omenti

the other dick fosbury

A few years ago I was having fun of  thinking about the fluid dynamics of a ball moving through the air.  The main focus was  beach volleyball, but also several other sports that use roughly spherical balls. Of these, the two with the most curious motions are table tennis and beach volleyball.   It came as a surprise as soccer (football in most of the world) has dramatic movements and baseball can be tricky, but they're not in the league of these two other sports.  People found it fun so I decided expand it into a college level class - a one semester physics of sport course for non-majors.

 

It was a very interactive affair,.  After a few weeks a few coaches and their assistants turned up,  offering things to think about.  After going through the Fosbury flop, one of the track and field coaches mentioned there had been a similar innovation in long jump, but it had been outlawed as too dangerous.  He pointed out you lose a lot of energy at takeoff by leaning back at takeoff.  The innovation was to halt the loss doing a forward somersault.  He couldn't remember the name of the athlete who did it, but thought it was in the mid 70s with World Athletics quickly banning it.

 

That evening I took pencil to paper and came to the conclusion it could increase distance traveled by as much as ten percent.  Ten percent is a change in sport where people will try crazy things for fractions of a percent.  Eventually around I found the athlete - a student from New Zealand who would have had the world record on his first attempt if he had slightly better form.  There were a few still images, but no video.  

 

With the recent passing and celebration of the life of Dick Fosbury I decided to look again.  Two years ago  an interview with a video of the jump was posted.  Long jump records would probably be over ten meters if it hadn't been banned.  The physics is simple - the leap was realizing you didn't have to lean back.

something deeply profound

David Epstein mentioned Hillel Einhorn's epiphany about happiness - something triggered by a fortune cookie.

beyond the equal sign

 

 

 

About ten years ago I devoted a good deal of time to learning a bit of category theory - an area of abstract mathematics that's become important to physics.  It deals with relationships rather than objects, abstractions and analogies, context, sameness, equivalence rather than equality, how you see patterns, and much more.  It can impact thinking in many areas beyond mathematics. It provides a rigorous framework that gives choice in how problems are approached and take you beyond  putting things in boxes that don't always make sense.

 

It's very technical, but Eugenia Cheng has written an excellent non-major freshman level book with the goal of getting art students to think about the world differently.  I suspect it will work for many other people - particularly those who have to deal with complexity and seeing curious patterns that are otherwise missed.  The Joy of Abstraction is one of those rare books that can change how you see the world.  I'd recommend getting a physical version and doing the exercises. (I don't buy books from Amazon and recommend you support local bookstores)

 

Steve Strogatz chatted with her on this episode of his Joy of Wh(y) podcast.  Listen even if the book doesn't strike you as interesting!

 

 

a celebration of homophones

For those who use the American mm/dd/yyyy,  it suggests 3.14   With an error of about 0.05%, it’s a good enough approximation for many uses.  And those who use the more common dd/mm/yyyy notion can wait for the 22nd of July for 22/7. At  0.04% from the mark, this one's a slightly better approximation. 

 

But just how much Pi do you really need? Eight digits is overkill for most engineering tasks. NASA uses 16 digits .. 15 to the right of the decimal .. to navigate around the solar system going beyond Pluto. The most I can imagine would be to construct a high quality circle the radius of the known universe, about 46 billion light years, to the diameter of a hydrogen atom,. You’d need 38 significant digits for the task.

 

xkcd has a piece on approximations that’s painfully true. For some back of the envelope calculations in astrophysics and cosmology you can get by with crude approximations.

 

 

It’s also Einstein’s birthday. Pie, being a homophone of Pi, seems to the a good bet for the day. I have it on good authority he loved cherry pie and ice cream, so there you go.

 

He’s one of a dozen or so people who have quotes improperly ascribed to them.  He was a joker and much of what he did say is often taken out of context.  But here's a real one I love. Shortly before his death he was asked to offer advice to young people.  His reply was published in Life Magazine.

"The important thing is not to stop questioning. Curiosity has its own reason for existence. One cannot help but be in awe when he contemplates the mysteries of eternity, of life, of the marvelous structure of reality. It is enough if one tries merely to comprehend a little of this mystery each day."   A. Einstein 1955

 

physics isn't about looking for some grand truth

 

It turns out we make it up as we go along and that's a foundational piece of why it's been so successful.

 

Apparently there's been some discussion about the connection of physics and some quest for an ultimate truth on a social media site I gave up.  A few people reached out and asked my opinion.  Starting down the path as a teenager, I thought science was after some deep and ultimate truth.  Some grand and beautiful underpinning of the Universe.  I was wrong.  It took time, but finally as a grad student I had come to know better.

 

What physics does is build mathematical models that describe observations. (why math works so well is another discussion - but it just does)  The hypothesis models that are robust enough to survive different experimental techniques and are predictive become accepted theory.   It's very much an iterative process and the models get progressively better with more predictive models coming into use as appropriate.  The Standard Model is the best we have for particle physics.  It has some warts, but it's far and away the most accurate theory devised.  General Relativity is the benchmark for gravity.  Both are under constant study and that study has led to radically different views of the Universe and how it has and will evolve. 

 

But to the question from the birdsite.  What is real?  Is a proton real and what is it made of?  What about the spin of a particle?  Is dark matter or the dark force real?  For the electron and the quarks and gluons that make up protons and neutrons, the most accurate and way to think about them is as excitations in fields that spread throughout the universe.  You, the computer screen you're looking at, your dog and your lunch are vast assemblages of field excitations.  It's far from intuition and these models are rarely taught until the junior year and that's usually at a high level as the math is a bit intense.  It may seem weird, but it's much more predictive than anything else. And gravity being spacetime that changes it's curvature around mass and mass that responds to the curvature of spacetime isn't exactly Newton's falling apples.  

 

Current accepted theory addresses serious flaws in earlier work. The earlier work was good enough given the sophistication of measurements, but tools improve with time.   It turns out most  (but not all!) of applied physics and engineering doesn't have to worry about these deeper descriptions as these are fields that usually apply to things roughly our size moving a low speeds.  There's an interesting notion called complementarity that tells you it's okay to use less precise, but easier to understand, models when you don't need extreme accuracy.  We live in a world where models with a Newtonian underpinning are usually valid.  

 

In a way it's how we perceive things. We only sense a tiny amount of the electromagnetic spectrum, our hearing misses quite a bit, our sense of time misses large and small intervals and our size is roughly between that of the Sun and a hydrogen atom.  We don't perceive much of the reality around us - we have to use our imaginations to figure out how to ask the right questions and see more deeply.

 

So we build models.  We don't work out the equations of motion, but we walk, throw balls, drive cars and so one. We have a fair amount of information for these simple actions.   Physical models quickly get more difficult.   How do you predict an earthquake and its severity?  When and where tornadoes will do damage?  We build approximations, but that's all they are.  Better models will come using the scientific process. Even if it's a messy undertaking, it's proven to be the best approach we have.  And there's a foundational principle you have to deal with  - one that Feynman famously articulated:

The first principle is that you must not fool yourself and you are the easiest person to fool.

More complex and social sciences come into play. Predicting markets and economic models don't have the twelve digit accuracy seen in the Standard Model .. they generally don't make it past the decimal point.  They're also a stab at reality.    Social models are often poorly constructed with bad data and analysis - a major "AI" issue these days.  And how we understand each other as people.   We all build models.  Hopefully they're good enough.

 

For a long time I believed you could change minds about the causes and threat of global warming through education.Silly me -  it turns out to be a fool's errand. We all have different views of reality.  We can witness an event and see differing realities.  These lenses can be very different and culturally driven.  Quite a bit of work has been done trying to understand how this works.  Recently a podcast chat with  psychologist Jer Clifton was recommended by Corinne.  I give it two thumbs up and recommend it to anyone who has to deal with other people in almost any capacity.  It may even make you think twice about how you interact with others.

 

revisiting eliza the next generation

a minipost

With all the talk about large language models and applications like ChatGPT,  I recalled a post from about two and a half years ago:  eliza the next generation.  I'd say much the same now - some of the predictions have come about.  Of course it's only been a short while and I played with prototypes, so there's that:-)

 

It was high pandemic and I was spending quite a bit of time talking with friends who were working on interesting things.  Among other things I started to become aware of progress in large language models.  While many who work in AI (a poor label, but you get the idea without me having to talk about a variety of fields and subfields as well as what's unrelated), believed them a diversion, there was progress in LLMs and some interesting applications were emerging.  We chatted about where it could go and what techno-social issues might arise. (disclaimer: I'm not a techno-solutionist)  About the time I posted, Emily Bender at the University of Washington wrote a terrific paper on issues with natural language understanding and if we can believe what comes out of LLMs.  Recently an article on her work appeared in The Intelligencer.  It's a necessary read if only to understand her octopus analogy. 

 

This morning Gregg Vesonder and I exchanged some text messages. He's a serious AI guy with decades of experience and is someone who thinks deeply about social technical issues.  One of his comments:

We’ve seen this scene before, blinded by the hype. My concern this time is that explicit or inadvertent bias in the data can mask some nasty behavior that can subtly or not so subtly mask really tragic manipulations.

Aka speech acts. 

I hadn't thought about speech acts much before.  The definition and examples given in the link are excellent tools for thinking about these things. 

 

I'm not against LLMs and there are some useful applications beyond writing horoscopes in white bro-speak, but the dimensionality of the space for abuse strikes me as large.  As Suw Charman-Anderson notes, we're already seeing it in publishing. 

 

'AI' is a very large and diffuse collection.  Some areas are useful and potentially revolutionary.  Others are potentially dangerous.  We really need to think about these things rather than just trying them out and seeing what breaks.  The things that break may be far too valuable.