A thoroughly sporadic column from astronomer Mike Brown on space and science, planets and dwarf planets, the sun, the moon, the stars, and the joys and frustrations of search, discovery, and life. With a family in tow. Or towing. Or perhaps in mutual orbit.

Showing posts with label predictions. Show all posts
Showing posts with label predictions. Show all posts

Science? You betcha.

I’d been thinking a lot about bets recently.
One evening a while back I was working late in my office on the Caltech campus, and I decided to take a break by dropping into a lecture by a Caltech economist (why not? I didn’t know anything about economics. No better time to learn) who had described experiments they had been performing demonstrating the efficiency of markets at collecting, analyzing, and making decisions based on diffusive knowledge. The specific example that still sticks in my head more than a decade after I heard this single lecture was the real-life case of a computer manufacturer trying to predict next month’s price of computers and printers. Several people in the company were supposed to be experts at these predictions, but they rarely got the right answer. The experiment that the Caltech economist ran involved letting everyone at the company who had any sort of knowledge about any part of the process participate in a predictive market, where they could buy “shares” in a certain price point, or essentially bet on what the price next month was going to be. If they were wrong, all of the shares that they bought were worthless, but if they were right they could win big. This was simple betting, but with a twist. The market price of the “shares” somehow reflected all of individual thoughts and hopes and speculations of what the price would be. No one thought that a computer was going to cost twenty dollars, so you could buy shares of “twenty dollars” for next to nothing. Of course, they were pretty much guaranteed to be worth absolutely nothing, which is less than next to nothing, so you would lose money. As you got closer to the price that most people thought the computers were going to be next month the cost of the shares rose.
All of this experimentation could be considered an interesting exercise except for one astounding point: the economists found that the market was better at forecasting the future price than the “experts” were. Somehow all of the individuals were able to exchange and synthesize information simply by buying and selling shares of prices and all of this information exchange led to better predictions than anyone else was able to make.
As I sat and listened to this that evening at Caltech I was shocked and astounded and excited. Scientists had always thought they had a monopoly on the best way to predict things (“the scientific method”) and yet here was a totally non-scientific method that seemed to lead to truth in a pretty clear way. It was a strange truth: not one proved with postulates and experiments, but one simply deemed the most likely.
Before getting too carried away with these ideas, though, I also learned at that lecture that markets aren’t perfect predictors and that the economists could run similar experiments where, with a few simple tweaks, they could make market bubbles and other odd effects. This fact would come as no surprise to anyone who has read a newspaper in the last few months.
I walked back to my office that night with my head trying to reconcile science and markets. OK, so perhaps this market approach couldn’t lead to truth in quite the way that the scientific method could, but maybe there was still a place for it. What about scientific markets for things that can’t quite be proven beyond sufficient doubt but that most scientists would be willing to bet a lot on. The first thing that came to mind was climate change. To most earth scientists, the only arguments about climate change are precisely how strong an affect it will be in different places. But somehow, because of the complexities of the questions, the public frequently sees the disagreement over the uncertainties rather than the fundamental agreements. Clearly, this was a place where a market could work. What about betting on the magnitude of climate change? Buying shares of the average temperature rise by 2050?
While this was all fun speculation, it also seemed obvious that given that the people were confused by the science, they were just as likely to be confused by scientists buying shares in a future temperature market. Clearly this was interesting, but going nowhere. But what about just using the ideas amongst scientists? If I could get all of the astronomers around to buy into a market on whether or not earthlike planets would ever been found around other stars, for example, I would have an effective way of collecting all of the disparate information that everyone had and coming up with the best prediction based on all of the data.
But then I realized that the idea could be taken one step further. Scientists could engage in what amounts to insider trading in markets! If I really believed, for example, that the climate was warming, and that, for the most part, the rest of the world was not dealing rationally with that fact, I should buy land somewhere in central Canada where it is right now too cold for most people and then I should reap the incredible returns when the land prices skyrocket because people can no longer live in Los Angeles anymore (as crazy as this sounds, I know one scientists who independently came to the same conclusion and bought himself some [currently] chilly property in Minnesota). A market does exist for climate change speculation, only it is a bit more indirect than simply betting on temperatures.
As a mere professor, I don’t have the financial means to follow up on my late night thoughts, but the ideas still continue to percolate in my head. I remain convinced there must be a way to figure out something about which you were certain but which was not generally understood and then use that knowledge to hit it big.
It is a testament to my general lack of financial thoughtfulness that after all of this interesting speculation and pondering about bets and markets and scientific insider trading that the best I ever came up with for my own personal attempt to hit it big with a piece of scientific speculation was a bet made in 2000 that someone would find a tenth planet before January 1st, 2005, with the winner of the bet to receive five bottles of champagne. Perhaps it is also a testament to my happiness at drinking champagne.
I think, though, my five-bottle-of-champagne market tells you that scientific markets, in the end, won’t work. I think that the best scientists are more motivated by being the one to discover and prove the truth than by being the one to guess correctly at the truth and profit off of it. Even though I am exceedingly certain that the world will be warmer fifty years from now I would still rather figure out a way to demonstrate the fact convincingly (or better: prevent it) than go buy Canadian land. And for those years between 2000 and 2005, when no tenth planet was in sight, I was not busy considering the financial plight of the loss of five bottles of champagne, but rather I was searching the sky, night after night, in the hope that when the champagne was drunk, it would be drunk in honor of the new planet I had found.

Vague hopes and just happenings

Last week, when writing about potential discoveries in 2008, I admitted to having some specific ideas, some vague hopes, and that some might just happen. I wrote about some of my specific ideas for the year. Here I’ll talk about the other two – and the much more common – types of discoveries.

The “vague hope” types of discovery are very different from the “specific idea”, or, to be more specific, the make-a-hypothesis-and-test-it types of discoveries favored in explanations of the scientific method in 8th grade classes.

The best example from my own past that I can give us this type of vague hope was the discovery of Eris itself (the Kuiper belt object larger than Pluto that caused the uproar over Pluto’s status and finally forced the demotion of Pluto to a dwarf planet. You can read much more about Eris on my detailed site). When I started scanning the skies almost a decade ago looking for large objects in the outer solar system, I certainly didn’t know what specifically was going to be out there, but I knew that there was a not bad chance that we would find something bigger than Pluto eventually (though I will admit that at one point midway through the decade I contemplated declaring the search over, thinking we had found all there was to find of interest. One of my students eventually talked me out of it). On January 5th 2005 (3 years ago today, even though it feels like a much longer time ago to me), I finally spotted the object that we first called Xena -- then temporarily became named 2003 UB313 and finally became Eris -- moving very slowly past the stars on my computer screen. It’s not that we had had a hypothesis to prove, just a realization that if you search more of the sky than anyone else has since the invention of the computer and digital cameras, you will certainly find things that no one has before. Will the thing you find be bigger than Pluto? Maybe yes, maybe no, but, as long as you are hoping, you might as well hope in the direction of bigger.

Now to 2008. The survey of the skies that led to the discovery of Eris and the other dwarf planets ended more than a year ago when we finally had scanned almost all of the skies that can be seen from our telescope at Palomar Observatory. But after spending most of a decade searching the skies for newer and larger bodies, it was hard to actually quit. What to do? Start over again. But this time I am doing it with the knowledge gained from doing it the first time, so this time we are doing everything – I hope – right. In practice, the most important thing that this means is that we are extending the survey to find extremely distant objects that we would have missed the first time around.

Why? Vague hope. Or perhaps it is better called “directed hope.” From our discovery of Sedna, which spends most of its time far far away from the sun, we realized that there might well be many many objects out at those distances, and that some of them could be quite large indeed. By “large” here, I am talking about something perhaps even the size of Mercury or of Mars (hope. remember: the key here is hope). I certainly don’ t have a specific scientific hypothesis supported by equations and calculations that some such object is out there, just a realization that it plausibly could be and that no one has ever done a thorough search.

So that could be one very exciting answer for discoveries of 2008: a Mars-sized body orbiting at perhaps twice the distance of Eris. Amusingly, by the current IAU definition, such an object would still be called a dwarf planet, though it would be a dwarf bigger than some real planets. If we really did discover such a thing it would probably re-light the planet definition fire, and we would all get to watch astronomers begin arguing once again.

To be honest, I have to admit that I think finding such a beast is a bit of a long shot. The main reason is that even though I am pretty convinced that such large dwarf planets are out there floating in the same region occupied by Sedna, they are likely to be too far away for our modest telescope at Palomar Observatory to see. Really, we would have to get pretty luck. But again: directed hope. It just might be there, and we won’t know until we look.

Finally, some discoveries are the type that I said just happen, though, really, that is not quite the right phrase. No discover ever just happens, I don’t think, but, sometimes, while you’re looking for something else, or examining something just because you’re curious, or trying hard to understand one little detail of something that just doesn’t make sense to you, something will pop out that you had no idea was coming.

The discovery of Sedna was like this. We were scanning the skies hoping that we might find something bigger than Pluto, knowing that we were bound to find many things than no one had ever seen before, but we never anticipated anything like Sedna.

When I first saw Sedna I wasn’t even convinced it was real. It was so faint that I thought it might just be a recurring smudge on the pictures I was looking at. The first email I sent out to Chad Trujillo and David Rabinowitz, the two guys with whom I was working, said “Not that I think it’s actually real, but the thing I might have just found is really really far away and really really big.” It was true, which is why we are now looking for things even bigger and further away.

Sedna orbits in a region of space that astronomers expected to be essentially empty, so it really had never occurred to me that when we began hunting in the skies something like Sedna might show up in our snares. Sedna is smaller than Eris by perhaps 20-30% and thus smaller than Pluto, so it didn’t get nearly as much press coverage as the Eris discovery, but, of the two, Sedna is by far the much more interesting scientific discovery. Eris’s main importance is less in the scientific knowledge that something bigger than Pluto exists but much more in the cultural importance in that it was the discovery that finally drove home the fact that Pluto is not a unique oddball at the edge of the solar system but simply one of the largest members of a much more extensive population. But Sedna tells us something we never knew before. What? It is still not clear; we’re working hard to understand all that Sedna says, but by the time we are done I hope that Sedna is the beginning of an intricate story of the birth of sun in a crowded cluster of stars which eventually caused Sedna to be peeled out of the inner solar system to become part of this new still unnamed region beyond the Kuiper belt. Stay tuned. There are other possibilities for what it might mean. But whatever it turns out to mean will be something we never even considered when we first started looking across the sky.

What might we discover unexpectedly in 2008? There is no way I can tell, of course, but I can at least give some areas of possibility, because all of these accidental discoveries come about with a lot of hard work to make the accidents possible. If I had to place bets on what project is most likely to lead to something like this, I would have to again say the new sky survey. Anytime you are looking over vast areas of sky in ways that no one ever has before your chances of having a good accident are high. But that is not the only project my group and I are working on these days – we’re thinking about storms on Titan, giant collisions and icy atmospheres in the outer solar system and more – so one of the other projects may sneak in as a long-shot. Or something new may come along unexpectedly.

Happy third anniversary of the discovery of Eris, and hope for discoveries – expected, hoped for, and accidental in 2008.

Discovery 2008

A friend asked me a few days before Christmas, “What big discoveries are you going to make in 2008?” to which my immediate reply was “I have no idea.” In thinking a little more about this question for the past week, I realized that the real answer is a little more complicated than that. I think that discovery comes in three different flavors. I think I should have said “I have a very specific idea about some, I have vague hopes about some others, and some might just happen.”
The first type of discovery is the classical type of science that we all learn about in eighth grade: a scientist has a hypothesis and sets out to prove it. Some of the most famous science stories of the past involve this sort of hypothesis testing: Einstein theory predicts that the sun’s gravitational field would deflect the positions of nearby stars in the sky; Eddington mounts an expedition to view stars during an eclipse and proves Einstein right; and relativity is instantly enshrined as the newest part of physics. In many ways this classical type of discovery is also the most respected. It is certainly the most dramatic. In thinking back over interesting discoveries I have made over the years I can think of none whatsoever that fit this category.
Why not? Partially it is because not a lot of actual science works in this way. In looking over the list, say, of Nobel prizes in physics, most of them are not for this type of eighth-grade-textbook hypothesis-testing science. Most science proceeds in one of the other two flavors.
Interestingly, though, for 2008, I am indeed working on one such type of hypothesis-testing discovery. It would not be nearly as dramatic as Einstein and Eddington or as universally encompassing as relativity. But, still, it’s a hypothesis I have that I am attempting to test. The hypothesis suggests that collisions between large objects like Pluto and Eris and others out in the Kuiper belt played a much larger role in sculpting these objects than anyone previously thought, and, in particular, the collisions removed much of the ice that these cold objects would have had in their interiors. The key to testing a hypothesis like this is to figure out a test, like Eddington realizing that an eclipse would provide the key measurements. For my hypothesis there are two tests. The first is to show that a giant collision indeed removes ice. The second is to show that an object that has a lot of ice didn’t have a giant collision.
The giant Kuiper belt object and dwarf planet Quaoar (for details on these large Kuiper belt objects see my dwarf planet website) looks like it indeed had a giant collision. The key piece of evidence is that it has tiny moon circling it which appears to be a remnant of the collision, much as our Moon is a remnant of a collision between a Mars-sized object and the Earth 4.5 billion years ago. In my hypothesis, this giant collision that happened on Quaoar long ago should have removed most of the ice inside of Quaoar, meaning that the interior should be much more rock than ice. I can test this prediction by figuring out how much Quaoar weighs and seeing if its weight is more rock-like than ice-like. And I can figure out the weight by using a method first used by Newton more than 300 years ago. If I track the path of the moon around Quaoar and calculate how much time each orbit takes I can use that information to get the weight. If Quaoar is heavy and rock-like, the moon will be pulled along quickly, while if Quaoar has more ice and is lighter, the moon will take longer to go around.
The moon of Quaoar is, unfortunately, so tiny and so close to Quaoar itself that it is impossible to see it with anything other than the Hubble Space Telescope. Last year I wrote a proposal asking to use the Space Telescope to track the moon, the proposal was accepted, and now the observations are scheduled for sometime in the spring (turnaround time is not fast!).
One of the nice things about hypothesis testing is that the answer can sometimes be instantly clear. As soon as the pictures from the Hubble Space Telescope are taken I will be able to quickly calculate the amount of rock and ice inside Quaoar and see if I’m right or wrong. If I’m wrong the whole hypothesis is dead and thrown out the window. If I’m right it will be time to go on to the second test.
The second test is to show that an object that has a lot of ice didn’t have a giant collision. Another giant Kuiper belt objet and dwarf planet, Orcus (which is about the same size as Quaoar, that is to say about half the size of Pluto), also has a moon, and we had observations from the Hubble Space Telescope about a year ago which allowed us to track the orbit of that moon and determine that Orcus is much icier than anyone had anticipated. The moon of Orcus might have formed in a giant collision, but the moon is a bit larger than the other moons of dwarf planets that look like collisional remnants. Perhaps, then, Orcus never experienced a giant impact and the moon is simply another Kuiper belt object captured from space. For my hypothesis to be correct, this moon must be captured, because otherwise my hypothesis would predict that Orcus should be rocky, not icy.
Happily, I think we can test whether or not the satellite of Orcus came from a collision or was captured from space by looking in detail at the composition of what is on its surface. Typical Kuiper belt objects (like one that Orcus would capture) appear to look very different in surface composition from collisionally formed satellites. The best way to look at the surface composition is to examine the spectrum of sunlight after it has reflected off the surface of the satellite. Like the satellite of Quaoar, the satellite of Orcus is small, faint, and close in, so looking at the spectrum is particularly difficult. We are thus going out to the Keck telescope – the largest telescope in the world – on the summit of Mauna Kea in Hawaii at the end of March to spend three nights trying to get a clear glimpse of the satellite. We’ll be using the fancy new Laser Guide Star Adaptive Optics system at the observatory which shoots a laser beam into the sky to help correct the smearing caused by the earth’s atmosphere. I’ll write in detail about that at the end of March.
Like the first test, the results of this second test will be apparent almost instantly. In only a few hours I should be able to figure out what the moon of Orcus has on its surface. If the moon of Orcus looks like it was made in a collision I will again have to toss the hypothesis out. But if the moon appears captured, the hypothesis will have survived two important tests, and it will suddenly have to be considered seriously. That doesn’t mean it will have been proved correct; other astronomers will look for other ways to test and challenge the hypothesis. But if it is correct, it will pass all of these tests and challenges and become part of our understanding of the outer solar system.
So while I said “I have no idea” to my friend who asked about discoveries for 2008, it is clear that, at least in this case, I have a very specific idea and definite observations to confirm or refute the idea. I presented these ideas at a scientific conference in December and suggested that I thought there is approximately a 15% chance that I am correct. I think that’s about right. My hypothesis would require a fairly major revision of some of our fundamental ideas about how the outer solar system formed, and such major revisions don’t come along too frequently, so if you felt like betting I would recommend betting against me. But, still, at least I have one idea of what might be a discovery in 2008.
Next week I’ll write about the other two types of discoveries and what else might happen in 2008.