A Lot of Knowledge Is a Dangerous Thing

I was looking at some polling about science over the weekend, and discovered that they helpfully provide an online quiz consisting of the factual questions asked of the general public as part of the survey. Amusingly, one of them is actually more difficult to answer correctly if you know a lot about the field than if you only know a little. I’ll reproduce it here first, if you would like to take a crack at it, and then I’ll explain why it’s tricky below the fold.

Choose only one answer– this is being recorded for SCIENCE!

So, what’s the problem? The correct answer is obviously “Satellites,” right?

The problem is that two of the answers are technically correct. GPS is based on a constellation of satellites sending out radio signals, but those satellites contain atomic clocks. And the atomic clocks in the satellites rely on magnets to function.

The key idea behind the atomic clock is that atoms have discrete energy states, determined by quantum physics, and move between those states by absorbing or emitting photons of light. Light, in turn, has a frequency associated with it that is determined by the energy of a single photon.

An atomic clock uses the discrete energy levels of an atom as a frequency reference. Unlike mechanical clocks which depend on the construction of the pendulum, or quartz clocks that depend on the properties of an individual quartz crystal, every cesium atom in the universe is guaranteed to be identical to every other cesium atom in the universe, with the same basic separation between energy levels. If you have a frequency source, and tune it so that it is at exactly the frequency that cesium atoms like to absorb, you can use that frequency to measure the progress of time– to be specific, every 9,192,631,770 oscillations of the light absorbed or emitted by a cesium atom moving from one hyperfine ground state to the other is one second.

GPS relies on precise timing– your GPS receiver determines its position on the surface of the Earth by measuring the time required for signals from at least three different satellites to reach its position. The orbital positions of the satellites are well known, and light travels at a fixed speed, so the travel time determines your distance from three of those satellites. That, in turn, fixes your position at a specific point on the surface of the Earth.

So, where do the magnets come in? Well, in order for the clock to work, you need to take a bunch of cesium atoms, expose them to the light that you’re hoping to use as a reference, and see if they move from one state to the other. In order to know this, though, you need to know which state they started in, and which state they ended up in. Magnets can help you do this state selection, as shown in this schematic of an atomic clock taken from the PDF file at the bottom of this NIST page:


The clock starts with an oven containing a chunk of cesium. Atoms from the oven stream out in an atomic beam, but the atoms are initially distributed over many states, when clock operation demands that they be in a single well-specified state. In older atomic clocks, and the clocks on the GPS satellites, the specific state of interest is selected out by using magnets to deflect the other states out of the beam, so they don’t enter the area where they are exposed to the light.

The physics behind this deflection is the same as the Stern-Gerlach experiment, done by Otto Stern and Walter Gerlach in 1921. Stern and Gerlach sent a beam of silver atoms between the poles of a tapered magnet, so that the magnetic field was stronger on one side than the other. The beam of atoms split into two beams, which they detected by letting them fall on a photographic plate.

As is typical of pioneering experiments, Stern and Gerlach actually measured something other than what they thought they were measuring– they thought they were just seeing an effect cause by an electron orbiting the nucleus of a silver atom, with atoms whose electrons orbited clockwise experiencing a different force than atoms whose electrons orbited counter-clockwise. but the planetary type model behind that picture doesn’t work. Instead, what they discovered was the intrinsic spin of the electron, which has its own angular momentum as if it were a spinning ball of charge. That angular momentum can take on one of two possible values, and in the case of silver, leads to a force on an atom between Stern and Gerlach’s magnets that is either up or down, depending on the particular electron.

Stern and Gerlach were fortunate that they were using silver, which only has two states. Other atoms have more complicated level structures, and split into more than two beams, but the essential physics is the same. This splitting is the basis for the state selection in an atomic clock– the cesium atomic beam passes through a Stern-Gerlach magnet, and then a pinhole blocks everything but a single state that happens to be bent onto exactly the right trajectory.

After the cesium atoms are exposed to the light source, they enter a second set of magnets, which again splits the atomic beam, with the atoms that have changed states being sent one way, and atoms that have not changed states sent another way. Measuring the fraction that changed tells you how close the frequency of the light is to the frequency the cesium wants to absorb, and allows you to correct the frequency to keep it right on target. That, in turn, provides an absolute reference for the GPS satellite to use in determining the time, which it encodes into the signal that it broadcasts to your GPS receiver.

So, while satellites are the really important part of GPS, and the correct answer to the question, the satellites wouldn’t work without magnets. And now you, too, know enough to be confused by a simple and straightforward poll question.

(Anticipating a possible comment: the schematic shown is for a Ramsey-type clock, with the atoms interacting with the light field twice. A comment in one of the linked references suggests that the actual GPS clocks may be Rabi-type clocks, with a single large interaction region (though I could be misreading that). I couldn’t find a picture of the clocks used in GPS, though, so you’ll have to live with this one.)

44 thoughts on “A Lot of Knowledge Is a Dangerous Thing

  1. You could also make an argument that Stars is valid, given that GPS uses 212 quasars to triangulate position of the satellites themselves. Quasars being accretion disks of super massive black holes, which are collapsed stars.

    I agree that would be more tenuous of an answer, but it also goes to the point of how complex the GPS system is, and how much of modern science is needed to create it.

  2. I’m surprised the answer isn’t “all of the above.”

    I assumed stars of some sort were used a la Sean’s comment 1, and I’m really surprised lasers aren’t a basic part of the mechanism too.

    (But then, I’m enough of a physics doofus that I don’t know why you couldn’t use a laser or some kind of xASER to generate a pure and steady enough sine wave to use that as your basic clock pulse.)

  3. I found the on-line quiz dreadfully easy, which puts me in the top 10%, but am I complacent? No, I find it depressing that the public understanding of science in the US is so poor. FYI, I’m from the UK. I believe I have a fair idea of my shortcomings regarding science as a result of a fair amount of reading science, the Dunning-Kruger effect and to paraphrase Donald Rumsfeld:

    There are known knowns. These There are [not many] things I know that I know.
    There are known unknowns. That is to say, there are [many] things that I know I don’t know.

    But there are also unknown unknowns. There are [probably an enormous number of] things I don’t know that I don’t know.

    I find that the more I learn, the less I know. And that assumes that none of the science I read will be overturned by some future paradigm shift. It must be a terrible thing to be a Professor who rightly understands that they know virtually nothing about everything.

    Note: ‘These‘ should be struck-out.

  4. I was all set to argue that lasers were even more important than the magnets, as the cesium clock I’m familiar with uses laser-cooled atoms and a laser-based atomic fountain-design (see link below). http://tf.nist.gov/cesium/fountain.htm
    But then, when I read more closely, I realized that I’m not certain which atomic clock GPS satellites use. Do you know if GPS is based on the older style of atomic clock using room temperature cesium (like you have described), or the new fountain-design?

  5. “You could also make an argument that Stars is valid, given that GPS uses 212 quasars to triangulate position of the satellites themselves.”

    Is the ITRF really tied to the ICRF?

  6. I love this, and it makes me think how easy it will be to figure out how to fix the dishwasher. : )

    It’s very exciting to learn how things work inside.

  7. Were Stern and Gerlach fortunate to use silver? If they saw a bunch of beams instead of just two, maybe it would give them something to ponder? 😉

  8. Excuse me! If the question was “what do you need to fly to your holiday destination”, wouldn’t the right answer be just “a plane”?
    Should I give an exhaustive list, like pilot, steward, ticket, air control service and, again, magnets, lasers, stars, etc?

  9. The obviously correct answer is “stars” since all the satellites, not to mention the Earth and us, have heavy elements that were all formed in stars. As the old Monsanto commercial says, without stars, life itself would be impossible. Or was that “without chemicals”?

  10. I think even people who aren’t laymen when it comes to GPS technology would have no trouble answering this. As the question and choice of answers are incorrect depending how pedantic you want to be it is obvious that the questioner is not an expert, has no deeper knowledge of the inner workings of GPS and would expect the more commonly known answer of satellites.

  11. It really just depends on who analytical and how far down the hole you want to go.

    Maybe Satellites is the correct answer, and Magnets would be the correct answer for what makes an atomic clock work…

  12. What is shown is a poll, not a quiz. And if it were a quiz, it wasn’t written very well. One would expect to see only one correct answer (and three incorrect answers) for a question offering four choices. Of course, I have seen test and quiz questions asking for the BEST answer (implying that all answers may be somewhat correct, with one being more correct than the others), but those were given by professors who really pushed us to draw upon all we have learned and not simply memorized, and the classes were not science or math, but other subjects more open to interpretation.

  13. I think a “global positioning system” can exist without magnets. Didn’t the old seamen use a sextant, a timepiece, and stars?

  14. The questions asks what GPS depends on to work, not what GPS satellites depend on to work. So, magnets are not a second correct answer. You could very well say screws are essential to the GPS network if you follow that logic.

  15. test taking 101: while there may be several correct answers, you are to choose the best* answer

    *what the test writer would consider the best of course! writing good test questions is much harder than answering them.

  16. @SacredAmoeba – I found it dreadful to read through the pretentious English posturing in your response. I’m an American, I also scored in the the top 10%, and I’m effectively a high-school drop out. I’ve also been all over England from London to Liverpool, and my experience is that England matches the United States pound for pound on many of the things her citizens like to feel superior about – intellect being one of those. You’ve got your fill of under-educated, over-fed knuckle-dragging hooligans on your little island – and it is worth noting that the most *mindless* television on American stations right now are reality TV and game-shows exported from the BBC. I’m sure the average teenage tart at a club in Liverpool knocking back lollipop drinks is just as dismally educated as her American counterpart, so let’s have an intellectually honest conversation about this – the problem reaches far further than the shores of the United States of America. Did you ever imagine yourself being schooled by an American high-school drop out? You can now add it to your list of life experiences.

  17. The question asks the GPS system relies on which of thiese to work. Magnets may work but can be wrong so can lasars and the information present for the Satalite might be wrong. But the Satalite will still work and your GPS will still work but be wrong or inaccurate. If Magnets are not working and lasers are working then satalites are working and your GPS wont work. The Satalite is the final say on all of the issues presented. If I have a GPS the laser, the magnets or stars wont make a difference because it wont work. I need a statalite to get the information to make my GPS work. The question could be what is neccessary for your GPS to work.

  18. I guess I need to say outright that the quiz thing is mostly a framing device providing me an excuse to hold forth on the physics of atomic clocks than a literal complaint about the quality of the poll questions. I thought that was a common enough convention not to need an explicit statement, but I have once again underestimated the Internet.

    Also, to answer the question from #6: Do you know if GPS is based on the older style of atomic clock using room temperature cesium (like you have described), or the new fountain-design?

    The last tie I had any certain knowledge of how the GPS clocks are set up was 1998 or so. At that time, they were definitely the older style, with magnets for state selection. I know this because a post-doc from our group took a job with a company that was trying to design a new system using lasers for the state selection and detection. Future GPS upgrades will almost certainly rely on lasers, but I don’t think they’ve implemented those clocks in the actual constellation.

    There are a number of proposals to put state-of-the-art clocks in space, mostly for fundamental physics tests. Because these would be in free fall, the now-standard fountain design would not work, but you can either pulse on the microwaves around stationary atoms, or use an old-style beam type clock with an extremely low atomic velocity to give you a long interaction time.

  19. I think you’re confusing “correct answer” with “best answer.” Most exams want the best answer, not the just the correct one. Satellites it the best answer.

  20. …and the scientists, engineers and mathematicians who designed the GPS system ate donuts while drafting the designs, so therefor GPS also depends on donuts! Yay! < >

    This is really, really dumb.

  21. Chad, I think you’re mistaking regular old comment slop for obtuseness. Although dog knows there’s plenty of obtuseness to go around.

  22. I’ve struggled with this problem in exams my whole life. This questions isn’t actually a very exam–many better ones exist on most any typical exam on any subject. The key to passing exams is figuring out what they are looking for which is I’d say far too often not even technically correct. You must not consider truth but rather a simple-minded perception of truth, or fail. Exams are generally tailored to a superficial memorization and comprehensive understanding is punished severely.

  23. That is what I explained to my prof when we had similar multiple choice questions and I got marks for that!

  24. The answer is A.

    Don’t confuse the ability to tell the time with either time itself, (good luck with that,) or with the ability to build a clock.

    These things are HOW the GPS works, but anybody can know that a GPS is determined by triangulation.

  25. Locally, extra knowledge can be a burden, but by adding an even greater understanding to what is in question, knowledge is not dangerous. It is helpful. In this case, if the poll (or quiz) becomes “harder”, maybe you should question the poll. For instance, an answer of “(E) Combination of above” would be more of a “best” answer.

    Polls and quizzes are designed to test a person or groups’ knowledge. With incorrect results, it shows that the people need to be re-educated (like the middle of this article does). If the answers were more evenly spread, and the people giving answers were professionals, it would accurately show that the poll was flawed (given that a “correct” answer was expected). Not that knowledge is “dangerous”.

  26. True genius overlooks the obvious.

    Choosing Satellites basically acknowledges the premise that there is underlying technology being utilitzed for a composite whole.

    People just don’t normally feel the need or want to think in those terms. Science isn’t as important to them as the intended effect.

  27. As a quantum optical physicist who works with Cesium atoms on a daily basis.. this is a stupidly worded question.

  28. Also, many satellite use electro-magnets for steering due to travelling through the earth’s magnetic fields.

  29. Hey, I’ve answered all 12 correct 🙂

    I feel smarter already and I’ll go and email my high-school teacher who kept telling me I’ll never know anything about anything.

    Oh, and I’ll be reading some of your older posts as well.

  30. So, now I’m stuck wondering if the GPS needs 3 satellites to help triangulate your position, then what determines each of those satellites position? Is it triangulated or is there something else used?

    As far as the poll and being a layman, satellite was the obvious first answer, but magnet was right there. Just seems like satellite is a better answer than magnets, but not the perfect answer.

  31. This is completely stupid. If GPS technology relies more on magnets than on satellites, why not just say that stars are more important? By this logic, it’s absolutely a valid answer. Why? After all, stars and their violent deaths lead to the creation and dispersal of atoms across the universe. Without a star to actually make elements through fusion, you wouldn’t have any cesium for an atomic clock, any iron/nickel/neodymium for the magnet, any fuel to launch it all into space, or even an earth for the satellite to orbit!

  32. I think the most correct answer in this case is satellites. Hey, we had lasers and magnets and stars long before we had satellites. But did we have GPS ? Nope. Of-course, thank you for enlightening us on how some of the things in a GPS work. I really had no idea that this was how the entire thing came together. Pretty cool : )

  33. “So, now I’m stuck wondering if the GPS needs 3 satellites to help triangulate your position, then what determines each of those satellites position?”

    They’re computed from ground-station data (carrier phase, mainly, I think) in the IGS05 reference frame.

  34. There is an even closer link to magnets since the satellites likely use reaction wheels to maintain their attitude. The motors used to spin up the reactions wheels contain magnets. I sure proved how my knowledge makes this soooo difficult too.

  35. There’s duct tape in there somewhere. Depend upon it, if duct tape everywhere failed simultaneously, so would the GPS.

    Someone with real knowledge should come forward to confirm it. Failing that, we’ll just have to go by the fact that no one will deny it.

  36. meh this is a poor example really a beter second answer would be “accurate time” the fact that its achieved through an atomic clock with magnets in it is a bit lame. you could say like “what does nascar depend on? Cars, magnets, calories, glue

    cars is the right answer but what about the magnets in the alternator of the cars? or the calories the driver mus consume to function? take any question like this far enough and you could say GPS depends on dumptrucks (to haul the ore to make the metal to manufacture satellite parts)

    my favorite example of these questions that are harder to answer if you know to much is ” what weighs more a pound of feathers or a pound of Gold?” the answer is neither or feathers depending on your assumption of units and familiarity with god using troy weights.

    another classic is “before mt Everest was discovered what was the highest point on earth?” most 5th graders see right through it and answer correctly “mt Everest” most adults try to think of what was thought to be the highest point up to its discovery

  37. It actually takes four satelites to give you an exact location, as three will give you a point on a map, the fourth is needed for elevation, also a question, if the satallites are so accurate in thier “timing” why are the clocks purposely set incorrectly on earth so once they are in orbit the are “correct”? I can only referance an article about relative time and how the closer to the earths mass, the slower (or faster I can’t remember) time is perceived. This was actually proven 50 years after Einstein proposed it. Seems even without satallites a gps could be done via earthbound towers, but without the perfect timing needed it just isn’t possible, we need the satellites only as the tallest towers, not the science behind them.

Comments are closed.