=talks
This is a record of a meeting in October 2022 between myself, Alyssa Vance, and Zvi Mowshowitz. I wrote it from memory afterwards, and edited for length and clarity, so it shouldn't be considered to have exact quotes.
--------------------------------------
Alyssa:
You have various
technological ideas, and a sort of vision for the future economy. What do
you see as the main obstacles to those being implemented?
bhauth:
From my perspective, the biggest problem is how investors operate. Most
early-stage investors are not very interested in technical due diligence.
Often, the only advantage of good ideas - including mine - is that they're
better from a technological perspective. If investors don't care about
technical due diligence, then that advantage means nothing.
I
understand how this happened: technical due diligence is sort of a public
good, and investors also found that the results of it weren't very accurate.
But I would say, their technical due diligence wasn't very accurate because
they didn't have the domain knowledge to choose good experts.
Engineers in US companies complain about how MBA directors come in and think
they don't need domain knowlege because "management and business" skills are
generic, and it doesn't go well. This is a similar issue, but for investors
rather than managers.
That's one reason Elon Musk is effective: he
learns some technical knowledge, and that lets him choose better technical
people. He's not a top engineer, and he's overconfident or surrounded by
yes-men, so he wastes a lot of time on flawed ideas he has - like his
Hyperloop design, or "overclocking" tunnel boring machines, or trying to do
beamforming with multiple Starlink satellites - but the end results are
still better than when American MBAs run tech companies.
Because
technical due diligence is sort of a public good, you can push investment
around by doing it. Bill Gates has been doing that intentionally with
Breakthrough Energy Ventures. He has Myrvhold pick top people like Lowell
Wood and professors, and they do, relatively speaking, very thorough
technical due diligence, and...the results are terrible. The BEV portfolio
is bad. As for Intellectual Ventures - how's that anti-mosquito laser
working out, huh? I think Bill Gates picked the wrong people because he
didn't get into technical details as much as Elon did.
Early-stage
investors regard the occasional Theranos and Nikola as part of the cost of
doing business, but they don't understand that their process is also
rejecting most of the good ideas early on.
--------------------------------------
Alyssa:
What's your overall
vision for low-CO2 energy production?
bhauth:
The low-CO2 energy
scenario I'm most optimistic about is: mostly PV solar and wind generation,
with some solar-thermal in particularly sunny areas, and energy storage from
Hydrostor-style compressed air energy storage or from some future flow
batteries. (Not Form Energy or ESS!) In this scenario, liquid fuels would be
made from biomass.
Natural gas is more expensive in Europe, but
also, solar and wind are about twice as expensive in Europe as the USA
because of the lower average sunlight and onshore wind siting issues. That
makes offshore wind competitive, while in the US it's about twice the cost.
Energy storage is also expensive. Still, I think solar plus wind plus either
certain compressed energy storage or future flow batteries will be slightly
cheaper than nuclear power in Europe, considering the costs we see at eg the
Hinkley Point C project.
However, seasonality of energy production
would then be a problem. Germany doesn't get much sunlight in the winter,
and sometimes the wind doesn't blow much for a while. The usual proposal for
that is storing hydrogen in underground caverns, but that's a lot more
expensive than nuclear - which is more expensive than coal, which is what
Germany is going back to now.
So, I think we'll see a lot of LNG
usage for a while.
Alyssa:
Why is offshore wind so much more
expensive? Oceans have stronger winds, which should count for something.
bhauth:
Yes, but the winds on land are relatively better with taller
towers, and the towers are pretty tall these days. Offshore wind can also
enable bigger turbines if the blades don't need to travel over roads, but
that doesn't matter too much.
So, offshore wind has some advantages,
but constructing supports in the ocean is obviously more expensive, and
that's just how the costs work out in practice.
Alyssa:
What
about lithium-ion batteries for energy storage?
bhauth:
Sure, a
lot of people like the idea of using lithium iron phosphate batteries for
energy storage, but there are some problems with that.
First, those
plans relied on battery prices continuing to go down, and they're actually
up in the last year.
Second, the current battery prices that get
reported in magazines come from the BloombergNEF survey, and that's heavily
weighted towards subsidized Chinese batteries for use in China. Not only
does China make a lot of batteries, but I think the prices for American
companies are less likely to be reported when Bloomberg asks. There's no
incentive to report them, you know. So, if you look at the price of
batteries in the survey excluding China, it's about twice their average.
Third, I think the lifetime of lithium iron phosphate batteries for grid
storage is being overestimated. People use cycle life and calendar life
separately, but when you cycle the battery, it breaks up the SEI,
which increases the aging over time for a while as a new SEI layer forms.
Alyssa:
There's lots of sun in the Sahara. Can you just put
solar there and send it to Europe?
bhauth:
Some people are
basically doing that! There's a project in Morocco using an undersea cable
to the UK, the Xlinks Morocco-UK Power Project. But sending power such long
distances, even with HVDC, is pretty expensive. Piping natural gas is
cheaper than moving electricity around. Also, if the goal is mitigating
geopolitical risk relative to getting natural gas from Russia, then long
undersea cables aren't ideal.
The lack of clouds in Morocco and
Algeria is especially important for solar-thermal power, since clouds
interfere with focusing sunlight.
Alyssa:
You said a national
HVDC network is a good idea. What do you think the obstacles to that are?
bhauth:
America doesn't even produce enough aluminum for that
anymore, but the biggest issue is definitely land rights. That was a big
problem with the California High-Speed Rail project too.
Alyssa:
The US managed to make the national highway system. What do you think
changed?
bhauth:
There was more federal land back then, and I
think the legal situation for eminent domain has changed since then.
--------------------------------------
Alyssa:
Increasing US
immigration is sometimes put forward as an EA cause area, but you seem less
positive about that.
bhauth:
As you know, real US median wages
were basically flat for decades. GDP went up, but typical workers captured a
shrinking percentage of it. A lot of Americans think that's because
immigration kept wages down, and I think that view has some merit.
Alyssa:
What about immigrants with in-demand skills?
bhauth:
If you look at the fields that companies supposedly desperately need to
bring in immigrants for, like Material Science or Aerospace Engineering,
there are Americans with Master's degrees having to go into other fields
they're less interested in, because while there aren't that many graduates
in those areas in percentage terms, there are still more grads than jobs. So
every person you bring in for those roles is probably taking a job from an
American. Companies have to say they can't find Americans for a job to bring
in H1Bs, but they're lying, they don't actually try.
Considering the
state of American universities, I'm not going to have much sympathy for
those companies until they're desperate enough to consider hiring people
without college degrees. Some of the best people don't - it's just that
their application is buried in ones from incompetent people who send 100
applications a day, and gets thrown out by HR.
Alyssa:
A lot of
companies were started by immigrants.
bhauth:
Those companies
were started by certain types of immigrants from certain areas, and they're
being used to justify increased immigration of very different types of
people, like broke migrants from
Venezuela who don't have the resources to, say, start SpaceX. The political situation is such that, on the margin, increasing
immigration won't bring in many more people of the types cited as
justification. I also think that in many cases, the slots available for
certain opportunities, like entrance to elite universities, are the limiting
factor.
Alyssa:
Still,
shouldn't bringing in people increase the available jobs? If you double the
number of Americans, then if anything, it seems like the economy would more
than double, because of the better economy of scale, greater specialization,
increased research, etc.
bhauth:
Sure, if they bring a
proportionate amount of capital. What's the net worth of the US, including
the value of land and natural resources? Let's say it's $500k per capita.
OK, how many immigrants bring $500k with them?
Even if they did,
there's also the
culture issue.
Alyssa:
It's possible to take immigrants
selectively, and I think there are some high-quality people that America
wants.
bhauth:
Sure, I know people who should be given
citizenship but can't come here. But while being selective might be
practical in theory, it's been politically difficult in America.
Also, if you look at the criteria proposed, it's things like "having an
advanced degree". If you require that, then you'll get Indian guys with
diploma mill PhDs. If you go further and require scientific papers, you'll
get Chinese guys with paper mill publications. You'll get people who game
the system, and that's exactly who you don't want.
Zvi:
What
criteria would you use, then?
bhauth:
I'd give people the choice
of Ludum Dare or winning in my roguelike.
--------------------------------------
Alyssa:
What do you think is
needed to make US manufacturing more competitive?
bhauth:
The
former CEO of Honeywell, David Cote, wrote a book called "Winning Now,
Winning Later". In it, he said the key to success as a CEO was to find some
way to invest a moderate amount in the future while still giving good enough
short-term results to keep your position long enough to take credit for the
investments.
Executives at publicly-traded American companies don't
want to make long-term investments - they want short-term profits for either
their stock options or a promotion. Jack Welch is sort of the ur-example. He
was considered a genius, perhaps the greatest businessman of all time, but
now, GE is worth less (inflation-adjusted) than it was in 1990.
In
the long run, investors got screwed and GE is a shell of its former self,
but Jack Welch made his money and got out, and investors didn't realize how
bad things were until 7 years after he retired.
By the way, Jack
Welch had a PhD in chemical engineering, so it's kind of ironic that he
ended up being the inspiration for MBA practices that a lot of engineers
hate. Xi Jinping studied chemical engineering too, and while I didn't study
chemical engineering, I have done consulting work for it. If you put the
three of us together, I guess we'd be Curly, Moe, and Larry.
Outsourcing to China was one way of getting short-term profits. Once you're
committed, the Chinese government might start squeezing you, requiring tech
transfer or applying indirect taxes or requiring sales. But that doesn't
show up in quarterly reports until after those executives have already moved
on.
The lack of long-term investment in western publicly-traded
corporations is why privately-owned companies are now doing better. Asian
companies also seem more inclined to invest long-term. See eg
this and
this.
Some people like to say, "Actually, US manufacturing
productivity and production has done well despite the drop in employment!"
But those statistics are no good. A lot of companies do final assembly in
the USA, and deliberately overvalue that final assembly to meet legal
requirements for being "made in the USA". Also, some of the supposed
increase in productivity was just semiconductors getting better.
Because of the lack of investment by public companies, these days, when
stuff is made in the USA, it's often by a privately owned corporation. In
some ways that's fine, but I'm concerned about the effect that'll have on
concentration of wealth.
Another problem is construction costs. The
cost of building a factory or chemical plant in America is generally a lot
higher than in China, and that's a significant factor in siting decisions.
With homes, too, you've probably seen that house construction costs have
gone up quite a bit in the past couple years.
Alyssa:
Why do you
think that is? Why is construction in America so expensive?
bhauth:
Well, that's a complicated topic. Do you know the
Construction Physics or
Pedestrian Observations blogs?
Alyssa:
Yes, I read both of
those.
bhauth:
They have some good posts, but obviously they're
not a complete explanation.
For public projects, I think a big issue
is lack of technical competence on the government side. Even if government
isn't building something itself, it still needs to be able to write
contracts. These days we see a lot of bad contracts with litigated change
orders that end up being really expensive. That's why defense industry stuff
is cost-plus now, and you know how that is.
Zvi:
Why do you
think the government lost that capability?
bhauth:
Normal
institutional rot, I guess? Why does that happen in general? If you have a
good solution, I'd be interested. Maybe the AI alignment people can actually
contribute something: I think institutional alignment is a related problem.
Zvi:
Hah, unfortunately I don't think they've made much progress
on a solution to AI alignment yet. I think institutions just need to be torn
down and rebuilt sometimes. That seems to be the traditional solution.
bhauth:
One possibility I've thought about is competition between
redundant departments. For example, why have
a
Department of Education? Why not have 2 Departments of Education, that
compete with each other?
Zvi:
Interesting idea. You could even
do a libertarian version, where multiple private FDAs compete, and they get
money from the government when the drugs they approve are good, or fined when
they're bad.
bhauth:
I hadn't thought of that.
--------------------------------------
Alyssa:
You've talked about
chemical companies charging much higher prices for some chemicals to
external customers than internal customers. That seems like a market
failure. What do you think causes that, and what would change it?
bhauth:
I think the root problem is oligopolies, but the effect of
oligopolies is amplified by complex multi-stage production processes.
Imagine a 10-step process. At step 1, material A is made, and there's an
oligopoly that leads to slight underinvestment in production.
Zvi:
Underinvestment relative to perfect competition?
bhauth:
Right. So A is slightly more expensive than it should be. Then, the same
thing happens at step 2. At each step, the prices increases from oligopoly
compounds. Some economists have said that oligopoly rents aren't that bad,
but they're not accounting for this compounding.
Zvi:
So, you'd
break up those chemical companies?
bhauth:
I certainly wouldn't
start with chemical companies. Chemical plants need to be big and it's hard
to break them up; there's a conflict between minimizing oligopoly rents and
economies of scale. Also, material costs aren't usually the most important
thing these days. The first thing I'd break up is hospital chains, and
hospitals that bought private clinics. But that would be a big fight.
Zvi:
There's an argument for that kind of policy, but that's not the
type of thing we want to pursue at Balsa Research.
bhauth:
Right, you want to "pull the rope sideways".
Zvi:
I think the
usual position among economists is that monopolies in the US aren't actually
reducing consumer welfare.
bhauth:
Yeah, I know, but I don't
think the reasoning or models involved are very good. I'll leave that
argument to Matt Stoller for now, but...Zvi, you like video games, right?
When EA buys an indie game company, are the fans of that company's games
happy about the merger?
Zvi:
Is that a new EA cause area?
bhauth:
Hah.
Zvi:
I don't think people are happy about
Electronic Arts buying game companies, no. But maybe Paradox Interactive?
bhauth:
That's a lot smaller.
Zvi:
True.
--------------------------------------
bhauth:
Hey Zvi, did you
happen to see my
blog post on stop signs when it was linked by MarginalRevolution?
Zvi:
No, I'll take a look. Hmm. Yes, I think this would actually
work.
bhauth:
The comments were mostly that "this is too
complicated, drivers wouldn't understand it".
Zvi:
Oh, that
would definitely happen sometimes, but even then...
bhauth:
The
failure modes aren't that bad?
Zvi:
Right.
bhauth:
So, this idea seems worth at least trying somewhere.
Zvi:
Sure.
bhauth:
Why do you think it won't be?
Zvi:
There are
way too many veto points. In a lot of places, it's hard to even put in a
bike lane or public restroom.
bhauth:
Right.