unexpected layover.
Speaker 1 (30:35):
Yeah, you had a Buffalo layover anyway, Sorry, Buffalo, You're great.
But you know I would I would take one for
the team and be willing to give up my phone
a little bit if it meant you know, game integrity
and things that are kind of safer for the game. So,
you know, I think, But just to wrap this up,
(30:55):
I think there are existential threats. We need to address them,
and we need to figure out kind of the best
way forward because I think you and I both feel
that poker is a great game and a game that
can really kind of help people think better, make better decisions,
and be great if people could safely continue playing in
all in all forms, and I hope that happens.
Speaker 2 (31:17):
Speaking about existential threats, you want to talk about open
AI real quick?
Speaker 1 (31:21):
Yeah, So now we've been talking about existential threats to poker,
and after the break, let's talk about open an AI
and revisit some of those existential threats day a while ago.
(31:49):
I think the last time we spoke about the concept
of p doom was what a few months back, right
where we talked about the risk that AI poses to
the world. Last week, in the last week, there's been
a lot of news coming out of open Ai, a
lot of new departures and a lot of fundra saying.
Speaker 2 (32:11):
So yeah, open ai raised six point six billion, we
are told by our reliable producers, at one hundred and
fifty seven billion dollar valuation. It is losing a lot
of money. By the way, this hardware is expensive, compute
is expensive. The commercialization has not been great so far,
but people are betting on the upside outcome of this
being you know, maybe the leading AI company. But it's begun.
(32:35):
You know. Open ai began its life as a nonprofit
co founded by.
Speaker 1 (32:41):
Rest and Peace and nonprofits.
Speaker 2 (32:43):
Yeah, co founded by Elon Musk and Sam Maltman. You know, look, Evon,
he does an impressive array of things. I know that's
gonna like really piss off half our audience, right, but
like the fact that this isn't even his main thing.
It's not Tesla or SpaceX, and that he has open
AI and and neurallink, and for some reason it decides
(33:03):
to invest his time on Twitter slash x instead. But yeah,
so open I has been in various formal and informal
ways moving further and further away from this notion of
being a nonprofit. You have had a great number of
the alignment people, meaning people that are so alignment is
an AI term for is AI compatible with human values? Now?
(33:28):
Whose values are these? And what's it mean to be compatible?
Are big open questions. But basically Open Eye is giving
up on much of the much of the pretense I
suppose of being an AI safety company, and instead, you know,
Sam Baltman is just saying the benefits outweigh the risks.
(33:50):
In a recent post, for example, he claims that we
will fix the climate, establish a space colony, and have
the discovery of all physics, and these will be commonplace
things in the age of AI with nearly limitless intelligence
and abundant energy. He claims, you know, when I talked
to when I talked to in twenty twenty two, I
guess it was he's like, yeah, I think that AI
(34:12):
will solve all of poverty. So he's he's kind of
a DJA. I think I don't mean that about his
personal life. I just mean that, like he is willing
to make this big gamble, not because he thinks said
it's to completely safe. He is clear that there are
these risks, but I think he thinks that the rewards
outstripped the risks.
Speaker 1 (34:31):
And in the meantime, I mean, I'm not sure where
all this limitless energy is coming from. Because in the meantime,
we've got Microsoft making plans to reopen Three Mile Island
the nuclear power plant, because frankly, our power grid cannot
keep up with the demands of open AI and chat
GPT and what is happening there. So I think that people,
(34:52):
you know, when we talk about existential risk, where we're
talking about a lot of different things, but like the
energy demand is crazy. Like I'm very pro nuclear power,
Like I think that, you know, we should be building
a lot of safe nuclear power plants. Like I think
that that's a great way of how the world kind
of reach energy neutral status. But it's crazy that the
(35:16):
first time people have actually like talked about and thought, Okay,
you know, maybe maybe we can reopen this this power
plant that's been shut down.
Speaker 2 (35:26):
Changed it, like give it some corporate confinity or something.
Don't call it three Mile Island.
Speaker 1 (35:30):
Let's see right, No, that's a really bad name. But yeah,
the first time that they're actually like, they're like okay,
Like before they're like no, no, no, like three Mile Island,
horrible disaster, blah blah blah. And then they're like, oh,
open AI needs it. You know, we need it for
being able to do memes with chat GBT. Hey, everyone
ask chat GPT this. Okay, let's let's reopen this. So
(35:52):
it's just, you know, it's just absolutely backwards and to me,
like a really perplexing state of affairs that he would say, oh,
you know, you know, we're solving everything. But so far,
what I see is that they haven't solved a lot,
and most of what they've done, you know, is actually
things that have been both physically polluting the world and
(36:14):
just polluting my head by creating an Internet full of
just absolute like hallucinatory bullshit that is really difficult to
parse through. Makes goop. By the way Google results now,
I don't know about unit, but like I'm looking for
a new search engine because it's unusable. I'm like, what
the fuck? You know? I try, I try to search
(36:36):
for something. It's good to searches, and it's just I
just want it. Can I just rant about this for
one second? It seems like you want to join me now, Like,
first we have this AI summary, which is usually wrong,
like it's actually given me like absolutely incorrect information in
that summer and I'm like, wait, no, I know this
is not true, this is not at all. And then
it's all these anyway, and now it's just completely unusable.
(36:58):
I'll let you interrupt me, because you look like you
have something to say.
Speaker 2 (37:01):
No, I don't. I mean, so a lot of it's
driven by ads, right, but yeah, the aioorviews are often
not very good. And the whole point of search engine
is I don't want your fucking summary of it. I
want to go to a website, right, I want to
choose which website I go to. Just give me a
list of good fucking websites, Google, And instead you have
this like AI summary, which I think is a product
that is frankly behind both chatchipt and Claude, which is
(37:26):
the anthropic product. And I think Google is terrified because say,
Google developed the transformer algorithm, right, Google has harnessed a
lot of the talent but they've had for years, they've
had like a brain drain and more. Why that is
for internal cultural reasons, I won't speculate, but like, but
(37:48):
I wish that Google didn't like shove this mediocre version
of AI for its once very wonderful search product in
our in our face. That that really annoys me.
Speaker 1 (37:56):
Yeah, but we're it really annoys me too. And you know, listeners,
please recommend a good search engine that you actually trust,
that gives reliable results instead of Google, because it's just
it's just killing me. But this kind of drek, I mean,
that's one of the things that we're seeing so far.
It hasn't revolutionized humanity. It's created a lot of drek
(38:18):
or slop as as some writers have called it. And
you know, the the Internet is it's you know, it's
more difficult to use. And when I go to look for,
you know, answers to things, oftentimes I can't figure out.
Unless I go to like, you know, Bloomberg right or
The New York Times or just a site that I
(38:40):
know to be trustworthy, I can't even figure it out.
And even then, if I go to ESPN, they have
articles that have been written by chatbots right by chat
GBT by other AI kind of large language models, and
they often create fake names and fake photographs. And I'm
singling out ESPN because I know it doesn't I don't know,
(39:01):
you know what other sites do it, but it's it's
a problem where I, you know, before I was thinking
of p doom in terms of like the world's going
to help and we're going to not survive, but also
there's like a p doom of my brain and of
the usability of the Internet. And I'm like, what the
hell is happening right, Like this is really cognitive load
(39:23):
is going up. It's over taxing me, and like I
don't want to have to deal with this, Like just
give me real articles, real information, real people, like things
I can trust, and don't give me. Don't force me
to have to like figure out you know, two paragraphs
in Wait, this is definitely not written by a human,
and I can't trust anything it says.
Speaker 2 (39:44):
You know, Look, when Sam Altman says we're going to
solve all of physics, I mean that that to me. Uh,
I'm a little skeptical about that, right. I think we
kind of end up in like many universes where AI
is a highly useful product maybe not good for society,
maybe not. But like, I don't know, I don't really
(40:05):
think that we're going to solve all of physics through
deep learning, right, I mean, I, you know, to extrapolate
out to these super human accomplishments from doing human things, well,
I think it's I think the jury is still out
and where that will occur or not. I mean, look,
should your pea doom be going up as a result
of this, I suppose though for me it kind of
(40:28):
seems like the inevitable game theory optimal out come anyway, right,
is that people pursue profit in the capitalist system, and
that you know, and that it's the role of the
public sector decide if, how and when we want to
regulate AI companies. Gavin Newsome vetoed and AI regulation bill
actually had passed the California Congress.
Speaker 1 (40:48):
So yeah, that was that was kind of crazy, and
his justification for doing it was kind of crazy. You know,
he said that it's only targeting certain companies. That's fine, Okay,
pass it and then do another one to target other companies.
But yeah, I think that I think that this is
a really kind of interesting situation that we're in right now.
(41:09):
I'm curious to see how it evolves. But I wish,
you know, I wish that we weren't taking such a
toll on you know, collective brain power, on environment, on
other things as we wait to see what regulation will
happen and what will actually happen. Do you have a
percentage for your new P doom or not?
Speaker 2 (41:27):
Really?
Speaker 1 (41:27):
You know, we're not going to put numbers.
Speaker 2 (41:29):
Given enough percentages today, I feel like I give the
exact number.
Speaker 1 (41:33):
That's fair, that's fair. But yes, I now have two
P dooms. I have my PE doom with the capital D,
which is kind of the world destroying, and I have
a lowercase P doom, which is my brain blowing up
because of all of the bullshit that's happening on the Internet.
So this is just my play, Like, let's let's try
to reverse some of this, and let's try to make
(41:54):
the internet back into a usable experience, because otherwise we're
going to have to it's about to say, otherwise we're
going to have to just keep go to libraries again
and like just rely on paper. But we can't even
do that because libraries have become inundated with books that
have been written by any guys and librarians have to
know try to figure out what the hell to do
(42:15):
with this. So it's a problem that's just permeating everywhere,
and I'm not happy about it. And on that note,
I think, I think we've talked a lot today about
about some interesting trends, and I'm very curious to see
how all of these different things will develop. And I'm
I would like to be an optimist and say that
(42:36):
things will get better, but I don't know.
Speaker 2 (42:38):
I used to, and I think I'm an optimis I mean,
think about how nurotic most people are, right compared to
most people, I'm an optimist.
Speaker 1 (42:45):
All right? And these are two optimists signing off on this.
Speaker 2 (42:48):
Week's Risky Business.
Speaker 1 (43:03):
Risky Business is hosted by me Maria Kannakova.
Speaker 2 (43:06):
And me Mate Silver.
Speaker 1 (43:08):
The show is a co production of Pushkin Industries and iHeartMedia.
This episode was produced by Isabelle Carter. Our associate producer
is Gabriel Hunter Chang. Our executive producer is Jacob Goldstein.
Speaker 2 (43:20):
And if you want to listen to an AD free version,
sign up for Pushkin Plus. For six sinety nine a month,
you get access to ad free listening. Thanks for tuning in.
Speaker 1 (43:29):
J
Pushkin. Welcome back to Risky Business, our show about making
better decisions. I'm Maria Kanakova.
Speaker 2 (00:29):
And I'm Nate Silver. Today on the show will be
giving you a degenerate gambler's guide to the election, including
a big proposed election bet which has not been reciprocated yet.
Speaker 1 (00:39):
After that, we're going to talk about the threat of
pots and automation to one of the things that Nate
and I hold dear to our hearts, poker and online
poker specifically, and then.
Speaker 2 (00:51):
We'll re evaluate p doom, the probability of severe misalignment
from artificial intelligence in light of recent news at Open AI.
This is a good episode, Maria. I feel like we
were kind of hitting all the basic angles today.
Speaker 1 (01:10):
I feel like we are too, so so let's uh,
let's get it started. Nate, I couldn't help but notice
a recent exchange on Twitter that you that you engaged in,
As so often happens, your views on Twitter were challenged.
I know this is this comes as a shock to
everyone every time I hear it. You know, people don't
(01:32):
always agree with you, and this particular challenge was about
Florida right and about the likelihood what percentage by what
percent Trump was going to carry Florida. So do you
want to do you want to talk us through a
little bit what happened and why we're why we're starting
our show with this particular Twitter exchange.
Speaker 2 (01:53):
Yeah. So we published an article at Silver Bulletin late
last week pointing out that that Florida's kind of close
in the polls, Harris is down by about three and
a half points, similar to Trump's margin of victory in
twenty twenty, and that maybe the Harris campaign, you have
infinite money, it's worth actually investing in the Sunshine State.
(02:15):
The American investor Keith Erboy, He's like, you're an idiot.
Trump's gonna win Florida by This is not the literal
verbatim version, right, You're an idiot. Trump's gonna win Florida
by at least eight points, if not ten or twelve.
And I'm like, okay, bro, how much are you willing to, like,
actually wager on this? And he says one hundred K,
which seems like a seemed like a good amount to me. Right,
(02:37):
it's a serious it's a serious adult bet. And by
the way, there's a chance that Trump will win Florida
by eight points or more if you look at polymarket,
about twenty five percent. But I'm willing to take one
hundred k bet where I win seventy factions of the time.
And so I said, okay, you got to send me
a contract though, because you've been sketchy lately, and I
don't really you know, he's not like a known I mean, ironically,
(02:59):
like known gamblers are like more reliable for payout stuff, right,
But I don't want him saying like, oh, the election
was stolen. I don't want him saying, oh haha, I
was just joking. So need some type of third party
to step in and create a contract or at least
create an obligation expectation. So far that has not happened.
We had a little bit of negotiations by back channel email.
(03:22):
The mutual friend said I'm willing to do this and
then and mister Riboy has not replied yet, so I'm
not sure kind of how much more time I want
to give my I don't like giving people free options, right.
You know, if there's a poll that comes out this
week showing Trump way up in Florida, then all of
a sudden, his bet becomes not plus TV but better,
but Yeah, So anyway, I think this is an honorable
(03:45):
way to resolve differences, right that if you actually have
to pay a tax for your bullshit on Twitter, right,
seventy five percent chance of losing one hundred K, then
that to me seems like you know, I think the
world will be better off if people spent more time
betting and less time trolling on Twitter.
Speaker 1 (04:02):
Absolutely, I think that there should be a bullshit tax,
for sure. And just so that the the audience is
up to date, this is not the first time you've
been challenged. To date, not a single person has actually
taken you up on your Nobody ever, actually, nobody has
goes through Yeah, nobody has Heath ever done.
Speaker 2 (04:24):
To a further step than most people do, right, So
give them credit for that. But like, but yeah, they
almost always back down because if I'm making a bet,
they're typically carefully. I'm not trying to eke out you know,
one or two percent ev bets right, in part because
I have a lot on the line anyway as far
as the election goes, right, So I'm not trying to like,
but if you're gonna give me a seventy five twenty
(04:46):
five coin flip for one hundred K then I feel
like I would be negligent to pass up on that.
Speaker 1 (04:52):
Yeah, absolutely, So you know, tune in and uh and
we'll we'll see how that goes. But the news for
people who do want to make bets like this not
on Twitter with Nate Silver, but in general on elections.
We talked a few weeks ago about Calshi, which is
(05:13):
a prediction market that actually operates within the United States
real money bets and Calshie had been trying to for
years basically make their way through the US court system
to make election betting legal in the United States, to
make those contracts legal here, and they had prevailed, and
(05:35):
then the order was frozen, right, so their options on
the election went live for a few hours and then
they were no longer allowed to continue while the Court
of Appeals reviewed this, and they have now ruled in
favor of Calshie, and so people in the United States
(05:56):
are once again allowed to bet on election outcomes. I
actually got a very funny email from Interactive Brokers, which
is an online trading platform on which I happened to
have my retirement account, and they sent an email to
everyone saying, hey, guys, like you can now bet on
(06:16):
all of these election things we're going live, So it's
something that immediately it seems like a lot of platforms
were ready for this and are really excited to start
taking these sorts of bets. But we actually had some
listeners who were interested in our takes on a few
different bet ideas. So one listener, David Paul, wrote in
and asked, would we bet that twenty twenty four voter
(06:36):
turnout will be higher or lower than in twenty twenty
and why? And I thought that that was actually kind
of an interesting bet proposition because you and I have
discussed this a little bit in terms of how close
an election is perceived to be and what the turnout
is as a result of that, and also obviously how
(06:56):
passionate people are and how engaged people are, right, and
different different subsets of the population are more likely to
actually go out and vote with their feet at the
polls as opposed to vote on Twitter talk about how
you know one candidate or the other, but then they
don't actually go and cast a ballot. So what do
we think about this? What do you think about this?
Speaker 2 (07:18):
Well, let me ask an important question. Are we talking
about the raw number of voters or the share of
the eligible voting population, I don't know.
Speaker 1 (07:27):
Let's let's let's discuss because because I actually don't know
what has happened with voter registration. Has that gone up
with you know, Taitay's endorsement and all of that. Are
we actually seeing a much larger number of voter registrations
than we saw four years ago?
Speaker 2 (07:43):
Look, so, first of all, I would probably bet on
turn up being a bit lower than in twenty twenty,
especially with you know, if you are accounting for population growth,
then it might be kind of a tass up. But
if you're looking at the share of the population in
part because like twenty twenty was a bit of an outlier,
sixty six percent of eligible voters cast ballots. That was
up from sixty percent in twenty sixteen and fifty nine
(08:06):
percent in twenty twelve. And this is data from the
University of floor or as Election Lab website. But we
have had an increase, is you know? So turnout in
the twenty twenty two midterm was forty six percent, which
is high for a midterm, but down from twenty eighteen. Right,
you know, when it was Biden versus Trump, people were
distinctly unenthusiastic about this selection. With the events of the
(08:28):
summer with Harris taking over for Biden, then there's more
enthusiasm now. But but you know, twenty twenty turnout was
really high, in part because it's a pandemic. There were
more ways to vote than ever before. And I would
vote on I know, sixty three percent of the voting
eligible population instead of sixty.
Speaker 1 (08:42):
Six I suppose, so you would put your money on
on lower turnout. And to be clear, we're not actually
making a bet, We're not actually putting money on the
line because this is not a topic that either one
of us has probably a lot of conviction. And do
you have to.
Speaker 2 (08:58):
Actually have a model. So the model predicts, oh, there
we go, one hundred and fifty six million, six hundred
and eighty five thousand, six hundred and eighty people exactly
will vote. Okay, I obviously it's fake precision. Let me see, uh,
let me see what the numbers were in twenty twenty.
They say one hundred and fifty eight uh point four
(09:23):
million people voted, and I'm predicting our mouths one fifty
six point seven, so that would be off by just
a million or two.
Speaker 1 (09:32):
Okay, but lower but lower. So so your your gut,
your gut was correct, and now the silver bulletin model
has validated your gut at least as of now. And
these numbers might change right as you get more data.
Speaker 2 (09:46):
Yeah, no, I mean, so what we look at is
polls asking people how enthusiastic they are about voting in
the election, and people actually say they're pretty enthusiastic. However,
turn out with just really high in twenty twenty, So
until we see more confirmation of that than we're expecting
some reversion to the mean.
Speaker 1 (10:04):
Great, well, if we're if we're going down that kind
of Djen rabbit hole. I actually think it's in interesting
because it does kind of touch on different pressure points
of the election. So some other ideas, what's our over under,
for one, major news outlets will actually call a winner.
I think that's an interesting one because I don't think
either one of us particularly feels like the election is
(10:24):
going to be over on November fifth, given how close
it is and given the the way elections have gone
in recent years.
Speaker 2 (10:32):
Yeah, look, I mean the over under might be you know,
November fifth is a Tuesday, maybe Saturday, excuse me, Thursday morning.
The seventh. I mean keep in mind that, like the
networks send to be really conservative about making these calls,
except for Fox News, which weirdly called Arizona for Trump
when it was great premature to do so. But leave
that aside for a moment. Yeah, I remember doing in
(10:53):
twenty twenty. I was kind of like like in the
boiler room, it's during COVID doing like these are PEATV
hits over and over for ABC News, and like the networks,
it was obvious. I thought by like midday Wednesday that
Biden was going to have enough past to victory, but
the networks didn't call it until Saturday morning. But I
don't mean to make light of this. I mean, like,
(11:14):
you know, another disputed election outcome where Trump is sowing
doubt about the results, that would not be great for
the US. I think people are not quite pricing in, uh,
what's going to feel like if you have it, if
you have an unclear, ambiguous, disputed election result.
Speaker 1 (11:33):
Yeah. I mean, I'm actually hoping that we do get
a winner announced on the fifth or you know, on
early morning of the sixth. I hope that we do
have a very clear cut election, because I think that
that is something that we need because we have you know,
we've had some close calls and we've had a number
of disputed elections in recent memory. So so that's not
(11:54):
necessarily a great place to be when you're looking at
kind of when you're looking at democratic out outcomes and
peaceful transitions of power. No.
Speaker 2 (12:03):
I mean, look, you don't need that much of a
polling er to have a pretty clear all right, if
Trump beats his polls by three points, then he wins
all the swing states. Likewise, for Harris, you could even
win Florida, which was a source of dispute with Keith
or boy. So, yeah, you know, on what other prep
bets do we have? Yeah?
Speaker 1 (12:24):
So, so there were a few others. One was in
which of the swing states will polling to be turned
out to be the most accurate and in which will
it be the least accurate? So do you have as
ours our pole guru, do you have a sense of
how the polls are doing this year?
Speaker 2 (12:41):
I mean, maybe Georgia for most accurate. The polls have
tended to be pretty accurate in Georgia in the past.
And also like like in Georgia, like you kind of
know everyone's going to vote, right, you have like the
evangelical whites, the African American population, and like you have
the you know, suburban so it's not like it's just
a matter of kind of racing to see who can
get more turnout, and that tends to make the polling
(13:02):
a little bit easier versus states where you have like
a lot of swing voters. But I'll go with Georgia
as the most accurate polling state. Do you have an opinion, Maria, No, I.
Speaker 1 (13:12):
Don't because I don't know. You know, I don't follow
the polls as the way that you do. I just
look at your outcomes, and I tend to trust you
obviously when it comes to pulling red when it comes
to reading the polls, so I just kind of look
to see what the prevailing wisdom is. But I don't know,
you know, I think that it's one thing to say, oh,
(13:34):
you know, these are high quality polls, these are worse
quality polls, you know, et cetera, et cetera. But I
think it takes expertise that I, frankly do not have
to be able to say how accurate is polling within
a state.
Speaker 2 (13:46):
I'm so torn on deferring to X, you know, because
I want to how your friends like Keith r who
clearly is outside his depth in this particular enterprise. A
lot of the time the experts are full of shit
and the non experts if I'm more full of shit? Right,
Life is hard. It's hard to make predictions about things.
Speaker 1 (14:03):
It is hard to make predictions. And let's do a
shout out to the work of Phil Tetlock, who was
one of the first people to show how hard it
is to make predictions and how you know experts are
also often wrong. And but if we can't actually I'm
curious to hear the answer to this. If we can't
defer to experts, then you know what, who do we
(14:23):
defer to? I mean, there's obviously, you know, going back
to what we started off with, you know, with prediction
markets and places like cal shan As. We know you're
paid advisor for polymarket. I just I hate having to
throw that disclaimer out every single time, but but I
think we do have to throw it at every single time.
(14:44):
Our prediction markets a better place, Like what do we
defer to? Then if we can't, if we say, like
you know, you can't really defer to experts, how do
you figure out you know, how do I form an
opinion on this.
Speaker 2 (14:55):
Yeah, look, obviously my fan of prediction markets. I think
I think one of if you read Phil Telllock's book
that you reference, We're Expert Political judgment. I mean, you
know that actually forced experts to make probabilistic predictions, right,
Like what's the chance that the USSR will break up
in the next ten years might have been a question
that he asked. For example, when you encourage people to
do that, then you actually create more accountability, You create
(15:17):
a paper trail because the most part like experts don't
actually have incentives to make accurate predictions, you know what
I mean, they have incentives to be ambiguous so they
can't be proven wrong. And so you know, stuff like
what we do at Silver Bulletin is actually quite unusual,
and if experts were forced to come up with actual
answers then they would do they would potentially be a
(15:39):
little bit better, although I think we'd also expose people
who are not very vigorous in their thinking.
Speaker 1 (15:43):
Yeah, so then maybe we would be able to have
kind of a hierarchy of expertise, which I think is
something that is very would be very very helpful, at
least to someone like me the other thing that happened
in the news, kind of putting our little prop bets aside.
Obviously we had the VP debates since you and I
last spoke. Did that make any difference whatsoever? I know
(16:05):
you've said over and over and over that you know,
the VP debates don't really matter, piece don't usually matter.
Have we seen them matter at all in the last
week or no change.
Speaker 2 (16:15):
There's been oddly little polling of any kind. I think
the polsearch just all decided to take the weekend off.
I don't know, man, you can let me say something
that will get clipped onto the Twitter, right, Like, I'm
not into Tim Walls really, I mean he's fine, but like, uh,
you know, first of all, there's all this discussion about like, oh,
(16:37):
you don't want to the term sain washing. It's like, oh,
you don't want to make Trump Advance seem insane, and
he's like he's doing this whole nice guy thing. And
I think Vance is just objectively like a better debater
than Tim Walls. And Tim Walls was obviously extremely nervous
for the first fifteen minutes of the debate, which is
a part that people often remember, like it won't matter
I don't think very much of the end. The polls
(16:57):
showed that people thought the debate was a tie basically,
but I don't know, you know, it's kind of you
have your one job thing and you should really nail
that debate. And he I think it was tactically not
very good at going after JD. Vance and I'm I'm
just not into it. I don't really understand, you know what.
(17:17):
There's like feeling of euphoria after Harris replaced Biden, and
I think people therefore overrated everything she did in that
period after that. And if it were me, I would
rather have, say, the extremely popular governor Pennsylvania, which is
the most important tipping point of the state.
Speaker 1 (17:35):
Yeah, well, let's see what ends up happening, and after
the break, we'll talk about betting on something else, which
is betting on cards and poker bots. Nate, if I
(18:02):
remember correctly, way back when you made a living professionally
playing poker online playing limit hold him? Is that correct?
Speaker 2 (18:13):
That's correct? For preire to two for three years, I
guess three years. Yeah, my main source of income was
playing fricking party poker.
Speaker 1 (18:24):
And when was this let's put it on a timeline.
Speaker 2 (18:27):
It's two thousand and four to two thousand and six,
and then then I had a year where let's not
tell them, but I actually lost money in two thousand
and seven after the game's cut. So the US Congress
passed a bill called the U I G E A
EUG U G. Yeah, and once that happened, the game
started to dry up and Party Poker closed. And but anyway, yeah,
(18:47):
I'm actually I was able to change my whole career
by playing online poker basically.
Speaker 1 (18:53):
Okay, So the reason that I asked about this is
this past week, we've had several blockbuster investigation pieces come
out about bots in online poker and bots as kind
of an existential three to the game. So the first
of these came out on September twentieth, and it was
(19:14):
a piece by kit Chilo, who has been doing this
deep dive investigation into a massive, massive Russian bought. He
calls it an army. So usually in the online poker
world we talk about bot farms, but when you read
this piece, you realize that, holy shit, army might actually
be a legitimate way to look at this. So he
(19:38):
had kind of found a few breadcrumbs that then led
him to poker forms to people who actually were involved
in this bot operation. And it turns out that this group,
originally this group of students, then it grew much bigger,
is working, you know, in the depths of Siberia, and
(20:00):
these are stats nerds, college students who need extra income
and they want to get better at poker, at online poker,
and so they start kind of playing around with basically
what you know today we'd call solver software. Then they
start developing little bots to try to implement some of
these ideas. Fast forward to kind of the capabilities that
(20:21):
these bots can actually be deployed onto poker platforms, and
these bots start being used and start winning money on
online poker sites in great numbers. And then it turns
out that some online poker sites might even be deploying
bots themselves in order to generate artificially larger player pools,
(20:43):
because if you're an online poker site, you don't particularly
care if the person playing the rag is a little
bot or a human as long as you get the money.
And so then after this Bloomberg piece came out, a
second piece came out by Jonathan Robb and he investigated
further and actually named names of some online poker companies. Anyway,
(21:03):
this is a very long introduction, so we'll talk more
about this later, but it seems like we see quite
the existential threat to online poker and to players and
it like yourself who look at online poker as a
way of making money. So what what do you think?
You know, have you been following this poker BT explosion
at all and have you given any thought to kind
(21:24):
of the state of online poker play.
Speaker 2 (21:27):
I mean, first of all, I don't look at online
poker as a way to make money. I look at
it as a form of recreation, I suppose, because like
to me, it seems like a somewhat unsolvable problem that
you have solvers now that can that can give you
(21:48):
a game theory optimal solution to a poker hand, never
mind anything more advanced about like doing data mining particular opponents.
You know, to me, unless you're interested when you're actually
having somebody like physically on a camera or something. Then
you know, combined with the fact that, like as it
says the Bloomberg article, it's kind of a Darwinian struggle
(22:11):
to begin with online poker, right, people are already playing
pretty close to optimal that I just don't. I wouldn't
want to have to live my life now trying to
make money playing poker online. And and there are home
games that people say, but I just think the risk
is pretty is pretty high.
Speaker 1 (22:29):
Yeah, And when we when we look at kind of
these further investigations. So this second article that I alluded to,
the Jonathan rob piece in Poker Pro, he actually found
kind of this one site, jack Poker that seemed to
be almost entirely creating their playerpool out of bots, which
(22:50):
is kind of crazy, right, And he you know, signed
up as a real person with real money, and then
he noticed these crazy patterns in their tournaments, et cetera.
Went down that rabbit hole and actually saw that they
were connected to their user interface was connected to this
corporation that Chile had wrote about, the Bot Farm Corporation BFC,
(23:13):
and what kind of they're deep play software, which is
kind of these bots that they were deploying. And if
it's true that sites like that are actually kind of
embracing bots in some ways like that to me is
quite worrying. By the way, time for my own disclaimer,
as people know, but I should say I'm an ambassador
for Poker Stars, which is also an online poker site.
(23:35):
Poker Stars was not implicated in any of this, but
you know, I should, I should throw that out there.
Speaker 2 (23:42):
I mean, like there are incentives to not have perceived
or actual cheating. I suppose it just that, you know,
I mean when you can like, I don't know. I
mean it's I tried to play poker against chatchipt once.
Speaker 1 (23:55):
Really, how did that go?
Speaker 2 (23:57):
It's very nitty. No matter what options you give it, it
always just checks the pot down, even if it has
like the nuts. So you know, maybe we still have
some time for humans to be better than Chetchapet at poker.
But like, but you know, it's an asymmetric arms race.
It's just hard for me to imagine that someone who
is connected to a computer can't find efficient ways to cheat. Basically, yeah,
(24:22):
I mean.
Speaker 1 (24:22):
This is clearly a real problem, and there are I
do think there are solutions. By the way, you can
usually figure it out when you're looking at deep analysis.
So Nate, as a statistician and someone who does this
kind of analysis, you can probably appreciate that if you
have a database of you know, hundreds and thousands of hands,
you can start seeing is this a person or is
(24:44):
this a computer that has some sort of regular type
of output and you know, flagging and getting rid of
those things immediately is a is a big one the
other one, And I don't think anyone wants to do this,
but I think it could. I think it's potentially kind
of the future of a safe poker environment is what
(25:05):
if we play on camera? Right? Like, what if you
have those types of checks where you actually try to
monitor who's playing, how they're playing, et cetera. And even
that's not perfect obviously, but there are certain things that
you can start doing. But it takes a lot of resources,
and the incentives, unfortunately, are often with the players. And
as you said, it is an arms race, like we
(25:26):
can learn to try to figure out, oh, what's this
kind of bot look like? What does that kind of
bot look like? And then someone comes up with something better,
someone who deviates or you know, something that looks more
human like like the Turing test. Right, every single year
becomes more and more likely that we'll actually have someone
be able to beat the Turing test. I think all
of our listeners probably know what the Turing test is,
(25:48):
but do we want to pause for a second to
explain what that is for people who might not know.
Speaker 2 (25:56):
Yeah, I think probably most people know. But the Turing
test is basically, can you program I mean it's actually
anticipates chat GPT pretty well, right, but can you tell
if you're having a conversation with a human being versus
a computer? And yeah, no, I mean like CHUTCHIPT, I
mean it has signature for example, frequently says as a
(26:18):
large language model, which is something that like. But the
thing even that though, like, you know, so if you
do play online poker and you're playing a lot of
hours and sometimes you get a prompt saying prove that
you're a human, right, do some task now that proves
that you're a human and not just a bot. But
the thing is, though, like you could intentionally introduce some
degree of like randomization. I mean, one thing about cheaters
(26:41):
in general is that most cheaters are really greedy, or
at least the ones that get caught right, and so
they cheat too much. When if you were a smart
cheater and poker could just kind of get away with
the edge cases where you magically happen to know in
the spot where you're a different what your opponent has
on the river, And instead people get very greedy, but
I just think it's kind of asymmetric warfare. In the end,
(27:02):
life poker is good, though, yes, more life poker.
Speaker 1 (27:06):
I agree. I think life poker is certainly a for know,
though not altogether safe. You know, we've we've had live
cheating scandals as well.
Speaker 2 (27:14):
And this has always bugging me at like poker is
quite forgiving. I think of cheating other bad behavior and
ways like yeah, it should just be banned for many
casino for lifetime or for five years and for things
like that, and instead they're kind of like slap on
the wrist at the time. Yeah, you know, Isaac Heckson,
I talked to you for the book, one of the
best players in the world, and told me that his
(27:37):
ROI and online poker is about ten percent, which is
not actually very good, or said under ten percent in
the single digits, right, So forever, forever, thousand dollars he
invests in playing a tournament, that's one of the best
players in the world, you know, maybe he has an
expected return of fifty or sixty dollars from that, Like
that's actually, given the variants of poker, not a very
(27:58):
good deal, right, And if he one of the ten
to twelve best players in the world is only making
that much, then I I don't think very many people
are really beating online poker sites without cheating in the
long run.
Speaker 1 (28:09):
Yeah, I think that that's that's very sobering. I did
not I did not realize that that I could communicated
those numbers to you if if one of the best
players in the world says that, but I will caveat
that was saying. Mike also plays in a lot of
much harder games, right, He's probably not playing in some
of the lower buying tournaments that you know, you and
(28:29):
I would would play in. He's probably playing the high rollers.
Speaker 2 (28:33):
But still, it's pretty it's pretty frightening.
Speaker 1 (28:35):
Yeah, for sure, for sure. So I'm very curious to see,
you know, what the fallout is going to be of
all of this, whether sites are going to address it,
whether they're you know, what the what the changes are
going to be. There was you know, for instance, I
the last poker live Poker I played was European Poker
Tory PT Barcelona, and at that stop that happened right
(28:57):
after there were some cheating allegations. We we we don't
have to go into into details on those, but of
live poker and of certain methods of being able to
kind of see cards live at the table. And so
what Poker Stars announced at the stop was some new rules,
(29:21):
including no phones on the tables, no phones on the rails.
And I think that, you know, things like that actually
show that, Okay, we're we're trying to take some of
these threats seriously as they're evolving. And I went to
step further and I said, you know what, like we
should ban phones at the table period, We should ban
(29:41):
sunglasses at the table. You know, there are certain things
that like we just know that there's cheating technology that
can be used in those instances. And I love having
my phone, but you know, I'm willing to put it
on the floor and to not touch it for an hour,
Like I'm okay with that, And I actually think it
would be very good for everyone's mental health to not
(30:02):
have your phone on you.
Speaker 2 (30:03):
Twenty four se day, Day twenty six of fifty two
with the Worldter's of Poker, You're playing in some eight
hundred dollars freeze out ten handed with the guy from Buffalo,
who I mean is taking fifteen seconds to fold. I mean,
you want your fucking phone, don't you?
Speaker 1 (30:20):
I do want my fucking phone, but for the greater
good of the things.
Speaker 2 (30:24):
Why I'm making fun of Buffalo by the way, we
love Buffalo.
Speaker 1 (30:29):
Oh, Buffalo has taken some flack from us in the past.
I just remembered because you were.
Speaker 2 (30:33):
Forgot the unexpected layover.
Speaker 1 (30:35):
Yeah, you had a Buffalo layover anyway, Sorry, Buffalo, You're great.
But you know I would I would take one for
the team and be willing to give up my phone
a little bit if it meant you know, game integrity
and things that are kind of safer for the game. So,
you know, I think, But just to wrap this up,
(30:55):
I think there are existential threats. We need to address them,
and we need to figure out kind of the best
way forward because I think you and I both feel
that poker is a great game and a game that
can really kind of help people think better, make better decisions,
and be great if people could safely continue playing in
all in all forms, and I hope that happens.
Speaker 2 (31:17):
Speaking about existential threats, you want to talk about open
AI real quick?
Speaker 1 (31:21):
Yeah, So now we've been talking about existential threats to poker,
and after the break, let's talk about open an AI
and revisit some of those existential threats day a while ago.
(31:49):
I think the last time we spoke about the concept
of p doom was what a few months back, right
where we talked about the risk that AI poses to
the world. Last week, in the last week, there's been
a lot of news coming out of open Ai, a
lot of new departures and a lot of fundra saying.
Speaker 2 (32:11):
So yeah, open ai raised six point six billion, we
are told by our reliable producers, at one hundred and
fifty seven billion dollar valuation. It is losing a lot
of money. By the way, this hardware is expensive, compute
is expensive. The commercialization has not been great so far,
but people are betting on the upside outcome of this
being you know, maybe the leading AI company. But it's begun.
(32:35):
You know. Open ai began its life as a nonprofit
co founded by.
Speaker 1 (32:41):
Rest and Peace and nonprofits.
Speaker 2 (32:43):
Yeah, co founded by Elon Musk and Sam Maltman. You know, look, Evon,
he does an impressive array of things. I know that's
gonna like really piss off half our audience, right, but
like the fact that this isn't even his main thing.
It's not Tesla or SpaceX, and that he has open
AI and and neurallink, and for some reason it decides
(33:03):
to invest his time on Twitter slash x instead. But yeah,
so open I has been in various formal and informal
ways moving further and further away from this notion of
being a nonprofit. You have had a great number of
the alignment people, meaning people that are so alignment is
an AI term for is AI compatible with human values? Now?
(33:28):
Whose values are these? And what's it mean to be compatible?
Are big open questions. But basically Open Eye is giving
up on much of the much of the pretense I
suppose of being an AI safety company, and instead, you know,
Sam Baltman is just saying the benefits outweigh the risks.
(33:50):
In a recent post, for example, he claims that we
will fix the climate, establish a space colony, and have
the discovery of all physics, and these will be commonplace
things in the age of AI with nearly limitless intelligence
and abundant energy. He claims, you know, when I talked
to when I talked to in twenty twenty two, I
guess it was he's like, yeah, I think that AI
(34:12):
will solve all of poverty. So he's he's kind of
a DJA. I think I don't mean that about his
personal life. I just mean that, like he is willing
to make this big gamble, not because he thinks said
it's to completely safe. He is clear that there are
these risks, but I think he thinks that the rewards
outstripped the risks.
Speaker 1 (34:31):
And in the meantime, I mean, I'm not sure where
all this limitless energy is coming from. Because in the meantime,
we've got Microsoft making plans to reopen Three Mile Island
the nuclear power plant, because frankly, our power grid cannot
keep up with the demands of open AI and chat
GPT and what is happening there. So I think that people,
(34:52):
you know, when we talk about existential risk, where we're
talking about a lot of different things, but like the
energy demand is crazy. Like I'm very pro nuclear power,
Like I think that, you know, we should be building
a lot of safe nuclear power plants. Like I think
that that's a great way of how the world kind
of reach energy neutral status. But it's crazy that the
(35:16):
first time people have actually like talked about and thought, Okay,
you know, maybe maybe we can reopen this this power
plant that's been shut down.
Speaker 2 (35:26):
Changed it, like give it some corporate confinity or something.
Don't call it three Mile Island.
Speaker 1 (35:30):
Let's see right, No, that's a really bad name. But yeah,
the first time that they're actually like, they're like okay,
Like before they're like no, no, no, like three Mile Island,
horrible disaster, blah blah blah. And then they're like, oh,
open AI needs it. You know, we need it for
being able to do memes with chat GBT. Hey, everyone
ask chat GPT this. Okay, let's let's reopen this. So
(35:52):
it's just, you know, it's just absolutely backwards and to me,
like a really perplexing state of affairs that he would say, oh,
you know, you know, we're solving everything. But so far,
what I see is that they haven't solved a lot,
and most of what they've done, you know, is actually
things that have been both physically polluting the world and
(36:14):
just polluting my head by creating an Internet full of
just absolute like hallucinatory bullshit that is really difficult to
parse through. Makes goop. By the way Google results now,
I don't know about unit, but like I'm looking for
a new search engine because it's unusable. I'm like, what
the fuck? You know? I try, I try to search
(36:36):
for something. It's good to searches, and it's just I
just want it. Can I just rant about this for
one second? It seems like you want to join me now, Like,
first we have this AI summary, which is usually wrong,
like it's actually given me like absolutely incorrect information in
that summer and I'm like, wait, no, I know this
is not true, this is not at all. And then
it's all these anyway, and now it's just completely unusable.
(36:58):
I'll let you interrupt me, because you look like you
have something to say.
Speaker 2 (37:01):
No, I don't. I mean, so a lot of it's
driven by ads, right, but yeah, the aioorviews are often
not very good. And the whole point of search engine
is I don't want your fucking summary of it. I
want to go to a website, right, I want to
choose which website I go to. Just give me a
list of good fucking websites, Google, And instead you have
this like AI summary, which I think is a product
that is frankly behind both chatchipt and Claude, which is
(37:26):
the anthropic product. And I think Google is terrified because say,
Google developed the transformer algorithm, right, Google has harnessed a
lot of the talent but they've had for years, they've
had like a brain drain and more. Why that is
for internal cultural reasons, I won't speculate, but like, but
(37:48):
I wish that Google didn't like shove this mediocre version
of AI for its once very wonderful search product in
our in our face. That that really annoys me.
Speaker 1 (37:56):
Yeah, but we're it really annoys me too. And you know, listeners,
please recommend a good search engine that you actually trust,
that gives reliable results instead of Google, because it's just
it's just killing me. But this kind of drek, I mean,
that's one of the things that we're seeing so far.
It hasn't revolutionized humanity. It's created a lot of drek
(38:18):
or slop as as some writers have called it. And
you know, the the Internet is it's you know, it's
more difficult to use. And when I go to look for,
you know, answers to things, oftentimes I can't figure out.
Unless I go to like, you know, Bloomberg right or
The New York Times or just a site that I
(38:40):
know to be trustworthy, I can't even figure it out.
And even then, if I go to ESPN, they have
articles that have been written by chatbots right by chat
GBT by other AI kind of large language models, and
they often create fake names and fake photographs. And I'm
singling out ESPN because I know it doesn't I don't know,
(39:01):
you know what other sites do it, but it's it's
a problem where I, you know, before I was thinking
of p doom in terms of like the world's going
to help and we're going to not survive, but also
there's like a p doom of my brain and of
the usability of the Internet. And I'm like, what the
hell is happening right, Like this is really cognitive load
(39:23):
is going up. It's over taxing me, and like I
don't want to have to deal with this, Like just
give me real articles, real information, real people, like things
I can trust, and don't give me. Don't force me
to have to like figure out you know, two paragraphs
in Wait, this is definitely not written by a human,
and I can't trust anything it says.
Speaker 2 (39:44):
You know, Look, when Sam Altman says we're going to
solve all of physics, I mean that that to me. Uh,
I'm a little skeptical about that, right. I think we
kind of end up in like many universes where AI
is a highly useful product maybe not good for society,
maybe not. But like, I don't know, I don't really
(40:05):
think that we're going to solve all of physics through
deep learning, right, I mean, I, you know, to extrapolate
out to these super human accomplishments from doing human things, well,
I think it's I think the jury is still out
and where that will occur or not. I mean, look,
should your pea doom be going up as a result
of this, I suppose though for me it kind of
(40:28):
seems like the inevitable game theory optimal out come anyway, right,
is that people pursue profit in the capitalist system, and
that you know, and that it's the role of the
public sector decide if, how and when we want to
regulate AI companies. Gavin Newsome vetoed and AI regulation bill
actually had passed the California Congress.
Speaker 1 (40:48):
So yeah, that was that was kind of crazy, and
his justification for doing it was kind of crazy. You know,
he said that it's only targeting certain companies. That's fine, Okay,
pass it and then do another one to target other companies.
But yeah, I think that I think that this is
a really kind of interesting situation that we're in right now.
(41:09):
I'm curious to see how it evolves. But I wish,
you know, I wish that we weren't taking such a
toll on you know, collective brain power, on environment, on
other things as we wait to see what regulation will
happen and what will actually happen. Do you have a
percentage for your new P doom or not?
Speaker 2 (41:27):
Really?
Speaker 1 (41:27):
You know, we're not going to put numbers.
Speaker 2 (41:29):
Given enough percentages today, I feel like I give the
exact number.
Speaker 1 (41:33):
That's fair, that's fair. But yes, I now have two
P dooms. I have my PE doom with the capital D,
which is kind of the world destroying, and I have
a lowercase P doom, which is my brain blowing up
because of all of the bullshit that's happening on the Internet.
So this is just my play, Like, let's let's try
to reverse some of this, and let's try to make
(41:54):
the internet back into a usable experience, because otherwise we're
going to have to it's about to say, otherwise we're
going to have to just keep go to libraries again
and like just rely on paper. But we can't even
do that because libraries have become inundated with books that
have been written by any guys and librarians have to
know try to figure out what the hell to do
(42:15):
with this. So it's a problem that's just permeating everywhere,
and I'm not happy about it. And on that note,
I think, I think we've talked a lot today about
about some interesting trends, and I'm very curious to see
how all of these different things will develop. And I'm
I would like to be an optimist and say that
(42:36):
things will get better, but I don't know.
Speaker 2 (42:38):
I used to, and I think I'm an optimis I mean,
think about how nurotic most people are, right compared to
most people, I'm an optimist.
Speaker 1 (42:45):
All right? And these are two optimists signing off on this.
Speaker 2 (42:48):
Week's Risky Business.
Speaker 1 (43:03):
Risky Business is hosted by me Maria Kannakova.
Speaker 2 (43:06):
And me Mate Silver.
Speaker 1 (43:08):
The show is a co production of Pushkin Industries and iHeartMedia.
This episode was produced by Isabelle Carter. Our associate producer
is Gabriel Hunter Chang. Our executive producer is Jacob Goldstein.
Speaker 2 (43:20):
And if you want to listen to an AD free version,
sign up for Pushkin Plus. For six sinety nine a month,
you get access to ad free listening. Thanks for tuning in.
Speaker 1 (43:29):
J
Hosts And Creators
Maria Konnikova
Nate Silver
Popular Podcasts
On Purpose with Jay Shetty
I’m Jay Shetty host of On Purpose the worlds #1 Mental Health podcast and I’m so grateful you found us. I started this podcast 5 years ago to invite you into conversations and workshops that are designed to help make you happier, healthier and more healed. I believe that when you (yes you) feel seen, heard and understood you’re able to deal with relationship struggles, work challenges and life’s ups and downs with more ease and grace. I interview experts, celebrities, thought leaders and athletes so that we can grow our mindset, build better habits and uncover a side of them we’ve never seen before. New episodes every Monday and Friday. Your support means the world to me and I don’t take it for granted — click the follow button and leave a review to help us spread the love with On Purpose. I can’t wait for you to listen to your first or 500th episode!
Stuff You Should Know
If you've ever wanted to know about champagne, satanism, the Stonewall Uprising, chaos theory, LSD, El Nino, true crime and Rosa Parks, then look no further. Josh and Chuck have you covered.
Dateline NBC
Current and classic episodes, featuring compelling true-crime mysteries, powerful documentaries and in-depth investigations. Follow now to get the latest episodes of Dateline NBC completely free, or subscribe to Dateline Premium for ad-free listening and exclusive bonus content: DatelinePremium.com