November 01, 2016 at 12:30AM

Today I Learned: 1) ...a few things about Salem, Massachussets. It was named after "salam", as in "peace". It was once the capital of the Massachussetts Bay Colony, until a governor's son drowned in the river near the city, and the governor decided to move the capital to Boston to avoid memories of his son's death. Salem also used to be one of the more wealthy, thriving ports in New England. Much of modern-day Salem occupies what used to be ocean, which was filled in with at least one New Hampshire mountain. List continues. 2) A joint iGEM team from Ludwig-Maximillian University and the Technical University of Munich just presented a new technology for 3D printing tissues. Current standard organ-printing technology involves printing alternating layers of hydrogel scaffolding and cells. The scaffolding provides structure for the cells until they're nicely adhered. Then the hydrogel is blasted or chemically washed away, which is fairly destructive. The Munich team built several cell lines that display biotin on their outer membranes. They print by injecting those cells (free-floating!) into a media containing tetrameric avidin variants*. The biotin on the cells sticks to the avidin in the media, and the avidin (being tetrameric --> having four binding sites for biotin) acts as a chemical bridge that causes the cells to stick together. They clump up almost immediately, and form vaguely rigid structures, or at least as rigid as a pile of cells is going to get. They can print pretty nice-looking 2D shapes, and they've demonstrated the ability to print two layers, with cells sticking in their printed layer quite well. The printer the team used was a modified Ultimaker 2. They modified it with a special printing head and a custom bed for holding petri dishes, both of which can be printed by the vanilla version of the Ultimaker for a conversion cost of about $50. * For those who don't remember from previous TILs, avidin and biotin are notable proteins mostly because they bind together really really strongly. 3) Chinese medicine has a regulation problem in China. Chinese medicinal substances occupy a weird liminal space between food and medicine, and as such somehow don't fall under the regulatory purview of any Chinese regulatory agencies. That's bad, because they are, in fact, taken as both food *and* medicine, and it turns out that it's not enormously uncommon for Chinese medicine to be contaminated with heavy metals or Aspargillus (which excretes some pretty nasty toxins).

October 30, 2016 at 11:31PM

Today I Learned: 1) In the late Victorian era, there was a very wealthy woman by the name of Isabella Stewart Gardner. She married into yet more wealth, and spent a good amount of it patronizing art of every kind. Her house was a veritable art gallery in itself, and when she died, her will stated that her house be turned into a museum. The will also specified that if any piece of art were removed from its original position, the entire estate would be put up for auction and the proceeds donated to Harvard. You can still go there -- it's a beautiful collection of collections, ranging from classical sculpture (imported from Europe) to medieval furnishings and paintings (also imported from Europe) to some of the most avant-garde painting of Gardner's time. 2) When the printing press was first invented, most printed books had very large margins so that illuminators could come in after printing and add illustrations around the edges. It was a big deal when printers first started adding illustrations directly by print, and there were artists who specialized in print art. 3) It's possible to 3D-print molds for microfluidic devices, using relatively inexpensive printers. An iGEM team from Uppsala, Sweden, used a 3D printer to manufacture microfluidic rigs for heat-shock transformation of bacteria. Including the cost of the master and chip, they managed to bring the cost of transformation down moderately (from something like $2.5/transformation to something like $1.4/transformation). The chips looked really nice, too.

October 30, 2016 at 12:07AM

Today I Learned: 1) There are human cell lines that aren't adhesive! Some kind of immune cell lines can be grown free-floating in media. 2) Glucose meters work by measuring the activity of glucose oxidase (GOx) or, more rarely, glucose dehydrogenase (GDH), which both metabolize glucose into other sugars. Both reactions are redox reactions, and the resulting change in oxidation state can be measured with a properly-built sensor. The more glucose, the more GOx reactions per second, the higher the amperage through the sensor. I also learned that you can make a similar device to detect ethanol using AOx (alcohol oxidase) instead of GOx. An iGEM high school team from China built such a device, which could detect alcohol levels in water with reasonable accuracy. They didn't test it in blood. 3) You can swap the active sites of different proteases, and it will work (at least sometimes).

October 28, 2016 at 11:13PM

Today I Learned: 1) The E. coli genome apparently forms a solenoid-like helix with a turn size of about 100,000 bp. This means that stuff like sigma factors that recruit other enzymes to specific locations in the genome can also affect locations that appear to be very far away, but are physically close to each other in the cell. Note that the diameter of this solenoid is significantly bigger than the diameter of an E. coli. I believe this works out because there is also higher-order coiling of the genome. Also note that this is old news to scientists who study Eukaryotes. I just didn't know that it also happened reliably and repeatably in bacteria, or that the bacterial genome formed such a huge loop. 2) The single most important chemical in modern dry-cleaning is perchloroethylene (or PERC). It's incredibly cheap, incredibly good at removing stains, and somewhere between midly and very toxic (and probably a carcinogen). It's nasty stuff to work with, but it's very difficult for dry-cleaners to get away from. France recently put a ban on PERC, set to go into effect in 2022. The iGEM team of Paris-Bettencourt decided to interview every dry-cleaner in Paris, in person, to determine what effect the ban has had so far and what dry cleaners are looking to replace it with. It turns out that a) a lot of dry cleaners are still using PERC, b) a lot of dry cleaners have moved away from PERC, c) by region, PERC use is negatively correlated with immigrant populations, d) a nontrivial number of dry cleaners report feeling sick around PERC, and e) a lot of dry cleaners don't really know how they can survive without PERC -- it's just too damned economical. 3) Bacteriosins are a class of toxic protein secreted by bacteria to kill or inhibit closely-related species. They are much more effective than standard lab antibiotics, and after fifty years of use in the food industry, we still don't see resistance to them in any meaningful amounts. Unfortunately, they're also quite narrow-spectrum, and they aren't particularly useful medicinally in part because they break down very quickly in the stomach.

October 28, 2016 at 12:11AM

Today I Learned: 1) One way to think about cancer -- it's a reversal, within a single organism, of the evolution of multicelluarlity. I don't think this really tells me anything strictly *new*, but it's a nice way of crystallizing a lot of ideas about cancer in a succict way. 2) Saw some really old piano-like instruments today, including a pianoforte, harpsichord, and clavicle from around the 1730s. Seeing those instruments in person really puts Mozart and the other classical composers in context. They're dainty instruments, relatively delicate-looking and very ornately decorated (though admittedly I was looking at some very wealthy persons' instruments). If I were Mozart and I was given one of those to work with, I bet I'd come up with something classical-sounding too. 3) There was a brief period in ancient Egyptian sculpture where the principal subject of the sculpture would be shown more than once in the same sculpture. I saw one example from a high-ranking official's tomb showing the guy, the guy again with a different kilt, and his wife and children. Nobody is really sure why they did this.

October 27, 2016 at 02:28AM

Today I Learned: 1) ...a plausible reason a single-celled organism might evolve multicelluarlity. Yeast cells in the wild have an occasional mutation that prevents their cell walls from splitting apart when they divide. The result is that they form almost-clonal clumps. Lab yeast been selected to not clump, so they live almost entirely as separate cells. Question: under what circumstances might that clumping feature be selected for? Here's one. There are some nutrients that yeast can only absorb after they're digested outside the cell. For example, yeast can break down extracellular sucrose (which they cannot process directly) into glucose and fructose (which they can process directly) by expressing the enzyme invertase on the outside of their cell walls. The problem is that much of the fructose and glucose they produce floats off into the surrounding media, where it is inaccessible or worse (it could be eaten by a competitor! *Gasp!*). A group at Harvard proposed that clumped cells might have a selective advantage in a sucrose-rich media, because having a bunch of clumped cells together raises the local concentration of available fructose and glucose quite a bit. They raised single cells of yeast in a range of concentrations of sucrose and glucose, then, based on which ones grew and which didn't, backed out a model of sucrose digestion and diffusion. They added clumping to their model, and predicted that clumped yeast should be able to survive on less sucrose than the non-clumped variety. They generated some clumping yeast mutants, tried seeding *those* in sucrose media, and their predictions panned out. There are a lot of other possible reasons to start building a multicellular colony -- for instance, it makes you potentially much harder to eat -- but this is a very nice experimental exploration of one such reason. You can read the rest (for free!) here in PLOS Biology: http://ift.tt/2dZ3jG4 2) Bose is the undisputed master of the noise-cancelling headphone industry. This in itself would make a fine little TIL, but I *also* learned today that Bose *also* uses noise-cancelling tech to make super-smooth suspension for driver's seats for truckers. It turns out that taking vibration from a truck for hours and hours every day isn't very good for you, so making a driver's seat that doesn't vibrate is a big deal. 3) What the rice plant looks like. It basically looks like wheat.

October 25, 2016 at 12:34AM

Today I Learned: 1) If Python throws an error at you involving receiving instances instead of something else, you probably forgot to put a "self" argument in a method declaration. 2) ...a little bit more about ring species, which I mentioned yesterday. I didn't realize that the bit where the species at the two ends of the geographic range can't interbreed is a *requirement* of a ring species. I thought it was just a thing that *could* happen. I was wrong. Along with that, I learned today that ring species are incredibly rare. Wiki claims there are only four known ring species (three kinds of birds and a tree), and that three of them are controversial (the tree is definitely a ring species). 3) Markov-Chain Monte Carlo (MCMC) is a class of algorithms for drawing random samples from a distribution where you can't explicitly write out the distribution, but you can evaluate the density of that distribution at any given point. A key property of MCMC algorithms is that they're memoriless -- once initialized well, consecutive samples returned by MCMC are independent. Today I learned a bit about how several different flavors of MCMC actually work. Classical Metropolis-Hastings MCMC starts with a "walker" at some point. In each iteration of the algorithm it takes a random walk in any direction, then probabilistically decides to either stay there or go back, based on how high the density of the underlying function it's walking over (the higher the density, the more likely it is to say; intuitively, this means it will spend more of its time where the density is high). Adaptive MCMC works the same way, except that when the adaptive MCMC takes a walk, it isn't totally unbiased in its direction -- based on past results, it tries to learn when there is covariance in the underlying function (imagine a narrow ridge in the function's landscape) and changes what directions it will walk in accordingly. This is to save some computational effort when it encounters very narrow ridges, which are apparently pretty common. Vanilla Metropolis-Hastings would spend tons of iterations jumping off the cliff to either side and rejecting the new value, so it takes it a long time to move around on those ridge; Adaptive MCMC spends most of its time just walking around on the ridge. The MCMC Hammer (a very popular Python package for MCMC) works by having many walkers, all working in parallel. At each iteration, each walker picks another walker and takes a random step towards (or away from?) that walker. This solves the ridge problem much the same way as the adaptive MCMC, but I think it's a bit more robust on certain kinds of functions, and it's *much* more parallelizable for a multi-core system, since multiple walkers can be updated in one operation.

October 24, 2016 at 01:51AM

Today I Learned: 1) A ring species is a kind of species undergoing spatial speciation where the species varies considerably, but continuously, across its geographic range. Typically a ring species varies enough from one end to the other that members of the populations at opposite ends can't interbreed. It's called a "ring" species because in some cases, the geographic range bends back so that the populations at either end actually overlap, but are still distinct because they can't interbreed. 2) One of the common traits of developing market (i.e., third-world country) is high local currency volatility -- many developing markets have a large degree of political or financial uncertainty, which leads to rapid and frequent changes in the relative value of the nation's currency. 3) If you 3D print a part and it comes out warped, you can sometimes fix it by heating it with a blow dryer and twisting it back into shape. Thanks Andrey Shur!

October 23, 2016 at 03:07AM

Today I Learned: 1) A number of chess endgames in which one player only has a king and the other players only have a small number of pieces, have been solved, though some solutions are difficult to execute. Two knights, for example, cannot win against a king. Two bishops can, but it's difficult to execute without accidentally tieing. 2) "Wort" is a word, but not the word I thought it was -- wort is, according to Wiki, "the liquid extracted from the mashing process during the brewing of beer or whisky". 3) ...about Margaret Sanger. When I first heard the name, I thought perhaps she was the inventer of Sanger sequencing, which had me really excited to learn about her. Turns out she's just one of the most important sex educators in the 20th century and basically started birth control as an institutional thing in America. Oh well.

October 22, 2016 at 03:08AM

Today I Learned: 1) People really notice (and care) if a conference has bad coffee. 2) Janelia Farm (one of the coolest biological research institutes out there -- if you don't know about it, you should check it out!) has an endowment from HHMI of about $19 billion. That covers something like 50 very well-funded labs, operating with some ridiculous equipment. Just a good number to know in terms of calibrating institute expenses. 3) I really like the style of the current director of the FBI, James Comey. Check out his prepared remarks on Clinton's recent email scandal -- it's clear, precise, balanced, well-structured, and addresses a lot of possible questions an audience member might want to ask. Really my kind of informative prose. http://ift.tt/2dX27AB

October 21, 2016 at 02:47AM

Today I Learned: 1) Learned a bit about insect immune systems. For one thing, I learned that hemolymph, the goey stuff that insects pump around their body, doesn't carry much oxygen. It mostly seems to be for moving nutrients around... and for immune system function. 90-95% of the cells in the hemolymph are killer cells, which detect, track down, and ingest foreign cells. These are insects' primary line of defense against pathogens. The other major immune system component insects have is a system of anti-microbial peptides, or AMPs. AMPs are a diverse class of small proteins produced across the eukaryotes, but they're particularly important to insects. AMPs act as antibiotics in a number of ways, and are secreted en masse when the insect's immune system detects an invasion. 2) Remember that weird nested symbiosis I mentioned a couple of TILs ago? With the bacteria that lives in a bacteria that lives inside an insect? Today I learned a bit more about that triple symbiosis. The insects (it's actually a whole clade) in question are mealybugs, a clade of tiny little fuzzy sapsuckers, meaning they live on sap sucked from host plants. This isn't a very nutrient-rich diet, so it helps to have symbiotic bacteria that can metabolize unusual carbon sources. The first-level endosymbiont is Tremblaya, a bacteria that has been reduced to almost organelle status -- it has a massively reduced genome, even lacking the standard tRNA synthetases found in, say, mitochondria. The endosymbiont of Tremblaya, in contrast, is much closer to a normal bacteria in genome content. The second-level endosymbiont varies pretty widely across mealyworm species. It's worth noting that having a bacteria living inside a bacteria is extremely unusual. As far as I know, there aren't any known cases outside of insect cells. It's thought that Tremblaya can acquire an endosymbiont because of its unusual cell membrane. See, it's so massively reduced that it no longer has its own genes for synthesizing a membrane and cell wall. It relies on the mealyworm to produce a membrane for it. This means that from the outside, it effectively looks like a eukaryote -- and the symbionts that live in Tremblaya happen to be well-adapted to invading eukaryotic cells. 3) ...about preadaptation. Preadaptation is when a species just happens to have a trait that makes it very well suited for an environment it hasn't encountered before. When the preadapted species *does* encounter that environment, it tends to do quite well. You could consider any successful invasive species an example of preadaptation. Another example is thought to be lobe-finned fishes, which have far, far fewer bones in their fins than ray-finned fishes (most fish are ray-finned). I don't know why they evolved lobs, but it seems to have been really useful to have a small number of finger bones when the first lobe-finned fish started to crawl up onto land -- which is why we have five fingers on a hand instead of fifty (lobe-finned fish have order of ten digits per hand; ray-finned fish have more).

October 20, 2016 at 02:08AM

Today I Learned: 1) ...the "true" definition of molecular cooperativity. Cooperativity is (can be?) defined as a property of a chemical reaction in which one reaction makes a second more likely. 2) ...a few bits and pieces of Norse mythology. There are a surprising number of giant worms (snakes) in Norse myths. Also a lot of wolves, ravens, and eagles. All three of the latter were in some sense signs of war -- not good things. Wolves, for instance, didn't have their modern connotation of nobility. They were considered savage, bestial things, full of hunger and spite. 3) You can, in principle, build a high-gain, high-accuracy amplifier from a high-gain, low-accuracy amplifier coupled to a low-gain, high-accuracy amplifier.

October 19, 2016 at 02:49AM

Today I Learned: 1) At Caltech, if you don't register for classes, you will be classified as not-enrolled, which can cause you to lose building access priviliges. As far as I can tell, that is the only consequence of not registering. 2) In HTML, the "a" tag is is the tag defining a hyperlink. Why "a"? Seriously, why "a"? I have no idea. 3) The word for a tile used in a mosaic is "tessera", or "tesserae" for plural.

October 17, 2016 at 02:10AM

Today I Learned: 1) We've had jellyfish in space since 1991. A bag full of polyps was sent to up with a space shuttle, and has been growing in microgravity ever since. The jellyfish were recently brought back to Earth. They did not handle gravity very well -- they were pretty bad at maintaining orientation in true gravity. 2) Spaghetti is... trickier to cook in tomato sauce than the other pastas I've tried. Sauce doesn't soak into the pasta as quickly as boiling water, which means there's this awkward and rather lengthy period where the spaghetti remains stiff, which makes it a) hard to mix in properly to where it can absorb sauce better and b) hard to mix the bottom, which causes it to tend to burn. 3) Never run "conda upgrade --all" without superuser privilages. Just like with pip, if conda detects that it needs to upgrade itself, it will download the newer version of itself, uninstall itslef, and re-install the new version. Except that if you don't have superuser priviliges, it fails during the reinstallation bit, leaving you with a broken, half-uninstalled conda that's really difficult to clean up after.

October 16, 2016 at 04:24AM

Today I Learned: 1) Linked traits were well known to breeders and horticulturalists by Darwin's time (and probably long before then), though there was no explanation until geneticists realized how genetic linkage works. Relatedly, when Darwin wrote "On the Origin of Species...", it was widely believed that domestic animals tended to gain floppy ears from disuse, since they did not need to raise them in alarm as frequently as their wild cousins. (In fact, droopy ears seems to be linked with other traits of youth, which tend to be selected for through domestication.) 2) The Feynman Sprinkler is an interesting physics puzzle. Imagine a standard garden sprinkler. It spits out water, and the resulting momentum transfer causes the sprinkler head to spin. Now imagine you reverse the flow of the pump providing water, so it sucks in rather than spitting out. Put it in a tank of water. What happens to the reverse-sprinkler? The result is... confusing, at least to me. Think about it for a few minutes, then check out the Wiki page on "Feynamn Sprinkler" and see if you can make sense of it. 3) ...how to use CSS classes! Somehow I keep re-learning this over and over again, but I always have to re-learn it anew whenever I take on a new HTML project.

October 15, 2016 at 01:57AM

Today I Learned: 1) There is an insect with an obligate endosymbiotic bacteria -- that is, a bacteria that lives in the insect's cells, and provides important functions to the insect -- that itself has an obligate endosymbiotic bacteria. More details to follow. 2) Relatedly, there's a well-known nasty effect that happens when one organism becomes a required endosymbiote for another (usually it's a plant or animal that takes on a bacterial endosymbiote) that poses a serious danger to both organisms. See, endosymbiotes have to deal with having an incredibly small effective population size, since a) they can't really get access to outside populations to recombine genetically, and b) they undergo a nasty population bottleneck whenever the host reproduces. It turns out that the effects of genetic drift become much more severe with small populations. If the effective population size is small enough, drift and random mutation will overcome selective pressure, and the symbiote's genome will literally start to degrade on evolutionary timescales. See, population size and mutation rate are intimately linked, and both contribute to the maximum size genome a species can stably maintain through selection. The higher a species' mutation rate, the smaller the biggest genome they can maintain over many generations. Smaller population sizes effectively amplify the effect of mutations on genome degradation. There's some thought that this is basically what determines genome length -- species adapt to have genomes that are sizes their mutation rate can support, basically. That's why RNA viruses, which have insane mutation rates, also have insanely tiny genomes. Probably. In the case of endosymbiotes, the symbiote evolved to cope with a reasonable mutation rate/effective population size, but gets hit rapidly with a much harsher mutation rate/effective population size. The net result is that the symbiote's genome gets quickly whittled down to a small fraction of its original size. An extreme example is the ancestral predecessor of the mitochondria, the genome of which was whittled from a (presumably) normal-sized bacterial genome to a 10-20kb circlet. This puts the host in a real pickle. Now it depends on an endosymbiote, but that endosymbiote is rapidly degenerating. Thus begins a mad scramble to somehow make the symbiote functional before it completely loses any ability to function, killing both symbiote and host. This is presumably why gene transfer from symbiote to host is so common (it's been observed in just about every sufficiently-old endosymbiotic relationship) -- it keeps the endosymbiote functional while reducing its genome size, making it less likely to decay to uselessness. 3) ...what a young termite queen looks like. It's a surprisingly elegant looking insect, at least from afar -- it has long, diaphanous wings, making it look almost like a squat dragonfly, but with a flight pattern more like a moth.

October 14, 2016 at 03:49AM

Today I Learned: 1) I like persimmons! I'm not sure if I'd ever actually had a persimmon before, but today I had one pretty much fresh off a tree, and it was pretty darned good, for a fruit. I'd describe the taste as kind of a mellow, sweet, orange/mango flavor. 2) There are parasitic transposable elements. I mean, all transposable elements are parasitic on the host genome*. There are, however, also transposable elements that have lost their protein-coding regions. They aren't actually fully functional without *another* transposon around to snip them out and move them from place to place. There are some really clever adaptations transposons have evolved to combat parasitic transposable elements, including a whole class of RNA transposons that get transcribed, leave the nucleus, produce a virion-like coat from a protein monomer, reverse-transcribe themselves, produce and attach an integrase to themselves, and move back to the nucleus for insertion. * this is contentious -- some scientists, most notably those from the ENCODE project, think that transposons are kept around as drivers of genetic variation. However, today I learned that the wider molecular evolution community suspects that this is rubbish, and that transposons are truly junk DNA, in the sense that they do not contribute to host fitness. 3) Apparently what we call "transgenderedness" today was relatively common in Native American societies. Of course, any time someone talks about "Native American societies", they're talking about a huge swath of cultures, some separated by centuries or millennia of divergence, so keep in mind that everything in this particular TIL is a MASSIVE generalization, BUT! The point is that many, many Native American societies had between three and five standard gender concepts. Typcially these were roughly equivalent to one or more of "male", "female", "male in female body", "female in male body", and "male and female". The non-("male" or "female") genders (referred to as "two-spirits" in modern parlance) were considered anywhere from perfectly normal to special and honored. Interesting how different other cultures' views of gender can be from our own... yet also interesting how ultimately circumscribed those gender views are. It all really boils down to variations on "male" and "female" -- even accounting for cultures quite different from ours, the human race hasn't explored the space of possible genders very thoroughly.

October 13, 2016 at 02:24AM

Today I Learned: 1) When ISIS first invaded and took ground in Iraq, they did so largely on the back of a massively successful media campaign. ISIS only had a few thousand troops, but they put out a number of well-produced battlefield videos on Twitter and Facebook and other platforms that made them look like a much bigger threat, essentially magnifying the effects of a string of small victories. As a result, they were able to force the surrender of Iraqi forces an order of magnitude larger than their own. 2) There is a transposable element*, called the P-element, that was acquired by the fruit fly D. melanogaster very recently (within the last couple hundred years). It confers a whole constellation of nasty phenotypes on its bearer, including a high probability of sterility. When discovered, the P-element quickly became a classic example of the spread of a selfish element in a naïve population. Also, since the invasion, D. melanogaster has developed an RNA interference pathway to shut it down. More recently (between 1800 and 1950!), the P-element invaded another fly species, D. simulans (which I assume is named so because it's similar to D. melanogaster?). Evidence points to a single horizontal gene transfer event being behind the invasion. * a transposable element is a gene which does nothing but move and/or copy itself into other parts of the genome, found almost exclusively in Eukaryotes. It's hotly contested whether or not transposable elements are functional in some way, if they're just selfish genes that their hosts can't get rid of. 3) Viridis, the current most-popular colormap in matplotlib, is very similar to Matlab's current default colormap, parula, but has been tweaked so that a) the gradient is visually smooth across the entire range of the colormap and b) it doesn't lose contrast if you're red-green colorblind. For more details, see http://ift.tt/2dZgvvp.

October 12, 2016 at 03:06AM

Today I Learned: 1) Here's a little algorithmic puzzle: You have a point (x,y) somewhere in the unit square, where one corner of the square lies at (0,0), the other at (1,1). You want to "jiggle" the point by moving it around randomly, in any direction, in a kind of Gaussian way. What algorithm do you use to make sure the point doesn't leave the square? Can you make it efficient? It's a trickier puzzle than it looks. It's not capital-D Difficult, but it's tricky. Try it. Ok, so here's an algorithm I used: 1) Pick a random angle theta, uniformly sampled between 0 and 2*pi radians. Then draw a number N from a Gaussian centered at zero. Try adding N*sin(theta) to the point's y-coordinate and N*cos(theta) to the point's x-coordinate. If that point leaves the unit square, draw another N and try again. Today I learned that this isn't a good algorithm. It works most of the time, but if you have a point in the corner and you get an angle facing the wrong way, it goes infinite (or close to infinite) and breaks. For example, imagine the point is at (1,0) and you pick 45 degrees as your angle. Now, no matter what N you choose, it'll be outside the unit box. There's a simple fix -- re-choose the angle every time you re-choose N, and before too long you'll get a good combination. Can you come up with an algorithm that isn't as wasteful? 2) A timing belt (or cam belt) is a geared belt that physically synchronizes various valves in an engine. If it fails, the engine can do really nasty things involving terms like "catastrophic failure". Apparently it's also possible for a timing belt to be off in such a way that causes accelerator input to the engine to be delayed, though I don't see how that's actually possible. 3) Mantis eyes may change color over the course of the day? I noticed that my mantis's eyes were dark, almost opaque. This made me a little worried, because usually I can see right to the back of the eye, where there are little itty bitty pupil-like spots. Later, though, it went back to normal, so I guess it's not a problem? I'm going to have to keep an eye on this.

October 11, 2016 at 01:11AM

Today I Learned: 1) You know those fancy prepackaged salads with multiple lettuces of different colors? Apparently those lettuces are grown together, in parallel rows, so that they can be harvested together and processed and packaged in the field for shipment. The effect is quite beautiful. 2) Last night I ran an experiment evolving agents in a series of iterated prisoner's dilemmas*. The agents were limited to "reactive strategies", which are strategies in which each agent probabilistically determines whether to cooperate or defect based on only the last move of the opponent, where the cooperate rate for each possible previous move can be tuned between 0 and 1. Each generation, each agent plays an iterated prisoner's dilemma against some fraction of the other agents (one at a time). The total fitness of each agent is calculated, then a new population of agents is randomly selected from the original (with duplication possible) where each agent's probability of being picked for the next generation is proportional to its fitness. Finally, each new agent's paramters are jiggled slightly to add variation to the next round of selection. Here are a couple of lessons I've learned from the results: 2a) Taking a random (Cauchy, in this case, which is like a bell curve but with greater probability of a large result) move in one axis and then another random move on the other axis isn't even *close* to a good way to generate random spread around a point (this came up when generating mutations between generations). 2b) Tit-for-tat, which is generally considered a really, really good strategy in an iterated prisoner's dilemma, doesn't seem to be stable in my evolution simulator. If you start with all of the agents clustered in the "cooperate after the opponent cooeperates; defect after the opponent defects" corner of strategy space, they pretty quickly start to spread out. Notably, I got a cluster of strategies at around generation 20 that defected a lot, and they seemed to do much better than the tit-for-tatters. By about generation 65, the agents were totally spread around strategy space. I've read that tit-for-tat is actually really vulnerable to "mistakes" in gameplay, as happens when the agent's probability of cooperation in response to cooperation isn't quite 1. If you have two agents that copy the other one with 99% fidelity, then eventually one of them will mess up. When it does, they get stuck in a vicious cycle of betrayal, which isn't very profitable for either one. Or maybe I've added too much mutation, and nothing is actually stable. Or, alternatively, my population size (100) could be too small to overcome genetic drift with the level of selection I'm applying. 3c) Some possible evidence to support the "too much mutation" hypothesis -- in a much longer simulation (5000 generations), it looks kind of like there's a ton of wandering around, without much in the way of stable strategies. I only printed out population snapshots every 50 generations, so it's hard for me to say right now exactly how the populations are moving, but they're definitely shifting around a lot between snapshots. A common -- but not universal -- trend is that defecting when the opponent cooperates is a highly profitable strategy, wheras whether or not the agents cooperate after a defection doesn't seem to matter much. I don't like the way I handled mutation in last night's run, so I've fixed it and will start another simulation tonight, this time sampling every generation (and perhaps not running quite as long). * I'm assuming most of my readership knows what a prisoner's dilemma is; if I'm wrong about that, let me know and I'll explain in the comments. 3) My new favorite party music playlist: http://ift.tt/2d8WCNW

October 10, 2016 at 04:13AM

Today I Learned: 1) Concentrated calcium chloride tastes... bad. Like, burning chemical bad. Like "this should not be in contact with living tissue" bad. Don't do it. 2) If you try to make cheese using a standard feta cheese kit, but with soy milk instead of real milk, it doesn't curdify. It turns into a rather thick, pudding-like mix that's kind of halfway to setting... but it doesn't get there. Next time, I will try adding some calcium chloride, which supposedly helps harden cheeses if you put it in at the right time (I didn't). 3) Numba, a just-in-time compiler package for Python functions, has a lot of limitations. It can compile classes, but only if you give it a bunch of type information about what kinds of inputs the class is going to get, and it doesn't support all of the features of the compiler. It also can't compile methods of objects -- only free-standing functions. Even then, it's a bit tetchy. I think it's back to Cython with me if I want to get this code to run much faster....

October 09, 2016 at 04:22AM

Today I Learned: 1) "Peccable" is a word. It means "capable of sinning". It was used much more widely in the early 19th century, according to Google, and fell out of usage somewhere around 1950. 2) Rennet tastes like... well, it's hard to describe. It's subtle, slightly aromatic, slightly acidic, slightly salty, and a little broth-like. Honestly, it tastes a bit too much like the smell of minipreps for my taste, as delicious as they sometimes smell. 3) There are gated communities for RVs.

October 08, 2016 at 04:24AM

Today I Learned: 1) You know the lemonade machines that some restaurants have, where you can see the lemonade swirling around in a tank? It turns out that, at least in many cases, that lemonade is not what you get in your drink. It's just there for show. Your drink is actually mixed from a contraption underneath the counter. This deceptive device actually lies at the center of an interesting turn in patent law, which basically says that you can get a patent even if the device you patent is immoral or deceptive (or, as it turns out, even illegal). A few more details here: http://ift.tt/2dTHa90 2) There are salts that are liquid at room temperature. I only know about one, and I only sort of know its structure -- it's a somewhat complex organic, much bigger than the usual salts you hear about in a chemistry class. 3) There are attempts right now being made to chemically sequester CO2 from the air, potentially converting it into methanol or other useful fuels. Unfortunately, the process isn't efficient at atmospheric levels of CO2 (even though we're now well above 400 ppm...), so it's really planned for use in CO2-emitting plants as a kind of air scrubbing technique. It's also really, really hard, and we don't have a good way to do it yet.

October 07, 2016 at 01:52AM

Today I Learned: 1) SSD drives (and presumably flash drives, too) have an interesting problem in that they read and write in entire blocks of .5-1 megabytes. When a write happens, the entire block is actually read, modified, and written back in. That's particularly annoying since flash memory wears out with use, and too many read/write cycles to a single memory location will break it. Accordingly, SSD manufacturers have a lot of tricks for reducing write requirements. Also, SSDs come with a bit of extra memory onboard. When blocks break, they can detect the fault, permanently lock out that block, and allocate to one of the reserve blocks instead. 2) So, how long do SSDs last? This (http://ift.tt/2cWbdkE) delightful experiment from last year pushed six commercial SSDs to their deaths. According to the authors, SSD manufacturers guarantee their drives to about 20 TB of read/writes, or 20 GB/day for three years. In fact, they perform well for much, much longer -- of the six drives tested, none hit a single error until around the 100th terabyte. *That* drive continued to funciton, albeit with increasingly smaller storage capacity, until between 800 and 900 total writes. Other drives performed better, logging hundreds of terabytes before their first failure and the best writing over a petabyte before finally crashing for good. Not bad. Thanks to Chris Lennox for finding this! *Just a warning, when I left a tab with this article open for a while, it started sucking up memory and processing time. By the time I killed it it was using several gigabytes of RAM. Not sure if that was a bug or some kind of background process, but I thought you should be aware. Can anyone replicate? 3) There is a species of parasitic wasp that lays its eggs in caterpillars... you know, there are a lot of great evolutionary stories that start that way, or at least very similarly. Anyway. This *particular* species has a weird quirk that their young hatch together as a brood and then immediately mate with any wasp around, which is usually siblings. Yes, it's odd that they're so inbred, but there you have it. This mating system has an interesting consequence for optimal sex ratio choice. If all of your children are mating with your own children, then it turns out the best sex ratio strategy is to have almost entirely female children, and just a couple of males that can fertilize them all. That way, you have the most possible opportunities for grandchildren. This is, in fact, what this wasp does -- it can control the male-to-female ratio of its broods, and it prefers to lay broods with lots and lots of females. That changes if the wasp lays a brood in a caterpillar that already has another set of eggs in it. In this case, the wasp's male children have to compete with the first mother's (few) male children. The payoff for having males is suddenly higher, because it increases the chances of fertilizing all the copious numbers of females that will be around, which is really efficient. The second wasp will lay a brood with a higher ratio of males to females. As more and more wasps lay their eggs in the same food source, the ratio of males increases, eventually approaching 50%. Amazingly (or, perhaps, quite expectedly), you can calculate the game-theoretic optimal ratio of males in a brood for any number of previous broods layed in the same caterpillar, and the behavior of this wasp matches the theoretical results quite well.

October 06, 2016 at 04:01AM

Today I Learned: 1) There is a limit, called the diffraction limit, on how small a thing you can see using light of a given wavelength. Today I learned that the diffraction limit only applies at focal distances greater than a couple of wavelengths. So if you could build a lens and camera less than a few hundred nanometers across, you could, in principle, build an optical microscope that could see things below the diffraction limit for visible light. Except not really, because to do so would require materials with negative and/or imaginary refractive indices, for reasons I do not understand. But you *can* build metamaterial lenses for longer wavelengths, like radio. 2) DNA bases, in the usual narrative, pair beacuse of hydrogen bonding between bases that sort of fit together. That's not really true, exactly -- most (order of magnitude 50%) of the energy in a base pair is actually in stacking bonds between adjacent bases (they have some sort of favorable shared pi-bond between aromatic rings). The base pairing works largely because when two bases don't sort of fit together, they physically bump into each other and mess up the structure required to get the benefit of base stacking. Something I'd idly wondered about was why you can't get relatively good bonding between pyrimidines (Ts, Cs, and, in RNA, Us), which are much smaller than purines (As and Gs). They're small, so they shouldn't clash with each other, right? Today I learned a plausible explanation, which is that pyrimidines don't actually contribute much, if at all, to base stacking -- it's the second aromatic ring on the larger purines that actually makes the base stacking interaction, so every base pair needs a purine to "work". 3) One of the well-known (in the business world) places where growing businesses fail is the so-called "Chasm". The chasm is the gap between a company that can produce a product in small volume for early adopters and a company that sells to the mass market. Apparently it's quite common for a product to get really enthusiastic support early on from a small pool of early adopters, only to be completely ignored by the vast majority of consumers once the company tries to scale up. On a similar note, economy of scale doesn't always apply. Service-oriented industries, or other industries where a large amount of human intervention is required, are a good example.

October 05, 2016 at 02:03AM

Today I Learned: 1) So, starfish may not be radially symmetric after all -- one of a starfish's arms always has a higher density of nerves, and it tends to favor that arm over the others. Arguably, it is actually a head. Today I learned (although apparently not for the first time?) that if the head arm of a starfish is cut off, it can shift that functionality to another arm! 2) We have some delightful printable labels for labeling tubes (and other things) in our lab. They're particularly nice because they tolerate cold well, which is important for labeling tubes that are going to be stored at -20°C or -80°C for a long time. Today I learned that though they are quite cold-tolerant, they are not particularly water-tolerant, and care should be taken not to get the adhesive wet before freezing them down. 3) Fire spreads significantly more quickly uphill. I guess that makes sense, because uphill is for hot air kind of like downhill is for water.

October 04, 2016 at 02:25AM

Today I Learned: 1) Qiagen midiprep kits* come in at least two varieties, which use mostly the same reagents but different columns. I also learned that you can buy columns, which are, as far as I can tell, by far the limiting reagent in the kit. That makes buying midis much, much cheaper. *these are kits for extracting plasmid DNA from medium-sized numbers of bacterial cells. For small numbers of bacterial cells, you use a miniprep kit; for large numbers of bacterial cells, you use a maxiprep kit. 2) A yearly subscription of Asimov, widely considered the best magazine in short-form science fiction by a fair margin, costs less than $40. Moreover, you can get a pack of old magazines at $15 for 12 magazines. Not to sound too much like an ad here, but I'm rather surprised how cheap sci-fi short stories are if you're not buying them in book form. 3) The tracrRNA in the CRISPR system (which is the bit of RNA that is recognized by Cas9 and causes an associated gRNA to be loaded) has to have the right sequence, not just the right secondary structure. With a few exceptions, chainging the sequence in a structure-preserving way destroys the function of the tracrRNA.

October 03, 2016 at 02:00AM

Today I Learned: 1) Some games that look at first blush like prisoner dilemma games can be solved by agreeing to alternately cooperate and defect, basically taking turns exploiting each other. If the payoff matrices are right, you can make more money trading off like that than you can by cooperating all the time. 2) The Lotka-Volterra equation, which describes population dynamics for a simple predator-prey model, is mathematically identical to the replicator equation describing evolutionary dynamics of replicators with different frequency-dependent strategies. 3) There are, in addition to vacuuming robots, swiffering robots and mopping robots. I hear they're not very effective (so says Andrey Shur).