Word doc available here
Pics might be up soon.
"Stanford students and professors alike expressed frustration when they encountered unnecessarily complicated vocabulary in written text, and considered overly verbose writers less intelligent because of their diction" (http://www.stanforddaily.com/2010/02/19/eloquence-or-elitism-the-concept-of-sounding-smart/).
I came; I saw; I didn't sleep much.
This term, I learned the extent to which rote memorization sucks. For the first time since debate camp before my senior year of high school, I had a schedule that was too much for me (in no small part because the rote memorization wasn't fulfilling). I took Bio (interesting, but the memorization killed it), Chem (still awesome), Algorithms (very useful, though I still don't understand it as well as I should), and Section Leading (like TAing. Very fulfilling). Because Bio disappointed me so much, I don't plan on taking any more for a while -- it's a waste when there are so many good classes at Stanford.
I was the sole chaperone of my Palo Alto High School debaters when they went to tournament in Vegas for a weekend. They did well at that tournament and at the others that they attended.
Through Queer Straight Alliance, I planned events that focused on queer women and on international queer issues. I also planned Queer Formal, and I headed up the Queer Coalition endorsement for student government elections. It's like I'm a politician. Or a lobbyist.
By planning Hackathon, a 24 hour coding-for-nonprofits event, I functionally raised $200,000 in one day for global public health nonprofits.
Immediately after Hackathon, I went to the Stanford High School Debate Tournament. I brought them internet. After that, my reception was vaguely messianic.
Anne Bartlett said to donate money to Valentino Achak Deng (http://www.valentinoachakdeng.org/) rather than Save Darfur (their $300,000,000 hasn't gone to help people). During my Chemistry midterm, Winona LaDuke told me that computer science was a fine vehicle for social change. Valerie Jarrett made me proud of my country.
I broke my bike (not too bad).
I cooked food for 50 every week.
I filled up my email (don't worry -- it'll still get where it needs to go).
I took pride in my attire, and not just because I avoid anything that was made in a sweatshop (that is a big part, though).
I taught nonprofits how to effectively send email and postal spam. A small amount of effort on my part saved them hours and let them do their jobs more effectively. Go computers!
The Vagina Monologues reminded me about the slave labor conditions used to extract coltan, which is used in electronics like most (all?) laptops and cell phones.
Avatar was awesome. Since 3d is the way to present information, I have decided to stop consuming 2d information like videos and 1d information like written text. Now I just need to see a 4d movie so that I can stop consuming 3d information too.
I thought that Cormac McCarthy's "The Road," a post-apocalyptic novel, was optimistic. Similarly, I thought that Faber and Mazlish's "How To Talk So Kids Can Learn," about effective teaching, was optimistic.
Next term, I'll be taking classes on computational biology, computer systems, jewish modernist philosophy. I'll also be taking a class by the person who made public key cryptography, which keeps the internet secure, a lecture series on tech law, and a small statistical programming lab.
Next year, I'm planning on being an RA in my frosh dorm.
This summer, I turned down a job at Google along with $13,000 and all of the perks associated with being an employee there to work with InSTEDD, an organization that uses computers to prevent diseases and disasters, in Cambodia.
MWF 9-950: Chem33
MWF 11-1215: Bio42
MWF 315-405: CS106a
TuTh 11-1215: CS161
M 445-605: CS198 Monday Meeting
Tu 7-8p: prepare for LaIR Hours
Tu 10p-Midnight: CS198 LaIR Hours
W 10-1050: Chem Section
W 215-305: Bio Section
Th 9-950: CS106a Section
(Th 1-6: Bio44x)
Tu 315-6: Cooking Dinner
Tu 8-10p: Stanford Debate Meeting
W 7-8p: Prep QSA Meeting
W 8-9p: QSA Meeting
W 830-10p: Dance Marathon Meeting
Th 630-930p: Palo Alto HS Debate Meeting
Sa 1p: ASSU Tech Meeting
Weekends:
Queer Formal
Hackathon / Stanford HS Tournament (same weekend)
Stanford Debate Tournament x1
Palo Alto HS Debate Tournament x3
In other words, I had something at 9am 4 days of the week, and I was teaching one of those days, so I had to get up an extra hour earlier to be fully awake and prepared. I had an event that I was planning or actively participating in on 6/10 weekends, so I couldn't catch up on my sleep or get any homework done on those weekends -- and that doesn't take into account the speaker events, arts events, or co-op chores that took my time on the other weekends.
The schedule dominated my life. I didn't stop doing my other extracurriculars (they fill in the gaps in this schedule and my weekends), but I didn't go to as many speaker events as I would have liked to (you'll notice that this verbose letter was a bit shorter than the last one), and I didn't feel happy about my schedule during the term.
Last term, my schedule was very hard, but I felt like there was some reason to the madness. I was the conquering hero, taking three CS classes (and some other classes), including two that people think of as very hard, and I did well in all of them and maintained the level of extracurricular activity that I wanted. I was proud.
This term, I scrapped through. I wasn't the conquering hero, but the soldier in Vietnam: in a hostile land (the land of premeds and problem sets), not quite sure why I'm there (I'm an engineer, not a biologist!), and up at all hours. The problem sets were really annoying.
Because I was busy every day of the week and had an event or two to plan or attend every weekend, I could never do my problem sets ahead of time, so I would do almost all of my weekly assignments in bio, chem, and CS161 the night before they were due, so I didn't usually get more than 4 hours of sleep before Fridays (CS161) or Mondays (Chem). With Bio, I eventually learned that TA office hours were very helpful, so I stopped pulling all nighters for those. And I did try to do my CS161 ahead of time. But I don't think that I ever got more than 4 hours of sleep before a Friday or Monday. With CS198, I had 6 students and I had to grade all of their programs and problem sets (8 total throughout the term), and each program or problem set would take between 6 and 8 hours to grade. I could sometimes get this done ahead of time, but this usually also meant less than 4 hours of sleep.
I already talked about how I had no time to catch up on sleep on the weekends. As a result, I took a 20m-2h nap almost every day between noon and 3, and I would often also take a nap at night. There were also very few lectures that I managed to completely stay awake through, though there is a big difference between lectures where I was sleep deprived and nodded off intermittently throughout the lecture and ones where I had gotten 0-2 hours of sleep the night before in addition to being chronically sleep deprived and I would fall asleep and wake up 20-30 minutes later. I also learned that, with massive sleep depravation, cell phone alarms don't quite cut it. I slept through my alarm for my 9am class (chem) a few times (in addition to sleeping through my alarm when I would take naps). The second time that happened, I downloaded an alarm on my computer so that I could blast music at full volume with 0 chance for technical error (I didn't always sleep through my alarm. Once my phone turned off mysteriously. At least two or three times, my alarm didn't go off when I set it. I think that my particular model of phone only lets a particular alarm go off once per day, so resetting its time does no good. I didn't actually sleep through my alarm more than twice), 0 chance that I wouldn't wake up (I can make my laptop loud enough to wake up the building), and 0 chance that I could fall asleep after getting up once (I can't turn my laptop alarm off from my bed). Twice, I also voluntarily slept through my bio class because I was so sleep deprived (once after dropping Bio44x, once after the weekend that contained both the Stanford HS Debate Tournament and Hackathon, both events requiring about 24 hours of time during my weekend. Consecutively), which meant I could sleep until 2.
I put more effort into staying awake with professors that were really good. This worked fairly well with the neurobiology professor (especially because bio was 11a-1215p). This didn't work very well with the first chem professor or with the CS161 professor. Thankfully, anything taught in lecture can be learned by reading the book (um. I don't actually think that I opened my bio or chem books. The cs161 book was good, though. There was also an optional book for chem that was written in plain English and explained the concepts, and that was very good) or the lecture notes (the lecture notes for bio and chem were both fairly good. Especially since the professors in bio sort of wrote the book, and the lecture notes were the important things extracted from the book. Both classes tested exclusively off of learning the material in the lecture notes and not at all off of the book).
During the first few weeks of the term, I was taking Bio44x. The class itself was moderately interesting. Throughout the term, there would be 4 lab rotations: genetics, developmental bio, molecular bio, and biochemistry. I started in developmental biology. We played around with eggs and sperm from sea urchins.
However, during the first few weeks, I was also taking a 21 unit schedule and I wasn't taking it easy with any of my extracurriculars. At the start of the third week, there was a bio44x assignment due, and I was scrapping just to get it finished. I hadn't taken much time for relaxation in the past two weeks, and I still couldn't get everything done. While my statement from last term that my only limits are exhaustion and the physical limits of my endurance, my workload in the first few weeks surpassed those limits.
After I dropped bio44x and went down to 17 units, the work load became manageable.
Halfway through my Sophomore year, Bio42 (along with Chem33) featured my first closed-book exam at Stanford. I was disappointed.
There are a lot of premeds at Stanford (at least, among frosh and sophomores). Med schools still have required classes (even though they teach you everything you need to know in med school so the required classes don't mean a lot), so hundreds of people that aren't interested in biology or chemistry take the introductory bio and chem classes. I think that the bio 40 series has decided that its mission is to weed out those premeds. While chem deals with these premeds gracefully, bio uses tactics like rote memorization. I'm disappointed that so much of bio42 is ruined, not for any educational reason, but because of misguided institutional structures.
I'm not opposed to closed book tests in principle (I would be, but chem33 was so graceful with closed book exams that I reconsidered). However, if there's any question on a closed book exam that could not practically be asked on an open book exam, then it should not be asked on a closed book exam either. That means that any question that is trying to determine whether or not a student correctly memorized a page of information would be ruled out because a simple lookup (whether in a book, a sheet of notes, Wikipedia, or Google) would easily give the answer.
We're living in the information age. Soon enough, everyone will have internet on their phones. Even absent that, anyone within range of a wifi network literally has the world's information flowing through them. Computers can memorize information better than me, so I'll let them take care of that. Since computers can't yet surpass my problem solving skills, though, schools are responsible for making me a good problem solver.
Bio42 exams do not meet my criteria above. Some of the questions do test problem solving ability, but about half of the questions would pose no problem given a sheet of notes. For instance, on the first exam, there one multipart question worth 5-10% of the total exam that just asked about identification of certain proteins by name.
Because the institution of bio42 hands out the grades, I did very poorly (I would guess, more than a standard deviation below the median) on the first exam even though I had a decent understanding of the material and had done well on all of the problem sets. (Aside: looking at exam grades in bio41 and 42, it appears as though my score was about the median for the first exam in bio41. The second exam in bio41, and every exam after it, has a median substantially higher. It appears as though bio41 is where people learn that the bio core is about rote memorization, and that the median is much less dependent on the difficulty of the material and much more on whether or not the student knows what's coming).
After that, I started playing the rote memorization game, and I got about the median grade on the second exam and on the final. I didn't get an A because that would be nearly impossible after my performance on the first midterm and because trying to beat Stanford premeds at the rote memorization game is not something that I can really do (at least, not without sacrificing my commitment to extracurriculars and my other classes, both of which are much more important than winning at rote memorization). I ended up with my first B (*cough* minus) at Stanford.
Aside from the institution of bio, I actually liked bio42. It taught me a lot of things about how the world works.
It was taught in different units, and each unit would have a different professor. And after each unit, there would be a pizza lunch with that professor.
The first unit was cell bio. The emphasis was on how a protein gets from point A to point B within a cell. It was decent. Professor Martha Cyert, who works a lot with yeast, was the professor.
The second unit was developmental bio. Dev bio isn't really my thing. Professor Craig Heller, who works a lot with thermoregulation, taught these few lectures. Fun fact: Sonic Hedgehog is an actual protein (http://en.wikipedia.org/wiki/Sonic_hedgehog).
The third unit was immunology. This was the first unit that I thought was really interesting. We had a general overview of the immune system, including general responses (skin keeps pathogens out. Acid and antimicrobials can take care of some pathogens. Fever, diarrhea, vomiting, and inflammation can help too), then we talked about B and T white blood cells (they both originate in the bone marrow, but T cells mature in the thymus. B cells do cleanup and eat any foreign substances that they see. T cells can help B cells and can kill any virus-infected cells. B cells feel around for big chunks of foreign substance. T cells look for little broken off pieces -- the stuff that goes wrong when a virus infects one of your cells), immune cell specificity (immune cells have a "constant" region and a "variable" region. The variable region determines WHAT pathogens it can deal with, so we randomly make several billion antibodies with different variable regions hoping that we'll have the one that we need when a pathogen comes. The constant region determines WHERE the antibody is. So, an antibody with a different constant region might be responsible for taking care of parasites versus taking care of bacteria), and problems with the immune system (AIDS and autoimmune diseases).
I was also surprised by how much of the material was new for the other students. A lot of things that I had known for years because my mom was a doctor and I have been interested in health for a while are apparently not widely known (ie, that fevers are created by your body to fight off diseases).
The professor was Pat Jones.
Neurobiology was the best unit. The professor, Robert Sapolsky (http://en.wikipedia.org/wiki/Robert_Sapolsky -- check out some of his talks on YouTube, and check out some of his quotes from lectures he's given), was awesome. He managed to fight back against the institution of bio42: in each of his lectures, he would focus on really understanding a few things and how to problem solve with them rather than just throwing a ton of disparate information at us, and he told us what 3 or 4 things we should know in each lecture (typically big conceptual things) and he made sure to tell us what we didn't have to know (typically, memorizing the nitty gritty names for everything). Sapolsky is also very charismatic.
Even though my CS knowledge might not have directly helped, I did feel right at home with the material. I guess spending the last year and a half dealing with information makes it easier to think about well developed information processing systems, regardless of whether they're silicon or carbon.
The big concepts:
1) The brain has to do a lot of work to maximize the signal to noise ratio.
It wants to make sure that every signal actually gets through and isn't mistaken for noise. In CS, we do this by doing everything in binary so that we only need to decide between 2 signals, and we also encode extra information (that way, even if you can't make out a small part of the information, you can figure out what it was based on the rest of the signal).
2) Signals within one neuron in the brain.
The neuron itself is normally very negatively charged because proteins are negatively charged. When positive ions (sodium) flow in from other neurons, it makes the neuron less polarized (less negatively charged). Then, there is positive feedback, and tons of positive ions finish the depolarization. Then, a stream of positive ions goes down to the dendrites so that it can pass the signal along to other neurons. Finally, the neuron hyperpolarizes itself (makes its charge really negative) so that it won't fire off again immediately.
3) Signals between several neurons in the brain.
When a neuron fires, it releases neurotransmitters (ie, glutamate -- MSG, which is excitotoxic for babies who don't have a very big blood-brain barrier -- or aspartate -- the fake sugar aspartame. Sapolsky recommends John Olney's work on why aspartame and glutamate is dangerous because Olney is the leading person on excitotoxins in the world -- melatonin, dopamine, adrenaline) to all of the neurons that it's connected to by its dendrites. That neuron can either let in positive ions or negative ions. Positive ions will depolarize the connected neuron, making it more likely to fire. Negative ions will hyperpolarize the connected neuron, making it less likely to fire. Neurotransmitters are used A TON, so they need to be easily recyclable and made of cheap materials (which means that receptors which detect the neurotransmitters need to be very detailed, which means very expensive).
A lot of drugs work by tricking neurons in the brain at this level. For instance, when the brain doesn't have much energy (ATP), it leaks some of its energy-drained ATP back to the pre-synaptic neuron to tell it to stop working so hard (so the energyless ATP works as a retrograde -- going backwards -- neurotransmitter), and when the brain has lots of energy, it leaks some ATP back to the pre-synaptic neuron to tell it that it can learn more and work harder. Caffeine tricks the brain into thinking it's always leaking ATP, so basically the engine will run even when it's dry. Hallucinogens tell the brain that it's getting signals that never even existed (versus caffeine that just sort of amplifies existing signals) because hallucinogens look similar to neurotransmitters. Amphetamines increase the release of dopaminic neurotransmitters (which is the same thing that happens in schizophrenia; a schizophrenic is clinically indistinguishable from someone on amphetamines).
4) Long term potentiation (learning; forming connections).
Learning means forming stronger connections with some neurons. This could mean learning some fact; it could also mean triggering a fear response whenever you see a big bear. Some of the excitatory (positive ion) receptors that respond to things like glutamate trigger other really excitatory (calcium rather than sodium) receptors, and the big inflow of calcium will lead to the neuron becoming more excitable and more strongly connected to the pre-synaptic neuron (the neuron that sent the signal).
The addictive part of drugs works at this level. There are both excitatory and inhibitory pathways for neurotransmitters, and when you take a drug frequently, there will be a lot of that drug present, so you'll .learn' the inhibitory pathway more strongly, so your natural state will be to inhibit the effects of that drug (regardless of whether or not you're taking the drug at a particular time).
Alcohol also works directly on LTP by interfering with one of the excitatory receptors. Yes, it's true: you learn less when you're drunk.
5) Neural Networks (brain as a whole)
This was one place where CS helped a little bit extra since I've actually implemented artificial neural networks. The brain doesn't work too differently from a neural network on a computer (the main difference is probably more the hardware than the actual implementation. In the brain, each neuron can act on its own and send and receive signals on its own. Every neuron is like a really small processor. Every individual processor in a computer is a lot smarter than any neuron or any few neurons, but since the brain has a ton of neurons, it's as if you had a whole room full of fast computers that could each interact with each other quickly rather than every virtual neuron being centrally controlled by one processor).
Knowing the material about signals between several neurons and about LTP is all you need to know about the mechanics of neural networks. In terms of thinking about it, though, one thing to know is that, while there are a few bits of knowledge stored in individual neurons, most of it is stored in the network structure as a whole. For instance, where your eye connects to your brain, there are neurons that can detect light of a certain color at a certain location, and there are neurons that detect lines (ie, if a bunch of adjacent neurons are all saying "yes, there is a dot here," then a line neuron will say "yes, because there is a dot in these points, there must be a line here"), and there are neurons that detect motion. However, there is no neuron that detects the particular dot and line pattern of your grandmother's face (http://en.wikipedia.org/wiki/Grandmother_cell).
It works by sending signals down a network. One neuron doesn't encode the whole; one neuron will get signals from each of the different parts of the whole, directly or indirectly, such that when a certain set of neurons fire, it triggers a memory. That's why something can be on the tip of your tongue. If you know that a certain memory is within a certain location in your brain, there are a bunch of different things that could potentially be in that location. Depending on how many different networks you're accessing, you'll be closer to or farther from a given memory. For instance, you might have a network that is triggered whenever you experience impressionism, and one that is triggered for visual art, and one that has to do with dismembered ears. If you're trying to think of "Van Gogh," triggering parts of any of those networks might help because Picasso has to do with each of those. "Impressionist art" would be triggered by Picasso and Van Gogh and Monet.
Fun Fact: "people don't even use 90% of their brains!" = 100% false. People use their brains all of the time.
6) Autonomic Nervous System (the brain regulating the body)
The autonomic nervous system is everything that isn't voluntary. It is composed of the sympathetic and parasympathetic nervous system. The sympathetic nervous system (SNS) is "fight or flight". The PNS is "rest and digest." They each excite parts of the body that help their respective functions and inhibit parts of the body that don't.
7) The Parts of the Brain.
There are three parts of the brain, each of which interact with each other. The .reptilian' brain deals with basic bodily functions. The .limbic,' or emotional, part of the brain deals with things like happiness and fear. The cortical, or higher order thinking, part of the brain deals with critical thinking and conscious thought.
7) Hormones (the mechanism of sending signals to the body).
For a long time, people thought that testosterone was the cause of aging (which ignores half of the population). As a result, people used to get monkey testes implants (including a pope) and they used to get testosterone injections in water (testosterone is a steroid. Steroids are fatty. Fat is not water soluble. Thus, these injections didn't work).
Then they discovered that glands can't act independently (well, mostly) -- if you cut them out and put them in a petri dish, they don't really know what they're doing. They need the pituitary. Thus, people started thinking that the pituitary was the master gland.
Then they discovered that the pituitary doesn't know what it's doing (well, mostly). Two rival scientists spent decades grinding up cow brains to discover that the hypothalamus controls the pituitary (the hypothalamus is really small, so they needed a lot of cow brains). After 22 years of work, one of them made the discovery. Two weeks later, the other one made the same discovery. A year later, there was another similar discovery by the scientist that made the two-weeks-later discovery. Then, a month later, the other scientist made the same discovery.
Hormones are basically like neurotransmitters that go in your blood and affect stuff that isn't directly close to each other (whereas neurotransmitters can only work on two neurons that are adjacent).
Fun fact: the WHO estimates that breastfeeding is the best contraceptive in the world and has prevented more births than all other contraceptives combined (it secretes prolactin). It doesn't work in the West because we don't breastfeed like normal people do. In the West, women nurse for a short time every once in a while and then stop breastfeeding after 6 months. Elsewhere, babies nurse for 1 minute, then stop for 30 minutes, then start up again, repeating without cessation (at, at night, women will sleep with their babies so that the babies can breastfeed in the middle of the night) for 4 years. As a result, outside of the West, some women might have a total of about 15 periods in their life: they start puberty later, get menopause earlier, and breastfeed. In the West, women have closer to 300 periods. It is theorized that this difference is responsible for some of the health problems unique to women in the West.
In the lunch with Sapolsky, I learned about his research at Stanford and over the summer. At Stanford, he has, in the past, been researching stress.
Now, one of the cool things in his lab has to do with toxoplasmosis. Toxo is a neurological parasite that can be very harmful to pregnant women, young kids, and adults with compromised immune systems, but doesn't do much to healthy adults. Up to 1/3 of people globally, and about 10% of people within the US have toxo infections (it never really goes away once you get it). It might cause depression, anxiety, and schizophrenia. It's found in cat feces. The lab discovered that rats with toxo are attracted to, rather than afraid of, cats. Toxo knows how to track down the part of the rat brain that has to do with fear of cats (not even just fear in general or cats in general. It's very targeted) and change it. That's really hard to do. They're now trying to figure out how it manages this feat.
Over the summer, Sapolsky studies baboons in Kenya. These baboons are at the top of the food chain in their niche, so they only have to spend about 2 hours each day working. The rest of their time is dedicated to developing social structures to make each other stressed. Thus, they're the perfect model for studying stress in humans.
The last unit was mammal physiology. The professor was the same as for dev bio - Professor Heller. The material was intrinsically interesting, even if the lecturer was less charismatic than Sapolsky. I don't feel like I learned a ton of overarching concepts, but I did learn a lot of nitty gritty facts.
A few fun facts:
The heart spends 1/3 of its time in systole (as in systolic blood pressure) and 2/3 of its time in diastole (as in diastolic blood pressure). In systole, it is beating. However, the body doesn't only need circulation 1/3 of the time. To make flow constant, the circulatory system has a few innovations. Veins (and arteries, to a much lesser extent) are elastic (veins can have as much as 70% of total blood volume in them). These elastic fibers let the heart store some of its pumping energy in potential, stretched, energy, which gradually will be released during diastole. In addition, there is a limit on the output rate -- capillaries are small. To demonstrate this, Professor Heller started out with a hand water pump (the water went all over the floor in the lecture room. He was cautioned to clean up carefully so as to not short circuit the fancy electrical stuff that they use to hook up to the big projectors. I was in the front row.). With just that, the flow went out during the 1/3 pump time. Then, he added capillaries -- he hooked the hose up to a syringe, so that only a certain amount could go out at any time. Better, but still not enough. Then, he put a balloon-hose between them. With that, the flow was constant.
The lungs absorb oxygen through alveolar sacs. However, not all sacs are the same size. Because P=2T/R (pressure = 2 times surface tension divided by radius), that means that we need surfactant to decrease the surface tension in the smaller sacs. To demonstrate the necessity of this, try hooking up a large balloon to a small balloon. All of the air will go out of the small balloon into the big balloon (from higher pressure to lower pressure). That's because the pressure is higher in the small balloon because it has a smaller radius. You can also think about how it's harder to blow up a balloon at first, and when it gets bigger, it gets easier.
The bladder and kidneys evolved out of the gut of fish.
Cold blooded animals use about 10 times less energy than hot blooded animals.
Cooling is a very good indicator of physical fitness. Professor Heller photographed dogs at the very first leg of the Iditarod, a very long dog sled race across Alaska (or something) and predicted that the ones with the best cooling at the very beginning would finish the best and that the ones with the worst cooling would do very poorly. The 4 that had the worst cooling dropped out of the race. The 4 with the best cooling all finished in the top 15. He has done similar things with athletes. He measures their output under normal conditions, and then he gives them something that helps cool them (ie, he puts their hand in a small, air conditioned chamber, and it manages to cool their whole body) and they perform drastically better when they can cool themselves.
In the Dominican Republic, there's some genetic difference that causes less di-hydro-testosterone (DHT; really potent testosterone) at or around birth (see http://en.wikipedia.org/wiki/5-alpha-reductase_deficiency). Thus, some young boys don't develop noticeable testes or penises because those are triggered by a lot of testosterone. However, at puberty, because there is a ton of testosterone, their testes descend, and their development continues like any other genetic male (except they don't get male pattern baldness). In the Dominican Republic, the prevalence is about 1 in 90, so it's seen as a normal thing, at puberty, for some .girls' to start being .guys.' It's called "Guevodoches," slang for "huevos a los doce" or "balls at age 12."
Chem33 was pretty cool. The professor in the first half (professor Stack) was very awesome. The professor in the second half was less charismatic, but still good.
I really like how the course was taught. Its entire emphasis was problem solving using chemicals. In Bio, you had to memorize 50 pages of powerpoint slides per lecture to get an A. In chem, you needed to know and be able to apply the concepts on one page, sometimes front and back, worth of diagrams of chemical reactions per lecture. It helped if you played around with your chemistry modeling kit so that you could see what the reaction looked like in 3d. They even abstracted away most of the math: you need to know how to intuitively get a rough estimate of rates of reactions and how to use equations when provided with them, but we didn't have to memorize a bunch of formulas. I probably memorized less than a dozen specific things about chemical stability or reactivity and less than a dozen different types of reactions, but I understand problem solving with those, and I understand why they work, so if I were presented with a problem that I hadn't heard of before, I would probably have a decent chance of solving it, and I'm also much more prepared to incorporate more chemical facts into my existing theoretical framework. In other words, I feel like I actually understand the material.
Apparently, a lot of premeds take chem33 and find it a lot harder than bio. The proposed reasoning behind this is that premeds know how to do memorization, but problem solving is something new and foreign. I wish all classes emphasized problem solving.
As discussed in the bio section, because of this emphasis on problem solving, chem33 has good tests even though they're closed book. It's like a CS test: the material tests you on what you know by giving you problems that you've never seen and probably never thought about. However, you have all of the skills to work through it, so by the end of the test, you feel like you learned something new.
My one critique is that their first test was time pressured, which they fixed for the second test and the final. I disagree with time pressured tests because, even if you know the material intuitively, a time pressured test means that you won't have a chance to go back over your work to check it for errors. On the first test, I got about the median (~60%?), even though I had a strong understanding of the material, and at least half of my errors were just mistakes because I couldn't check my work. In CS, few of the tests are time pressured.
An algorithm is the overall strategy for solving a problem. The implementation of an algorithm will be different depending on the programming language, but the overall framework will look the same regardless of whether the program is in C or C++ or Java or Python or any other language.
CS161 is dedicated to the analysis and design of algorithms.
Analysis means looking at it and seeing how fast it is on average and in the worst case. "Fast" is measured based on the size of the input. When you're only analyzing a dozen numbers, you would have to actually try to make an algorithm that would be slow.
However, once you get to a few dozen, "exponential" algorithms start behaving slowly. I had a conversation earlier with Owen Zahorcak, my former debate coach, about genetics. His thought was that, since computers are getting faster, we'll eventually be able to use "brute force" (looking at every possible solution and seeing which ones are desirable) to analyze genetics. To do that, we would need to come up with every possible gene rearrangement and see what a body with those genes would look like. The problem with that, however, is that even though computers are getting faster at an exponential rate (about twice as fast every 18 months -- see Moore's Law), there are too many genes for any time in the foreseeable future. Every gene has 4 possibilities (a,c,g,t). There might be between 20,000 and 30,000 protein-encoding genes in the human genome. To look at every possibility, that would be, conservatively, 4^20,000 possibilities, or 10^12041 possibilities. As a comparison, there are less than 10^81 atoms in the observable universe, and the universe has been alive for less than 10^18 seconds.
It's hard to think about what that means, because that's a very big number. Lets say it takes a day for one computer to figure out one protein encoding given one sequence of genes. Current computers can do it in a few days, but to do anything really big with it, you'll need a lot more analysis -- how does that protein go throughout the body? What about the exponential number of interactions that protein has with other proteins physiologically? Etc. Thus, I think that one day is a reasonable number. Even though all of those numbers will be different for many proteins given any two different gene encodings, let's pretend that when one computer can figure out everything to do with one protein encoding on a gene, it can also figure out all of the other proteins, so it takes one computer one day to reduce the 10^12041 possibilities by 1. The hard thing to grasp is that that doesn't mean it will be 10^12040. It means it will be 10^12041 -1. 10^12041 means a 1 with 12,041 0s after it. Another way to think about it: 10^6 = 1000000, or 1 million. The exponent there is 6. Double the exponent to 10^12, and you get 1000000000000, or 1 trillion. When you multiply two numbers, the exponents only ADD, and when you divide, the exponents only subtract. So, 10^6 * 10^6 = 10^12. 10^12 / 10^6 = 10^6. What this means is that if you have a trillion computers crunching those numbers, 10^12 computers doing nothing other than calculating genes all day of every day, it would take 10^12041 / 10^12 = 10^12029 days, a time much longer than the life of the universe. If you had 10^81 computers (one computer for every atom in the observable universe) computer for 10^16 days (as long as the universe has been alive), that's only 10^97 done, which means there would still be 10^11944 remaining. In other words, if you had an impossible number of computers computer for an impossibly long time, you would only solve a drop in the bucket of the human genome using brute force.
You can probably see why something as big as solving genetics using brute force would be intractable to do using brute force. To give you more of a feel for when exponential problems become intractable, think about it like this: 2^10 is about 1000. 2^10 = 10^3. A computer now has a few gigahertz of processing power, meaning it can do something like 3 * 10^9 calculations per second. If you want something to be instant, then it has to be able to use 100% of your processor for less than a second. In other words, "instant" means less than 10^9 calculations, so instant means less than 2^30 calculations. 2^40 means it takes 10^3 seconds, or 1000 seconds, or 16 minutes. 2^50 means it takes almost 12 days. 2^60 means it takes almost 32 years. In other words, only very small exponential problems are doable.
Polynomial problems are problems where, if the input is size n, then the running time is n^k (versus exponential problems where the running time is k^n). If our dna problem was backwards and required 20000^4 rather than 4^20000 problems, that would only be 10^17 problems, so if you had 2000 computers (and could split up the problem between them), could solve the problem in a day.
For most problems, we aim for polynomial time solutions with a small k. So, n^2 (quadratic) or n^1 (linear) problems. It's hard to do faster than linear time because if the only thing you do to each input is take one look at it, you have a linear time solution. To do faster than linear time, you need to have your input so organized that you can literally say that most of it is unimportant without even looking at it.
Linear time solutions that can't be easily divided become non "instant" when your input size is bigger than a billion, but they're tractable in most instances, and if you can split it up among a bunch of computers, it's good in most instances.
Higher order polynomial time solutions, though, are less nice. If you have an exponent to a power, the two exponents multiply. 10^6^2 = 10^12. An input of size 10^5 (100 thousand) means 10^5^2=10^10 operations in a quadratic solution, which means 10 seconds. Higher order polynomials are annoying, but they're often used because they're the best that we can get (and, often, we can prove that it's impossible to solve a given problem in less than that runtime).
There are, however, problems that can be solved in sub-linear time. For instance, when you have a dictionary, you don't have to look through every word before you find the one you want. Since it is sorted alphabetically, you can examine one word in the middle of the dictionary. If your word starts with "k" and the middle word starts with "m," then you can ignore everything to the right. Then, go halfway between the start and your "m" word. Now, you have reduced your input size to 1/4 of the original in only two tries. This is called "logarithmic" time, and the particular algorithm that I just described is called "binary search."
A logarithm is the opposite of an exponent, so as slow as exponents are, logarithms are that fast. In other words, if you have an input of size 2^100, that only takes 100 operations in logarithmic time (you would only have to look at 100 words to find the word you want in a dictionary that had 2^100, or 10^30, or a thousand trillion trillion trillion words, in it). Logarithmic time is really fast. A lot of times in algorithms, the time will be something like "n log n," which means that you need to do something logarithmic a linear number of times. For instance, imagine you needed to look up n different words in a dictionary that was size n. Then, you would need to do n lookups, and each lookup takes log n time. n log n is a lot faster than n^2.
So if logarithmic time means that you can solve half of the problem in one operation (or 1/k of the problem, where k is some constant), what's faster? Constant time. Imagine that I had the dictionary memorized such that, even if there were 2^100 words in it, I could tell you the definition of any of them without even looking (or if I could tell you what page it was on without even looking). That's what constant time looks like. What's crazier is that this is actually possible. The usual way of achieving this is to use a "hash map." A "hash" takes a number from a big universe (ie, all possible words, including both "the" and "xxfyfz-lllllnk," since we can think of each word as some number) and calculates it into a number in a smaller domain (ie 673). This means that multiple numbers from the big universe might map onto the same smaller number (ie "the" and "ffffqrln" might both may to 673 depending on your particular mapping algorithm), but since there are only a few thousand words in English compared to 26^8 = 10^11= 100 trillion possible words with less than 8 letters, if we make our "small" domain big enough, we can make sure that it's unlikely for any two real words in the English language to map to the same small number. As long as our hash function only takes a few operations to do (ie, take a big prime number and multiply it by a different number a different number of times depend what the word is), that means that, no matter how many words my dictionary holds, I can tell you the definition of a word in the same amount of time.
Another way to think about constant time that I learned about in my Google interview in December: imagine you have a sorted array (array means list) of n numbers that are drawn from the uniform distribution between 0 and 1 (in other words, if you have 100 numbers, you'll probably have one number close to 0, one number close to 0.01, one number close to 0.02, one number close to 0.03, etc, because every number between 0 and 1 has the same probability of coming up). You can look up numbers in this array in constant time because you know about how far into the array a number will be. Say, for instance, that your array has 100 numbers and you're looking for the number 0.25. 0.25 might not be exactly a quarter of the way through, but you know that it'll be very close. If you're at 0.249, then you can probably just look at the next number in the array (you can do this by performing the same lookup given the distribution and the restricted domain). Thus, you can do something similar to a binary search, but because you know the distribution, you know that you'll get close. In other words, if you know the distribution of words in English, you don't have to go to the middle of the dictionary to search for "knapsack," you can go right to the "kna" page. If you're like google and you know the distribution of word usage in language, I would imagine that you could use this trick to get constant time lookup using a sorted array rather than a hashmap (though hashmaps are probably still used in practice. And, really, this trick is more or less a hashmap where the hash function is the "I know what the distribution is" function).
We looked at a few main ways to design algorithms.
"Divide and Conquer" means that you use "recursion." In order to solve one big problem, you'll split it into a few smaller subproblems, do a small bit of work, and then get your solution. To solve each subproblem, you'll split it into a few smaller subproblems, do a small bit of work, and then get your solution to the subproblem. And so on until the problem is really easy to solve. For instance, to sort an array, you can use the following algorithm: If there are only two numbers, put the smaller number first. Otherwise, split the array in half. Sort the first half. Sort the second half. Merge the two split halves by looking at the smallest element in either half and adding that to the merged array.
"Greedy" means that you make a myopic decision that seems like it will be good now, and you hope that it will work out for you in the future. For instance, if you're trying to find the shortest path between two locations (and you can only go between two points that are on your map. Ie, if you're driving, you have to follow roads, and you can think of .locations' as intersections between streets), you can start at one location, keep track of all of the partial paths, and keep expanding out from the start, always starting from the shortest path, and it turns out that the greedy algorithm stays ahead, so the first path that you find will be the shortest one.
"Dynamic Programming" means that you're trying to solve a problem that has a lot of choices (dynamic programming is often using where the number of choices means that a brute force solution would be exponential). However, not all choices are as good as all others. The reason that exponential problems are so bad is that, if you have 3 steps, each of which has 4 choices, then there are 4 * 4 * 4 = 64 possibilities. What if I only had to pursue a subset of those possibilities? What if , after looking at the first choice, I could definitively say "Option 2 is best." Then, I would only have 4 * 4 = 16 choices remaining. The key insight of dynamic programming is that you can more easily explicitly figure out the solution to a smaller subproblem and throw out all of the bad choices.
We also did a few other things. We did "linear programming" which is a method of approximation that uses linear algebra to solve matrices quickly, we did some analysis of randomized algorithms, and we also looked at a bunch of specific examples of each of these types of problem solving.
CS161 was hard. I don't feel like I understand the material as well as I want to (mostly because the material is really important and this is the first CS class that I didn't feel like I mastered by the end). Part of this could be that I didn't have as much time to read the book as I wanted to and that I was nodding off in lectures. It could also be that I learn best when programming, so doing a lot of proofs that algorithms were correct or fast wasn't quite as good of a teaching mechanism for me, though I'm not sure what the alternative would be.
Part of it could also be less of a lack of understanding of the material and more just that the final was hard. When Professor Roughgarden gave out last year's final, he said "I give out tough finals. The median last year was an 80." Dramatic pause. "Out of 150." This year, the median was closer to 60/150. I'm not sure exactly how I did -- I got an A, so I probably got a good deal above the median, but I don't feel like I got much higher than 80 or so. It was really hard and really time pressured. Most people didn't put anything for all of the questions and didn't put complete answers for the rest. There were 8 questions, and a lot of them were similar difficulty to problems that we would spend 2 hours on for homework. In other words, we would need to compress 2 hours of thinking into 20 minutes to do as well on the final as we did on the homework.
Because the material is so important and I didn't do as well (in terms of understanding) as I would have liked, I plan on doing a lot of independent study of algorithms. And taking a lot more classes.
On a brighter note, there was an algorithmic problem presented to me last term that I just figured out the solution to. Imagine you want to pick out a random piece of data from a data stream that could be arbitrarily long. For example, if you wanted to pick out one frame, at random, from a video, and you didn't know how long the video was. You can't just store the whole video and then randomly choose after you know how long it is because it could be bigger than you can handle.
The solution? Imagine you only had enough room to store one frame. Store the first frame. Then, you know that if the video is only two frames long, you have to replace the first frame with probability 1/2, so do that. Then, you know that if the video is only three frames long, you have to replace whatever frame is currently stored with probability 1/3. Etc. If you do the math, you'll see that, at step 3, the first, second, and third frame each have equal probability of being stored. The same goes for every step -- and you can do this even though you're only ever storing one frame.
This was my first term section leading CS106, the introductory CS class at Stanford. Section leading means that I lead a small weekly discussion section, grade the assignments from each of my sectionees, attend meetings, attend classes, and have LaIR hours. The LaIR is a computer lab where students can come for CS help any time from 6pm to midnight Sunday through Thursday.
It was rewarding. I could see them get better as the term progressed. Some of them did less well towards the end since they were busy, but everyone was really smart.
Though grading was annoying, I gained an appreciation for the creativity within computer science. Each of my students would solve the problems presented in the assignments in a different way. Some parts were common, but a lot was unique. I looked back over my code from when I had taken cs106a, and there were a couple of places where my sectionees did something much more elegantly than me. They made me proud.
I did get stuck with the 9am section, but my sectionees had fairly good attendance. I brought food and tried to be energetic, and I guess it worked.
Early in the term, there were two two-hour meetings per week for the new section leaders. To find a time that all of us could make it meant that most of them started at 9 or 10pm. The meetings were practical discussions of teaching. It was especially nice that they included discussions of the gender disparities in CS education.
I plan on continuing in the future.
I took the team to three tournaments this term.
On 1/22-1/24, we had the Golden Desert tournament. Three teams went, including one first year debater. It was in Vegas. And I was their only chaperone.
I didn't touch bases with Paly at all during the Stanford tournament because of my lack of sleep and my double-time-commitment during the tournament.
On 2/12-2/14, we went to the Berkeley tournament. Berkeley is always fun. There are also a ton of people there, so I got to see some folks that I hadn't seen in a while like Owen, my former debate coach.
On 3/5-3/7, we went to Harker. They treated us well (they gave us good food).
We did fairly well. Nate and Chloe, a team that was trying very hard to get to the tournament of champions, didn't quite make it, but they did do well at Golden Desert, and Nate did well with another partner at harker. I guess they had some partner issues that they were never quite able to resolve. Abby and Grace had a very laid back attitude, but they did very well at each tournament, getting to elimination rounds at Golden Desert and Berkeley. Alex and Steven pretty much won the novice division of all of the tournaments that they attended this year. Too many other debaters to list all of the awesome things that they've done, but it feels good to be their coach.
Also, there are a lot of other kids who are very dedicated. At each tournament, I judged rounds in addition to coaching, and there were a few teams that would really follow up with me about how to be better debaters.
I'm excited about debate next year. In the spring, there will be a new batch of novices to train, and my old batch will finally be committed to varsity debate. Two of my sophomores will be going to Michigan 7 Week, the debate camp that I went to before my junior year. I'm really proud of their commitment: one of them is a first year debater, but he has really worked above and beyond what he had to, and his work is showing. They will ROCK next year. I am sad that one of my debaters who wanted to go to Michigan 7 Week won't be able to because of finances.
I ended up giving one of the debaters more help on their application to Michigan than I originally said that I would. His parents were very thankful, and they sent me a box of chocolates.
Next year, the current head coach, Ben Picozzi, is graduating, and I will become the head coach of policy debate. I've also been talking to a debater who was accepted early to Stanford, and he seems on board to be my assistant coach if he decides to come to Stanford. He seems to be a good debater, to compliment my skill set (he's more policy oriented, whereas I'm more critical / philosophical in my debate style), and to be responsible. Fingers crossed!
On 1/20, I was part of a Haas Center for Public Service focus group as a representative of .activist' groups on campus. The questions that they asked seemed a little bit weird -- almost like they were trying to feel out my philosophy towards service more than places where I need the Haas center's help. They are, overall, a good organization, though, and I'm sure they'll do well with the information that they gathered.
Our next series of events was on the week of 2/1. We had an International Queer Rights week. On Monday, we showed the movie Fire, about Queer women in India, along with dinner and discussion. The interesting part of Fire is less the movie and more the reaction to it -- there were riots. On Tuesday, we had a discussion with Professor Matthew Sommer about historic queer issues in China. He studied court cases and found evidence (and, to an extent, implicit acceptance) of openly queer people in ancient China. That is not to say that there weren't issues; just a different set of issues.
On 2/4, we had a slight change of pace and did a "Queer Women's CoHo [Stanford's coffee house] Takeover." The flier was cool.
Our other big event of the quarter was Queer Formal, an alternative to prom-style events. The speakers that we rented broke halfway through, so we had to relocate for the last hour or so (and Events and Labor Services charged us for the speakers. Urgh). Next time, we'll probably get our own speakers. The event itself was pretty good, though. People enjoyed the Lady Gaga theme, the pictures, the chocolate fountain, the food, the deserts, and the dancing while it lasted.
Since I'm the financial officer of QSA in addition to its cochair, I had to go through the .special fees' application process. This year's senate went rogue and decided to cut costs at the expense of student groups and without much student input (as a result, only one of them is even running for reelection, and a lot of angry people have decided to run for senate), so our budget will be a bit smaller than last year.
Close to the end of the term, I realized that noone started to organize the Queer Coalition, the organization that gathers together the Queer groups on campus to make an endorsement of people running for student government. In the past, after the folks were elected, endorsing groups (Queer Coalition, Students of Color Coalition, Women's Community Center, etc) dissolved. However, with this year's senate fiasco, Queer Coalition and some other groups are planning on continuing throughout the spring and next year to ensure that the people we elect are held accountable for their campaign promises.
Because this happened close to the end of the term, everyone was busy. While I got a lot of people to agree to sign on and help as much as they can, a lot of the impetus has been solely on me to make things happen -- ie, interviews, getting pictures from candidates, getting pledges from candidates. I am very grateful that people will step up when I'm out of my league. Particularly, QSA's art person has agreed to design the flier that we'll use, and everyone has been responsive when I've asked their input about anyone.
It does feel a little bit weird that one year ago I was applying for endorsements and now I'm handing them out. It also showed me how hard it is to be an interviewer. People can put together pretty words, but I know that they got coaching, and how can I determine actual commitment? What question can I ask that will tell me if the person that I'm interviewing is a good person? What question can I ask that will tell me if they are organized or if they understand policymaking?
I'm a director in Dance Marathon Hackathon, a 24 hour event during Dance Marathon, an AIDS fundraiser, where Computer Science students program for nonprofits. The event took place on 2/6, so Hackathon involved a lot of organization in the first month of school. We had to get together advertising to attract hackers. We put up fliers all around, sent out emails, and tapped into Dance Marathon's big network to help us out.
I also got a few minutes of fame as people interviewed me about Hackathon: http://www.stanforddaily.com/2010/02/08/dance-marathon-raises-money-for-haiti-aids-awareness/ and http://www.stanforddaily.com/2010/02/05/lending-a-helping-hand/
One slightly annoying thing was slow responses to email. As such, folks didn't get things done until the last minute. We made Hackathon tshirts, but noone responded to my email for about two weeks, so I had to make the tshirt design on my own at 4am before the day we wanted to get the design done. At that point, my position was that I came up with something decent, the time for debate had passed, and if they wanted something else, they should feel free to design it and email it to the rest of us. No one seemed to object to my design, so I had Dance Marathon's graphics person touch it up, and we had the shirts made.
There was also some drama among the organizers. Seemingly small debates, such as whether or not to charge a registration fee to hackers, became long and divisive because of philosophical differences regarding the relationship between Dance Marathon and Hackathon. The folks who saw the two as separate events thought we should not charge a fee, and the others thought that we should.
My thoughts were that, because Dance Marathon and Hackathon were part of the same event and all of the dancers had to pay a registration fee, it would be weird to say that a certain group at the event didn't have to pay, and even though Hackathon has sponsors to pay for our unique expenses, we still should help Dance Marathon pay for joint expenses such as event space and food. Hackathon wouldn't exist absent Dance Marathon, and they provided continual support throughout the year (shirts, fliers, word of mouth, web hosting...), so I certainly see Hackathon as a part of Dance Marathon rather than as an independent event.
I was less organized than I should have been with my project. InSTEDD got back to me in December saying that they would be interested in having a Hackathon project (and that they had an internship available) whereas we had been working on the details of our other projects for several more months. As a result, the project wasn't as well defined as some of the others, and the hackers had more freedom to design a tool on their own. There were five hackers working on the InSTEDD project, and all but one of them was fine with the freedom. The other one ended up switching to another project.
The event itself was awesome. I didn't really participate as a hacker -- I ended up doing logistical work throughout the event. I made sure that the hackers were well fed. I organized a Super Smash Brothers tournament among the hackers. I raffled off some prizes. In the post-event surveys, the hackers seemed happy about how the event turned out, so I'm happy about what I did.
It also gave me a chance to experience Dance Marathon. Hackathon was 80 people in a room working together for nonprofits. Dance Marathon was closer to 1000 people celebrating (by dancing for 24 hours) the culmination of a year long fundraising effort. Computer scientists, in general, lack the same type of energy that attracts people to Dance Marathon, so immersing myself within Dance Marathon was refreshing.
I could feel myself as a part of a larger community dedicated to service. In an earlier letter, I wrote about how Dr. Larry Brilliant said that a community of service was necessary because the truly important projects will fail 8/10 times. He said that you needed that community to keep you working hard, because those two successes are important. Dance Marathon showed me what he meant. Being a part of Dance Marathon showed me that I wasn't alone. The community made service more rewarding and more fun.
I mentioned earlier that there were philosophical differences about how connected Hackathon was to Dance Marathon. After experiencing the community of service, I can understand just how inseparable the two events are. One of the core missions of Hackathon is to show that there is meaningful and interesting public service for computer scientists to do. Being part of a community of service does just that. In the post-Hackathon surveys, people indicated the same. They wanted more of a connection with the dancers. They were fine with the registration fee.
It was also exhausting. Because I was helping to plan the event, I didn't get 8 hours of sleep coming up on the event. Because I had to help with the Stanford High School Debate Tournament immediately after Hackathon, I didn't catch up on sleep after the event ended. I did manage to sleep through a few classes to make up for it, though.
To give you an idea of the impact, we had about 80 people coding for 24 hours, which is almost 2000 person hours. Nonprofits pay $20/hr-$100/hr (yes, it's obscenely high, but it happens more than you might think) for tech help. Thus, the one day event raised between $40,000 and $200,000 of work for nonprofits.
A few weeks later, the Dance Marathon and Hackathon directors had a last joint meeting, and we talked about next year. I was the only person interested in being a Hackathon director again, so next year I will be the sole director of Hackathon. I'll recruit some staff to help me out, but I'll take on all of the broad organizational duties. It'll be great!
Towards the end of February, Garrett Neiman, one of the Dance Marathon directors, contacted me about a nonprofit that he had founded, SEE College Prep, which gives SAT prep courses to high schoolers who are typically underrepresented in college. He wanted my technology advice. I guess that, because I went the extra mile to help Dance Marathon and saw Hackathon as part of the same event, we made a connection, and he thought that I would respect the work that he was doing (and he needed technology help). I'll probably help him out next quarter.
I did a few scattered projects for the student government tech team, but not a ton. A lot of what I did was just maintaining the website and making some posts every once in a while.
I only went to one tournament at the very beginning of this quarter -- the others conflicted with either my or my partner's schedule.
I don't think that I'll debate next year. That means that I won my last debate round ever -- and on a criticism of capitalism, no less.
On 2/7, Stanford had its high school tournament, which acts as our main fundraiser of the year. Because this was at the same time as Dance Marathon Hackathon, I missed out on the first day and was exhausted on the second day. However, I also made internet happen at Stanford for the first time ever. I had to make a script to create hundreds of wireless guest accounts because Stanford doesn't provide any way to make masses of accounts. I also needed to take some annoying precautions to make sure that other students don't pirate while using my guest accounts (my solution was to give the accounts to coaches and have them sign something to take responsibility for it).
The internet is increasingly necessary for debates. First, a lot of teams are starting to debate paperless (ie, they have their evidence on their computers) to save costs and tress, and if they need to get an up-to-date copy of the evidence, they might need to download it from a server. Second, judges are putting their philosophies up on http://judgephilosophies.wikispaces.com/, so having the internet will help debaters adapt their speaking style and argument choice to the judge. Third, teams disclose their strategies on http://debatecoaches.org/wiki/, so having the internet means that there will be fewer debates that are bad simply because one team was caught off guard. Information makes everyone a better debater.
On 2/20, Stanford Association for International Development (the folks who put on Food For Thought last year) put on a conference on development in areas that had been in conflict. I wasn't very impressed. The people presenting didn't seem to do much other than drop the names of the famous people that they knew. The content was all obvious, like talking about how developing institutions were important.
On 2/27, there was a panel with Valentino Achak Deng, Anne Bartlett, and Terry Karl about ethnic violence.
Professor Karl emphasized that the issue is not ethnic hatred, but a desire for political power because invoking ethnic hatred is a very powerful tool.
Valentino Achak Deng talked about his own experiences living in Darfur and, now, starting up a school system (see http://www.valentinoachakdeng.org/).
Anne Bartlett, the director of the Darfur Centre for Human Rights and Development, was probably the most emphatic. She problematized the language used to talk about genocide. The Sudanese government uses the term "insurgents" to avoid blame because they cast the genocide as similar to what the US is doing in the War on Terror. She also discussed the current efforts: almost none of the $300,000,000 raised by Save Darfur went to actual people in Darfur. We need a lot more political pressure, and we need programs like Deng's that materially benefit the victims of Darfur.
On 3/1, there was a Sophomore Faculty Night. I was paired with someone who researched the history of science, but they ended up not showing up. The next night, Terra had its faculty night, and I invited Mehran, my advisor. They were some good chats.
On 3/3, Winona LaDuke, Nader's old running mate, gave a talk. It was at the same time as my chem midterm, but there was an alternate time, and they let me take it then.
The talk was good. I had seen her talk at the University of Oregon a few years ago also, which is why I went to such lengths to attend her talk this time. She has a new website: http://honorearth.org. Its recent literature section has a lot of cool stuff, particularly their Sustainable Tribal Economies publication (http://honorearth.org/sites/honorearth.org/files/Sustainable-Tribal-Economies-HTE.pdf).
One thing that I like about her is that she makes the pragmatic appeal for .hippie' policies (in addition to the hippie appeal). In her case for wind power, she described how
1. 57% of energy is wasted between the point of production and the point of consumption because our generation and transportation mechanisms are so inefficient
2. Global warming will reduce global GDP by 20%
3. Her community had 50% unemployment, yet they were sending 25% of their money out of the community ffor energy.
When she bought a 75 kilowatt wind turbine to power a school, she was addressing each of these concerns. Producing the energy locally makes it more efficient. Using wind rather than coal means that there isn't CO2 in the production of energy. By getting a small turbine, she was able to use people from her own community to build the turbine.
Similarly, if you compare a seed indigenous to her land to a GMO seed, the organic was universally better (in addition to the numerous environmental benefits -- sequestering carbon, not having fertilizer runoff into local water supplies, not leading to monocultures, etc). When you're dealing with complex systems, everything has tradeoffs. GMO seeds are more .efficient' because they sacrifice nutritional value, frost resistance, drought resistance, and wind resistance for size and calories. In other words, they look good when you sell them, and they grow well in greenhouses when you have the resources to give them perfect conditions. For real farmers, that set of conditions does not exist, and the Monsanto crops fall down in the wind, dry up when there isn't enough rain, and give people diabetes.
While I think she was making a point about *industrial* agriculture being bad, she also talked about the environmental cost of eating industrially produced meat. Eating one kilogram (2.2 pounds) of industrial beef produces more greenhouse gasses than driving a car for 3 hours while leaving all of the lights in your house on (Sustainable Tribal Economies 20), and industrial meat production accounts for 18% of global greenhouse gas production, in addition to causing deforestation of the rainforest.
One thing that I try to do when I meet cool people is ask them for career advice, since I still don't feel completely justified in going into computer science. She said that her wind turbine needs a computer and that there is plenty of opportunity for CS to help build a more sustainable world.
On 3/4, Valerie Jarrett, one of Obama's senior advisors and the chair of the White House Council on Women and Girls, gave a talk. I got good seats because I was in the ASSU.
In my past letters, I had talked about politicians giving content-free talks that make me ashamed of them. Jarrett actually made me feel good about having her in office. She cares about helping people. She didn't tell long winded jokes with famous people (though she did know Obama for years); she actually talked about the important issues.
Go Obama!
I took my bike in for a tune up after the gear changing cable got a little bit stretched out, and I was amazed by how much smoother the bike as a whole was. I guess it needs a little bit of oil and a little bit of air every once in a while.
One thing that I would like to do is learn more about my bike. I think I have most of the mechanics figured out, but I still can't quite pinpoint some issues that I have. For instance, sometimes when biking, there's a feeling like I'm pulling on a cord and then it twangs when I put pressure on the right pedal. It happens regardless of what gear I'm on. It only ever happens with the right pedal. If I only bike using the left pedal, the right pedal doesn't twang. But there isn't anything that I can see that gets pulled. It isn't noticeably causing any problems, but I just wish I knew what it was.
This term, all of the baking slots were taken up, so I decided to cook. Every Tuesday, I got together with 3 other people and made dinner for all of Terra. We each came in with a menu two times for the term, and we would all cook it. One of my menus was tofu, stir fry, beans, and rice. The other was Purim themed. The first one turned out better than the second.
It was a good experience. There were a few people on my cook crew that were more experienced cooks than me, and I learned from them.
This term, there weren't any bands that I listened to more than others. Instead, I mostly listened to Pandora, an online radio where you give it a song or artist that you like and it finds related music. It works pretty well.
My gmail account is now 93% full, which means that 6936 / 7438 MB are used. Compare that to the 77% as of the last letter.
What surprised me this term was how much of that isn't spam. I am on a gazillion mailing lists, but I actually respond to a lot of the email that ends up in my inbox. Between Jan 1, 2010 and Mar 28, 2010, I have participated in 926 email threads (for those of you who don't use gmail, an email thread, referred to by them as a conversation, is everything that shares one subject at one time. So, if myself and 3 other people are discussing A "Very Promising Hackathon Project" over email and we end up having 36 emails that span a month, that counts as one thread). In other words, I send emails in an average of more than 10 different threads per day. In that same period, I have received 6447 email threads, and I have read more than half of that.
Last time I talked about shaving and pride. Continuing the trend of the little things mattering, I have looked towards the clothes that I wear this term.
Clothes carry messages in their imagery, in their production, and in their history. What does the shirt commemorate or advocate? Where was the shirt made, and by whom -- did you participate in its creation? Did you wear the shirt at an important event, or did someone you know?
When I was taking my bio final, I wanted something on that said, "regardless of how I do on this final, I am proud of what I'm doing." I wore made-in-America socks; used jeans so that, regardless of their original production, I knew that my dollar would support Goodwill rather than a sweatshop; a made-in-America blue-collar overshirt from my Dad; and a sweatshirt that I bought in a fundraiser for a student organization that does service in Latin America and that I hand tie-dyed. I had my notes in my made-in-Seattle backpack. I wore my Hackathon tshirt: a shirt that I designed, that a local business printed, that was made in America, and that commemorated an event that I planned in order to catalyze service for international public health among Stanford computer science students.
I feel proud that I can get behind the message in my clothes. Often, there is a tendency to divorce individual actions -- purchases, words, careers -- from politics. It feels good to politicize a part of the personal sphere, regardless of how small a part it may be.
Dance Marathon, the 24 hour dance fundraiser for AIDS in Africa that Hackathon is a part of, is the biggest student organization on campus. As a result, they have the resources to do things manually. During the 1/10 meeting between Dance Marathon and Hackathon, they were talking about filling out 1000 or so mailing labels manually.
At that point, my experience as the campaign manager for the county commissioner campaign of Chris Crew, my former debate coach, came handy. One thing that Jessica Bradley, Chris' wife, taught me was how to use Mail Merge, a Microsoft Office tool for doing mass mailings. Since I hadn't used Mail Merge for over a year (I was campaign manager the summer before going to Stanford), I had to rediscover all of the intricacies and bugs in Mail Merge, but I figured it out and showed the Dance Marathon folks how to use it.
They were very excited. One of the people in Dance Marathon, Garrett Neiman (very cool person), started a nonprofit called SEE College Prep that gives college prep to underrepresented high school students. Garrett organized a meeting where I showed people from a few nonprofits and student groups on campus how to use Mail Merge and how to send personalized mass emails.
I put up a guide on mass emails at http://stanford.edu/~samking/guides/mass-email.html
Don't get me wrong, I believe in public service because of its intrinsic value, but I won't pretend that the accolades are unimportant.
I was discussing my plans for the summer -- do I want to program for InSTEDD or for Google? -- with some friends and mentioned that I was leaning towards InSTEDD despite all of the perks (including pay) associated with Google, one of them said, "Hey, we know Gandhi!" Another friend commented that she was glad that I was going to save the world some day. And whenever I show someone how to use mail merge for their nonprofit, they can't stop talking about it. People interview me for Hackathon, and the Haas Center for Public Service takes my picture to put up on their website (I think it's going up next term?) and asks me to participate in a focus group about how the Haas Center can better catalyze service.
It is a little bit weird, though. There is a large community of people at Stanford interested in service, but it seems like not all service is praised equally. Whenever someone praises me for my service, it seems like it has to do with computer science -- mail merge, or Hackathon, or InSTEDD. Third parties haven't given me that same level of recognition for being a debate coach or for being a chair of Queer Straight Alliance.
I imagine that the same is true of a lot of service at Stanford. A lot of Stanford students tutor or raise money for good causes or spend their summers volunteering or make sure that women or students of color or queer students on campus feel accepted, and I don't think that they get the recognition that they deserve. We're all in this together.
On 2/21, the Women's Community Center put on Vagina Monologues, a feminist play about appreciating vaginas rather than pretending that they don't exist.
The last time I saw it was a few years ago at the University of Oregon. I don't remember a ton about that rendition (except for the part that Kawa, my sister, was in), but I do know that Stanford's rendition had a few different parts -- it's an evolving play.
The one thing that struck me this time was the .moans' skit. Part of it is a series of racial and ethnic jokes. The rest of the play was fine, though.
One cool thing about the play is that the proceeds went to a shelter for the victims of relationship abuse and to a program that helps women in the Democratic Republic of Congo. STAND, a student group on campus, was also using the opportunity to raise awareness about our individual responsibility for the military dictatorship in DRC. A lot of high tech stuff (game consoles, computers, cell phones, etc) uses coltan, and DRC is the center of global coltan production. Even though some electronics manufacturers, like Sony, claim to not use coltan mined in DRC, if you look at the amount of electronics that they produce, it's very nearly impossible for them to do otherwise (that is, even if they don't buy it directly from an organization that advertizes that they buy from DRC, the coltan almost certainly originated in DRC).
Selling coltan is one of the main ways that the military dictatorship gets the money to stay in power. In general, buying raw materials from failed states is one of the main factors that keeps them failed states. If the international community is willing to buy raw materials (oil, diamonds, coltan...) from dictators, then that means they're willing to trade with anyone. That means that if a military coup takes power, they know that they'll have international financial support. In other words, it means that we can't use trade as a weapon for good.
Aside from keeping dictators in power, coltan production is immoral in itself. In sweatshop labor, there's at least the formalism of freedom -- workers could, in theory, work somewhere else. There is no such formalism in DRC. If you don't follow along with the military, there's a good chance that you'll be raped or killed.
That's why I'm so resistant to buying a new cell phone or laptop. I don't think that I should be ok with actively supporting slave labor.
I wish that I knew of a cell phone or laptop manufacturer that had an ethical means of production.
On 1/9, I saw Avatar in 3d with some friends from Sophomore College. It was pretty cool. 3d is really the way to tell a story. Especially in as pretty an environment as exists in Avatar.
Some people complain that the story wasn't original (white man goes to foreign land, betrays the native people, falls in love, and redeems himself. Featuring evil corporations, benevolent scientists, and cool aliens!). I didn't mind. I also don't care much about .originality.' Once I read that there are 12 archetypal stories, and every story falls into one of them, thus nothing is original. I'm also not very critical of movies in general. I feel like a lot of people spend too much time predicting what's going to happen and otherwise evaluating movies while they're happening rather than just experiencing the movie. I prefer to just relax and watch. I'll notice if there are any particularly exclusionary messages or if there are any interesting messages, but I never saw the point in evaluating the unoriginality of a movie except insofar as it relates to understanding allusions that the movie makes.
Avatar probably wasn't original, but it was aesthetically pleasing, had an ok message, and told a story well.
My one critique is that it shouldn't have had a happy ending. Avatar is a corporate dystopia (as opposed to a statist dystopia) where pollution has ruined Earth and greed is causing an interplanetary corporate resource war, but the valiant heroes stop the evil corporations from going too far. There's no "this is what your destructive ways will get you" moment.
On 2/19, I saw Shutter Island with Thomas and David. It included a critique of the poor treatment of people with mental illnesses, but that was tangential to the movie as a whole. The movie wasn't bad.
I saw a youtube video by Taylor Mali, a teacher and slam poet. http://www.youtube.com/watch?v=0xuFnP5N2uA is a more polished version. http://www.youtube.com/watch?v=RxsOVK4syxU is a more emphatic version. I like the second version, even though it lacks the subtitle, "If things don't work out, you can always go to law school."
(in case you didn't watch the video, teachers make a difference)
Mali also has some funnier videos such as "The The Impotence of Proofreading" (http://www.youtube.com/watch?v=OonDPGwAyfQ).
Settlers of Catan is a game that celebrates colonialism of a small island called Catan. It's a fairly small and simple competitive strategy game.
Magic: The Gathering is the classic trading card game. I played it a little bit in elementary and middle school. My roommate last year resparked the interest of a bunch of people in Magic since he's actually good at it, and that continued to this year.
Munchkin is a game that I used to play in middle and high school. The folks who I play Settlers and Magic with wanted a new game, so I invested in Munchkin. Even though it lacks some of the Munchkinly fire that it had in middle school (mostly because I was playing with folks who were themselves Munchkins, who memorized all of the rules, who knew most of the cards by heart, and who would always try to cheat), it's still very good.
Super Smash Brothers is a multiplayer cartooney fighting video game. A few people in the dorm have an old version of it, and I play with them occasionally.
I read "The Road" by Cormac McCarthy and finished "How To Talk So Kids Can Learn" by Faber and Mazlish.
Parents:
Don't read "The Road." It's a very good book, but you'll be a wreck from start to finish.
Do read "How To Talk So Kids Can Learn." It has a ton of practical advice on helping kids learn.
My review of The Road:
In The Road, there is a father and son walking down a road in post-apocalyptic (humans are the only remaining species; the sky is darkened by ash) America. One person cited on Wikipedia says that The Road is the best environmental novel of the century because it emphatically shows just how important the biosphere is. Even though that's only the backdrop for the book that comment made me think about the importance of setting. Cormac McCarthy was able to focus in so deeply on the relationship between father and son because there was nothing else. There is no job or school to keep either one of them busy. There is no community. There is no natural beauty to distract or to sustain. There is little hope of survival apart from one another -- why go on living when the entire world and everyone you knew in it was dead? The apocalypse means that the father is the son's world and vice versa. Thus, McCarthy has an opportunity to explore that relationship in depth.
From the perspective of the father, McCarthy explores two competing values. The father must care for his son's physical wellbeing, doing everything necessary to ensure his survival, no matter how ugly. The father must also care for his son's moral wellbeing, acting as a moral role model and instilling good values in his son. The two values compete when there are limited resources (and any economist will tell you that resources are always scarce. Otherwise, they'd be out of a job since economics is the study of allocating scarce resources). Food is always scarce, so when they encounter a starving person on the road, the question is whether they stop the person from starving now or stop themselves from starving once they run out of food.
In the end, I feel like McCarthy comes out on the side of instilling ethics. At the start of The Road, the boy wishes that he were dead. As the novel progresses, the boy grows more strongly into his moral framework, and he has a reason to continue living: he's one of the good guys; he's "carrying the fire." One time, the father and son come upon a starving elderly man. The father is reluctant to give him any of their food, but he does so at the request of the son. After they leave the elderly man, the father tells the son that he'll be thinking about how they gave the elderly man their food once they run out of food. The son replies, "I know. But I wont remember it the way you do" (174). The father ensures their survival, but by the end, the son ensures that they both are an ethical force, even when it means his own suffering.
I also think that the book is, overall, optimistic. The apocalypse is not an event in the book. It is the setting. Thus, the desperation caused by the apocalypse in The Road does not represent McCarthy's outlook on humanity. The book answers the question, "if the world is lost, what should our outlook be?" Throughout the book, there are very few instances to feel pessimistic, so long as we take the premise that the world is lost as a given. In other words, from the perspective of the boy, born in a world with no sun, there is still hope. The boy is not alone. The boy has experienced tasty food. When the elderly man says that the boy will "get over" whatever it is that makes him do good, the father replies, "no he wont" (146). There is struggle, and the world is cold, but humanity is not.
I read a good review of the book in Slate: http://www.slate.com/id/2151300/. It provided background on some of the literary allusions that McCarthy makes that were lost on me. McCarthy's descriptions of nature makes more sense in light of Hemmingway.
The movie is nothing like the book. The book has large, cold expanses with action every once in a while. The movie, in order to have all of the action contained within the book, sacrifices all of the empty space. The book was about the empty space.
How To Talk is something that you should probably read on your own so that you can get a lot of examples of how to identify and deal with the feelings that are often at the root of the issue. One tip that can stand alone without too much difficulty is to be more descriptive than prescriptive when giving praise or criticism. As Daria says to her mom in an early episode, "say, you have a friend who responds to everything you say with, .That's great!' This insincere reply is the same whether you saved a life or killed a bug, and thus becomes .negative reinforcement,' causing you to withdraw from that person or persons" (The Lab Brat). Generic praise like that -- praise that could be substituted in for any number of circumstances -- only teaches the child to look to authority figures for praise rather than themselves. If, however, you put the focus on what they did, and you simply describe the thing that they did that was good, the child will praise themselves. For criticism, give some praise, and then describe what still needs to be done.
The person sitting next to me lived in California and was a student at Santa Clara State in Psychology. She was reading a book by Sapolsky, who was the neurobiology professor in Bio42. He had given a guest lecture at her school, and she decided to read more about it.
When I moved to the lounge, I helped a math major with a linear algebra problem.
On the train back to Stanford, I was in the arcade car because the lounge was fairly full. Also camped out in the arcade car were two high schoolers heading down to LA for spring break. They were from Elma, Washington -- where Chris Crew, my old debate coach, ran for county commissioner. They were seniors that were going into the military next year. They were also kind of racist and antisemitic.
I guess one weird thing about the train is that the people who travel by train are fairly representative of the country as a whole, and the train is also a social experience. Those two things don't often coincide. Usually, people of differing beliefs don't really socialize about their political beliefs. Most communities (classes, businesses, activities, religious groups...) have some prevalent political beliefs or are mostly apolitical. Most places that don't have such beliefs aren't very social (travelling by air where there's a voice on the speaker telling you not to trust anyone, commercial places where people are busily trying to make their purchases). On the train, however, you have free discussion between people that have absolutely no common ground.
The train is transitioning towards more outlets. There's usually one car that has an outlet for every seat, but most of the cars haven't been upgraded yet. In the mean time, I still end up hanging out in the arcade car or in the lounge.
It is always nice to watch the dawn in the looking car. The dawn is pretty when I'm not seeing it as the result of an all-nighter.
CS110: Principles of Computer Systems
This class is the followup to CS107, which I took and this fall.
CS274: Representations and Algorithms for Computational Molecular Biology
I want to see how I like computational biology.
History287e: Jewish Intellectuals and Modernity
I was looking for a class about Levinas, an ethical philosopher who talks about an infinite ethical obligation to the other, and History287e was one of a few that featured him in their course descriptions. Levinas didn't make the final syllabus, but the class as a whole seems very interesting. The professor is visiting from a university in Israel, and it seems like there's enough freedom within the class that I'll be able to write my final paper (25 pages. Ugh) on something related to Levinas even though we won't do any official study of him in class.
EE190: Nuclear Weapons, Risk, and Hope
Martin Hellman is known because he came up with the idea of public key encryption (the thing that keeps the internet secure), but for the past few decades, he's been working on nuclear disarmament. He has an article called "Soaring, Cryptography, and Nuclear Weapons" (http://nuclearrisk.org/soaring_article.php) that gives some accessible analysis about how risky it is to be living in an age with nuclear weapons (ie, if we think that there might be a nuclear catastrophe sometime in the next 1000 years or that there is a 1/1000 chance per year of a nuclear catastrophe, that would be like going skydiving twice per week, or like being surrounded by 1000 nuclear reactors, and that would yield a 10% chance that a child born today would die of nuclear war). The class should be fairly interesting.
BioE80: Intro to Bioengineering
As per my discussion of Bio42, I like bio, but I dislike the way the introductory bio classes are taught. My hope is that bioengineering will have more of a focus on problem solving. The syllabus is similar to Bio42, but it has a project rather than a final, and the class as a whole seems fairly cool.
CS109L: Statistical Computing with R Laboratory
CS302: Progressive Tech Law
Stats166: Computational Biology (If BioE80 doesn't work out). I think I might be the only undergrad in the class, but I feel OK about it so far.
Taking CS161 made me think more about the different tracks in CS. The tracks are:
AI, Biocomputation, Graphics, Human Computer Interaction, Information, Systems, Theory, and Unspecialized.
AI teaches machines to do cool things that we can't easily program explicitly and probably can't do ourselves (like flying a helicopter).
Biocomputation deals with biological subject matter, like diseases and genes.
Graphics makes pretty, realistic, fast, easy pictures.
HCI makes programs that real people can use. Like games.
Information deals with things like securing information (encryption), compressing information, and ensuring information is clear (how to make sure that the signal that I send a hundred miles over the internet is the same as the signal that the server receives).
Systems goes into lower level programs and worries about the implementation details.
Theory stays at the high level and thinks about the broad structure of what a good program would look like.
Unspecialized means you take classes from all of them.
To some degree, I like Systems and Theory because Systems teaches you how to worry about every detail, and Theory teaches you how to think big picture.
My hunch is that AI, Biocomputation, and Graphics are somewhere between Systems and Theory since they're all about making fast, functional programs and require some degree of big picture and some degree of attention to detail. The difference is in the specific applications.
I'm not quite sure what the information track would look like. Looking at the requirements, it seems a lot like the unspecialized track -- a little bit of AI, a little systems, a little bio, a few electives.
The HCI track seems to have more to do with psychology than programming. How to make a program that is intuitive to use. I will certainly want to take some HCI classes, but I feel like one of the reasons that I like CS is that it is a very simple system that can be used for abstract problem solving. Specializing in psychology tackles the complex system of the human psyche. I get enough of that in politics; in CS, I would rather situate myself in the backend and let someone else make my programs usable by the general public.
I have some degree of interest in everything but graphics (many of my favorite games use simple 2d graphics. Then again, the CGI in movies is really cool). I might end up fairly close to unspecialized. The AI track is basically CS221 (AI; taken), CS229 (Machine Learning; planning on taking), plus some scattered other classes and electives. Information is CS124 (languages and information; planning), CS145 (databases; planning), and scattered other classes. HCI is CS147 and CS247 (the two HCI core classes. I'm taking at least 147) and scattered others. Systems is CS140 (operating systems; it's a right of passage. I have to take it) plus scattered other courses. Theory is CS154 (I probably won't take it, but that's because I've heard that it's like a slightly harder version of CS103, which I thought was very easy), CS261 (the second algorithms course; definitely taking), and scattered other courses. Biocomputation seems like it requires some depth in the field, but I feel like if I take the classes that interest me within each track, I'll already have fulfilled most of the requirements for all of the tracks because a lot of the courses that I listed specifically as ones that I'm planning on taking are listed in the .scattered other courses' of the other tracks.
I'm trying out biocomputation next term. I'm also taking a systems class, so I'll see to what extent I like systems (so far, I have really liked CS107, the previous systems class. Partially it's because I'm the type of person who likes thinking about the details in a program). I feel like Theory has a lot of mathy classes that I won't be very interested in even though they're very important, so I'll probably take a lot of theory classes, but not enough to be on the theory track. Same deal with information and HCI. I could see myself in AI also.
I'm applying to be an RA in East Flo (my frosh dorm) next year. I think that I want to be an RA because:
-I want to be a positive role model
-I like helping and taking care of younger folks, even in a limited capacity
-I want to incite my residents to gain an appreciation for public service
-I want to be a part of and help make the SLE community (the frosh in East Flo are mostly in SLE, the humanities program that I was in last year)
I'm still interested in public health. However, I don't want another term like this one. I don't want to take classes that strongly feature rote memorization. I don't want to take classes that are institutionally sub-par (see my discussion of Bio42) when there are plenty of amazing classes that I want to take but won't have time to. If I still want to do med school after I finish at Stanford, I can always take the premed classes then.
There are just too many premeds at Stanford. A substantial portion of them don't really care about the content of the med school prereq classes, and they're just in it for the grades. Thus, to be a premed, I would have to compete against people who only care about grades. As Michael Franti says, "Only a rat can win a rat race." Worrying about grades is not a rat race that I want to participate in for another two and a half years. I want to take classes because I'm interested in them and study them because I learn something new.
I already did my technical interview with Google in December. On 1/19, I interviewed with two different people that Google thought would be good matches for me -- people that I would work with over the summer. I talked to Barnaby James, who works on Google Sites (a platform for people to easily make their own website), and Ivan Posva, who does lower level stuff for Google Chrome and Chromium OS (ie, his history is in making Java Virtual Machines. One of my projects would have been helping make the Javascript engine work really fast).
It would be really cool to work on Chrome. I would have been improving a product that myself and millions of other people use every day. And I would be doing a type of programming that I have really enjoyed. And there are all of the perks associated with Google.
However, after getting the project from Google, I decided to go with InSTEDD. When I told the person that I had been talking to from InSTEDD, he said that I didn't sound very happy with my Google project. It wasn't that. Google did a good job pairing me with a project -- as I said, I love both the style of programming and the end result. It was more that I needed to see the perfect project from Google before I could actually do something as unthinkable as turning down an offer for an internship at Google (and a good deal of money). Only then could I appreciate the qualitative differences between working for Google, doing the most amazing thing that I could do in the corporate computer science industry, and working for InSTEDD, doing the most amazing thing that I could do as a computer scientist for global public health.
I guess, to some extent, I decided to take the advice of the entrepreneurs from the social entrepreneurship class that I took last term and buy a ticket to Cambodia (well, mostly. They mostly went to Latin America and Africa, and they mostly bought one-way tickets). Yes, I turned down 3 amazing meals a day, interesting speakers, an institution that will ensure that my internship experience is fun by putting on events for interns, and about 13k, but there's something more important, and this summer I'm going to continue finding out exactly what that is.
One thing that helped me feel more secure about my decision was Dance Marathon and Hackathon. I really felt what it was like to be part of a community dedicated to service (see the Hackathon section for more). Other people have mostly supported my decision, too.
Also, since I'll be in Cambodia, Kirsti Copeland, my academic director last year, found me a contact. One of the parents of a frosh in FloMo runs Pillows for Peace, a nonprofit that does development work in Cambodia. They have a program where people go to Cambodia, do some touristy things, and build a house. I'll probably take a few days off from InSTEDD to spend time with them. They're also seeing some of the sights that I want to see, like Angkor Wat.
Dad says that I should end my letters better.