Some lanes are more equal than others.

September 1, 2016 8:07 AM

These charts clearly show how some Olympic swimmers may have gotten an unfair advantage. A data heavy analysis of the swimming in Rio that shows that some lanes were faster than others. On the Washington Post Wonk Blog.
posted by OmieWise (30 comments total)

That's pretty compelling evidence. My thought was that there might be something wrong with how signals from the touchpads at the far wall were recorded (which shouldn't happen, they do test these things, but it could mean there wasn't an actual lane bias), but the analysis of the 50m rules that out.
posted by figurant at 8:34 AM on September 1, 2016

Wow, those graphs are very, very convincing. Especially the fact that the effect didn't show up at Beijing, but showed up for two different contests in the same Rio pool.
posted by RedOrGreen at 8:34 AM on September 1, 2016 [4 favorites]

Watching the swimming on TV I was struck by the potential for arbitrary readings of the electronic touch-pads. In a couple races the difference between first and second place was hundredths of a second - a difference that is largely logistical. As in, the one swimmer flexed her finger a little faster than the other and etc...

I would think there'd be another layer of analysis that could be undertaken to evaluate the lack of bias in the touch sensors. (Wouldn't they also show consistent error?) The presence of current in the pool would be evident in a floating object (or better still a string of floats extended from across the breadth of the pool.
posted by From Bklyn at 8:56 AM on September 1, 2016

The presence of current in the pool would be evident in a floating object...

Or dye, if the pool wasn't, you know, already green*.

*Yes, I know that was the other pool. Work with me.
posted by Capt. Renault at 9:09 AM on September 1, 2016 [1 favorite]

I wouldn't be too worried about bias due to the touchpad instrumentation. A hundredth of a second is forever in electronics time. FINA/Omega could easily (and used to) time to the thousandth, like some other sports. I'd guess you could read out microseconds without too much trouble, and tens of nanoseconds if you're willing to deal with RF effects.

The reason to round to the hundredth is due to pool construction: a thousandth of a second translates to just a few millimeters at race speed. No pool is constructed to that tolerance (and even if it was, thermal expansion/contraction would quickly ruin any perfectly calibrated lane lengths).
posted by lozierj at 9:13 AM on September 1, 2016 [10 favorites]

The presence of current in the pool would be evident in a floating object...

Not if the current isn't at the surface, and not if the current is caused by swimmers' motions in the pool.
posted by Kutsuwamushi at 9:17 AM on September 1, 2016 [8 favorites]

What kind of electronic error could possibly produce the results in these graphs, though?
posted by straight at 9:38 AM on September 1, 2016

Beijing had a specially designed advanced pool that which is why it had no current problems.
posted by w0mbat at 9:41 AM on September 1, 2016 [1 favorite]

In 1972, Sweden’s Gunnar Larsson beat American Tim McKee in the 400m individual medley by 0.002 seconds.

That's 0.0008 percent, if my numbers are correct. That's twice the percentual margin compared to when Sweden's Thomas Wassberg beat Finland's Juha Mieto in an Olympic 15 km cross-country race by 0.01 seconds a few years later, after a 42-minute race with staggered start (the other guy in that clip isn't Mieto, obviously, but a Norwegian skier who started 30 seconds before Wassberg. He got the bronze medal).

The skiing federation changed the rules after that race as well. Sneaky swedes.
posted by effbot at 9:41 AM on September 1, 2016

This is a really excellent description of research design and how to take into account a number of confounding variables.
posted by MisantropicPainforest at 10:08 AM on September 1, 2016 [5 favorites]

There are too many missing pieces in the article to put together (from my BarcaLounger) what the cause might be but it strikes me as one of those riddles that would be very very satisfying to solve.
posted by From Bklyn at 11:00 AM on September 1, 2016

just as a thought experiment: the pool is 50m long (permanent pools are often 51.5m or 53m long to account for 1 or 2 moveable bulkheads), 25m wide (10x2.5m wide lanes sometimes plus a gutter which we are going to ignore) and a minimum of 3m deep. So the water in the pool is likely to weigh a minimum of:

50 × 3 × 25 × 1000 × 2.2059 = 8,272,125 pounds

Moving some of the water will give said volume some one rum which will be lost to entropy in the friction against the rest of the water in the pool. Wave effects are not likely to be significant here given the lane ropes. So creating a current would require moving most of, if not all of, the water in the pool. At least until viscosity & boundary effects soak up the extra energy. That leaves a swimmer needing to measurable move a mass of water some 40k times their own mass, still more than 5000x that of the whole heat.

It's worth noting that when filling anOlympic sized pool takes days to chemically balance it even with the aid of all of the pumps (which get a lot more water movement per joule than a swimmer would) party because of the difficulty of stirring the bloody thing.

As an aside I think it's also worth noting that measuring to the 100th is actually 3 times the precision used for determining a false start for the relay legs. during the relay last 3 swimmers are simply required to be in contact with the block at the moment the previous swimmer touches the wall. Each of the blocks has a pressure switch in them and the leaving swimmers time of departure is compared to the time of the incoming swimmer tripping the touchpad. Anything up to -0.03seconds (early) is still legal.
posted by mce at 11:03 AM on September 1, 2016 [3 favorites]

What are the fluid dynamic effects of 8 x 200lb swimmers diving into the same end of the pool at once, with great speed and force?

Sure the water may be motionless when nothing is happening, but what happens when you add, oh I dunno, swimmers?
posted by JoeZydeco at 11:25 AM on September 1, 2016 [1 favorite]

Weird. Is it possible there's some centrifugal force from the earth's rotation that creates a slight circulating current? The Beijing pool is oriented east-west, and the Rio pool appears to be oriented SW-NE. (Obviously these aren't the only two pools in the world.)
posted by stopgap at 11:51 AM on September 1, 2016

In textbook physics land:

f=ma, a=f/m
m(water) ~= 5000×mass(swimmers)

Therefore, assuming 100% efficient energy transfer from swimmers to water yield the average acceleration of the pool water as a body to be:

a(swimmers)/5000
Assume point of entry to be the fastest and most synchronized point in the race, assume average entry speed to be roughly 5m/s (clean racing the middle of the pool rarely much exceeds 2m/s) and assume the pool absorbs the entire 5m/s we might accelerate the water an average of 1/1000 m/s
posted by mce at 11:56 AM on September 1, 2016

The coriolis effect. But if that's in play, it should be measurable. (Responding to stopgap.)
posted by klanawa at 11:57 AM on September 1, 2016

Coriolis should only apply to something in motion relative to the rotation of the earth (and shouldn't apply to the stationary water in the pool), but I don't know if centrifugal force has any effect on water within something the size of an Olympic swimming pool. I would think that's still not big enough to create any circulation.
posted by stopgap at 12:02 PM on September 1, 2016

Coriolis should only apply to something in motion relative to the rotation of the earth (and shouldn't apply to the stationary water in the pool),

It's enough that the water can move within the pool; it would be theoretically possible to have a current set up by the Coriolis effect in a swimming pool. However, a current due to the Coriolis effect should be clockwise in the southern hemisphere, and counterclockwise in the northern hemisphere. The opposite was seen in Rio (data suggests a counterclockwise current), and also at the 2012 US Olympic trials in Omaha (data suggests a clockwise current).

I don't know if centrifugal force has any effect on water within something the size of an Olympic swimming pool.

I'm not sure which way you think centrifugal force would be acting; could you clarify? To a first-order approximation, the surface of the earth is not a perfect sphere, but bulges out slightly towards the equator in such a way as to, roughly account for centrifugal force; if it didn't, the oceans would flood all land towards the equator, and leave purely dry land around the poles. Something that is level with the surface of the earth (for some appropriate definition of "surface") has, in a sense, already taken centrifugal force into account and neutralized it.
posted by DevilsAdvocate at 12:24 PM on September 1, 2016

And some swimmers got advantages of multi-million dollars of r&d in creating swimsuits that allowed them to swim just the tiniest amount quicker
posted by 2manyusernames at 1:45 PM on September 1, 2016

It's enough that the water can move within the pool;

I'm pretty sure the water has to be moving in relation to the rotating reference frame (the earth) for there to be any Coriolis force. Your description of clockwise for the southern hemisphere and counterclockwise for the northern hemisphere describe the deflection caused by Coriolis force on water flowing inward (e.g. toward a drain), but without any initial movement of the water, there won't be any inertial/Coriolis force.

The shape of the earth doesn't account for all of the apparent reduction in gravity due to centrifugal force. But after thinking about this more, I think the only effect would be that the water at the northern corner of the pool in Rio would be some imperceptible amount deeper than the southern corner, presumably well within the tolerance for real-world construction of a level floor anyway. As gravity and centrifugal force are balanced, there wouldn't be any acceleration, so I can't think of any way this would translate into rotation or convection.
posted by stopgap at 1:53 PM on September 1, 2016

I can't find the reference, but the timing pads aren't the culprit. The technology is there to make them much more accurate, but the nature of pool construction means that there is an official allowance in pool length, even from lane to lane. If they went to thousands of a second they would be in a situation where the timing intervals would be smaller than the allowed length difference between the lanes considering the speed at which they swim.
posted by thecjm at 4:15 PM on September 1, 2016

FWIW this could have been caused by a current speed of approx. 4 cm per second or 1/10 MPH.*

That's not *very* fast but it does seem like it would be fast enough to be noticeable via the float test or other similar and simple tests.

Note rounding blah-blah-blah. These are approximate, not exact numbers.
posted by flug at 5:33 PM on September 1, 2016

I can't find the reference, but the timing pads aren't the culprit. The technology is there to make them much more accurate, but the nature of pool construction means that there is an official allowance in pool length, even from lane to lane.

lozierj linked to it upthread.
posted by Etrigan at 6:23 PM on September 1, 2016

You can see videos of the float tests here. There is certainly no current there moving at anything like 4 cm per second. The float might have moved about 4 cm in an entire *minute*. But don't take my word for it--watch the float test videos yourself.

That article also gives a possible alternate explanation for the time variations in the end lanes--depending on which side the swimmer breathes on, the swimmer can see adjoining swimmers going one direction and not going the other. Not sure if that is the complete explanation but it is certainly more compelling than currents that pretty obviously are not there at all. If it is currents at all it is something a lot less obvious than simply one giant counter-clockwise current going on all the time.
posted by flug at 8:24 PM on September 1, 2016 [1 favorite]

Any swimmer who has come up to an international level will have been trained to swim on alternate sides in freestyle by any coach who is worthy of the position. For every 6 strokes, you breath twice: 1-2-breath-1-2-breath.

Taking a breath on a freestyle stroke necessarily slows you down (which is not nearly as important on any other stroke. You can get away with breathing every other stroke or every 3 strokes on the butterfly or breaststroke, but it's hardly revolutionary. The backstroke is a free breath any time you want outside of the push off the wall)

At the highest levels of achievement, you could conceivable, without having to take a single breath, fail to exhaust your O2 reserves in the 50m free before the end of a sub-25 second race. In longer freestyle distances it might be worthwhile to only breath to one side (every 4, 6 or 8 strokes. certainly not every 2) if you were an untrained superman who'd never been in international competition before. In that case, being in lane 1 and breathing to the left would be worse than being in lane 8 and breathing to the left. No-one, absolutely no-one, goes to the Olympics with their training based on that. If you don't believe me, watch Katie Ledecky swim a 400m or an 800m.

If this analysis is right, there's a significant, interesting, problem with these pools.
posted by figurant at 10:44 PM on September 1, 2016 [2 favorites]

As usual, the instant you think "that would be a neat problem to work on for a while," you find that someone else has spent many years thinking about it. (By this I mean the general fluid dynamics here, not the specific issue of speed-ups in certain lanes.) The main article has a link to a neat review (Ann. Rev. Fluid Mech, Wei et al. 2014):
"the fluid dynamics of competitive swimming."

Without thinking about it too much, the idea that you might have some interaction between the forcing (provided by the swimmers) and the walls (which vary in construction from place to place) that could sometimes lead to a net current -- enh, it doesn't sound crazy to me. One thing that is not clear to me from the article is whether this "lane effect" increases with time; I would think you could determine this by looking for a drift in the difference between out/back times for individuals in a long enough race. It is not at all obvious to me that, even if there _is_ a current that gets set up in long races by wacky nonlinear magic, this current would _also_ be present in the short (50 m) races -- currents take a finite time to develop, and I would be surprised if they could be established in less than the ~25 seconds it takes for the 50m race to finish.

Put another way: if there is a current, and if it is established in some fashion by "forcing" from the swimmers, that current must develop over some finite time. I guess in principle the fastest timescale here is the acoustic (sound-crossing) time, but water is pretty incompressible so I would assume we can ignore sound waves. (For aficionados, I would say it is a "Boussinesq" or "anelastic" problem.) In that case the fastest time is some dynamical time, determined by the forcing (the swimmers) or the container. In theory maybe you could have some wave that developed in the containing walls and propagated very quickly; I honestly don't know. But the most obvious candidate for the "forcing" time is just the characteristic time of the swimmers themselves. So I would have guessed that at best you could set up a circulation over a time equal to a few "swim-crossing" times, not <1 -- so while it might be a big deal for races that last for many of these crossing times, it shouldn't be evident for very short races. Maybe?
posted by chalkbored at 2:17 AM on September 2, 2016 [1 favorite]

Any swimmer who has come up to an international level will have been trained to swim on alternate sides in freestyle by any coach who is worthy of the position. For every 6 strokes, you breath twice: 1-2-breath-1-2-breath.

no, not every swimmer.

the stroke can be seen during Ledecky's 800-meter swim at the 2016 Arena Pro Swim Series. Note how her left arms drags and stays on top of the water longer than her right and how she continuously breathes from her right side

Freestyle breathing with Michael Phelps, Katie Hoff, and Bob Bowman

(admittedly their one-sided breathing is considered somewhat unorthodox)
posted by mr vino at 3:17 AM on September 2, 2016 [1 favorite]

While the breathing-on-one-side theory sounds plausible, you'd need to explain how the effect isn't present at some events.
posted by Horselover Fat at 6:36 AM on September 2, 2016 [2 favorites]

this no secret. I swam competitively in the 70's. outer lanes have always been slower. center lanes the fastest.
posted by judson at 10:52 AM on September 2, 2016

this no secret. I swam competitively in the 70's. outer lanes have always been slower. center lanes the fastest.

This is different. They found that swimmers in lane #1 were consistently faster on their outgoing laps than their incoming laps, and swimmers in lane #8 were consistently slower on their outgoing laps than their incoming laps (and similar but lesser effects in less-outer lanes). This graphic shows what they're talking about.
posted by Etrigan at 11:05 AM on September 2, 2016