You are viewing shagbark

The Blog Monster
 
[Most Recent Entries] [Calendar View] [Friends]

Below are the 20 most recent journal entries recorded in shagbark's LiveJournal:

    [ << Previous 20 ]
    Wednesday, February 29th, 2012
    3:50 pm
    Quote of the day: Men's yokes
    Interesting quote of the day, from Goodbye Chains:

    For every man, there is a yoke into whose traces he will gladly leap -- whose burden he will carry unto death, even beyond if he could. A yoke fitted perfectly to his shoulders, and engraved with a name only he can know. Sometimes it reads "Hate", sometimes "Love", or "God", or "Country", or a thousand other worthies -- a lover, a cause, a memory, a dream.

    Women are not like this. But we can see that men are. This is the greatest secret that we know.

    Agree? Disagree?
    Thursday, February 16th, 2012
    1:16 pm
    The last smart President
    I don't mean Obama. I mean whoever gets elected in 2020.

    Almost everyone born in the U.S. after 1970 in any socioeconomic class for which becoming President is feasible has grown up using the Web. Almost anyone intelligent and interesting born in the U.S. after 1970 has grown up writing content for the web. Most of which is still out there.

    I used to sometimes hesitate before posting something that could be traced back to me, and think, "Having a permanent record of this opinion on the web will make me unelectable for anything." By now there are hundreds of my opinions archived all around the world that each individually make me unelectable for any U.S. public office. Such as the fact that I'm a brony. While the U.S. electorate was ready for a black or female President, bronies are still beyond the pale.

    We've already seen that, to be approved by the Senate as a Supreme Court justice today, it is helpful to never have rendered opinions on important subjects, and to have as small a paper trail as possible. This is more difficult for justices, as they have to make judgements and write things down.

    In 2020, if the typical presidential candidate is 60 years old, they will have been born in 1960. That's just about the latest an intelligent person could have been born in the U.S. and not grow up leaving hundreds of public posts and comments that can be taken out of context (or in context) to disqualify them from public office.
    Friday, February 10th, 2012
    2:36 pm
    Activism is not altruism
    Two teachers at Miramonte Elementary School in southeast Los Angeles were recently accused of sexually abusing students. In response, the school board fired... the entire teaching staff.

    Parents are protesting the response. Are they protesting because the school board made an insane overreaction? No; I heard interviews with parents on NPR today. Some are upset that they don't know the new teachers and don't want that much change. Some are upset that the school board disrespected them by not consulting them first. No one mentioned that firing an entire school's staff because of allegations against two of them is, maybe, extreme; or that employees should not be held responsible for allegations made against other employees in the same building. Everyone cares only about the impact on them personally.

    I find this aspect the most disturbing aspect of the entire affair. Hundreds of people are engaged in this issue as activists, and not one of them cares about doing what is right. Everyone is just looking out for themselves.

    I suppose this is how democracy works.
    Monday, February 6th, 2012
    3:37 pm
    Mutualistic scams
    If you're a college student, you may have gotten an email this week that reads like this:

    "You are invited to apply for a membership from the Wellgates Scholar Program based on nomination we have received. This is a global initiative to encourage and recognize academic excellence, leadership, and entrepreneurship among university students."

    Is it a scam? It doesn't appear to be a non-profit. The clear attitude of people on Yahoo Answers is that, if it isn't a non-profit, it's a scam.

    The organization may make money by selling your contact info to advertisers. But they do give merit-based scholarships.

    Why do we insist that people who help us out at no cost are "scamming" us if they aren't losing money doing it? Is broadcast television scamming you when they make you watch commercials? Folk economic theory doesn't seem to have a concept of "mutually-beneficial exchange".
    Sunday, February 5th, 2012
    2:59 pm
    Art and culture
    Follow-up to my last post -

    Writers have rules that they believe are timeless rules of art, that are actually rules of their culture. The dominant paradigm in any culture says that art consists of those things that reinforce that culture's worldview.

    The idea that stories must be based on a protagonist fixing an internal flaw may be based on a mistranslation of "hamartia" as "flaw" rather than as "mistake". The plays that Aristotle was talking about when he made that rule (e.g., Oedipus Rex) do not feature protagonists with fatal flaws; they feature protagonists who made mistakes. Oedipus Rex' mistakes were based on things he did not and could not know; so the entire story is not about a character learning and growing, but about the inevitability of fate. (Many Greek stories, such as the Iliad, do not fit Aristotle's model very well, suggesting that Aristotle was writing about how he wished tragedy to be rather than about how it was.)

    The entire point of Beowulf (a feudalistic story) is that Beowulf does not change. Change, in those days, was bad. People were not supposed to change. They were supposed to meekly accept their role in life. It's no coincidence that Western literature as we know it today did not begin until the 14th century, and started to mature in the late 16th century. The stories we tell today would have been deemed dangerously subversive under feudalism.

    (Exercise for the reader: Write an Aristotelian interpretation of the Who's musical Tommy.)
    Saturday, January 28th, 2012
    4:15 pm
    Why the new My Little Pony is 20% cooler
    An interesting exercise for an author is to watch several of the new My Little Pony: Friendship is Magic episodes, and then watch one of the older-generation My Little Pony episodes. The contrast will highlight the differences between good storytelling and animation (viz., MLP:FIM) versus bad (viz., original MLP). Here are some principles for writing for episodic television that FIM demonstrates. Some also apply to "paper publishing":

    Have no enemies

    Too many stories are driven by villains. There are many reasons this is bad:

    • A plot with a villain is reactive: The villain has a goal, makes a plan, and the hero reacts. You've already given most of the character-building material to an unlikable character; and you've begun by saying that bad people are smart and plan and do things, and good people are conservatives who try to restore everything to the way it was before.
    • The villain takes up half your screen time. That's okay in a movie, which is short enough to have one villain from start to finish. But in a TV series, it's wasted time spent developing temporary characters, unless the villain is a constantly-recurring character like Lex Luthor. And if the villain is, that's probably stupid. (There are exceptions. Like Joss Whedon (and MLP wrt Luna), you can turn your heroes into villains and your villains into heroes; then their development wasn't wasted time.)
    • If your characters are always fighting villains, we get to see only one aspect of their characters. What is the Scooby Doo Gang like at home? What do they do for fun? Where do they see themselves in ten years? Who knows? Not the show's authors.
    • Rooting for one side to win wastes cognitive energy and distracts from the real story. A football game is not art. "Who will win?" is the weakest possible tension to build a story on. The Lord of the Rings is not about whether Frodo will get the ring to Mount Doom. Of course he will. It's about what it will cost him. The finger is in case you didn't already get that.
    • It is inherently bad art, by which I mean art that, like a Harlequin romance, makes a person who is exposed to it more dysfunctional. We don't need more stories telling people to solve problems by categorizing the world into good people and bad people and defeating the bad people. They do that naturally.
    MLP has a few villain-based episodes. But the typical MLP:FIM plot involves a conflict between two characters who have conflicting personalities, or within one character who must choose between doing something she wants to do and straining a friendship. This structure forces the writers to write characters who have their own goals and obstacles, and to develop the characters further within the episode, without wasting any time on non-character-building plot.


    Have more than one central character

    Having the same character be the center of every episode encourages infantile writing and infantile viewing. (This applies only to genres with continuity and character arcs. It doesn't matter as much in comedy, mystery, or suspense; or in genres like action that are supposed to be wish-fulfillment.)

    • We have too many stories about the Chosen One, the one with enough Midichlorians in his blood, on whom all else depends, who will defeat the evil empire with his pluck and determination. I've seen too many smart people with Ender Wiggins Syndrome, who waste their lives because they read books that taught them to do everything themselves instead of building networks and learning to cooperate. Yes, the Chosen One is massively popular. So is high fructose corn syrup. We often like things that are bad for us. "Selling out" doesn't mean making money. It means making money by giving people the comfortable self-destructive crap they want, whether that's Big Macs, cigarettes, crack cocaine, Thomas Kinkade paintings, or yet another Chosen One.
    • Even if that character isn't the Chosen One, identification makes any series less of a thought experiment and more of a video game. And if the single central character isn't meant to be identified with, he or she naturally drifts towards becoming a daddy figure. The only reason to watch Doctor Who or read Sherlock Holmes is because there's one character who is fascinating. But that one character has to be the driving force behind most major plot developments; and this often leads him to be super-humanly competent and overshadow every other character.
    The single main character can be done without becoming daddy or Mary Sue (e.g., Monk.) But besides giving you more to work with, having a cast of multiple main characters gives you characters of equal weight that can be pitted against each other without requiring villains. (Also, you can sell six times as many plushies.)


    Don't have every main character in every episode

    Real people have lives, and their lives don't consist of nothing but endless get-togethers with the same group of people.


    Balance exploitation vs. exploration

    One of the central problems in control theory, machine learning, scientific research, and business administration, is how to trade-off exploitation versus exploration. Exploitation means using things that you already know. In storytelling, this means invoking things we already know about characters. Exploration means trying new things. This means putting our characters in new types of situations, or examining new aspects of their characters. Non-continuity animation, where each show is an independent entity, is almost 100% exploitation: Assign each character a schtick and a tagline. Show one character doing their schtick and saying their tagline. Play laughtrack. Repeat with another character. You might want to see what crass things Cartman says this week. But you won't write Cartman fan fiction, because you don't care what happens to Cartman, because nothing that happens to him will matter next week.

    Exploration is an investment in a character. It means spending screen time adding things to that character that you can use in the future, instead of using the things you already have to get a quick laugh. The Emmys are biased towards episodes that do nothing but exploit, because that gives the biggest emotional payoff within a single episode. But the awards ought to go to the scripts that came before, that built up the character for later scripts to use.


    Don't turn flaws into characters

    Established writers routinely tell beginning writers to make their characters interesting by giving them flaws. One reason they do this may be because many beginners write what are called "Mary Sue" characters: Overly-perfect characters that are the person the writer wishes to be. But I think the main reason they do this is that we live in a culture infected by Platonism. What I mean is that we live in a culture where people believe in the concept of perfection. "Perfect," unless applied to mathematical objects, is a nonsensical word. There is no such thing as the perfect food or the perfect town or the perfect woman. Yet most people believe, at least subconsciously, in Platonic ideals; and thus that any real-life individual is the Platonic perfect person, plus flaws. (See "Throw out Aristotle" below.) As a result, we have an abundance of flaw-based characters, whose defining characteristic is what is wrong with them.

    The characters in My Little Pony are distinguished by what they think is most important in life. What other TV show can you name for which you can easily say what each of the main characters thinks is most important in life? I can't think of any. Their flaws are not generally things they lack, but things they care too much about. And they are only flaws when you take one pony individually. When they work together, their "flaws" become the strengths of the group. This approach to defining characters is the main reason the characters are so likeable.


    Give your characters careers and hobbies

    Good exploration is character-driven, not initiated as a response to some villain's plot. It is something the characters come up with out of their own goals and career plans. So give them some! Applejack wants to expand the farm's business. Rainbow Dash wants to become a Wonderbolt. Rarity wants to become a famous fashion designer. Twilight Sparkle wants to become the Equestrian equivalent of a tenured academic. Pinkie Pie and Fluttershy don't have careers, but at least they have hobbies.

    One goal that sometimes drives characters in MLP:FIM - and not just the villains - is money. Other cartoons take place in socialist paradises, because writers think mentioning needing money will traumatize children or turn them into capitalists. Ponies have to work for a living.


    Everything advances character

    This was one of Walt Disney's rules. Everything that advances the story must advance character at the same time. If your script calls for an event, and you can't think of a way for every main character in that scene to act that says something about their character, throw the scene out and write something different. None of the generic "the ponies run from the bad guy" or "the ponies play with the bubbles" scenes that you find in previous-generation MLP. If a line of dialogue sounds like anypony could say it, throw it out - it's not conveying information about character, and that means it's not essential to your story. (By contrast, South Park has been running 15 seasons, and I still wouldn't be able to tell Kyle's lines from Stan's lines.)


    Characters are more than walk cycles

    I've read numerous articles wondering why grown men are drawn to My Little Pony.  Even though bronies have given the answer themselves, over and over again, nearly unanimously:  Because the characters are interesting.

    I think the resistance that everyone who isn't a brony has to this claim, is because animation is so impoverished in character development that we no longer can even imagine well-developed animated characters.  People think the cartoons they're familiar with also have interesting characters.  But they don't.

    The problem is that animation is generally made by animators.  And animators are visually-oriented.  Animators are very focused on how to visually express a character through their movements.  Read The Illusion of Life and you will have the animator's message pounded into you over and over:  Draw movements that express that character.

    What gets overlooked is two other crucial aspects of character:

    • Context-dependence
    • Semantic content
    By context-dependence, I mean that a character should act differently in different contexts and towards different people.  Most animated characters don't.  SpongeBob and Patrick, Fry and Bender, and Stan and Cartman act the same way towards everybody and in almost every situation.  Animators can get so focused on "expressing the character" that they forget that the character is more than just a collection of body movements, and they don't like to muddy the "clarity" of the character by varying it.  But a character should act differently in different contexts.  Like the MLP characters.  Rarity acts differently in the presence of customers, friends, family, among high society, or in the presence of influential people.  Twilight acts like a puppy dog whenever Celestia is around.  Applejack gets easily irritated by Rarity's fussiness or Rainbow Dash's self-centeredness.  Big MacIntosh speaks in complete sentences only when talking to family members.

    By semantic content I mean what is inside the character's head.  What do they want, what do they believe, what do they think about?  Hence the importance of careers and hobbies.  And also of opinions.  The MLP characters express their different opinions about each other, about life, about values, to an extent far beyond any other cartoon characters that I can think of.


    Throw out Aristotle

    The most frequently-cited rule of writing is one originally laid down by Aristotle: A story must be about a protagonist who is faced with a problem, cannot overcome that problem because of an inner flaw, and changes (although the protagonist may or may not overcome their problem). See Phil Dyer's rules of script structure for elaboration of this formula. It's the basis of stories from Star Wars to SpongeBob SquarePants. It works, but it isn't the only thing that works.

    (It also probably isn't what Aristotle meant, since it doesn't fit the famous Greek tragedies of his time. In Oedipus Rex, Oedipus' downfall is due not to a character flaw, but to a lack of information, and/or to fate. People have argued that hamartia should be translated as "mistake", not "flaw". But no translation of "hamartia" can make Aristotle's theory fit the Odyssey or most of the Iliad; so it could be that the entire Poetics was meant to correct, not to describe.)

    Some MLP stories can't be described at all by this structure. Some can, but it's like fitting Cinderalla's shoe on to her sisters' feet, Grimm's Brothers-style. The stories are not organized around protagonists overcoming internal flaws; they are centered around protagonists learning the value of friends and family. The plots could more usefully be said to have these structures:

    • The protagonist is presented with an opportunity to get closer to one of her goals. As she pursues this opportunity, she realizes the price she must pay to take advantage of it. She decides that friendship or integrity is more important than her goals, and decides to pass on the opportunity. She then gets the opportunity anyway as a karmic reward. ("The Ticket Master", "Sweet and Elite", "The Super Speedy Cider Squeezy 6000")
    • Two ponies don't get along, but eventually each decides that keeping their relationship is worth making compromises ("Sisterhooves Social", "Look Before You Sleep").
    • A secondary character becomes an obstacle to the primary character, usually while trying to help, by being self-centered. These episodes are often resolved by the secondary characters (who realize what they've done after disaster ensues) rather than by the protagonist, which is a big no-no according to dogma. ("Sonic Rainboom", "The Ticket Master", "Green Isn't Your Color", "Suited for Success", "The Best Night Ever".)
    Most episodes end with a "lesson learned", but they're not usually things you would think of as internal flaws. The writers deliberately subvert Aristotle in "The Super Speedy Cider Squeezy 6000", having Applejack instead say at the end, "I didn't learn anything at all! I was right all along!" (The Simpsons did it first, but more cynically, at the end of "Blood Feud", when Homer said there was no lesson to be learned and it was "just a bunch of stuff that happened".)

    Probably the Ponies' biggest break, not just with Aristotle, but with all of Western literature, is that episodes are often resolved by the actions of secondary characters rather than by actions of the main character. Western literature looks at one individual at a time. You may have a lot of characters, but each one is supposed to resolve their own issues. MLP:FIM has a different message: Sometimes you can't resolve your own issues, and you need your friends to intervene.


    Throw out Disney's Law of One Emotion

    Another of Walt Disney's dictums, which was also adhered to by Warner Bros. cartoons (possibly excepting Bugs Bunny at times) and is now nearly universal in animation, is that every character must at all times display exactly one emotion. Disney, like stage actors, had problems with people not being able to tell what the characters were feeling. This was partly because Walt Disney usually staged scenes with elaborate backgrounds and did not go in much for close-ups, and partly because the craft of animation was new to both animators and viewers.

    The insidious effect of this rule was that the writers learned to avoid writing scenes that called for complex emotions, and eventually learned to avoid writing characters that had complex emotions. The animation controlled the story, instead of the story controlling the animation. This is why Disney had great animation, but could never have produced The Last Unicorn or Grave of the Fireflies.

    The MLP:FIM animators said, "To hell with that rule. We are going to depict whatever emotions the story calls for, and assume that the viewers are smart enough to figure out from the context what the characters are feeling." And it works.


    Most of these boil down to the same thing: Stories describe and develop characters and their relationships. MLP:FIM is unusually rigorous about expunging anything that doesn't do that, and any obstacles to doing that. As a result, it has more interesting characters than most shows, despite some significant handicaps not enumerated here.

    You can get technical tips from the show as well (e.g., show the viewer what the episode is going to be about within the first 90 seconds; introduce a new scene or a new story element roughly every two minutes; do something amusing at least twice a minute). But I'm too lazy and inexperienced to cover that.

    Thursday, September 22nd, 2011
    3:27 pm
    1dollarscan: Part 1
    I've been scanning my books into PDFs for about 7 years now. It's tedious. This weekend, I packed up 47 books in a box and mailed them to 1dollarscan, for $156 fees plus $21 shipping.

    They won't "scan your books for a dollar", but they will scan them for only a little over 1 cent a page. That's cheap. I have a well-honed book-scanning system, but I couldn't chop, scan, proof, and correct them for under 3 cents a page even if I paid myself minimum wage. I thought about hiring somebody to do it, but don't know anybody who would work cheaply enough whom I would trust with the book chopper. I assume 1dollarscan is relying on the fact that 10,000 people will send them a copy of The DaVinci Code, and they will scan it only once. They're probably losing money on me, since the most popular book I sent them is "Fourier Analysis".

    If this works, it will be a great time-saver for me. I'm especially impressed by their willingness to scan color pages. The scanners I've used can only scan about 10,000 sheets before they develop microscopic scratches that ruin color pages; but they can go over 100,000 sheets between sensors and still do black and white fine. Sensors are the biggest problem in scanning books. Scanners are NOT designed properly to protect the sensor glass from scratches. The KV-S2046 is a vast improvement over the KV-S2045 in this respect, but it still SUCKS in an absolute sense. It is very expensive to replace the glass on the sensors - either buy a new sensor for $400, or replace the glass by hand - difficult at best as they are glued on; and usually not possible in America, as scanners are Japanese and have a thickness measured in millimeters, while American glaziers have glass sheets only in 1/16" or 1/32". The wrong thickness will ruin the focus.

    The second-biggest problem is the chopper blade. I have found no one who can sharpen one properly; and doing it by hand is dangerous, takes hours of hard physical labor, and gives poor results. It really requires an old-fashioned grindstone. They need a convex edge, not a straight edge.

    Lots of people complain that they could never do that to their books. These people think they love books. So I admit - I don't love books. I love what's in books. Loving books is like loving someone for their looks.
    Sunday, September 11th, 2011
    10:43 am
    Ben Franklin
    Yesterday I found out Ben Franklin founded the University of Pennsylvania, one of the world's best universities. Now, if you had founded Yale or Dartmouth or U. Chicago, that's probably what you'd be remembered for. Your tombstone would say, "Founder of Dartmouth".

    But I recently read a biography of Ben Franklin that did not have space to mention that he founded the University of Pennsylvania.
    Sunday, May 15th, 2011
    11:19 pm
    Biotech priorities
    The Johns Hopkins MS in biotechnology program, which I am enrolled in, has:


    • 3 courses on ethics
    • 5 courses on industrial biotech
    • 6 courses on pharmaceutical development
    • 8 laboratory courses
    • 11 courses on bioinformatics
    • 13 courses on immunology, pathogens, infection, and epidemiology
    • 14 courses on business development and management
    • 14 courses on genetics, genes, genomics, and DNA
    • 17 courses on complying with government regulations
    Wednesday, March 9th, 2011
    3:54 pm
    Pre-order your eBooks now!
    I got spam from Borders today encouraging me to pre-order a forthcoming eBook.

    But I'll take my chances. Somehow, I don't think they'll run out.
    Monday, September 27th, 2010
    10:49 pm
    Need more letters
    I'm trying to sign up for an account on pogo.com.

    shagbark... Username taken.
    fragbark... Username taken.
    DarthGarth... Username taken.
    DarthPikachu... Username taken.
    Steerpike... Username taken.
    TitusGroan... Username taken.
    Blargle... Username taken.
    Wargle... Username taken.
    jubjub... Username taken.
    Aargh... Username taken.
    Thursday, July 29th, 2010
    10:25 pm
    LiveJournal substitute?
    LiveJournal is no longer acceptable, since they now require people to watch (!) a video advertisement before they can read a blog entry.

    Any suggestions where to move to?
    Wednesday, June 9th, 2010
    5:32 pm
    Gross National Happiness
    I heard another story today, on NPR, about why we should measure Gross Domestic Happiness (GDH) instead of Gross Domestic Product (GDP). GDP, the argument goes, is a product of a crass materialistic age, and doesn't measure quality-of-life issues like time spent with family.

    Sounds nice. But GDP is used by the government, to help set policy in many ways. It's relevant to what I think should be part of the business of government: Keeping the country's economy running.

    The clamor for the government to track happiness instead of mere economic indicators implies that happiness, too, is now the government's business. And I find that creepy. Keep the government focused on economics. We should not give it the authority to look after our happiness.
    Monday, May 31st, 2010
    3:10 pm
    The Deepwater Horizon well has leaked 5 million gallons of oil into the Gulf of Mexico, we're told. How much oil is that?

    Enough to last the US for 20 minutes, according to the DOE.
    Friday, October 30th, 2009
    9:10 pm
    Today the White House announced the results of a study showing that $159 billion of the stimulus money they spent so far has created 640,000 jobs. They estimate that another $189 billion in tax relief has created at least 350,000 jobs.

    That's $350,000 per job. About half of the 640,000 jobs from the $159 billion were as teachers, who earn a little over $40,000/yr. That means that about 11.4% of the stimulus money found its way into the hands of people who otherwise would have been unemployed, before taxes.

    The argument against giving money directly to unemployed people is that then they wouldn't have an incentive to find jobs, resulting in inefficiency, as a large percentage of the unemployed-but-employable would remain idle.

    Assume that the cost to society of a worker choosing not to work is about $40,000/yr in lost productivity. In order for distributing $348 billion directly to the unemployed to have less than 11.4% efficiency, it would need to motivate 7.7 million people to stay unemployed who otherwise would have found jobs. As there are currently only 15.1 million unemployed people in the US according to the US Dept. of Labor, I find this unlikely.

    (This isn't a fair comparison, because the money "lost" by giving it to business owners is just redistributed, while the productivity lost by people not working is really and truly lost. But let's follow this idea through anyway.)

    In fact, you could guarantee that this wouldn't happen, by giving the money to fewer than 7.7 million people. Give 3.48 million unemployed people $100,000 each. Once you've given them enough to motivate them to stop working, you don't lose any more productivity by giving them more. Giving people much more money than they are accustomed to, increases the percentage of it that they will spend. And if they don't spend it, they put it into a bank, where someone else will spend it. Either way, the stimulus money still flows to the business owners; it just goes through the unemployed first this way, causing a little inflation in exchange for giving a much larger share of the benefits of the stimulus to its alleged target population. You'll have provided a higher standard of living, for 3.5 times as many people, on top of your business stimulus.

    Of course, instead of better schools, police forces, and road systems, you'd end up with an extra $348 billion worth of Playstations, movies, concerts, and internet porn.
    9:09 pm
    Monday, October 26th, 2009
    1:06 pm
    A lucky bug, and evolution
    Last Friday, I finally packaged up the quarterly release of JCVI's automatic prokaryote functional annotation pipeline, which looks at genes found in newly-sequenced genomes and guesses what they do; and distributed it to the other 3 sequencing centers for the Human Microbiome Project. As always happens when I release a new version, several minutes afterwards, I discovered a major bug that had been hiding in the code for years.

    The program takes each new gene and runs BLAST against a database of known genes, and produces a list of identifiers of genes resembling the new genes. It then takes these identifiers, and calls a program to look up all of the synonyms for these identifiers used in all the different gene databases. This lookup step takes 90% of the program's runtime.

    I found that the database lookup usually failed, because most identifiers didn't match the regular expression used in the lookup program to retrieve identifiers. Nobody had noticed this, because nobody had checked the database log files.

    I fixed the program so that the database lookup would always work correctly, and re-ran the program. It produced exactly the same output as before, but took five times as long to run.

    So instead of going dancing, as I'd planned, I spent Friday evening figuring out why this happened. It turned out that the class of identifiers that failed to match the regular expression were a subset of the set of identifiers for which the lookup didn't have to be done, because the previously-cached results would give the same results. Once I realized this, I was able to speed the program up more, by excluding more such identifiers, and avoiding the overhead of about a million subroutine calls that would eventually fail when the regular expression failed to match.

    A bug in a program is like a mutation. Bugs in a computer program are almost always bad. But this was a beneficial bug, which had no effect other than to make the program run much faster than it had been designed to. I was delighted to see this proof of the central non-intuitive idea of evolution: A random change can sometimes be beneficial.
    Thursday, October 22nd, 2009
    11:50 am
    From Zooillogix:

    Sunday, October 18th, 2009
    2:44 pm
    Wednesday, October 14th, 2009
    4:55 pm
    Another child murdered by the FDA
    Today I heard a heartbreaking story, in a recorded presentation by Roscoe Brady of the NIH, made at SENS 2 in 2005. He had a patient who was a little girl with Gaucher disease; and he had a treatment for the disease, which seemed to be safe in animal experiments. But the FDA wouldn't let him treat her until the primate experiments were done.

    By the time the primate experiments were done, her brain had been irreparably damaged by Gaucher disease - as everyone involved had known it would be. Somehow the FDA decided it would be better to ensure her death, than to take the risk that she might have a bad reaction to the treatment.

    The treatment works, by the way.
[ << Previous 20 ]
About LiveJournal.com