Bees tackle a seriously tough math problem
At some point in your life, whether you know it or not, you’ve tried to solve “the traveling salesman problem.” In fact, you’ve tried many, many times.
The classic version of the problem postulates a salesman out on the road, facing many stops ahead and looking for the optimal route that will reduce the distance traveled to the minimum. If the route is simple enough, intuition is generally enough to develop a near-optimal approach. But add a few more stops, some more options in routing, and this falls under that big group we discussed a few weeks ago — NP-C problems. That nasty class of issues where there seems no neat way to find the best solution except by testing them all.
Keep that in mind as you ponder the best way to grab those last minute gifts. You’re not going to come up with the best route to the book store, the cheese shop, and that place with the quirky jewelry. Just drive.
The good news is that real life rarely penalizes a less-than-perfect solution to this issue. Real world human beings may be years discovering all the potential shortcuts and side routes that can be taken to squeeze in that one last stop for the day, and unless the pressure for best route is “getting back to the cave ahead of something whose name includes the word ‘saber,’” it’s rarely fatal. Good enough is generally … good enough.
But better is still better. Every mile saved is gas in that salesperson’s tank or, in the case of bees, more calories returned for calories expended. As it turns out, bees confronted with multiple sources of food do improve their routes over time … but their number-crunching isn’t all that hard core:
On our array, bees did not settle on visit sequences that gave the shortest overall path, but prioritised movements to nearby feeders. Nonetheless, flight distance and duration reduced with experience. This increased efficiency was attributable mainly to experienced bees reducing exploration beyond the feeder array and flights becoming straighter with experience, rather than improvements in the sequence of feeder visits. Flight paths of all legs of a flight stabilised at similar rates, whereas the first few feeder visits became fixed early while bees continued to experiment with the order of later visits. Stabilising early sections of a route and prioritising travel between nearby destinations may reduce the search space, allowing rapid adoption of efficient routes.
So the routes got shorter over time because the first parts of the route became straighter, though bees did continue to experiment with different orders for later stops on the route … which isn’t a bad approach, especially if the initial positions are already near-optimal. Actually, what makes this study impressive isn’t just the test of insect-path-solving, but the technology that let the researchers tag and continuously track six bees closely enough to map their paths in very fine detail.
Okay, let’s take a direct path to more research ...
Google Street View as a proxy for race and income surveys
By their pickup trucks shall ye know them.
The American Community Survey is run by the Census Bureau not with the intention of tallying the number of people in the United States, but to fill in the gaps when it comes to determining income, race, religion, and other demographic features of towns and neighborhoods. The Census does this the old-fashioned way — they ask. But a group primarily from Stanford has a cheaper solution:
We show that socioeconomic attributes such as income, race, education, and voting patterns can be inferred from cars detected in Google Street View images using deep learning. Our model works by discovering associations between cars and people.
Using Google Street View, the team captured the make, model, and year of the vehicles parked alongside the street at the time Google took its Street View images. Then they employed some modeling that really seems pretty straightforward for the most part — significant presence of German cars equals high income neighborhood. But mapping their results against recent political results generated one value that they seem quite fond of repeating …
If the number of sedans encountered during a drive through a city is higher than the number of pickup trucks, the city is likely to vote for a Democrat during the next presidential election (88% chance); otherwise, it is likely to vote Republican (82%).
Which seems like another way of saying that most urban areas don’t have a lot of pickup trucks. And those that do are probably located far from the coast in blood-red regions.
The Stanford team suggests that the Google images could substitute for making an actual survey and could be more up to date — but this seems to offer a high possibility of some coincidental circumstance (no parking on a street during snow, vehicles from outside an area parked there because of a nearby event, no parking this side of the street on alternate Tuesdays, etc). And the examples given for the success of this model seem both trivial and facile.
Giving older adults a fighting chance against the flu
One of the reasons for high mortality among older adults who are infected by the flu is simply that their immune systems fail at step one — identifying and attacking the disease. A group from Yale has followed up on research in that area and determined that monocytes (a form of white blood cell) in older adults are short on interferons and simply fail to raise the warning flag that draws a full immune response.
Monocytes from older individuals contained less of the adaptor protein TRAF3, which is required for the induction of both interferons and the interferon regulatory transcription factor IRF8.
When white blood cells from younger individuals were deprived of TRAF3, their response to infection became similar to that of older patients. The good news here is the the opposite was also true — upping levels of TRAF3 or IRF8 got the immune system back in gear. Which is something that could be particularly useful in years like 2017, when the match between the flu vaccine and the flu most could encounter seems particularly poor.
What made the Ice Age get worse?
Around a million years ago, there’s a climate mystery. For some time before that, the Earth had been going back and forth, on cycles of roughly 41,000 years, between ice ages and warmer periods. The usual explanation for the regular pulses of cooling comes from the changing angle of Earth's axial tilt, which rolls between 22° and 24.5°, over a period of about … 41,000 years. Which fits pretty neatly.
Until the Middle Pleistocene, when everything falls apart. From that point on, the glacial periods are longer, with a period closer to 100,000 years. But the periods are less regular, with a tendency for gradual glacial buildups followed by relatively abrupt periods of melting. While many paleoclimatologists have looked to the skies for a solution in the form of the Milankovitch Cycles, reflecting complex interactions of orbital eccentricity, the fit is a lot less than perfect. The messy, severe, saw-toothed changes that happen after the transition to the Middle Pleistocene just don’t seem like the neat, more gentle swings of the Early Pleistocene.
A large team, mostly from the UK and Australia, has modeled the holy heck out of the available data in another bid to add some understanding to the just what caused this transition to longer, more severe changes.
We argue that neither ice sheet dynamics nor CO2 change in isolation can explain the MPT. Instead, we infer that the MPT was initiated by a change in ice sheet dynamics and that longer and deeper post-MPT ice ages were sustained by carbon cycle feedbacks related to dust fertilization of the Southern Ocean as a consequence of larger ice sheets.
MPT here is “Mid-Pleistocene Transition” and the answer the team has produced is a variation on one that’s been put forward before. I’m not going to pretend to understand all that’s going on here but — the Southern Ocean contains a couple of deep water upwellings that provide nutrients to plankton in shallow water. Fertilizing these big masses of moving water with additional minerals could result in more carbon being captured by the ocean in the form of biomass. So CO2 levels, and temperatures, drop. It’s an answer that’s a lot less neat than a lot of people would like to have, and the best I can suggest is looking at this illustration that shows time relationship between sea level, CO2, dust, and temperature.
Fracking where you drink
There’s not a lot of theoretical work here, but a pair of UC Santa Barbara scientists simply looked at the location of wells where hydraulic fracturing over the course of a year, and found that about half of them were within three kilometers of a well used for drinking water.
Furthermore, we identify 11 counties where most (>>50%) recorded domestic groundwater wells exist within 2 km of one or more hydraulically fractured wells stimulated during 2014. Our findings suggest that understanding how frequently hydraulic fracturing operations impact groundwater quality is of widespread importance to drinking water safety in many areas where hydraulic fracturing is common.
The purpose of fracking is to open up pores in the rock where the oil and gas is trapped. However, this shouldn’t pose a threat to the zones that produce drinking water if both the oil and gas wells and disposal wells were properly jacketed. However, if there is an issue—like a blowout or a spill—the proximity of these wells to water sources definitely warrants careful watch.
Is that a hand-ax in your pocket, or is that a … crocodile!
Some recent findings had suggested that our Australopithecine ancestors had chopped up some of their fellow African mammals. This contention was based on bones of the right age and location that seemed to show marks indicating the action of stone tools at work.
However, a trio of scientists—two from Germany, one from the US—have an alternative cause for those bone-scratches that may take meat off the menu for Lucy and her friends.
The results show that crocodiles were important modifiers of these bone assemblages. The relative roles of hominids, mammalian carnivores, and crocodiles in the formation of Oldowan zooarchaeological assemblages will only be accurately revealed by better bounding equifinality.
In other words, croc bites seem to leave behind similar marks as those which had been labeled as action of early humans. So, barring more definitive evidence, it’s hard to say that the folks down at Oldowan really were tackling large mammals.
Is inequality not just natural, but inevitable?
Some papers tackle a rather small idea. Some shoot a little higher. And for some …”big” barely captures it.
A group of Dutch scientists have put together a mathematical analysis of inequality. Not just inequality in nature like “this wolf happens to run faster than that one,” but inequality in society.
We show striking similarities between patterns of inequality between species abundances in nature and wealth in society. We demonstrate that in the absence of equalizing forces, such large inequality will arise from chance alone. While natural enemies have an equalizing effect in nature, inequality in societies can be suppressed by wealth-equalizing institutions.
There’s a statistical process description that appears frequently in evolutionary theory, one that’s based on a very simple idea called “The Drunkard’s Walk.” The model gets its name from a thought experiment.
Imagine a drunkard staggering his way along a sidewalk. On one side of him is a wall. On the other, a gutter. Though he may have several steps worth of freedom in either direction, if he walks far enough it’s inevitable that he ends up in the gutter—because he can’t end up on the wall. Think of “stagger left” and “stagger right” as the heads and tails of a coin the god of drunkards is flipping over and over. Eventually, enough of the gutter-direction values will build up.
Now, think about it in terms of some biological feature. Say … complexity. On the one side, there are limits. You can’t get much less complicated than a bacteria, or a virus. That’s the wall. On the other side there’s no real limit to how large your Blue Whale/Redwood tree might become. If species are simply on a random “drunkard’s walk” through complexity, you would expect to find a lot of creatures very near the wall of ultimate simplicity, but a spreading tail of complex stuff that reaches out to en-biggening. Which is exactly what you find. What can look from the outside like a “drive toward complexity” or a “drive toward increasing size” is just a bunch of genetic drunks, staggering along.
The same thing may be true of other forms of inequality, both biological and societal. What’s easy to pass off as the result of “working hard” or “business acumen” may be nothing more than the tip of that statistical tail. An economic genius is just someone who got the right coin flip more times than average. That doesn’t fit the model that our minds usually impose on these data. Which doesn’t mean it’s not true.
But, since there’s no limit over on the gutter of richness, those lucky enough to have been sent staggering in that direction are just as likely to keep going that way. So the tail gets ever thinner, but also ever longer. Unless someone builds a wall.
Over the past millennium, such institutions have been weakened during periods of societal upscaling. Our analysis suggests that due to the very same mathematical principle that rules natural communities (indeed, a “law of nature”) extreme wealth inequality is inevitable in a globalizing world unless effective wealth-equalizing institutions are installed on a global scale.
Today’s infographic is from Compound Interest’s Chemistry Advent Calendar.