omenti

the critical edge in a major sport

The other day I was in a Manhattan library wearing the t-shirt of my undergrad school.  As luck would have it someone noticed it.  She was from the same school - her Ph.D. was in mechanical engineering - and a most interesting conversation followed.  It turns out she happens to work on suspensions for one of the more famous Formula One teams. 

 

Formula One may be the most form of auto racing. Four wheels hanging out in the breeze and an open cockpit, its roots go back to Grand Prix racing about ninety years ago.  After WWII it was associated with some of the most famous car makers and drivers.  It has enormous popularly worldwide with something like a half billion people following any given race and over 300 million fans who follow most of the season. Good drivers make in the eight figures and some have career earning exceeding a billion dollars.  A massively important form of sports entertainment larger than the NBA and NFL combined. 

 

F-1  technology side has made huge advances over the years.  Thanks to tire, suspension and aerodynamic advances the cars can corner with lateral accelerations over six times that of gravity.  That's more than an astronaut experiences during re-entry and on par with the limits of a jet fighter.  Top speeds are on the order of 240 miles per hour on relatively twisty tracks.  Teams spend a lot of money - she said an average team might have 100 engineers, technicians, physicists and mathematicians - over a third with Ph.Ds. - working on every little bit of vehicle performance.  Cars get much better during the racing season not making improvements every race means almost certain failure.  She said a few American universities major talent suppliers - notably Caltech and MIT - but England is the real ground zero for car design and tuning talent from a half dozen schools.  

 

So there's this question.  How much success is the driver and how much is the car?  The feeling is the driver, even though operating a car under such conditions probably is an athletic performance, is less important than the car.  Put a great driver in a lesser car and they might improve the performance of the car a bit, but they'll never win.  A lesser driver in a great car probably won't win, but will place much higher one might expect.   But there's a human story and sport is all about human performance.  You probably won't see much interest in autonomous cars zipping around a circuit even though they might outperform humans.  

 

But there's another story.  The master designers and aerodynamics people are very important, but there are enough very good ones and enough viable approaches that the top half dozen teams are on fairly equal footing. the most important person on the team, and sometimes it is two people, is the strategist - usually a mathematician or physicist. 

 

The cars are rich data sources sending hundreds of channels of telemetry for analysis giving the state of almost anything you can image about the car.  Additionally tracks, weather, cars, tires, competition and tactics and strategy are all modeled ahead of time.  Variables are thrown into the mix that makes the strategist even more important.  Tires are a good example.  There are three levels of grip... the grippiest tires wear very quickly, the least grippy more slowly. You have to use at least two during a race meaning a pit stop with a tire change is necessary. Additionally there are two types of rain tires.  It's an exercise in modeling and game theory sorting out how this will play out. The most consistent factor in winning races appears to be the strategist.  Three of the top ten are women.

 

As the sport changes to stay relevant hardcore fans are being challenged to examine some of the race data during the race .. there's thought about making it part of the race itself, but keeping the competition level high and the fans engaged is the key point.   One thing she mentioned is heart rate and blood pressure telemetry from the pit chief as a car comes into the pits at 70mph stopping a few inches away as he has a few seconds to make the swap.  

 

There are a lot of interesting things you can say about this type of data and its analysis.  The fact that some believe understanding it is more important than the driver speaks volumes. Some of the teams monetize this by letting their largest sponsors in on their technics of dealing with rapid change and strategy and tactics.  A few of the analysts have been loaned to these sponsors for non-trivial sums to look into their businesses. 

 

It makes you wonder about other sports.   It would be interesting to create a plot with understanding of athletic performance (including psychological elements) along one axis and understanding of strategy along the other.   Understanding hydrodynamics created a revolution in swimming along the first. A deep data and model driven understanding of strategy has driven Formula One performance along the second.  How much potential does a given sport have? 

 

 

 

 

smoke rings, sports and beyond

Canals saw explosive growth in England during the first few decades of the Industrial Revolution allowing raw materials like iron and coal to fall in price by as much as eighty percent.   Improving canals became a major industry - they became straighter with locks and aqueducts coming into wide-spread use to deal with changes in elevation.  Then, around 1840, railroads became the thing and many of the canals were purchased for their right of way.  The expansion of railways led to two of the more dramatic financial bubbles in history, but that's another story.  The story here is about something really interesting that happened just before the railroads started taking over.

 

James Scott Russell found himself working on a large canal near Edinburgh, Scotland.  A newly minted engineer, he had been hired to investigate the movement of barges in the hope of making them more efficient.  Russell spent a lot of time observing and making measurements.  One day the tow rope between the mules and the barge snapped and the barge came to a sudden stop.  He watched in amazement watching mass of water in front of the barge's front bow 

“I was observing the motion of a boat which was rapidly drawn along a narrow channel by a pair of horses, when the boat sud- denly stoppednot so the mass of water in the channel which it had put in motion; it accumulated round the prow of the vessel in a state of violent agitation, then suddenly leaving it behind, rolled forward with great velocity, assuming the form of a large solitary elevation, a rounded, smooth and well-defined heap of water, which continued its course ong the channel apparently without change of form or diminution of speed. I followed it on horseback, and overtook it still rolling on at a rate of some eight or nine miles an hour, preserving its original figure some thirty feet long and a foot to a foot and a half in height. Its height gradually diminished, and after a chase of one or two miles I lost it in the windings of the channel. Such, in the month of August 1834, was my first chance interview with that singular and beau- tiful phenomenon which I have called the Wave of Translation.”

Afterwards he built a tank in a laboratory to conduct controlled experiments. In 1844 he published his results noting four things about these solitons.

° The waves are stable, and can travel over very large distances (normal waves tend to either flatten or steepen and topple.

° The speed depends on the size of the wave, and its width on the depth of water.

° The waves do not merge, larger waves overtake smaller ones rather than combining.

° If a wave is too big for the depth of water, it splits into two waves, one big and one small.

A few decades later  physicists began to work out the physics leading off in unimaginable directions.  But back to that wave moving down the canal.  If it was large enough you could surf on it.  And, with enough study and money you could build very repeatable waves to your liking.  It turns out a surfer named Kelly Slater did just that.  The December 17^th, 2018 issue of The New Yorker has a great article. 

 

It turns out solitons exist throughout the physical world - smoke rings, packets of light in fiber optic strands and much more.  On light packets - high speed data communication using optical fibers simply wouldn't work without solitons - the bits of information would blur into a muddle at high data rates. The math is a tad beyond the level of this blog - you'd see it in a second year undergrad physics class.  As people continued to work out why it is stable some insight came from that radical moment in 1953 when the computer went beyond a bicycle for the mind to becoming both a telescope and microscope.

 

Physics 101 classes usually have a vortex ring demonstration to illustrate some of the properties of vortex ring solitons.  Smoke ring guns are a lot of fun and fairly easy to build.  Turning them into a really great demonstration takes some effort. Dianna Cowern recently posted her version.  Her enthusiasm is pure wonderfulness.

Recipe Corner

 

The return of recipe corner - at least for today.  This was great last night

 Roasted Brussels Sprouts

 

Ingredients

° 1 pound Brussels sprouts - remove the outer leaves

° 1 medium sized onion - I had a sweet onion (Vidalia)

° 1 largish baking apple (Granny Smith) cut into half inch chunks 

° 1 tbl olive oil (doesn't have to be a great EVOO)

° 1/2 tsp grated nutmeg

° salt

° ground black pepper

° 1 cup apple cider

° 2 tsp butter

° 1 tsp brown sugar

Technique

° preheat oven to 350°F

° everything through the olive oil into a large baking pan.  salt and pepper to taste and toss to coat.  Cover with aluminum foil. 

° roast for about 25 minutes, uncover and stir

° temp up to 375°F and roast with no foil 'til the sprouts brown a bit - 10 minutes for me

° when the roasting begins bring the cider to a boil in a sauce pan over medium-high heat.  Add butter and brown sugar and stir 'til dissolved. Cook at a boil until it is reduced to the consistency of a syrup .. about 15 minutes for me.  I had about 4 tbl remaining. 

° Put the Brussels sprouts into a serving bowl, add the syrup and toss.. you probably need to adjust salt and pepper a bit before serving.

 

technology risks

Fast moving technology can create risks as well as opportunities.  Recently we saw the shutdown of London's Gatwick airport by illegal drone activity.  It almost surprises me this drone delivered service disruption has taken this long to happen, although it is a surprise it went on for so long. It’s a classic example of asymmetric economics - drones are cheap and expendable but the impact of their misuse is large and very expensive.  Completely  defenses are impossible and what can be mounted is usually expensive. 

 

Drone attacks been suggested and defenses from drones with geofencing, flying nets, to jamming guns to trained eagles have been tried with varying degrees of success.  Several were tried at Gatwick but all failed.  All of these can be effective against subsists of attacks, but it is easy to design around them.  Having a variety of good enough measures will be expensive and one suspects cottage industries supplying all of this like we saw in the wake of 9-11.    I don’t know if penalties are enough to keep some types away, but it may be the most effective tool. In any event drone regulation is going to see a big re-visit approximately everywhere now.

 

This is a dramatic and very specific example, but there are so many others where the cost of attack is many orders of magnitude lower than the costs of damage and/or defense.  Russian interference in the 2016 American election may be the most dramatic example to date.   Coming down the road the Internet of Things opens up a serious Pandora’s box of opportunities for the bad guys.  Many, perhaps most, of these will be cheap with poorly written software and no mechanism for patching against attacks.  You really don't want these in your home where they can impact your life.  Add to this devices that spy on you..  Google, Amazon and Facebook offer hardware that takes information from your home and makes it the property of the company offering the services.  Thinking twice about these devices is prudent. 

 

It has been said that technology is neutral - it's all the application.  I don't buy that.  Some technologies are bad by design.  We need to be considering that much earlier on than we've been doing. The problems can't be eliminated, but perhaps they can be contained. Measures that can be taken with legislation and education.   A good example of education is the Electronic Frontier Foundation's Gift Guide .. it's a good starting point and may make you think twice.

 

 

hope for nature and humanity

In grad school I came across The Limits to Growth - a book by the Club of Rome. Basically a report on computer simulations of exponential population growth and limits of the carrying capacity of the planet for a number of resources.  It struck me as bleak and depressing suggesting zero population growth was our only path out. I was curious about their model and found issues with the underlying data set and models. It wasn’t possible to describe population growth trends with simple exponentials. Something else was going on. They admitted some issues, but swept it under the rug sort of claiming “correct” data would fit their model. It began to slowly eat at me.

 

A few years later  I found myself traveling to Princeton regularly as part of a collaboration. It’s a great place to run into people and ask questions.  Eventually I found some demographers and learned at their feet. They weren’t terribly impressed by the book. They freely admitted there were serious issues, some of which may be existential threats.  At the same time there were some extremely positive trends.. Instead of simple exponential models there are periods of development with different population dynamics.  At first I had a my doubts, but their models fit the available data and fit some emerging trends much better than the CoR report.  There are a  variety of models, but they mostly boil down to three main stages.

 

The first is a period where the death rate is very close to the birth rate - sometimes exceeding it, sometimes falling short. Humanity was scattered around the globe and regions would fade in and out.  Eventually with trading centers becoming stable with very low population growth. The death rates were enormous.. for a family to continue to the next generation you generally needed more than six children.

 

The second stage is marked by the introduction of sanitation and successful disease prevention. Death rates drop dramatically, but people still maintained extremely high fertility rates.   The sanitation and medical knowledge is quickly communicated to populations at a similar stage. This results in enormous population spikes.  In the US the core part of this stage was from about 1920 to 1970.

 

The third stage sees women gaining education, opportunity and control over their bodies and fertility. Incomes are rising and families begin to focus putting more into fewer children. The fertility rate drops .. often dramatically.  Think Japan and parts of Europe. 

 

Different regions of the world go through these at different times and rates. One of the huge drivers for the second and third stages is urbanization. At first it  seemed counterintuitive, but when you talk to demographers and read their papers it becomes clear the growth of cities is the rocket fuel that enables the rest.

 

About fifteen years ago I organized a small conference on this at Aspen. I left with great hope and enthusiasm believing that lowering some of the other potentially existential threats would only make this process easier. In my mind the immediate low hanging fruit was working on global warming. It could have been, but I didn’t understand the inertia and control of the fossil fuel industry.

 

I could go on and on about these demographic trends and had intended to, but it would be book length and others are more expert in the area. I may talk about it in the future. The next stage is cloudy I can argue for at least four very different paths .. some of them pretty awful. But I recently came across a nice discussion of the saving nature piece..

 

Now for the good part.  Sean Carroll of Caltech interviews Joe Walston - the senior VP of Field Conservation at the Wildlife Conservation Society.  Joe offers a better summary of these issues than I could in an hour of writing  It's a bit over an hour long, but worth it if you have any interest in demographic trends, why they happen and where they could go.

 

Finally it's interesting these trends appear to be society independent.  We like to think how different we are, but in many ways, for good and bad, we're the same.  I find that comforting.

 

 

 

the other scientific revolution of 1953 that radically accelerated knowledge

In the Winter of 1953 Francis Crick interrupted the lunchtime patrons of the Eagle in Cambridge to announce he and James Watson had discovered the secret of life.   In fact it wasn't far from hyperbole - they had discovered the structure of DNA and the world was forever changed.¹  Something else world-changing happened later that year.  Without it we  wouldn't have advanced past the technologies of the 1960s.  Surprisingly one of the people behind it thought it cute, but hardly worth publishing.  

 

But first a bit of historical context

 

The Manhattan project began with blackboards and experiments.  Individual researchers had their desk calculators, but it soon became necessary to work out how neutrons diffused when a critical mass of fissile material came together. A group of scientist's wives came together to perform the repetitive calculations.  Next came IBM card reading calculators and then von Neumann became interested in developments in electronic computing at Harvard, Bell Labs and the University of Pennsylvania.  There's a deep and rich history from this time on, but the important point was the meaning of "computer" was shifting from a woman with a mechanical calculator to an electronic machine.²

 

In 1953  electronic computers were still rare, mostly one-off machines that were expensive and temperamental.   It was still a world of research and development. A few people recognized practical applications, but the machines mostly weren't ready.  Except for physics.

 

Physicists were heavily involved in the birthing of electronic computers focusing on specific problems that tended to involve military projects or fundamental research to drive the hardware.  Fundamental research was seen as the tip of the R&D effort and was often funded from military budgets which were large enough to support these efforts.   Enrico Fermi  had an interesting idea.

 

He was interested in nonlinear systems thinking they might have something fundamental to do with entropy and why time only flows in one direction.  He, John Pasta and Stanislav Ulam conceived a simple toy problem - sixty four identical masses representing the atoms in a lattice connected end to end by little springs representing chemical bonds. Unfortunately calculating the restoring forces is a beast of a problem.

 

A linear system is one that can be described by a straight line on a graph.  Consider a spring.  Pull it twice as far and it exerts twice as much force. The same is true when you pluck a string on a guitar.³  Nonlinear systems have a more complex behavior.  With linear systems you just add up the contributions of each element without considering the others.  Nonlinear systems are nasty - you need to consider everything at once.  You might be able to calculate very simple systems with a few elements, but after things become very complex.

 

The three physicists wrote down a mathematical description of the physics. Think of it as a strange sort of string that you pluck and then listen to as time goes on.  Solving it would require a lot of computer time.  The big iron at Los Alamos was called the Mathematical Analyzer, Numerical Integrator, and Computer or MANAIC-1 for short.  It had been built for thermonuclear calculations, but this trio had an enormous amount of credibility and it was clearly fundamental research. 

 

The three had no experience with programming, but Mary Tsingou was a mathematical physicist who knew how to reason with  MANAIC-1.  She set to work creating a program that became an experiment simulating how the lattice behaved for sixteen, thirty two and sixty four "atoms" with various spring forces.⁴  

 

Imagine of the frequencies coming out of the system in terms of sound.  They had expected a pure note that would shift  over to more chaotic noises and finally the equivalent of just a hiss.  That didn't happen. 

 

The lattice was "plucked" and a series of overtones appeared and dissolved into each other creating a complex mixture.   Then the whole process reversed and the overtones disappeared in reverse order until the original pure note reappeared and the whole process started over again.  

 

 

Here's an illustration.  Consider the top graph.  The purple points are just the first vibration mode of a simple string - think of it as showing what would happen if it was just a simple string.  The system we're talking about with equal masses and springs is in blue.

 

Even this simple little problem that had baffled some of the greatest minds had a beautiful and completely unexpected result and a new way of doing science had been developed - the computational experiment.  The computer had become not just a bicycle, but rather a telescope for the mind.

 

Much of science is impenetrable without computer experiments like these.  It's so common, I've done thousands, that you tend to think it's natural.  But it had a beginning and without it much of what we've seen from science after the mid 1950s and the resulting technologies wouldn't have been possible.  Inefficient cars and airplanes, no electronic revolution, no cellphones, no ....

 

Fermi thought it was cute, but not fundamentally important.  That may be true for the hypothesis they tested, but how they did it changed the world.

 

__________

¹ There were many others who were in on this, but Crick and Watson weren't exactly the sharing kind.  Most notably the x-ray diffraction work of Rosalind Franklin.  But she was a women and died before the Nobel was awarded anyway.  

² George Dyson (Freeman Dyson's son) penned Turing's Cathedral.  An excellent and very readable history of the efforts at Princeton.  That may be the most important early thread as the computer architecture used today was developed there.  Much of the slightly later development came out of Cambridge in England. 

³ Sort of ..  you can get it to behave non-linearly, but for now consider a spring that behaves like F = kx where x is the displacement, F is force and k is a constant that tells you about the spring's stiffness.

⁴ I don't know for sure, but have been told she programmed down to the iron in machine language. 

 

 

just what is that privacy thing?

At a birthday party the other day conversation turned to privacy.  When someone complained about the erosion of our right to privacy I saw an opening to ask a question:  

"When was the golden age of privacy in America?"

Among a dozen people there was nearly universal agreement around the time of the Bill of Rights.  After all those founding fathers ..  I just listened.  It turns out they were wrong.

 

The idea of privacy was almost foreign in Revolutionary America.  While the founders were arguing various rights for landed white males, the only nods to privacy were associated with the ownership of property - notably intellectual property.  People were encouraged by the Revolutionary government and later the Federal government to spy on each other to ferret out who might be breaking a boycott or somehow not supporting the cause.  Punishment came through social pressure like public shamming and sometimes worse.  

 

 You might think living in relative physical isolation as a farmer without telephones, cameras and Internet would give you serious privacy.  The rub was you had to deal with others in your community.  You had to buy and sell goods and services, most people went to church and so on.  Chances were you'd never travel far from where you were born so you were surrounded by people who knew quite a bit about you.  You were defined by your reputation and position. It probably seemed very normal to most people.

 

 Change first emerged not on farms, but in the cities among the wealthy.  With property came the idea of ownership and protection.  This extended inside the house where there might be servants  By the mid 1700s doors began to appear inside homes along with this radical new privy.  Dealing with one's personal matters became a private affair.  Jump ahead to the 1820s and even homes of the emerging middle class had private rooms and privies. 

 

The South followed suit, but it took time and was more isolated.    Travelers noted, with some shock, that as late as 1850 many houses were doorless inside and multiple families sleeping  in the same room.  Like New England change first came to the wealthy, but slaves and indentured whites were not allowed these luxuries.  The right to this kind of privacy was reserved for certain social classes and skin colors.

 

 

I've written about the role of technology and will write more in the future, but privacy stakes me as something of a paradox.  Many of us point to the erosion we think is caused by technology, but at the same time there have been great strides in other forms of privacy - keeping the government out of the bedroom and reproductive rights in general come to mind. 

 

I've had conversations with legal scholars and historians.  All told me they can't think of a simple robust definition even though all of us think we now what it is.   At a high level I think it is at the boundary of the relationships among yourself, the state, religion, your work, your family and community and technology.  We're each members of several larger wholes as well as being an individual.   Change any of these in some major way and there is a need to be societal  renegotions .   Technology has been something of a cauldron of change over the past hundred and twenty years or so. .  

 

 We've always had problems with the judgement of others ..particularly when it's capricious, mindless or evil.  With work some of these practices can be discovered although rectifying them can be difficult.  Now we have massive data collection schemes often using data  and data processing of questionable and untraceable provenance sitting in judgement.  People point to China and its citizen scores, but Google, Facebook, Amazon and others are doing similar things here.  As Cambridge Analytica taught us  sometimes all a potential user of this information needs is a relatively simple toolkit.   

 

Laws will always trail - except perhaps in authoritarian countries.   Europe is taking some small steps with GDPR at this point. China and other countries (and companies!) are moving in the opposite direction.  And many in the Western world believe privacy (whatever that means) is a human right. We are seeing regionalization and balkanization of the Internet and that may be just the beginning.  And remember those algorithmic reputation and judgement engines..   So many difficult questions are raised. 

 

the development of the concept of privacy

My sister uses old photographs as an ingredient in her art.  One archive is a collection of glass plate negatives, mostly posed studio portraits of Danes.  Danes in suits, Danes in dresses, Danish children in little Danish suits and dresses.  All of them with the same fixed dour countenance.

 

Their letters suggest a more positive outlook on life.  The problem, it seems, was a photograph required an exposure of about ten seconds. It was an expensive process that you only splurged for on big days - Easter, birthdays and weddings.  You have to get it right the first time. Few can maintain a smile that long, but dour is easy.   Try it.  

 

It was a huge improvement from the earlier daguerreotype.  There's a story that Samuel Morse of telegraph fame visited Louis Daguerre's studio in Paris.  Looking at examples of this wonderful new photography, Morse was struck that even the busiest sections of Paris showed no human activity.  Daguerre patiently explained twenty minute exposures were at fault.  The streaks of moving people, carriages and horses only offered a slight blur on the streets.

 

While the Danes were exposing plates my sister would ultimately use, George Eastman had been working on making photography practical.  Around 1885 he had a film fast enough that an average exposure outdoors took just a fraction of a second - a short enough time that a smile didn't change.  He kept working and developed a series of smaller and smaller cameras that were increasingly easy to use. In a few years he had a small camera that was so small and so fast that a subject might not realize their photograph was being taken.  And then came the Kodak Brownie..  a small, fast, very easy to use camera that anyone could use.  It cost one dollar - about $30 in today's money - and came with a roll of film. The camera shown is the Model 2 - an improved camera that was introduced in 1900 at a much higher price point - two dollars.   Snapshots became a national craze.   People were taking photos of others without permission.  One woman even found a sack of grain at a store bore her unauthorized image.  Kodak fiend and Kodaker became insults.  Tabloid newspapers made a good businesses publishing unauthorized society pictures along with those of women in compromising dress - bathing suits and worse. People were fascinated by widely available photography and were doing it on their own, but a boundary was emerging without a much guidance. 

 

The 1870s also saw the wide scale use of telegraphy.  Moving a message from point A to point B usually wasn't direct.  Instead a message from Boston to Minneapolis might first go to New York City where a telegrapher copied it by hand and gave a piece of paper to another telegrapher with a wire to Chicago.  A similar exchange would take place before a connection was made to Minneapolis.  Some of these messages had important business information and there was a good deal of leakage.  Some companies began to use ciphers as a result.  

 

The telephone began to spread beyond the boundaries of business after WWI.  Operators could and did listen.  Many local exchanges were party lines where anyone could listen to anything up to a dozen people might be saying.  The telephone also revolutionized dating.  Scheduled meetings in the parlor of the girl's family could now be arranged to be somewhere else.   The concept of the date as a destination came about in the twenties and the automobile only made it easier. 

 

Society was going through huge changes.  Much more information was available in a variety of forms. Sometimes we wanted it to be made available and sometimes we wanted to protect it.  In 1890 two Harvard Law student, Samuel Warren and Louis Brandeis published The Right to Privacy in the Harvard Law Review.  It's an astonishing piece.  Clearly written and still regularly cited.   The concept of a right to privacy was beginning to emerge in America.

 

It's a delicate balance that is in many ways a way of looking at how society comes to grips with changing technologies.  The historians I've talked to suggest no one talked about a right to privacy before the Civil War. Rather there was a concept of the home being a castle that an owner had a right to protect.  Information and who has access changed that. 

 

It's a double edged sword.  We want .. we need .. some of our information to be known in order to function in society.  When Social Security numbers were first issued, marginalized segments of society - African Americans for example - proudly displayed their number for all to see as proof of their being part of American society.  On the other hand moral crusaders were invading (legally) US Mail and in the 1950s what happened in the bedroom became an issue for the Supreme Court in Griswold v Connecticut.   Some historians refer to that event as accidentally confirming the right to privacy in America. 

 

And now we have the Internet and an exchange of a number of services in return for information about us .. information that we often know little about.  The information may rightly or wrongly describe us and it is often used without the correct context.  It might have implicit or explicit bias.  It is then used with a sea of other information about us and others to create a set of digital judgements which can be used for targeted ads or for darker purposes.  This gets fascinating and deserves another post or three.  

 

Technology usually outstrips the development of a social norm which usually precedes law.   It's messy and companies are now betting their futures on it.  It's interesting that one of the largest companies in the world has decided a right to privacy by  the end user of their products is a core value.  Of course it isn't completely bulletproof, but it is diametrically opposed to many of the other players.  All of this while people generally unaware of the issues.   

hunting serendipity through waylosing

I've been lucky at the art of getting lost just enough.  If you've walked with me while I'm thinking you've probably noticed my sense of direction and local awareness can be low.  If I'm by myself awareness comes into focus, but getting lost - just enough - is something I enjoy and can be a source of serendipity.   I should probably explain.

 

Alarm was the order of the day when Sputnik went up on October 4, 1957.  A big military threat from the USSR , satellites needed to be tracked. Some folks at MIT put together do-it yourself plans to enlist amateur satellite trackers across America.   All you needed was a knowledge of your latitude, a simple sighting device, a pivot and a clock.  Thousands of homemade sighting scopes were made overnight in home shops and within a month a factory was mass producing a kit that was set out to high school science classes around the county.  

 

It turned out a good clock was centrally important.  Knowing the time of a position reading to a second or two was necessary.  Shortwave radios were common in the 50s and satellite trackers were asked to tune to WWV in Fort Collins, Colorado to listen to the reference time signal.

 

A couple of physicists considered about the inverse problem.  What if you had a lot of good clocks in well defined orbits and could read their times?  That's enough (well, sort of) to figure out your position on the globe.  You'd have a global positioning system.  

 

It seemed like an extremely hard problem at the time.  You'd need to orbit atomic clocks and there were any number of issues.  But navigation was a hard problem - a really hard problem.  DARPA eventually went to work on it and, armed with boatloads of money,  made it happen.   Mix in the revolution from Moore's Law, add a few dozen years and navigation had been democratized.

 

Many of us get around following instructions from our GPS systems not paying great attention to what's around us.   Kids in the backseat watch videos or play games.  While this may be ok for some types of travel I think a lot has been lost.

 

I like to pay attention to areas I visit regularly to see change that is otherwise invisible.  There's an undeveloped and heavily wooded swamp next to where I live and, assuming I'm home, I take spend an hour or so every day around noon.  Over the sixteen year I've been doing this a richness of change from year to year rather than season to season emerges.  You get to know about certain animals and plants.  Stop and listen. Let your mind take flight. Try it at night with the fireflies and crickets knowing a coywolf or two is watching.  It's here that a lot of ideas seem to present themselves.  

 

Cities are great too.  I live reasonably close to New York City. If I have the time I'll go somewhere very unfamiliar and learn something..  I'm not a terribly outgoing type unless I know someone, but curiosity can take the lead.  It's not so much being a flâneur as it is looking, listening and asking questions.  

 

The same for universities.  Some of the places I've gone to school and worked encouraged cross-disciplinary collaboration.  Sit in on a departmental lecture outside of your field, ask questions afterwards, walk into people's labs, ask them what excites them now.  And for sports - you can learn a lot about a sport if you spend a bit of time with a competitor.  

 

Waylosing (a wonderful term I heard a few year ago) is very efficient in its inefficiency if you're after more than getting from point A to B.  And normal travel is getting more and inefficiently efficient.    

 

As you get good at it you find yourself asking questions .. why are things the way they are.  Start out with simple things for practice. Why are stores arranged the way they are?   What about the open lots or failed businesses along a route you take?  How many types of ants can your find in Central Park?     How does someone react when you give them a harmonica or a balloon:-)  A lot of this is mental exercise that just comes in handy.  There's a certain pleasure to working out something yourself.  Of course you don't need to - and can't - do it all the time, but it's fun when you do.

 

I worry that the technologies that isolate us from our environment lead us to making poor use of our  travel time (and time in general).  They can be very useful, but use them appropriately.  Other technologies  like cameras, binoculars and microscopes can extend your question asking capabilities.  Choose appropriately.

 

The weather is good this time of year for most of the readers. If you're planning a trip try the classic road trip stopping randomly along the way - get lost, ask questions and look.  The journey can be where serendipity lurks.  

 

Many of you have developed deep waylosing skills.  I'm up for walks...

 

Finally it has been observed that very efficient operations tend to stifle creativity.  That doesn't mean that inefficiency guarantees creativity.. There are a lot of issues . clearly a lengthy subject for another time.

 

 

 

machine learning without a net .. from sports to smart toasters

And I don't mean Cylons.

 

Every now and again you notice a few trends coming together even though they hardly seem related at first glance.  Someone asked a question today that turned out to be something I've thought a lot about over the years - a collision between tiny machine learning, privacy and an ocean of tappable ambient information.

 

Twenty years ago some of us were playing with sound. After spending a lot of time analyzing it we wondered what areas existed beyond voice recognition and music. There was a small "name that tune" excursion where we tried to match sung, hummed or whistled bars of music to - well - name the tune.  We were trying to turn the sound into note patterns.  It sort of worked, but the the problem of building a notation library was too difficult at the time.  The easy way out would be a hammer and tongs approach - namely matching played music with a library of played music.  We filed a patent on it, but never pushed hard as we didn't see the use case. Oh well...

 

Next I tried to recognize misbehaving car sounds..  Auto mechanics have great stories of people trying to make broken car sounds.  Unfortunately the problem was not small enough and I shifted gears when a ventilation fan in my office started squealing.  

 

Big heating, ventilation, and air conditions systems mostly run with boring industrial sounds, but a good HVAC mechanic can recognize when something's about to fail.   They also tune these systems for efficiency by ear - for some systems the sweet spot isn't far from the point where problems happen and every installation behaves differently.   With my trust Sony DAT recorder I made recordings of these systems in various stages of failure.  Then some fancy pants signal processing and, lo and beware, you can detect and analyze failure modes.   There was probably a lot of interesting work that could be done, but my organization evaporated around that time. 

 

I was fascinated with low power computing a bit earlier than the sound work.  How little energy does it take to do a computation?  (you can make some interesting entropy arguments and get down to basic physics.)  More practically what could be done with very low powered sensors that didn't need batteries or wires?   The interest took several directions - all good as you slowly begin to learn.

 

Then about five years ago it struck me that the HVAC problem could be done more easily with machine learning.  I've expressed reservations about the misuse of machine learning and regard some of it to be not very different from phrenology, but potentially very useful where you can understand it and its use.¹  I supervised a senior thesis project that attacked the HVAC project using machine learning and a microphone.  It showed some potential, but for me the learning was finally realizing how computationally cheap machine learning can be at the low end.  I also ended up learning a fair amount about machine learning as the best way to learn is to do it yourself and then teach.

 

Given Bitcoin mining efficiency and the massive machine requirements at FaceBook and Google it isn't all together clear, but think about unconnected machine learning on your iPhone and perhaps your Apple Watch.  In conventional computing the CPU deals with a large number of instructions compared to the type of processor used in machine learning.  ML processing chips have a large number of tiny processors that work in parallel. Most ML is just matrix multiplication.  While there can be a lot of steps, each is really simple .. generally just addition and multiplication.  A properly designed ML element holds the numbers in the matrices its working memory cache.  It's much more efficient in time and energy use to keep what you're computing nearby rather than having to constantly read and write it to memory.  You should be able to do a specialized bit of silicon with enough ML horsepower to do some rather surprising tasks while using very little power.  And you can do a lot on a smartphone if some of it's silicon was designed to support ML.

 

I think Apple understands this at a rather deep level.  They're very careful about how to efficiently deal with graphics processors which is shorthand for saying they have optimized how to do huge volumes of very very simple computations.  The chips in the current iPhones are optimized for the task.  They've build machine learning machines you can hold in your hand and, in theory, run ML applications that don't connect to the net and divulge information.  The new measuring app on iOS 12 is an excellent example of a local machine learning deployment. I don't know if it is being done on the watch, but there's no reason why they couldn't. Perhaps the biggest design advantage Apple has these days is silicon. 

 

machine learning in your hand rather than the cloud.

 

I can think of a number of applications that use the sensors on the phone or watch -- from health, to motion analysis (think sports).. why you could imagine wearing a watch and using a couple of iPhones on tripods or held by friends to analyze your volleyball, gymnastic, or whatever moves.  A personal coach or a serious tool is you already have a coach.

 

Exciting stuff, but I suspect it goes much deeper.

 

A back of the envelope calculation suggests you can do useful ML with micropower.  I'm sometimes asked to come up with senior projects in engineering classes (dangerous as I'm not an engineer).  I think you could do the HVAC work with an accelerometer or microphone and simulate the type of compute power that, if done in silicon, could run for months on a watch battery.  Maybe some very simple chemical analysis.   Or recognize a very small vocabulary -- perhaps five words - to control your toaster,  thermostat or anything else.  You could use a key word to make it listen or shine a laser pointer on it.  You could also build a very simple image recognizer that would detect someone's gaze.  Look at your toaster and say medium please.  

 

Way safer than saying medium please to a Cylon.

 

A need for privacy may drive this even deeper and away from the Net.  You can sit down and go through a few dozen applications in an afternoon.  It seems very rich and very latent.    Apple knows how to do silicon and low power.

 

So many possibilities.. perhaps even my dream "big data" project of listening to the woods. 

__________

¹ I've written and talked a fair amount on problems.  It can get technical and I won't spend more time on them here, but it's a serious problem as surveillance systems transition to judgement systems.

 

privacy, first class

With Facebook in the news, a lot of people have started asking questions about how much online privacy they have and how this touches them and their families.   Here's an excellent non-technical summary of Facebook issues and challenges by Zeynep Tufecki and a bit of work by the Pew Research Center on Americans' Attitudes about Privacy, Security and Surveillance. 

 

Much has been written pointing out Facebook users aren't Facebook's customers, but rather are marketed as the product.  That isn't exactly right.  Facebook is mining and selling user attention employing psychological techniques to keep users engaged in any way possible.  This piece of the business model goes well beyond Facebook.. YouTube, Instagram, and most other forms of "free" social media play the same game.

 

People have been losing trust at an increasing rate that could have a chilling impact on the growth of Internet connected services.  We hear about evil examples, but even well-meaning data collection can be problematic.  A decade ago Netflix ran a competition to find a better way to recommend movies.  They "anonymized" a dataset, stripping out user identification.  Unfortunately it turned out it was possible to re-identify given users, political affiliation and even sexual identify by adding a tiny amount of additional information.¹  

 

Strap in - let's dive into differential privacy.

 

Differential Privacy (DP from here on out) is a privacy definition that has been around for a bit more than a decade.  In the past two years the term has become well-known as Apple uses it to protect user privacy.  A simple statement is something like:

Consider two nearly identical databases.  One has your information and the other lacks it, but is otherwise identical. DP says the probability a query will produce the same result is about the same no matter which database is used.

DP tells you if your data has a significant impact on the outcome of a query.  It is safe for you to contribute if it doesn't.    Note that just tallying up the results leaks out a bit of information about you and does not satisfy DP to the letter of the law.  Each time a query is made a little bit of information about you leaks and this can add up giving clever, and even not so clever, folks a foot in the door.

 

One way DP can get around this is to inject a bit of noise that masks the content of a specific person.²  This will muddy the results of the query a bit, but you can calculate how much noise to inject to have an accurate enough result.  

There is a tradeoff between accuracy and privacy!

We're getting to the part that makes DP "interesting" (in a challenging way).  The more information you get from the database, the more noise has to be injected to keep privacy leakage small.  This loss of accuracy can be a big deal - particularly for Machine Learning training that can hit that database thousands or millions of times.  Once you've leaked the data it's game over.  User privacy has been compromised.  You'd have to destroy the database and start over.  You have to figure out a privacy budget - how many queries can you use?

 

Apple isn't in the business of selling raw or processed data to other parties - they're not mining your attention. They do collect a lot of information to run their services and make heavy use of DP.  To make it safer, and give them something of a firewall to certain types of spying from say a government, information is randomized on the device before being sent over the network.  Additionally applications are not allowed to harvest information from other applications and databases on the phone without explicit user approval and Apple's blessing.³ Even Apple can't re-anonymize internally.

 

For Apple's purposes DP makes a lot of sense although they need to worry about the privacy budget constantly. The bit attention harvesting social media platforms are another story.  Its fascinating to think about what form a privacy preserving social media platform would take and if there is any way to migrate current SM users to such platforms.  I don't think the current Facebook or YouTube business models would be viable with even modest privacy protection.  That leads to an interesting question for regulators - what is the business model budget for the trade off between business model and privacy? 

 

Thinking this through for other services is essential .. the Internet of Things for example. 

 

In about a month the EU's General Data Protection Regulation takes effect.  It has some teeth, but also has loopholes. It's a start, but people are talking about more serious protections.  One can imagine these will vary from region to region. I'm hoping that anything one here (anywhere for that matter) has serious input from independent privacy and security experts and very little from the companies directly involved.  One can imagine intense lobbying from the richest companies on Earth. 

 

Note:  This is way too short.  I give myself an hour and don't want to bore people.  Expect it to be part of a major society and technology discussion going forward.

__________

¹ a classic paper by Narayanan and Shmatikov

² There are a lot of ways to do this.  Usually the noise has a Gaussian or Laplacian distribution.  For the purpose of this note, just think noise.  The amazing thing is determining how much noise is needed can be made without explicit knowledge of the contents of the database.

³ This is in stark contrast with Android - read Zeynep's piece for a high level description.  Sadly at this point I wouldn't recommend using an Android device if you value privacy..  then again I wouldn't suggest using Facebook either.  We all have lines we don't cross.