Tuesday, September 18, 2018

Fusion optimism

Fusion is famously the technology that's always 20 years in the future.  And because of that I hadn't really seriously considered it as a realistic solution to climate change, sustainable energy, etc.  But recently I've become more optimistic about the prospects of fusion so I thought maybe I'd explain why I've changed my views.

What is fusion?  Well, it's a process that releases energy by combining elements together.  Atoms tend to resist this, so they have to be smashed together very hard in order for it to work.  In a thermonuclear bomb the fusion is triggered by a standard fission bomb going off to provide the needed energy.  Atomic weapons are rather hard to contain, though, and impractical to use as part of a power plant.  So for fusion we'd need to heat up some hydrogen very hot and keep it all together rather than expanding and escaping as very, very hot things like to do.

In the Sun our star's gravity is enough to keep everything together.  That works very nicely but sadly masses as large as the Sun play together with power plants every worse than atomic explosions do.  But gravity is widely regarded as the weakest of the fundamental forces for good reason and if it takes a impractical amount of gravity to keep some hot hydrogen contained it only huge but doable amount of electromagnetism to do the same thing.

Because the super hot hydrogen you want to use for fusion is constantly knock electrons off it's no longer a gas but an electrically active plasma and you can use a magnetic field to pin it in place.  For the temperatures needed not the sort of magnetic field that will fit on your desk but more one the size of a house.  Still, a lot more practical than atomic explosions or Sun sized masses.

It would be much easier if a reactor could be made smaller so that they could be build and modified quickly and cheaply.  But you need the magnetic field strength combined with its size to be enough to keep the super hot particles in place.  To be practical you need superconductors to make the magnetic field and those stop working if the magnetic field gets too strong.  Hence, with the maximum magnetic field our materials can sustain there's a minimum size.  Which has been expensively large.

Back in the mid 1970s the people running our fusion research program looked at the problems involved and figured that they'd need a lot of funding to iterate on building huge building sized reactors until all the bugs got worked out.  The came up with a set of set of plans for how quickly they thought they could get fusion working reliably depending on how big their budget for building trial house sized magnetic reactors was.  Here it is.

As you can see the level of funding eventually selected corresponds with enough money to pay a few scientists to think about the problem but with no giant house-sized magnets built to see if their ideas work, somewhat below what people in 1976 thought would never successfully create fusion.

Put in this perspective the people saying that we could have fusion in 20 years way back in the 1970s seem a lot more reasonable.  It wasn't that we would have fusion but that we could with a large enough investment.  An investment that never materialized.

Well, now in the 2000s the Europeans have gotten around to building a house sized magnet facility called ITER that, if funding keeps being poured in, should eventually get the kinks worked out.

But something really cool had happened since the 1970s.  This technology called Magnetic Resonance Imaging, MRI, has taken off the field of medicine letting us look into human bodies for diseases.  This is a huge business and a very practical and immediate one.  MRI machines want superconducting magnets handling intense fields and so lots of people have had very practical and immediate incentives to figure out how to create superconductors that can handle higher field strengths.  And so, they have.

A group at MIT came up with something they call the ARC reactor (groan) that uses a sort of superconductor that can handle fields almost twice strong as the superconductor used at ITER.  "Almost twice" might not sound like much but apparently the energy you can get out of a fusion reactor scales as the fourth power of the field strength and so there was scope to make the reactor as powerful as ITER in a tenth the volume.  And now they've formed a commercial company around this.

Just building the reactors isn't the end of the story.  There's a lot of engineering effort that has to go into dealing with issues like plasma instability, having the reactor materials deal with neutron damage, and so on.  But having reactors lets engineers and scientists try out solutions.  It's possible that one of these challenges might be insurmountable.  But there's no reason to believe that and if they aren't then we should expect to see practical fusion power, finally and actually, within 20 years. 

Tuesday, July 3, 2018

Book Review: The White Man's Burder

I recently finished reading William Easterly's The White Man's Burden: Why the West's Efforts to Aid the Rest Have Done So Much Ill and So Little Good.  Reading some reviews online before starting it I was expecting it to be much more of a screed than it actually was.  Easterly criticizes the efforts of the World Bank and International Monetary Fund to further the development of poorer countries but his nuanced criticisms are in sharp contrast to most of their critics.  There's a lot of discussion about all the complicated moving parts that go into a modern commercial society, all the problems with both causing trust and causing trust to be justified between commercial actors and how people have found ad hoc solutions in the absence of regular institutions, and how it's very hard to know in advance which interventions will actually make things better rather than just funnel money into someone's pocket.  He provides some fairly compelling statistics showing that development aid of this type doesn't help on average and explains a bit why there can be occasional negative consequences to match the occasional positive consequences.  I think in my earlier review of Dancing in the Glory of Monsters I gave an illustration of the damage that ill-considered aid can cause.

He talks a bit about programs that actually help.  Initiatives lead by residents of the countries being aided get a lot of praise.  But he also talks about accountability.  Having one organization do one thing and having their impact on that thing measured.  So if this had been written later I think he might have specifically praised Effective Altruism institutions like Givewell.

There's also a chapter on peacekeeping, which I found less convincing.  Steven Pinker had some pretty convincing statistical arguments in Better Angels of Our Nature that blue helmeted peacekeepers are a substantial net benefit.  Easterly, though, uses anecdotes like what happened in Somalia to argue that peacekeepers are useless, in sharp contrast to his statistics and scorning of anecdote in the aid chapters.

Overall I'd recommend this book, particularly for the dives into the nitty-gritty details of institutions in the third world.

Saturday, June 9, 2018

Waymo is getting serious

First of all, in case you aren't following this sort of thing, the Uber crash I mentioned in my last post on self driving cars is actually a lot worse than it looked at first.  According to the NTSB report the victim of the crash had been detected but
emergency braking maneuvers are not enabled while the vehicle is under computer control, to reduce the potential for erratic vehicle behavior. The vehicle operator is relied on to intervene and take action. The system is not designed to alert the operator.
 which makes me wonder if this rises to the level of criminal negligence.  Certainly there are currently no localities that are allowing Uber to test on their roads which might be the end of Uber's program.

Tesla has also been in the news with some crashes involving its cars running on autopilot.  But autopilot of the sort you normally have in a boat or plane or one of Tesla's cars isn't full self driving.  The pilot or driver is supposed to remain on alert and react if the vehicle starts to do something unsafe.  In a plane or boat normally this involves noticing and reacting in the space of a couple of minute or so.  But in a car you might need to react in seconds which makes this whole approach much more dangerous.

The NHTSA defines a number of levels of autonomy.

Levels 0 and 1 aren't a problem because the user is always going to have to be paying attention.  Well, no more of a problem than we're already dealing with.  Levels 4 or 5 are fine because the user doesn't need to pay attention.  The car may get stuck but it can be trusted to pull over to the side of the road while the driver prepares to drive.  But levels 2 and 3 where the driver has to remain alert while not doing anything which I don't think is very realistic.

In contrast to all of this, it looks like Google's self driving car spinoff Waymo is doing everything right.  They're going straight for level 4 or 5.  They're testing carefully and not taking shortcuts like Uber.  And they've been working on the problem since 2009 having just accumulated 7 million miles of driving which is pretty impressive.  Going by some stats off Wikipedia a fatal accident happens around once every 2 million miles and Google hasn't had one yet.  There have been some accidents, even serious ones like this one recently, but that was a vehicle swerving into an oncoming lane and hitting the Waymo car.  In every incident I've heard of the Waymo car was not at fault so I think we can say at this point that Waymo cars are safer than human drivers, on average.

But now Waymo has just ordered 62,000 minivans from Crysler.  They'd been giving rides for free to people around Phoenix but only with a human driver monitoring things.  But with tens of thousands of cars they're going to be looking to be going fully autonomous.  It looks like fully autonomous self driving cars are about to actually be a practical thing, at least in Phoenix or whatever city those 60,000 cars will be serving.

Tuesday, May 22, 2018

Nuclear power in space

I'm a little behind on my blogging but a couple of months ago NASA released some information on their new Kilopower small reactor for use in space.  So I thought this would be a good opportunity to write something about power generation in space. 

Most satellites and probes in the inner solar system use solar power.  This was one of the first practical uses of solar power back when solar panels were very expensive and now that solar panels have gotten much, much cheaper it's even more of a good idea.  The solar panels on your roof are built tough and will weigh about 10 kg for every square meter or, on the Earth.  At 1300 watts of sunlight coming in times an optimistic 20% efficiency for typical solar cells you've got 26 watts per kilogram.  According to Wikipedia more expensive spacecraft-grade solar panels will give you 77 watts per kilogram, but even then the majority of the expense is going to be lifting the weight of the panel into orbit.

Solar panels are great if you're as near to the Sun as Earth is but they start to work less well as you get further from the sun.  The incoming sunlight per meter goes down as the square of the distance from the sun so when you're out at Jupiter, 5 times as far as Earth from the Sun, you're only receiving 50 watts per square meter and your spacecraft grade panels are only generating 3 watts per kilogram.

For trips to the outer solar system NASA has traditionally relied on radioisotope thermoelectric generators or RTGs.  The idea is that you take some particularly radioactive isotope, such as Plutonium-238, and just let it generate heat by decaying.  These will actually generate about the same electric power per kg of weight as solar panels will at Jupiter's orbit but without the worry of always having to point your panels at the sun.  And as you travel beyond Jupiter to Saturn they become much more mass efficient.

They do have their problems, though.  In order to be intrinsically radioactive enough to generate such heat the fuel has to have a pretty short half life.  Plutonium-238 will half decay away in just 90 years.  That means all the natural P-238 is gone and any we use has to be synthesized at great expense.  It also means that half the power will be gone after 90 years.  Missions don't last 90 years but over the long distances of the outer solar system they can last more than a decade and the power output from these generators will go down noticeably on those time frames.

And then there's the problem of radiation.  To generate heat the material must be very radioactive by its nature and that means dangerous.  If there were to be an explosion during launch the plutonium might be lost.  The RTG is designed to stay as one intact unit in the event of catastrophe and sink to the bottom of the ocean but it still causes worry.

That's one area where a nuclear reactor rather than an RTG has benefits.  Kilopower and most other reactors mostly run of Uranium-235.  In small quantities U-235 is pretty safe.  Well, it's pretty toxic but toxic in the manageable sort of way that lead or mercury or bismuth are toxic.  It has a half life of 700 million years, quite a bit more than P-238.  Long enough for large amounts of naturally occurring U-235 to remain since the creation of the solar system.  And while it gives off a little bit of radiation from decay that decay occurs so slowly that you can mostly just ignore it.

Why do nuclear accidents like Fukushima release so much radiation then, if the uranium fuel that goes into the reactor is fairly safe?  Because reactors work by transforming uranium into other elements to release the energy stored in it.  And the elements that are created are frequently even more naturally radioactive than P-238 is.  So a reactor that has been running is just as dangerous as an RTG is.  But a reactor that has not yet been turned on is actually fairly safe.

So that's a clear reason to prefer a reactor over an RTG in the outer solar system where solar doesn't work very well.  What about closer in?  Mostly you would want to consider nuclear in cases where you need reliable power for things such as life support through the night.  On Mars the night is going to be 12 hours long and adding enough lithium-ion batteries to last that long would effectively be a 20 Watt/kg system, and since you need twice as much solar panel during the day to both charge your batteries and run your life support you're down to a net of 13 continuous watts of power per kg of solar panel and batteries.  Which is still better than nuclear.

But what about the Moon which rotates once a month, with nights over 300 hours long?  With that the heaps of batteries you have to pile under your solar cells only give you .8 watts per kg of equipment making it clearly less mass efficient than nuclear.  Is it worth the problem of dealing with radiation?  That's for NASA to decide.  And it would only be NASA that gets to make that decision.  Such lightweight reactors have to use highly refined uranium that could be made into nuclear weapons in the wrong hands.  So there isn't much chance of SpaceX or other private ventures getting their hands on these.

Wednesday, May 16, 2018

Falling fertility rates shouldn't be a problem forever

There are a lot of countries in the world with declining populations.  Historically this isn't unusual if there's famine or disease or war causing it but in the developed world this is mostly just people choosing to have fewer children.  In Japan it's all the way down to 1.46 children per woman which, since half of kids are male, means that every generation will tend to be three quarters as large as the one before it.

There are a lot of reasons for this.  People having better options than they did in the past seems like the biggest one.  There are differences within the industrialized world in many ways involving religionattitudes towards child care, and other thing.

But those are societal factors and there are also individual factors at play in a couple's decision to have kids.  People have different personalities and personality seems to have an effect on child bearing just like you'd expect.  Here's a study I recently saw linked on twitter.

So, for instance, people who are more Agreeable based on the Big Five personality traits social scientists normally use seem to be more likely to have children.  And these personality traits seem to be substantially heritable.  So if nothing were to change we should probably expect evolution to have its say and the population decline to eventually halt and reverse itself through natural means.

I don't actually expect things to turn out that way.  Rather, I expect technological improvements in the ease of having and raising children to make a bigger difference sooner.  And societal improvements would be nice too but I'm not counting on those.  How we're going to take care of large numbers of older people as they live longer without a large working population is going to be a concern for the next few generations but there's no reason to worry about humanity going extinct or anything.

Of course technological progress cuts both ways and maybe virtual reality will reduce childbearing rates further.  But again I expect the next generation being more like the sort of people who had kids in the last generation to eventually rescue things.

But the sad moral of the story is that we might have we might have gotten out of the Malthusalian trap.  Maybe even for quite a while.  But I'm pretty sure that in the very long run the human population will increase again until it's reduced to a subsistence level. 

Tuesday, March 20, 2018

Self driving cars are like airplanes

Yesterday, for the first time, an autonomous car killed a pedestrian.  It isn't clear that the car was at fault but we're almost certainly going to have an accident where the car was at fault at some point.  At this point autonomous cars haven't driven enough miles for us to know if they're currently safer or more dangerous than human drivers.  But I think they have the potential to be much safer in the long run for a combination of technical and institutional reasons.

Technically, cars can learn in the same way that humans can.  But while we humans are mostly limited to learning from the situations we encounter a fleet of cars can hope to learn as a unit.  Some accident occurs, engineers analyze it, and then no car in that fleet will make that particular mistake again.  It's reasonable to think about robo-cars trained on a body of experience far greater than a human could amass in a lifetime.

And I think that robotic car manufacturers have the right incentives to care about this too, for the same reason that airlines do.  People are just less inclined to trust other people guaranteeing their safety than their are their own abilities.  This is true when people choose to drive long distances rather than fly because they're afraid of some airplane disaster.  This is even true when they know that people like them are more likely to kill themselves driving than a professional pilot is to kill them flying.  It's sort of irrational but it's a fact.  And I'm pretty sure the same principle applies to autonomous cars.

For this reason airlines are scared of airplane crashes in a way most other industries aren't.  When a planes crashed into the World Trade Center on 9/11 lots of people became afraid to fly and started driving instead.  This caused 1600 extra car deaths in the year after 9/11, increasing the death toll of the attack by 50%.  Many airlines nearly went out of business.  So while any individual airline might be tempted to skimp on safety they're all terrified that their competitors will and then cause a crash that will seriously hurt everyone.  In many industries the regulated parties use their influence to make the regulator weaker but for the FAA the airlines do their best to make it stronger.

I think the same dynamic is liable to occur in the autonomous cars industry too.  People are just inclined to trust their own driving over a computer's unless there's very clear evidence that the computer is better.  As long as people can drive themselves on public roads autonomous car companies will be scared that an accident involving one of their competitor's cars will make people want to do just that.  So I expect that once the industry grows and stabilizes enough for a good regulatory body it will be pretty demanding.  And I expect that the companies that get involved will be fairly safety conscious about their autonomous cars even if they're lax about other things.

UPDATE:  Actually the crash that prompted me to write this actually looks pretty bad for Uber, but I still think the forces involved will make autonomous cars safer in the long run.

Monday, March 5, 2018

The Drake Equation again

I was walking in to work today and as I did I was listening to a nice podcast on the Drake Equation.  The Drake Equation is an estimate of the number of civilizations in the galaxy based on things like how many planets there are, how many develop life, etc.  I learned a lot in the Podcast but it reminded me of a post I'd been meaning to make about why I think the origin of life probably wasn't the hard part in creating us.  Also, I promise this post on the Drake Equation is more pleasant than the last one.

A graph:

Dates taken from Wikipedia's timeline of life and timeline of the future.

It was just a pretty short amount of time, geologically, from when the Earth cooled down enough for oceans to start forming until we have proof of the first life - just 120 million years.  And that's probably a conservative estimate.  But from there it took three quarters of a billion years for photosynthesis to arise.  Then one and a half billion until one bacteria swallowed another in such a way as to turn it into a mitochondria and then become a big complex eukaryotic cell.  Then another billion before we had real multicellular life. 

So just looking at the timelines involved it seems like the origin of life wasn't the hard part.  If you're interested Nick Lane has some excellent books about the biochemical difficulties of these steps and why life might have been the easy part but for the Drake Equation the important part is that becoming complex took a long time.

And it's also important that life only had so much time to become complex because in just another billion years the brightening Sun will heat up our planet enough to evaporate the oceans and then there's not much chance of intelligent life evolving.  And it's very lucky that photosynthesis showed up so early.  If the Earth's carbon dioxide atmosphere hadn't been broken down into oxygen then Earth might have had a runaway greenhouse effect by now.  And without oxygen in the atmosphere to form ozone we might have lost the hydrogen atoms we need for water to the Sun's solar wind.

Looking at that timeline makes me feel optimistic that the reason there don't seem to be any aliens in the galaxy is that evolving intelligent life is hard, rather than that intelligent life tends to meet a grisly end.

Fusion optimism

Fusion is famously the technology that's always 20 years in the future.  And because of that I hadn't really seriously considered it...