omenti

diversity

The past nine months has been hard for me. Three good friends are gone.  The first was Dewayne Hendricks - probably the smartest radio person I've met. Our weekly calls ranged from his love of sci-fi everything (then Sukie would get on the call as that's her area) to working out some interesting math relevant to new ideas for radio propagation to just listening to his wonderful stories.  He grew up and had to make his way in a racist country that hasn't learned much and is currently racing towards white supremacism.   His passing left a big hole in the Universe. 

 

Next was Rohini Godbole.  Standing four foot eight and maybe weighing seventy five pounds, she became a giant in the advancement of women in STEM careers in India. Both of us lived in Stony Brook (we received our Ph.Ds there), but did most of our work twenty miles to the East at Brookhaven National Laboratories.  She didn't have a car so we carpooled in my ancient VW talking about math and on the way out and physics on the return trip.  We'd leave a bit before five in the morning as computer time until eight and would pick up a bagel on the way.  She had a thing for cinnamon raisin.   It was on one of those trips where both of us came to an understanding of what Peter Higgs had said in his paper. She was talking about superconductivity when we had one of those oh-my-god-pull-the-car-over moments.  I'm so grateful for having known and learned from and with her.  India is a richer and more diverse country due to her efforts in educating and inspiring young women.

 

And then there's my sister Juliette Powell who passed last month at 54.  I've written about her on this blog. She's an excellent example of the need for smart women in tech companies. I learned a lot from her thinking  and she changed my mind on more than a few things.

 

All three possessed a special clarity of thought. All three were minorities who had to navigate multiple cultures, racism, classism,  and in two cases, sexism.  I'm convinced the fact they were different from me was key to our friendship and progress.  

 

Thinking about what's normal reminds me of the work of Lieutenant Gilbert Daniels at  Wright-Patterson Air Force Base in the 1950s.  In the early fifties new fighter jets were killing pilots - even those with serious WWII and Korean War experience - at a frightening rate.  The conjecture was cockpit complexity was getting ahead of people.  Daniels was just out of Harvard with a degree in statistics and an interest in the physical design of cockpits. To save cost and deal with a size budget dictated by aerodynamic design, cockpit dimensions had been standardized. Over 4600 active pilots had dozens of measurements taken. Daniels looked at what was considered to be the ten most important and looked for averages.   Not one of the 4600 were average in all ten (the definition of average in each was quite wide).  Picking the three most important he found that less than four percent of the pilots could be considered average using that much weaker criteria.  His work led to adjustable cockpit designs and death rates plummeted.  No such animal as "average" exists. I'm personally convinced difference are even broader when you look at how we think.  There is not such thing as neurotypical.

 

Last week I came across a preprint of a study that considered how culture changes how we see the world. Three groups of people - the Himba people in Namiba,  residents of a semiurban Namiban town, and people in the US and UK were asked to look at six optical illusions and report what they saw.  The divergence was dramatic.  One of the illusions was the (in)famous Coffer illusion.  Take a look - what do you see?  I'll confess to taking nearly a minute before I could see the circles.  About half of the Himba people only saw circles and the other half saw rectangles and circles. The Western population only saw rectangles and the semiurban saw a mixture.  It opens up so many questions.

 

These days I try to work with a mixture of people with very different backgrounds.  A youngish female cosmologist at the Perimeter Institute just sees things so differently than her male counterparts.  Not better - but different. Different is better when you're trying to make progress.  Organization that think and act otherwise are kneecapping themselves.

 

I was going to include something on average progress in learning and sport, but that would take too much time.  Later!

 

 

 

 

Juliette Powell

We met at one of Pip Coburn's events on Houston Street in about 2011 or 12 and have been friends since.  She had pretty much become a member of the family by choice and was closer to me than my blood relatives.  This morning a detective from the NY Police Dept called telling me she had died in her sleep.  I was listed as her relative to contact.  She was only 53 and was seriously ill last week.  Some of you knew her.

 

I'm shattered so I won't say much other than she lived a remarkable life - you can check her wikipedia page.

From last year with Sukie.

 

tools, models and the good-enough

Computer Science was considered part of Electrical Engineering departments when I was in grad school.  It gained importance and size and its link to EE became something of a historical artifact over the next decade or two.  While I had many computer scientists as friends and colleagues. I didn't appreciate the evolution until 2002 when a friend asked me to take over for his junior level CS class for a week while he dealt with a family emergency.  He suggested my experience with hardware and a few unusual projects would be useful so I decided to spend the week on neural networks.

 

There were an unusual number of blank looks - more than I get when I teach physics. Assuming it was me I asked a few students to hang around after class to tell me what I was doing wrong.  It turned out they didn't know how to read simple schematic diagrams - the sort of thing a freshman EE would know In the seventies and eighties a CS or EE student  probably learned it on their own as teenagers when many were amateur radio operators or built their own computers from the ground up.  I assumed they knew what simple logic circuits looked like and wanted to contrast them with op amps and analog computing.  The rest of the week was proved interesting in the best way.  I was visited by a recurring epiphany - one of those 'ah ha' moments that rhymed with early ones - namely the importance of abstraction.

 

The year before my little Audi TT was still very new.  Very few of my friends knew how to drive a stick shift.  The automatic transmission had abstracted the task of changing gears and they didn't have to think about it. That was just one part of driving a car.  Many of the features of my car were abstractions - I didn't have to think about them when things worked and problems might require someone who can drop below the level of abstraction and understand what's going on. 

 

We use models and abstractions for nearly everything. Physics - especially physics - uses them.  Newton's laws are just fine for almost everything we do on our scale.  It's a simple model that's relatively easy to use and you learn to think of reality in terms of its model when you're using it. The same goes for Maxwell's Equations and electromagnetism.  But when my friend Sarah is playing volleyball she doesn't concern herself with equations of motion and fluid dynamics.  She has a wonderful intuition and physicality that allows her to abstract the underpinnings that internalize the physics and get on with winning a match.

 

Just as programmers don't worry about solid state physics and quantum mechanics when they write a spreadsheet program, we don't worry about the programming when we use one.  You worry - or at least you should worry - about the limitations of the tool you're using.  Is it appropriate for the job? Most of us rarely ask ourselves if it's an appropriate level of abstraction, but that can be important - it's very important in math and the physical sciences.

 

Often different tools and different models apply to different domains. In physics we talk about complementarity.  You can have different models that work in different domains, but it's okay if they both work at some boundary. Quantum mechanics can reduce to classical mechanics as size increases and special and general relativity do the same as speed and mass-energy shrink.  You need to ask yourself what level of accuracy is needed. 

 

Large language models and generative AI have become the latest tool of abstraction.  gAI (I hate the general use of the term 'AI' as there are so many flavors and domains of use) doesn't know anything.  Through cubic money, legal and illegal information acquisition and a huge energy expenditures these tools stochastically construct narratives that are useful to many.  There are accuracy and cultural issues that you should sort out in deciding if they're appropriate for your needs.  Richard Feynman offers a warning: The first principle is that you must not fool yourself and you are the easiest person to fool.  It can become a question of trust and truth at some level.  

 

I've seen some unfortunate misapplications, but people are just learning at this point.  I've played with several of the tools recreationally averaging about an hour a day for a few months of non-trivial use. I haven't found a use that would save me time or give me insight in my own work That hasn't been the case with some other forms of 'AI'.  Your mileage may vary.  Some comments:

 

ChatGPT has been characterized by some as overly sycophantic and Grok is basically white-nationalist genocide.  Both strike me as uninterested compared with Claude - in particular Claude Opus 4. Claude's creators attempt to give it a personality of sorts.  It's far more conversational than the other tools I've played with. That feeling that you're dealing with a person can be dangerous, but it's fascinating to watch the evolution.  

 

An interesting area is religion. Claude Opus 3 was heavily trained in the Buddhist traditional and Claude Opus 4 even more so.  A friend in France has been a practicing Buddhist for over three decades and describes her interaction with Claude Opus 4 as enlightening, sometimes spiritual.   She told me of a high demand religion that is experiencing difficulties as members, using the Internet, are questioning  fundamental faith claims. They've contracted the creation of a tool to deal with these questions as they're often beyond the knowledge of their lay ministry.  So far multiples of tens of millions has been spent, but it isn't ready for prime time.

 

When I was a postdoc we lived nest to a few visiting Chinese scholars.  I became a close friend of the theoretical physicist Ting-Chang Hsien - one of the broadest thinkers I've encountered.  Taking a walk one night in the early eighties we speculated on what happens if a machine passes whatever sort of Turing test would exist to cover general knowledge.  It's wouldn't have to be sentient - in fact it would be an illusion.  Ting-Chang suspected the most successful computers of this type would be those that agree with you and make you happy - the sort of thing that Feynman warned us against.  He suggested we might start worshiping them as we've done with some other explanatory models in the past.    Remembering that chat reminds me of the wonderful animated show Futurama with its Robot God and Robot Devil.  The most holy relic was the 6502 - the microprocessor used in the Apple II.

 

 

 

 

the importance of art and nature

It was a chilly day in early Spring and I had only been with Bell Labs  for a few months when I came across Alan in Lord Stirling Park.  It's a lovely place and I was near the end of an eight point something mile loop.  I'd seen him going on the way out as he crouched by his homemade four by five inch view camera.  It was beautifully made with a very expensive German lens.  He showed me the controls and went back to looking at a tree stump as I walked on.  Three hours later he was still in the same spot when I came by .  Naïvely I asked how many exposures he'd made:

"Oh just one - but it was so beautiful!  In an hour the light could be right for another."

Years later I learned he was a physicist at another Bell Labs location.  We worked on a few projects together.  His patience and out-of-the-box creativity are remarkable 

 

One of my undergrad physics professors suggested walking in the woods,  not thinking of anything in particular before working on a difficult problem. You'd regularly see him just watching something on a tree or bush on campus.   He was also a big fan of art museums and was taking by "finding beauty in simple forms. " He also told us not to study for exams in classes that meant something to us. Just keep up and think and let the tests flow by as nothing other than mileposts.   Some of the best suggestions I've had.  When I've interviewed people for positions that demand a degree of cleverness I always ask about their hobbies.  It turns out that was a good idea.

 

The trigger for this  is new work from a group of Cambridge psychologists that, for the first time, offers empirical evidence that engaging in artistic beauty helps you escape the "mental trappings of daily life" and induce "psychological distancing" -  the zooming out on your thoughts to form connections and gain clarity.

 

The tests asked people to study simple ceramics as well as line drawings of them. Their simplicity demanded focus - something that might be difficult with a complex piece of art.   All of the subjects in the ceramic group tested higher on abstract thinking tests compared to those who looked at the drawings for short time.  The later group saw no improvement over their scores before the tests.  More startling were the results of those who regularly engaged in an artistic hobby. Their abstract thinking scores were even higher than the non-hobbyist group. 

 

Physics and math are the only areas I know with any depth.  Both are driven by beauty and curiosity.  Perhaps it's not surprising that many in these fields engage in artistic hobbies like art, music and even poetry as well as spending some time every day being alone with nature in some way.

 

Go to your local museum and look at a few things deeply, wander in a park or the woods and loose yourself.  Try drawing something.  Not only is it a good break, but your ability to think abstractly may improve!

 

the hit list

Since restarting the blog I tool a look at its statistics.  These 12 represent around 90% of visits a month after posting.   As always lots of errors that I don't bother correcting given my textbookish editing style.

 

Do bats echolocate in color?

G, A, F, ↓F, C

How much power do you need to go fast?

fire and ice

50 miles on a pint of peanut oil

solar powered hurricanes

the physics of rowing (dimensional analysis)

why does light bend and slow down in glass?

playing with dimensions (simple estimates of difficult problems using dimensional analysis)

moore's law and largish numbers

falling trees and sufer music

homer simpson and olympic swimming events

 

 

 

 

comments on innovation

A few weeks ago my friend Om began a piece on innovation with the torpedo bat.  As with anything Om writes, it's a recommended read.  I have an interest in innovation in sport, but will start with some more general comments.

 

I tend to see innovation as a step in a chain: discovery → invention → development → innovation.   These usually don't come from the same organizations and can span a large amount of time.  There's also a great non-linear mixing.  Discoveries can lead to other discoveries, inventions can lead to discoveries and so on.  I've been involved in the first three areas at different times.  In my mind innovation is a successful change that embeds itself in some area.  I've never been directly involved in that process, although things I've worked on have led to it. 

 

It isn't an efficient process and efforts to make it more efficient usually fail.  Research, the leading edge, is often broken into three areas:  curiosity driven research, high impact research and applied research.  Various labels are used, but curiosity driven research is where major change often comes from.  In the US it's much rarer than it once was as funders expect progress on a time schedule.  High impact is more common - research that is likely to achieve a stated goal.  And finally applied research can be thought of as the leading edge of development.  

 

In many disciplines innovation is the only segment recognized and richly rewarded by the public, but they're all difficult and necessary,

 

Witnessing the current US regime's attack on science and education is beyond disturbing.  I'll stick with the research side as that's what I know best, but the attacks on intellectual freedom, students, faculty and funding will lead to medium term programs and long term disaster for the country and beyond. As an example very little basic research on drugs is done in pharmaceuticals.  They take research from universities and the NIH and apply it.  An important step, but the pipeline is being destroyed as I write. And the attach on education in general will dry up human resources at all stages.  The price the nation will pay for acting on grievance will be enormous. 

 

With that off my chest I'll go back to sports.  The International Olympic Committee  talks about three principles: fairness, safety and inclusion. I've written about the importance of the ordering of these and the interpretation from science, so I won't expand on that here, but I'll add that entertainment is part of the picture. Some sports are big business as is betting on them. That can and does drive the ordering of the three basic principles. 

 

The quest to win and safety create a desire for innovation.  About ten years ago I became interested in the fluid dynamics of a volleyball in flight.  Princeton has a rich collection of sports science papers.  There are hundreds of papers on baseball, cricket, soccer, rugby, American football, hockey, tennis and the so-called Olympic sports.  There are dozens of papers on baseball ranging from the fluid dynamics of the ball in flight, neurology and vision (odd fact - of all occupations and spots, major league baseball players have the highest vision acuity by a good margin. They also have slower than average reaction times), kinesiology, and so on.    The physics of bat design and sweet spots goes back to the 1950s.  Innovation like the torpedo bat is the results of successful design and testing as the fundamental work has been around for a while. 

 

Running shoes, bicycle design and swimsuits are more controversial areas. All are now regulated for fairness although it raises the question of what a record means. Track bikes from before 1995 were much faster than current designs so how do you compare records. Innovation in shoe design over the past five years has rendered records many track events obsolete.  Some innovations are less controversial, although their impact has been great.   And there are flexible bars and floors in gymnastics, a steady progression of tennis racket design and so on.  

And then there's innovation in technique. The Fosbury flop in the high jump is often considered the single most revolutionary innovation in sport.  There's a good deal of innovation in coaching that comes from just trying things to the application of kinesiology. a better understanding of physiology, psychology and even sociology. 

 

With all of this innovation and change there's something more fundamental.  The passion, focus and drive of elite athletes.  Knowing a few Olympians has taught me a lot about that.   At the highest level of sport I suspect that's a constant that remains unchanged.   That fierce passion exists in many other endeavors, but sport offers a simple language most of us can understand giving us a way to celebrate excellence.

 

________

With all of the political craziness in the US, I've been listening to music and doing math as a diversion.  Here's a wonderful performance of the "Pa-pa-ge-na! Pa-pa-ge-no!" duet from Die Zauberflöte. . It's not often you see a coloratura soprano in the role and there's serious musical chemistry in play. 

 

K2-18b

There's a lot of excitement about the best signature of life yet seen on an exoplanet.  The plant in question is a sub-neptune about 2.6 times the diameter of Earth that orbit a red dwarf that's 124 light years away. It has an orbital period of 33 days and is in the habitable zone.  It has an atmosphere of carbon dioxide and methane, but there's a strong suggestion it's a water world.  A few years ago dimethyl sulfide was found.  That's significant because it's only produced by living organisms on Earth.  The latest work shows it exists in densities about twenty times greater than on Earth. Dimethyl sulfide is mostly produced by algae in the oceans on Earth and is partly responsible for the odor you associate with the ocean.

 

Lots of caveats..  while we don't know of other natural mechanisms that could produce a lot of the molecule it doesn't mean there aren't any. Also careful examination of signals sometimes (often) causes them to disappear.  Folklore is this group and a few other competitors have been bashing the data for some time and can't make the signal go away.   I've read the paper, but it's not my field so I can't really judge other than listening to a friend at Caltech and another at PI. 

 

It was seen with the James Webb Space Telescope.  A follow-on has largely been completed that would be useful for confirming the signal and looking for other interesting signatures, but the Trump regime killed the program along with most space science and astrophysics.  

 

A few years ago I had a long series of exolife discussions with a 35 year old friend .  My belief then - and this assumed NASA and JPL would continue to be functional entities - was that we'd have a strong confirmation in her lifetime.., probably not mine.  At this point it seems pretty hopeless as so much damage is being done to the programs

 

correcting a prediction

A huge miscalculation.

Given issues with uses of the technology I had discounted the impact of large language models on the economy in the near term, but given how it’s being used by DOGE to dismantle the infrastructure of government and by Trump’s white house group to design tariffs to swing at the global economy, I have to admit my error.  Of course a massive change that uses technology as a fuel requires someone to start the fire, some to fan the flames and those opposed to the fire to be disempowered. 

Of course one can learn from their errors.

 

 

 

 

Das ist nicht nur nicht richtig; es ist nicht einmal falsch!

The Office of the United States Trade Representative released the formula used to calculate Trump's "reciprocal tariffs.

My gosh!  Where does one begin?

The famous physicist Wolfgang Pauli was asked to look at a paper by a young physicist.  After a few minutes he tossed it away saying: "Das ist nicht nur nicht richtig; es ist nicht einmal falsch!"  (That is not only not right; it is not even wrong.)

 

There are constructions so completely without clue as to merit Pauli's comment:  the flat Earth, the Earth is the center of the universe, phrenology, or that π can be legislated to be 3.2  come to mind.

 

Trump's formula completely ignores services, foreign demand and a host of other important items. His real goal is probably more along the lines of a protection racket - a tool to compel fealty.  We shall see. I can imagine people around the world acting like Canadians have when it comes to American products and services.  Not to mention the psychology of feeling "poorer" in the US. 

 

What could possibly go wrong?

_________

Catching up.. five readers have written enthusiastically about the maapl.net search engine. No AI, no ads, open source, no tracking and free.   It's also faster than Bing and Google.  I've been recommending Kagi - a paid search service - but probably won't renew my annual subscription when it comes due this Summer as maapl fills most of my needs.

 

Finally congratulations to Kip, Andrew, Frank and Jheri who solved the puzzle I posted on the 23rd of March.

 

 

 

 

 

 

 

getting lost for the thrill of it

 Last week a friend and I were chatting about the tension between exploration and exploitation in technology.  The work of James March came to mind.  My old friend Norm had a few of us read his paper on exploration and exploitation for a few of our Friday afternoon sessions at Bell Labs.  It had set off a good deal of work in a number of fields and was relevant to the Labs as exploratory research was cut in favor of more exploitative work. March's paper was interesting as it suggested ways to model the tension between research, development and exploitation when resources are limited (they almost always are). It also brought up the fact that people have differing tolerances and tastes when it comes to the risks associated with each of these areas.

 

For the moment I'm more interested in concentrating on exploration as that's what I find most rewarding.  By many metrics individual exploration isn't worth it compared to sticking to a well-trodden (by you or your firm) path.  It's difficult and sometimes impossible to optimize and it frequently fails.  On the other hand if you're willing to learn from failure it can be powerful and an education plus every now and again you have an insight or discovery. 

 

Some of  us find the discoveries or insights - and even the path of being somewhat lost - fun and even thrilling.  It's been suggested there's a dopamine hit that helps drive this in some.  It's not something we can do all the time as it's generally not productive, but the discoveries and flashes of insight are what can bring change.

 

There's a good deal of work that suggests interdisciplinary at the individual and/or organizational level provides richer explorations with higher success rates.  It certainly seems to be the case in the physical sciences.  Those who make the greatest strides are often those who have come up through varied pathways.  Creativity appears to be linked to interdisciplinary as well as focus.

 

Doing something is much richer than just watching.  It's been noted dementia rates are much lower among those who are constantly doing things that require learning.  It doesn't seem to matter what it is, but has to be something that's mentally challenging.  Learning to knit, a sport you haven't done, math you don't know how to do, how to draw, play a musical instrument .. anything as long as it's hard.  

 

There are false novelties.  Social media gives a stream of novel things to watch giving people dopamine hits. There's a bit of an addiction. Perhaps we're wired for it. Another false novelty is sugar.  When we needed the calories sugar provided a hit.  Now we don't need the extra calories, but for many of us the attraction of sugar can be unhealthy.

 

Back to creativity.  I might be wrong, but I've come to believe it's something you develop and enhance.   There are several measures of it.  You have to take them with a grain of salt as it's not exactly clear what's being measured, but there's a general agreement among these methods that creativity has been dropping in teenagers and college students since the late 90s with a very large drop beginning around 2010.  A sobering conjecture is heavy use of social media is taking away from other forms of the work of exploration.   It's also been noted that this drop isn't seen among those who are doing tasks that require constant exploratory learning.

 

So turn off the GPS every now and again and take find new paths.  Surround yourself with interesting people who are dissimilar.  It's something of a luxury to do much of this in most work environments so take up new hobbies and keep pushing yourself.