Friday, November 29, 2013

Physics and Testosterone Part 3: Jumping off Cliffs

Much of the education of a physicist or an engineer revolves around problem sets. To do a problem set, the student needs to understand some concepts to "set up the problem". There will be some calculating machinery that the student must have mastered. Problems can be difficult in a variety of ways. Often, the path from concept to solution won't be obvious. It might not be obvious which principles might be applied to a particular problem. The calculations required might be very messy. Insight or intuition may be required- you might have to guess the answer to figure out how to get to it. But often, solving a problem requires a leap of faith. You might have to work for a half hour before you know whether you've chosen the right approach to the problem. So it helps to have some self-confidence, to be ok with not knowing whether you're on the right path or totally lost. Other times, you just need the right tool, and maybe you have to invent the tool.

Real math or science doesn't come with answers in the back of the book. A researcher might work for years without knowing whether their efforts are leading them down a blind alley. The exquisite feeling you get when you've solved a really hard problem is why people become physicists, mathematicians, and engineers. It's the feeling of having eyes where once you couldn't see.

Having her self-confidence assaulted by every problem set in grad school was a challenge for "K", the applied physics Ph.D. I wrote about in Part 2. But there was that one problem set in high school that stumped everyone else in the class, but which she solved. Once you've tasted that success, you don't forget it.

Learning to ski was a revelation for K. You'd take the ski lift up to the top of a mountain, and somehow you'd end up at the bottom.

There are a lot of ways to get to the bottom of any slope. Some people like to do traverses. Some people go straight down. My method was to swallow the sheer terror, point my skis downhill, and power through some turns and some slides. I've never been a great skier so I'd get half way down and land on my butt. Gradually, I figured out how to avoid falling.

This is apparently a typical male's approach to skiing. A touch of reckless self confidence lends itself to this approach. Just watch some teenage boys on a ski slope if you doubt it.

K realized that she didn't have to ski like the guys. The part of skiing she enjoyed was carving turns. To carve a good turn, you have to put your weight downhill, which at first feels insecure, but in practice gives you more control. And having good technique gives you real confidence.

Realizing that she could approach problem sets her way really helped K get through those difficult problem sets. It was OK that she felt like she had no idea what to do while many of her male colleagues just pretended to know how to do them.  There was nothing wrong with focusing on skills and carving away the difficulty. And not break anything.

Tuesday, November 26, 2013

Physics and Testosterone Part 2: Study Hall

In the fall semester of sophomore and junior years, my work-study job at Princeton was to tutor freshman engineers in a study hall sponsored by the engineering school. The study hall had been created a few years before to help freshmen survive the shock of learning calculus, physics and chemistry, all at the same time.

We had a variety of students seeking help. I began to notice a distinct pattern in the sort of help that was needed by women and by men.

The typical interaction with a woman at the study hall went like this:

Woman: "I have no idea how to do this problem!'
Me: "Tell me about the problem"
Woman: [[ detailed explanation of problem ]]
Me: "How do you think you should attack the problem?"
Woman: [[ detailed plan for solving the problem ]]
Me: "Sounds good"
Woman: "Oh thanks so much, you've been so helpful!"

The typical interaction with a man at the study hall went like this:

Man: "I can't get this problem to work"
Me: "Tell me about the problem"
Man: [[ bizarre, complicated, and wrong explanation of the problem ]]
Me: [[ detailed explanation of problem that woman student just told me ]]
Man: "Oh"
Me: [[ detailed plan for solving the problem that woman student just told me ]]
Man: "Really?"
Me: "Would it kill you to try?"
Man: "but [[botched calculation]]"
Me: "Might want to check your signs"
Man: "Hey, I knew it would work!"

Of course, not every student was typical. I remember one freshman woman in particular. She would come in with a male friend. They were taking the sophomore level physics and math courses. This posed a problem for me, as I was also taking the sophomore level physics course, albeit the physics major track, rather than the engineering track. I was barely a week ahead of them. I used my "tell me about the problem" strategy, which at first seemed to satisfy them. But after a few weeks, their questions got more difficult and I was having more difficulty. So I suggested to them that the study hall wasn't really meant to help with sophomore level courses.

A few years later, I found myself a classmate of this student in grad school at Stanford; she went on to get a Ph.D. in Applied Physics. She never quite forgave me for "kicking them out of study hall".  Her story tomorrow.

Monday, November 25, 2013

Physics and Testosterone Part 1: Captain Kirk

The New York Times Magazine had a really interesting article by Eileen Pollack about women in physics last month. And the week after, the Nobel Prize was announced, and the recipients were, no surprise, men. It got me thinking about how gender and physics. I happen to know a lot of physicists, some Nobelists, and a fair number of women who are physicists. One or two of the women might win a Nobel one of these years, but the odds aren't so good.

Thinking back on my training in physics, I realized that I have some stories to tell that might shed some light on the effect of gender on the development of scientists, engineers, and technologists, and how to do better.

My sophomore year at Princeton, I took the physics-major track physics courses. For Electricity and Magnetism, we had a professor fresh out of Caltech, who we all called "Captain Kirk". As rumor had it Captain Kirk borrowed his curriculum from a graduate course at Caltech which had a track record of producing Nobel Prize winners. The textbook was completely inscrutable and the problem sets were pretty much impossible.

Looking back on it, I'm pretty sure that if the curriculum ever produced Nobel Prize winners, it wasn't because it did a good job of teaching the material. More likely it was effective because it did a terrible job of teaching the material. Which had 2 consequences:

  1. All us smart-ass physics students quickly realized that we weren't nearly as smart as we thought we were. We were unaccustomed to the fear of failure, and it motivated us.
  2. We formed groups to work on problem sets together and taught each other the material.
  3. Except for the freshman who miraculously did all the problem sets on his own.
What struck me from Pollack's article was a quote from Meg Urry, a professor of Physics and Astronomy at Yale:
“Women need more positive reinforcement, and men need more negative reinforcement. Men wildly overestimate their learning abilities, their earning abilities. Women say, ‘Oh, I’m not good, I won’t earn much, whatever you want to give me is O.K.’ ”
Maybe Captain Kirk's course was really designed to discourage us. Filled with testosterone or conditioned by society, the guys among us were stupidly overestimating our capabilities and we needed to be brought down to earth. We had all been solo stars in high school, and we needed to to be forced to work with our peers. We needed to be broken down so that we'd be more open to new ideas.

Probably the one woman in our study-group didn't need to learn those lessons. More positive reinforcement could have helped her more. (She ended up getting the physics degree just fine and went to med school.)

Despite Captain Kirk's hopes, no one from the class has won a Nobel Prize, yet.

Monday, November 18, 2013

Google Books and Black-Box Copyright Jurisprudence

Last week, eight years after the first lawsuit was filed to stop the Google Books Project, Judge Denny Chin finally ruled on the core merits of the case. The decision is being widely hailed on one side as a "tremendous victory for fair use" and on the other side as a "fundamental challenge to copyright". But these are short-term perspectives. I think that the long term impact of the decision may turn on the acceptance of Chin's approach to technology's transformation of copyright, which I would characterize as Black-Box Jurisprudence.

In my view, the core holdings about fair use were never in much doubt. The argument saying that indexing or lexical analysis or data-mining of books always requires the permission of a rights holder was never very defensible, or even seriously argued. A holding that display of snippets was not fair use would have made scholarly writing in the digital age impossible; a decision the other way on snippets would have been swimming up a judicial stream. But fair use is always a weighing of factors, and the untold story in the Google Books case is about the factors that didn't get weighed.

The reason that Google got sued in the first place was less about "what Google did" than about "how Google did it".  Google made huge numbers of copies of books without permission of the rights holders. Judge Chin's ruling said, effectively, that all those copies were incidental to the fair use.
[I]f there is no liability for copyright infringement on the libraries' part, there can be no liability on Google's part.
In the end, it didn't matter how Google did what it did. In Judge Chin's analysis, copyright is concerned only with the ends, not the means. Copyright seems not to be concerned with what happens inside the black box.

Chin is not alone in this approach. His opinion follow's Judge Baer's ruling in the Hathitrust case, which featured a ringing endorsement of the Library's fair use
I cannot imagine a definition of fair use that would not encompass the transformative uses made by Defendants' [Mass Digitization Project] and would require that I terminate this invaluable contribution to the progress of science and cultivation of the arts that at the same time effectuates the ideals espoused by the [Americans with Disabilities Act].
But for, me, the surprise in Baer's opinion was his transformation of the Arriba Soft case into a broad license for infringement. In that case, display of thumbnail images by a search engine was held to be fair use, and the copying of the images in the course of producing thumbnails was held to be necessary for the protected use. Judge Baer wrote that the fact that the images were on websites available for anyone anywhere to download was not relevant to the analysis, which he then applied to Google's scanning and OCR of physical books.
Although Plaintiffs assert that the decisions in Perfect 10 and Arriba Soft are distinguishable because in those cases the works were already available on the internet, Aug. 6, 2012 Tr. 19:2–4, I fail to see why that is a difference that makes a difference. As with Plaintiffs’ attempt to bar the availability of fair use as a defense at all, this argument relies heavily on the incorrect assumption that the scale of Defendants’ copying automatically renders it unlawful.
Baer thus reduces and equates Google's million dollar scanning operation with Arriba Soft's one line of code because they're in a fair use black box.

The Black Box approach to copyright can cut both ways. In Chin's dissenting opinion in the Aereo case, he wrote that it didn't matter that Aereo had engineered a way to use completely legal technical means to stream television signals over the internet.
In my view, by transmitting (or retransmitting) copyrighted programming to
the public without authorization, Aereo is engaging in copyright infringement in clear violation of the Copyright Act. [...] The system employs thousands of individual dime-sized antennas, but there is no technologically sound reason to use a multitude of tiny individual antennas rather than one central antenna; indeed, the system is a Rube Goldberg-like contrivance, over-engineered in an attempt to avoid the reach of the Copyright Act and to take advantage of a perceived loophole in the law.
In the Aereo case, Chin argued that since the end result of Aereo's engineering was a system with copyright infringing intent, the under-the-hood details of Aereo's system were not compelling. (Read James Grimmelmann for more on this case and copyright arbitrage in general.)

So when presented with cases where copyright law and technology collide, Chin has more or less adopted a consistent approach that isn't inherently pro-copyright or pro-fair-use.

If Chin's ruling had focused on the infringing means (i.e. massive copying) rather than on the fair-use ends in the Google Books case, Google could have gone back to the drawing board to devise a non-infringing means to accomplish the same ends. It would have been more expensive (à la Aereo), but the plain fact is that ten engineers can run technical circles around a thousand lawyers. In the end, Google would have lost the battle but would be far ahead in the war.

As the case now stands, while Google has a free hand to go back and improve and expand its scanning operations, it is still constrained in what it can deliver. For example, since Chin's decision cites the lack of advertising on snippet result pages in his fair use analysis, Google can't put advertising there without risking another $100 million lawsuit. Another innovator in the space can't go and do things differently without worrying about another judge's fair-use analysis.

The advantages of a black box legal approach is its practicality. Judges don't have to understand the intricacies of technology in order to decide legal questions. Technical processes are opaque for business reasons, too. But perhaps more importantly, a black-box approach to copyright law means that engineers can't use clever hacks to get around copyright.

The danger of the black box is that it pretends that technology doesn't matter, that code isn't law. Copyright law is rooted in technology, that of the printing press, and turning it into an abstraction that can also govern digital media wile ignoring what goes on behind the curtain is a dubious project. A complex enterprise like Google Books is a long journey from inception to delivery. Imagine if highway safety was addressed by regulating total travel times. Does it make sense to regulate a new technology like airplane travel in the same way?

Perhaps there ought to be a fifth factor in fair use analyses of systems more complex than a printing press. In addition to the usual four factors, Judges could also be weighing whether the steps involved in accomplishing a fair use would stand under their own 4 factor analysis. In the Google Books case, the analysis could have incorporated a weighing of the scanning operation by itself. Similarly, Aereo's meticulous adherence to legal means could weigh in favor of a fair-use determination.

My worry is that in other situations, perhaps with technologies we haven't imagined yet, the black box legal approach will end up with very wrong technical results. And then we'll be stuck, waiting for Congress to fix things. Look at what's happening as digital surveillance collides with crypto-security. There, the courts have uniformly refused to look inside the black box of the NSA, and the results may end up being disastrous.

(Gary Price has a thorough opinion round-up at Infodocket.)

Enhanced by Zemanta

Friday, November 15, 2013

Blogifying a Book

On Flatland the Blog, I'm turning a book into a blog. It's Flatland.

The ostensible reason is to promote our test campaign of Unglue.it's Buy-to-Unglue campaigns. It'll be a while before it's ready to launch for real, but I've found that there's no substitute for having real users try things out. One ungluer managed to find 2 different bugs within three minutes, partly by virtue of the 'ě' [LATIN SMALL LETTER E WITH CARON] in his username. Another user was the first to ever try changing their e-mail address to something invalid while having a username containing '@'. For some reason our unit tests didn't foresee these possibilities.

But really, I've been fascinated by the possibilities of the read-write book. We now have lots of ways to save and share annotations, but in most cases, this is done as a networked overlay on top of user-immutable texts. The annotation layers in Readmill, or in Kindle, live in their respective network.

Another effort to spread annotations over the web is Hypothes.is, which is trying to use standards to break the annotation layer out of closed networks.

I don't think there's anything wrong with networked annotation layers, but there's another technical direction that's been largely unexplored. What if a user's annotations are stored in the digital file that packages the ebook? This has the effect of restoring the individuality to copies of a book. The annotations could then be shared by sharing the file, the same way that pencilled annotations in a printed book might be shared privately. An anti-facebook, if you will, for an era when everything in the network layer is sure to be scanned by the NSA. And it also changes the dynamic of sharing a file in a library.

So what I want to do is collect comments and put them into the Flatland ebook that we're producing. I spent a fair amount of time producing a clean, attractive EPUB file from public domain scans by Google and Project Gutenberg, but I'd like to do more. The idea of turning the book into a blog occurred to me because Flatland's chapters were the right length, and, well they're curious, and need comment in the modern context.

So read along with me on Flatland the Blog, leave comments and suggestions, and at the end maybe something interesting will come out of the experiment!
Enhanced by Zemanta