I learned, 7/8/16

This is the story of a cat video that showed me something about how to do science.

I intended this week to tell a simple story about why we like fizzy drinks. Carbon dioxide bubbles are produced by bacteria and yeast as they spoil food, so you can imagine that the ability to detect this would be important for survival. And in fact, all mammals have receptors attached to their taste buds that are specific to identifying carbon dioxide. They clearly help us survive.

So why do we humans seek out fizz, when in principle we are evolved to avoid it?

That’s a question about motivation, a kind that’s harder for science to answer. In the articles I read, researchers speculated that humans may like fizz for the same reasons we like hot peppers, perhaps as a way to display fitness by showcasing our ability to tolerate discomfort. It’s the gastronomical equivalent of spreading our plumage to attract a mate.

The scientists support this speculation with experimental data: When animals are offered sparkling water in the lab, they refuse it. We also know that humans often have to be acculturated to soda or seltzer before they develop a taste for it. So perhaps cultural learning plays a role in overcoming our inherent aversion to the sour, slightly painful taste of carbonation.

It’s a lovely story. But is it true?

Not entirely convinced, I decided to take a slightly different tack: searched YouTube for videos of cats drinking seltzer. And while plenty of videos show animals freaked out by bubbly water, videos of cats, dogs and rabbits drinking soda water were actually quite easy to find.

The first thing that I find interesting here is that a bunch of cat videos on YouTube clearly show that the research scientists were wrong. Or, more precisely, that the scientists were making inferences about real behavior from a very small number of lab subjects, under a very unnatural set of conditions. Cat videos, in this case, serve as a useful check on the arrogance of researchers who believe that their work in the lab means they understand their world.

 

 

Yet, beyond the curiosity that a cat video can undercut a scientific hypothesis, they also present a fascinating opportunity: These cat videos potentially offer a better way of doing science.

What is the purpose of carefully controlled experiments that are so hard to generalize to the real world? What if, instead of doing their original lab experiments, scientists decided to learn about animal behavior by simply asking YouTubers to upload videos of their cats drinking seltzer?

The crowdsourcing approach to science would be cheaper than lab work. It would test the preferences of many, many more cats than are possible in a research lab. And even though scientists would complain that such videos create a “biased” sample (not representative of a random selection of cats), that doesn’t matter when testing the veracity of this particular hypothesis.

In fact, I would argue that any hypothesis that explores the limits of behavior, or the breadth of behavior, would benefit from studying these sorts of huge, hastily-organized groups.

Human behavior is magical in its diversity – just look at our variations in preferences for food, entertainment or sex. The traditional approach of science is to strip away as much of the variability as possible, to study the core of what is human by studying the “average”.

But what if there is another way? What if cat videos made by random people across the world give us the opportunity to explore behavior that could never be captured in a lab? What if asking the crowd to run experiments is more productive than running them ourselves?

It’s axiomatic in science that tight control is the first requirement for a good experiment. But that represents just one way of doing science, one that’s made oodles of sense in a world where communication and coordination is expensive. We’ve never had a tool with the reach of the internet before, and there is ample evidence from other disciplines that reams of low quality information can be much more valuable than a few, carefully selected data points.

To me, internet communication, as exemplified by cat videos, has the potential to not just improve the way that information is disseminated, but how knowledge is generated in the first place. Knowledge generated not just by scientists in lab, but by all of us working together.

Science could be made better by bringing in the non-scientists to participate. It’s enough to make me want to break out the bubbly.

I learned, 7/1/16

How on earth did the venus flytrap happen?

A venus fly trap closes its jaws in as little as 100 milliseconds. This is nuts, given that the plant has no muscle fibers or nervous system. So the plant is rightly famous, lauded both by Charles Darwin (“one of the most wonderful plants in the world”) and creationists (as proof of “evolutionary fantasy”).

So how does it work? And why?

The traps themselves are what mechanical engineers call a “bistable system” – they can exist happily in either the open or snapped condition, but not in between. Think of a snap bracelet, straight as a rod until you apply a little energy in the right place… it magically curls around your wrist.

An insect wiggling the inner hairs of a flytrap triggers a small but critical change in the plant that pushes the trap from its initial state to past its intermediate point, causing it to snap shut. Now, a bistable system isn’t that hard to design – snap bracelets are often made out of recycled tape measures, which aren’t all that complicated. But the speed is impressive.

To close, the venus flytrap has to send a signal from the hair, to a (still poorly understood) system that changes the mechanical properties of the trap, pushing it past its point of instability. And it has to do so before the insect leaves. So the flytrap has evolved the ability to send electrical signals an order of magnitude faster than most other plants. All plants move a little – think of how plants will turn its leaves to face the sun – but in terms of speed, the flytrap is a champ, closer in speed to an insect than a grass.

Optimizing both the mechanics and the electrical system seems like an awful lot of evolutionary trouble. Why bother? And why is the flytrap so unique?

Selective pressures push plants towards canivory when there aren’t micronutrients to sustain them. A flytrap can fix carbon through photosynthesis just like other plants. But the bogs it inhabits in North and South Carolina are covered with peat moss and almost completely lacking in available nitrogen and phosphorous. And plants need fertilizer to thrive. So evolution drove plants to capture from the air what they could not take from the soil.

There are other carnivorous plants in those bogs, ones that rely on passive techniques such stickiness to capture small insects, absorbing their nutrients after they die. The flytrap hit upon a system that could capture large insects like spiders and ants as well.

It’s not a great system – a trap can only close two or three times before it becomes jammed with the undigestible exoskeletons of its meals. Flossing may be an evolutionary bridge too far. But as snap traps evolved around 65 million years ago, it seems to be plenty good enough for survival.

I learned, 6/24/16 (Brexit edition)

I am a big fan of the idea of the Wisdom of the Crowd – the idea that averaging the desires of a diverse group yields the best answers. Crowds routinely beat experts in the stock market, idea generation, and bookmaking. So should I believe that UK crowd chose correctly in the decision to Brexit?

It’s an emotionally loaded question for me, because I’m a big fan of democracy. Even a populist. But, crap, I can’t help but think that the will of this crowd was horribly, horribly wrong.

So what gives?

There is an idea, supported by recent research, that crowds are great when people make their decisions in isolation from each other, but do a piss-poor job when social influence is present. Social destroys the diversity of the crowd opinions, and thus destroys the multiplicity of thought that bring crowds their wisdom.

Without social coordination, crowds act like a giant computational engine, averaging all of the information taken in by each person’s slightly different perspective. With social coordination, crowds can be persuaded by the loudest narcissist in the room, repeating the same trope as if it wisdom.

With our own election coming up, this does not make me happy.

The good news, I hope, is that the internet sees to be making crowds smaller. With Facebook we all get our own private echo chamber. But maybe in the aggregate our echo chambers are numerous enough and diverse enough that they will average out into something approaching rationality.

I’m an optimist. And at least I can make a case for hope.

I learned, 6/17/16

This edition answers a question that likely bothers no one but me: Why are soda dispensers in restaurants so damn big?

Soda machines are, by volume, mostly ice bins. The soda syrup and carbonated water are either tucked in back or below the counter; the bulk of what you are looking at is a giant hopper for storing and dispensing ice.

So why store all that ice? Why can’t it be made and dispensed in real time?

Being science nerdy, I ran the numbers. The answer is physics, specifically the giant amount of energy that it takes to freeze water. Water is famous for having both a huge heat capacity – a large amount of energy is required to heat or cool it. And it has an even larger “heat of fusion” – the energy that goes into organizing the water molecules into little ice crystals.

If you wanted to make ice “on demand”, you have to supply a *spectacular* amount of power. To fill up a 20 oz cup from “on demand” ice as fast as you can from the hopper would require about 100 kilowatts of energy. That’s big time.

How much is 100 kilowatts? More than your house draws. With that kind of power you could run 100 hair dryers simultaneously. You could charge your iPhone 6 in 200 milliseconds. You could open your own Tesla Supercharger station.

Note that the number is power (“how fast”), not energy (“how much”). It’s another way of saying that the slow step in making ice cubes is not the cubing part. It’s getting the electrons into the ice maker.

In practice, when you pour all of the power from a conventional 120V, 20A circuit into making ice, it will blast out cubes at a maximum rate of about 1 cube every five seconds. Since most machines don’t draw the maximum power avaiable, it’s more like 1 cube every 10-20 seconds. To get around the power issue, restaurants generate ice throughout the day and night, and store it in the giant hopper behind the soda dispenser. Soda machines are big because if you tried to make them smaller, you’d melt all the wiring in the restaurant.

I’m probably alone in this, but I think that’s kind of cool. Something to think about at lunch tomorrow.

I learned, 6/10/17 (vacation edition)

Back from Hawaii and starting to engage the world again, I thought I’d share this crazy tidbit: the upper elevations of Hawaii’s Haleakalā volcano are almost completely barren of plant life, as the arid soil is constantly baked by the sun. Yet in the crevices between the volcanic rock live a range of endemic insects, such as a moth that through evolution has lost its ability to fly. With almost no plants to dine on, it survives during its pupal stage on a diet consisting solely of organic debris blown up from shore by the wind. Damn.

Life is amazing – even under conditions that border on ridiculous, it still manages to find a way.

I learned, 5/27/16

The onset of Alzheimer’s disease may share the plot of Captain America: Civil War. (No spoiler).

Stay with me on this: A study, done at Harvard, proposed a startling hypothesis – the plaques of beta amyloid that are the hallmark of dementia may actually be an immune response. The idea is that beta amyloid proteins are released by the brain during an infection and form around invading bacteria, capturing them in a web of goo that looks strikingly similar to the way Spiderman captures a bad guy.

To test this, the scientists injected bacteria directly into the brains of mice. The mice formed Alzheimer’s plaques as a response, and at the center of each plaque was a single bacteria, trapped in its sticky web. The plaques had saved the brain from the bad guys, but – as with a crisis averted by Spiderman or another Avenger – not without leaving the area riddled with collateral damage.

So what happens if your brain doesn’t have a superhero to protect it? The scientists ran the same test on “knock-out” mice, genetically engineered to produce no amyloid protein. The brains of the mice were exposed to bacteria, yet contained no plaques. But these mice were far more likely to die of infection, defenseless against the marauding villains.

If Alzheimer’s is actually produced on purpose, as a defense mechanism against bacteria, then doctors would be left with the conundrum from the plot Captain America Civil War — treating the condition might involve getting rid of the superheros, leaving patients vulnerable to attack. We are damaged either way.

Yet there is hope, as the hypothesis opens up new avenues for prevention – if we can tighten up the blood-brain barrier in the elderly, we may prevent the infections that cause release of amyloid in the first place. Prevention is better than treatment, hands down.

But if we can’t prevent it, we need to train Spiderman to clean up his act. It happens in the movies; maybe we can pull it off in real life too.

I learned, 5/20/16

Once a year, a few nights after a full moon in the warm summer waters, coral will spawn. An entire reef will bloom sperm and eggs en masse, filling the water around them with pink and white. The timing has to be perfect: If a coral is out of cycle with its neighbors by only 15 minutes, it has almost no chance of reproductive success.

Biology is beautiful.

I learned 5/13/16

To a computer, soccer is harder than chess. A lot harder. The world’s best soccer playing robots still would easily lose to a group of four year olds (check out the video, which is most amusing if you have watched little play matches.)

The expert guess as to when robots may be able to beat the best humans at soccer is roughly 2050.

The robots may be coming for us, but they appear to be taking their time.

I learned, 5/6/16

How the highest profile biotechnology failure of the 1980s led to a longer ski season.

The company, Advanced Genetic Sciences (founded 1979) was the world’s first agricultural biotech company, and their proposed product was, scientifically, one of the coolest things I have ever heard of.

To understand their technology, you first have to know about how hard it is to make ice.

Water, it turns out, doesn’t actually freeze at 0°C – at least not at any reasonable speed. When cooled below freezing, water molecules continue bouncing around at random, and the first crystal of ice forms only when a cluster of molecules randomly find themselves in just the right ice-like orientation. From there, a template is available, and the rest of the molecules glom on fast. But at 0°C, or even -10°C, that first crystal is really, really slow to form.

Enter the bacteria Pseudomonas syringae. This little yellow bacteria is everywhere – it covers much of the plant matter on the planet – and its success is attributable to a protein it synthesizes that speeds up the crystallization of ice.

P syringae makes a protein that is spectacularly good at nucleating that first crystal of ice. And it makes it because it is a vicious killer – the ice crystals form little needles that grow outwards from the bacteria to puncture the cells of the plants, giving P syringae food to eat. And it is so good, it can crystallize water quickly at only -2°C. (To see it in action, click on the linked video.)

But what is good for this bacteria is bad for world agricultural production, which loses billions of dollars of produce every year when an early frost – a frost that forms on leaves only because of P syringae – takes out a crop.

So Advanced Genetic Sciences set out to make a mutant of P syringae with a damaged ice-forming protein. The idea was to spray the genetically engineered bacteria onto the leaves of critical crops. There, the engineered bacteria would outcompete natural bacteria and prevent ice from forming on the sprayed leaves. No ice, no damage.

It seemed like an ideal first trial for a genetically engineered organism. First, there were other “ice negative” strains of P syringae in the wild – it was completely natural for this protein to become defective through mutations. Second, since the ice-forming protein is critical for the bacteria to get food, it seems nearly impossible for the modified organism to outcompete natural bacteria outside the lab. The risks were essentially zero.

However, the world outside of science didn’t see it that way.

First, the folks at Earth First and other eco-activist groups didn’t take kindly to the story of man messing with genetic code, and the details of the science were not that important to them.

Secondly, in 1982b, just before the first request was made to the FDA for field testing, a plant physiologist at Montana State University discovered that P syringae were in fact responsible for the nucleation of ice in clouds. These bacteria quite literally made it rain. And so while the corporate scientists were almost certainly right in their analysis that there was no risk in modified P syringae escaping from fields and going native, if by chance they were wrong the down side would have been a dramatic change in the pattern of global precipitation. This caused pause.

But only a pause. Despite warnings of global catastrophe, Advanced Genetic Sciences was allowed by the FDA to test the bacteria in the open in 1987. And the sky did not fall. No bacteria were found outside the test area. Better, the test worked – the sprayed potatoes and strawberries fared better in a frost than untreated controls.

However, that wasn’t enough to save Advanced Genetic Sciences. Because in all of that effort – all of the science, the planning, the regulatory submissions, the publicity – no one had bothered to calculate whether these genetically modified organisms would work better than chemical treatments that were already in use. And they didn’t. The bacteria were far, far more pricey than existing alternatives, yet not really any better. The entire effort was, from an economic perspective, a complete waste. Advanced Genetic Sciences sold to a competitor the next year.

OK, deep breath. So what does this have to do with skiing?

In order to perform these first tests with genetically modified organisms, the company had to learn how to grow the naturally occurring bacteria. And they got very good at this – good enough that someone had the keen idea to sell P syringae to ski resorts to improve their ability to make artificial snow in relatively warm weather.

Not only did P syringae work for snowmaking, they were economical too. Dead P syringae were loaded into a snow machine and blasted into the air with water, and nearly all the water fell to the earth as fluffy white powder. Even at just a few degrees below freezing.

So thirty years later, anyone who skis on artificial snow is probably skiing on the legacy of this first, failed biotech company. The product, Snomax, today is purified bacterial protein (rather than whole killed bacteria), but it’s a direct descendent of that original invention. And it is used worldwide, in every country that has skiing.

I guess that the moral of the story, if there is one, is that if you are working on something cool enough, good things will happen down the road. Just maybe not what you were expecting.