omenti

mean free paths - from pinafore to removing risk to custom design

It was a beautiful late Spring day and I found myself of singing a song from The Mikado as I strolled down West College past the Conservatory.  I'm not a good performer, but my rendition must have been recognizable as a chap walking next to me took the one of the other parts and we were doing a bit of harmony as our paths became a formation.  We stopped for the light, stopped and continued until we had finished.  I smiled and nodded.  He reciprocated, but then he started into the first song from  Pirates - We sail the ocean blue.  He had a strong voice and I followed him in whatever harmony I could muster.

 

We crossed the street and continued a block to the intersection next to the conservatory and more magic - two others joined and we continued walking until the song was over smiling broadly and nodding before we parted.

 

Oberlin, Ohio is a tiny town and we had all probably overshot our destinations as we sang together that day late in April 12 years ago.   It was a powerful manifestation of how serendipity - at least a certain type of serendipity - is possible when the mean free path between people with similar interests is small.

 

In physics mean free path roughly means the mean distance between objects that may collide - the mean free path between molecules in a very high pressure gas is less than that of one at a low pressure.  There is also a social mean free path and it is one of the reasons why cities, Silicon Valley, the restrooms and central atrium of Pixar, the old Bell Labs, and any number of other concentrations of talent are important.  Bring people within a few meters of each other and they interact.  Bring passionate and talented people within a few meters of each other and you may just get the unexpected.

 

These clusters may not be stable over time as people migrated - Brooklyn has become one of the East Coast regions that shrinks the mean free path between creative residents. Take a look at the Outlier video Om has featured for an excellent example of what a low mean free path can mean for someone interested in something novel in dungarees.  

 

Brooklyn has become so important for a certain type of young creative that the new Pixar building in California is known in some quarters as "Brooklyn"...  

 

Now back to Oberlin ..  There is enough talent in the Music Conservatory that it is possible for someone without great means (me) to have a piece of music composed.  You need to be tapped into the roots of the beast - I was an adjunct professor there for several years as part of a joint AT&T Research/Oberlin College Music Conservatory project.¹  This position gave me a rich interaction with students and faculty often into the wee hours of the morning with drink and good food.  Ideas would bounce around and people began to sense a few new possibilities that would normally remain hidden.  These things can and do happen over the Internet, but their cross section - the probability of them happened and being recognized as interesting - is much lower.  Our online environment, rich as it is,  is much less sophisticated and rich than the real world

 

Under the right conditions the Internet can decrease the mean free path between parties. Long distance mentor relations and even a few projects.  The problem is finding people well matched for serendipity and creativity.  A large amount of research has gone into creativity in individuals and groups.  To date online creativity - the connecting the dots/serendipity kind - is still a loooooooong way from face to face human interaction.  

 

But the Net can help creative enterprises - often at crucial movements.  Om has recently talked about Kickstarter.  Now most projects proposed on Kickstarter fail, but well thought out and articulated projects do succeed providing enough money to get them over a critical initial hump.  

 

Perhaps just as important is a funded project also gives insight into potential market size.  I know someone with a fairly well known podcast.  Some people had encouraged him to make a transcript of all of the shows aired to date available for sale.  Unfortunately the problem with this is he had about sixty hours of material and even the cheapest transcription services would introduce a risk of several thousand dollars if people didn't purchase the final product.  There was also an issue of understanding how to price it and if a print or ebook version should be made.  

 

He decided to turn it into a Kickstarter project and learned (a) there was a market large enough to support the transcription and editing effort, and (b) the price he picked was considered reasonable by those who bit - two absolutely vital pieces of information acquired with almost no risk.  His success gave me paused and made me think of a few small projects I've seen fail due to incorrect estimates of these questions.  

 

The Kickstarter approach may be superior to funding it yourself even if you have the money as it provides a sandbox to answer potential market size and price questions

 

One of you (the Jheri of the Jeri/Jerry/Jheris I know) happens to be very tall and hard to fit - not only clothing, but also airline seats, countertops, chairs and so on.  She introduced me to her taller friend Colleen.  Both of them don't think about their poor fit that often as you get used to your environment and learn to compromise, but being different can take a physical toll with bad backs being far too common in that crowd.  It is good to have friends like this as they provide a rich problem space you may not have considered.

 

I decided to work on a simple cutting board for Colleen who happens to love to cook.  It turns out there have been numerous studies on work surface positioning for almost every body type and physical condition.  It became clear a raised wooden board would be the best bet and I had one made for her as the cost of hardwood is so high that a woodworking shop could do it at a lower cost than I could and they would probably do a better job.   The owner of the shop became interested in the idea of making this a product and Colleen put together a page promoting it as she loved hers.²  A bit of asking around got the attention of a well-known women's tv show and they wanted to feature it.  The problem was they also demanded a copy be given to everyone in the audience in a special "tall" show.  In the end the cost of doing that was greater than the profit made selling these online.  The product remains as it is custom made for the user, but I don't think they have made  profit on it.  

 

In retrospect this may have been a great Kickstarter project.  Although it is custom made by a small shop the more important function of Kickstarter would have been gauging the size of the market before any major effort was undertaken.  It may also have allowed a volume purchase of hard maple as well as specialized shipping materials.

 

The other project that should have been done on Kickstarter was an attempt to make very long inseam jeans for tall women.  The designer spent about a year online lining up customers, finding a boutique manufacturer and so on.  He rounded up about 2,000 firm commitments that they would buy - enough to allow an initial low volume production run and sell them at a reasonable price.  Based on these commitments he went ahead with a somewhat smaller run to err on the side of caution thinking 500 or so would bring him close enough to break even that he could proceed.  Unfortunately only about 25 orders materialized and he lost a large amount of money.³  The beauty of Kickstarter is that not only must a threshold be reached to start a project, but the money will be there once the threshold is signup period has successfully concluded. 

 

We are moving to a world where individualization and shared design and intellectual property will become much more common.  The Maker movement provides some fuel for this and grassroot interaction street fashion is likely to become important. Sites like Etsy as outposts for the tiny manufacturers and spaces like Pinterest for matching interests with the makers.  There will probably be a lot of evolution and new thinking in this area - current sites and tools aren't up to the task.  We may be at a diamonds on the beach period.

_________

¹ At once the least musical and most mathematical faculty ever associated with the place

² It turns out the standard countertop height in the US is 36" and was determined in an ad hoc fashion when cabinets began to be mass produced for the great suburban expansion after WWII.  This turns out to be optimal for a 5'4 to 5'5" cook.  Significantly shorter or taller cooks will have ergonomic issues and those who are seriously away from average can experience pain bending or an inability to comfortably reach the surface.  On the tall side the threshold is around 5'10 - certainly cooks 6' or taller should use a higher surface.  The shop own and I both thought this meant a market might exist. If you think you might use one I can give a strong recommendation to AWP.

Before settling on a simple raised table I was interested in an adjustable board.  It presented a challange so I attempted to crowd source some solutions from readers of another blog of mine  here and here adding some notes on design. Some ideas were submitted, but nothing earth-shattering, so I went with a simple table.  There are times when brute force is appropriate.

³ It turns out 20:1, 50:1 or even 100:1 ratio between people who say they will order a product and those who follow through is common in apparel

________________

And a recipe.  I needed chocolate and pistachios recently.  In order to manage my weight and deal with Sukie's chocolate allergy a few neighbors had a happy evening.

 

Chocolate Pistachio Cookies
 

Ingredients

° 185 g  (6.5 oz) shelled pistachios

° 28 g (2 tbl) unsalted butter

° 115 g (4 oz) unsalted butter, sliced into tablespoons

° 115 g (4 oz) bittersweet chocolate, chopped - I used Lindt

° 1.5 tsp vanilla extract

° 60 g (1/2 cup) AP flour

° 35 g natural cocoa powder

° 1/2 tsp baking powder

° 1/2 tsp fine grained sea salt

° 100 g (1/2 cup) white sugar

° 2 large eggs 

° 1.5 tsp vanilla extract - you need the real stuff - mine is rum based

° 120 g chopped milk chocolate - again I used Lindt ..  reliable for baking

Technique

* Place one oven rack at the center of the oven and preheat to 350°F

° Line two cookie sheetswith a silicone mat or parchment paper.

° Spread the pistachios on an unlined baking sheet and bake for ten minutes until golden brown, Remove from oven and immediately add the 28g butter and stir well. Set aside to cool.

* Melt chocolate and remaining butter together in a microwave oven. Stir together until smooth and set aside to cool slightly as you beat the eggs. While the chocolate is melting combine the flour, cocoa, baking powder and salt in a food processor and pulse to blend completely. 

* Combine the eggs and sugar and beat until they become pale yellow and light in texture. Stir in the chocolate. When it is mostly incorporated, add the vanilla. When the vanilla has been incorporated, add the flour mixture and stir gently to blend. When the flour is almost completely absorbed, add the pistachios and milk chocolate folding into the batter. It will be dark and shiny with a texture similar to marshmallow Fluff.

° Spoon about a heaping tablespoon (or 1.5 inch ice cream scoop) of dough onto the sheet tray for each cookie, arranging them in a 3X4 pattern on the tray. Bake the trays one at a time on the center rack in the oven for 10-12 minutes. The cookies will rise slightly, develop a shiny crust and appear to be just set. Let them cool for two minutes and then loosen the bottoms from the lined tray. Let them finish cooling on the sheet trays or move them to racks depending on the final texture you want - experiment the first time.

 

 

 

sharpen your dot connection skills...

Scientists are skilled dot connecters as that is one of the more powerful tools of the trade.  Years ago Greg, another physicist, recommended visiting "companies that make stuff"and see if they'll give you a tour of their process.  Making physical things presents multiple challenges including human factors and social challenges and, if you get someone who is proud of what they're doing, you can dive in and ask a lot of questions getting some real depth in the process.  

 

Over the years I've visited companies that brew beer, cartoon, make rocket engines, fabricate ICs, build large gas turbines, do precision wood work, build submarines, bake cakes, make ultrasound machines, pull wire, smelt copper, make exotic steels, make potato chips, make electric motors, make dresses (many companies here - it is a rich area), craft bespoke shoes for athletes, build exotic batteries, make LED lamps, and build bikes (that is a small list - it has been going on for years).  

 

All of these tours have been worthwhile and some of the visits have given serious insight for problems I have worked on years later - the potato chip factory visit provided just the right clue in thinking about mesh networking. Although the reasoning is a bit difficult to explain, all of a sudden the network problem became obvious in light of the clues I had learned watching a quality control step at the Utz factory in Pennsylvania.  

 

Post 9/11 it has been more difficult to get tours (I tired for a few years to get a tour of the place that makes water tanks for buildings in Manhattan before having success) and many companies are giving up the fabrication phase at this point.  But don't let that stop you - there are a lot of places that are thrilled to have someone take an interest.  Except for the Mars company (M&Ms) - one I will probably never crack...

 

I've learned a lot from bicycles including their design and fabrication.  Simple machines  hold a special interest as you can easily wrap your head around the integrated design.  So this 1945 film on the fabrication of Raleigh bikes is particularly good.  I'm keeping this short today as I spent the 17 or so minutes watching.  Comments from the company:

Our resident historian (HR Manager Frank Ellis) tells us that the bike was a “Low Gravity Carrier” (a Butcher’s Bike to you and I) and “The guy putting ball bearings in the bottom bracket took me back….he had that authentic piecework dance…people used to say that if you worked at Raleigh long enough you started doing the dance in your sleep!” 

Frank, who carried out time and motion studies when he joined Raleigh as an apprentice in the 70s, said: “The hub lacing shot towards the end was great to see. It used to be done by hand by women and the first frames capture the classic fanned spokes that the women achieved with a simple flick of the wrist…try that one and see if you can get them anywhere near as evenly spaced! I’m always telling people the top women pieceworkers were faster than the automatic hub lacer, but I might be exaggerating just a little.”

Give it a view - great stuff!

 

 

Bike manufacture has changed considerably in the 65+ years since then and that is another story...  I've been involved in it a bit myself and for someone without a mechanical engineering background it has been a terrific education and has expanded the pool of dots in my quiver.

 

Nothing like ending on a mixed metaphor...

 

 

a pre-review

Jon Gertner's new book on Bell Labs has just been published and a few people have written to ask what I think about it.  I haven't read it yet, but am very curious as I was part of that organization from the end of its golden years through its demise.  It was a unique and amazing place - a fundamentally different approach to "innovation" than what is common today.  It isn't clear to me which approaches are better or worse - the old Bell Labs approach doesn't square with how companies work these days.

 

The old AT&T was a regulated monopoly divestiture in 1984.  A few percent of its income was channeled into running the labs - known as the Bell Telephone Laboratories and, through a series of regulatory consent decrees - funding was incredibly stable and sufficient to insure very long term research and development projects.  After a decree in 1956, inventions were licensed under very liberal terms to the rest of the nation and the Labs had become something of a national laboratory.  

 

I won't go into the history of the place and will be interested to see how it is charted in Gertner's book, but the core DNA of the place needs to be explained.

 

Innovation had a very specific meaning.  It didn't mean invention or discovery, but rather was discovery taken through stages of development and finally implemented at a scale that would change society.  

 

The Labs was primarily advanced and applied development with a core basic research arm (area 10 - later renamed area 11 about the time I joined it).  Fundamental researchers were presented with an amazing array from curious problems.  The notion was not the standard approach of engineering where you attack a problem by tallying your resources and try to build an optimal result, but rather something a bit more zen.

 

In the research unit you defined and attacked problems by looking for the unknown and then trying to understand as much as your could about it.  You had to go deeply into uncharted territory not knowing if your efforts would pay off or if you had the right intellectual tools for the job.

 

This approach - solving a jigsaw puzzle not by identifying your pieces and putting them together, but rather looking for the missing pieces and trying to solve a puzzle by understanding what was missing - required a multidisciplinary approach.  It is close to the approach of physics and basic research had a high percentage of Ph.D. physicists.  Really good ones.  Their specialties were not as important a their taste in problem solving.  I was an experimental particle physicist.  Bell Labs never dabbled in particle physics, but that was "the right stuff" and they saw an immediate match.

 

We worked with computer scientists, mathematicians, statisticians, mechanical engineers, electrical engineers, opticians, chemists, acoustics experts, material scientists, psychologists, neurologists, biologists, musicians and artists to name a few backgrounds.  The central laboratory at Murray Hill, New Jersey was built around a five story corridor a quarter mile long with a sometimes maze-like structure of smaller buildings.  Going to lunch or visiting someone else's lab would force interaction with others. It was said a researcher going to lunch was like a magnet rolling by iron filings.  It was a brilliant architecture and aspects of it have been duplicated elsewhere.¹

 

Some of the projects had very long horizons - digital communications took about four decades and thousands of researchers years of effort.  This type of time frame is only seen universities and national laboratories these days and is rarely directly coupled efficiently with commercial development.

 

But the next step in innovation - and one where the majority of Bell Labs people worked - was development research and engineering.  This is enormously difficult to get right. Most of the projects were defined by either Bell Labs management based on needs of the Bell System or by the government for national defense.  Many were difficult projects that advanced the state of the art by an order of magnitude and this was something of a model for DARPA, which came about in the 1950s with a clearer focus on military needs.

 

As the applied research and development proceeded it was productized and turned over to Western Electric - the manufacturing arm.  People in the various units spent time with each other.  Small branch development labs were built into Western Electric factories and the development folks often worked directly on the lines to understand the problems.  

 

The approach wasn't as clean as I describe it and there were tensions that had to be dealt with, but it was an innovation engine that changed the world.  It even paid a lot of attention to education creating everything from science and engineering ciricula to identifying promising students - mentoring them and getting them through their Ph.D.s in the best universities at no expense.

 

A very different approach than what is used, and perhaps needed, today.  The model today has universities and national labs doing the basic work with companies mostly engaged in short term development - real corporate research is not common.  Even long term development is rare^.2  True corporate research is mostly dead as is an integrated multi-disciplinary approach.  

 

Companies identify the pieces of the puzzle they have and try to build something optimal rather than trying to understand the missing pieces deeply.

 

As the Labs started to crumble in the late 80s I switched my focus from physics to something more applied - broadband to the home.  The problems were how do you do it and why do you do it.  The later lead to work in digital music and both led me to the notion that I *really* needed to understand social research.  I immersed myself in a human computer interaction department and hopefully helped out a bit with a bit of experimental rigor, very different thinking and a lot of naïve questions that turned out not to be so naïve.  I learned a lot in the process, but the efforts of the group were never coupled to the company at that point.  We did a lot of neat stuff that is finally beginning to be realized now.

 

I still try to search for projects that are looking for a bit of deeper understanding when I consult.  There are a lot of diamonds on the beach these days - as there probably always are.  The approach is rare and I find it odd that I'm sometimes seen as much broader than I see myself.  It is really just a bit of curiosity and the luck of having an amazingly rich chance to work in an environment that makes places like Apple and Pixar look lame when it comes to "multi-disciplinary".  I think there is great value to bits and pieces of the approach as suspect we'll see more of it as the tech world matures and energy becomes a more pressing issue.

 

I don't know what picture the book will paint.  I was witness to an amazing dinosaur and hope some of the better features of the place can be replicated.

 

We had a crisp definition of innovation. The physics people had a crisp definition of data and information.  I find the physics defnition at odds with common usage and I'll post on that in the near future - I've hinted at it before.  Data isn't basic...

 

Om recently commented on creativity talking about John Maeda of RISD.  This caused me to think of some comments from the professor of a graduate physics course I took as an undergrad.  The professor was famous for his creativity having been sprinkled with Swedish holy water.  Somehow he wanted to get over a few points on being creative - he thought you could inspire it.  

 

from my notebook:

 

° be curious about anything and everything. expose yourself to different people and ideas

° don’t be frightened to try the unknown. Many solve puzzles by figuring out what they can do with the puzzle pieces they have. A better approach is to figure out the missing pieces and solve a puzzle you don’t know anything about by learning about them.

° remember it must be play

° the delight is not in finding the answer, but in coming to it yourself

° be totally honest

° deeply know everything about your problem

° real creativity simplifies – aim for simplicity as it is beautiful

° get into something so deeply you forget everything else. you need large pieces of uninterrupted time

 

much more on these later, but back to the old Bell Telephone Laboratories for a bit. When I first started I was delighted to learn that my first director was something of an aurora expert.  He was invovled in the Telstar communications project and was involved in the beginnings of creating the field of what is now called space weather. He would have loved this time lapse of the aurora taken from the ISS a few days ago...

 

 _______

¹ Pixar uses centrally located washrooms and has a design that reminds me of Murray Hill on a much smaller scale.

² IBM, Intel and Microsoft do longer term work in the tech sector. Apple has a ten year horizon for some projects, but anything longer than three years is very rare in industry. One of the longer term interests was video telephony. Here is something on developments in color video telephony .... in 1930.

the best part is how it intimidates the guys...

A few years ago I helped a young relative learn how to operate a manual transmission.  The idea was that used manuals were inexpensive and it isn't that difficult.  It turns out there was a bit of serendipity as far as she was concerned - none of the guys she dated knew how to use one and this fact gave her enormous pleasure.

 

In 2010 only 6.7% of cars and light trucks sold in the US had manual transmissions.¹  As manufacturers continue to drop manuals from their lines a backlash of sorts has developed and a few niche models are doing very well with people who love to shift by themselves.

 

Cars are fascinating examples of user experience and user interfaces.  Development is glacial by the standards of the consumer electronics industry - most manufacturers are well into design of models four and five years out and there are ideas that reach out nearly a decade.  Power trains are enormously expensive to engineer and are not only used in several platforms of a maker, but are sometimes used in other makes.  The laws of countries and physics both weigh heavily on design.  Mix into this economics and an attempt to create a desirable looking vehicle and a good user experience and you are left with an optimization process with thousands of variables.

 

Automobiles and light truck are remarkable examples of industrial design.  They are Fantastically complex pieces of engineering, they represent the most complicate mechanical objects most of us own.  Over the years they've become more trouble-free, safer and, at least by some metrics, more efficient.²  We take it for granted, but the interface is very similar and predictable.   Furthermore most cars can accommodate a wide range of human body types - generally from the 5% female to the 95% male, but some expand the range on one side or the other.  I was impressed that Colleen could get into my smallish Audi Coupe as she's at the 99.99% male level in height.  She can't drive it as her right leg can't get past the steering wheel, but there are cars she can drive and someone who stands more than a foot and a half less can easily drive the same car after fiddling with the seat a bit.  In theory both configurations offer protection from crashes.

 

Small badges can craft brilliant vehicles for specific niches, but the costs of low production runs mean the buyers have to be well heeled.  A few years ago I had the pleasure of meeting a few of Ferrari's top engineers including the engine guy who was largely responsible for the user experience of the drivetrain.  His title was very different, but he was in charge of how it felt to the driver - a defining feature of the brand.  Even the exhaust note of each model is carefully tuned and part of the art of engineering compromise.

 

As production volumes of a model increase this complex design process and comes with a cost.  Two years ago I was thinking of trading my 2001 TT for the current version.  I test drove one.  A lovely engine and transmission, but the cockpit displays and user interface were jarring and, in a word, wrong.  It had a frightening array electronics and several byzantine interfaces.  Compared to my simpler cockpit was like fingernails screeching a path across a chalk board.  It was like using Windows XP after having used an iPad.  I decided to keep my current car and worry about keeping it running for a long time.  

 

Many cars are much worse.  

 

There is very little flexibility and major change comes slowly.  The industry tends not to optimally modularize. Electronics constitute 20% to 30% of the production cost of a car and that figure is increasing rapidly.  But the electronics that are part of the user interface are rarely optimal and the pace of car design evolution is not in synch with consumer electronics. It probably makes sense for the makers to user modular assemblies that are swapped out regularly to refresh the car, but to date efforts have been lame - witness the Ford/Microsoft effort.  There is an enormous opportunity for innovation and partnership in this area, but there would be a clash of engineering culture.  Steve Jobs told Walter  Isaacson that he'd love to work in this area if he could start fresh.  My guess is he would have had to start his own car company to make it work - or at least to have complete freedom over interior design at the bare minimum. 

 

 In addition to cockpit electronics, this decade will offer fantastic opportunity for fundamental change as the nature of power trains is shifting for the first time in nearly a century.  At this point pure electric cars are basically a platform that is closely related to a conventional car powered by an internal combustion engine - largely because radical change would be too expensive and volumes are currently to small to allow creativity.  But the ability to move battery packs and motors around will allow a much richer set of possibilities.³

 

It seems likely that the US, Europe and Japan will not be the primary drivers of design. Indeed the next billion cars are likely to be destined for use in China, India and some rapidly growing third world countries.  These cars may be much smaller than what we are used to and may have less range and much lower price points.  Serious work is only in its infancy and will roll out over the next three or four decades.  All of the big manufacturers see this as perhaps the core piece of their long term strategies.  The megacity car or the post-internal combustion car...   Population growth and the need to leave fossil fuels behind will be powerful forces.

 

Of course there are some of us who believe human power and human electric hybrids have a place in the mixture.  It will be interesting to see how all of this evolves and if it changes the notion of where people live and what mobility means as much as the automobile did.  

 

So I'll be using a mixture of human power and an IC/manual transmission car for awhile.  At some point the IC vehicle won't be practical for me and I'll get a pure electric.  Hopefully there will be a niche that couples me to the act of driving as well as a stick shift and a clutch does now.⁴

 

_______

¹ Government figures show automobiles came in at 9.0%, but light trucks were much less and manuals are non-existent in SUVs and minivans.  In Europe the numbers are nearly reversed and automatics are comparatively rare.

² Engine efficiency per unit of displacement, power to weight ratio, fuel consumption per unit of power, and overall power train efficiency.  Where they have fallen down is in overall efficiency for moving people and payloads.  There have been improvements, but nothing like the other improvements suggest - it turns out cars and trucks have also become much heavier and this erases many of the technological gains.

³ Designers tell me that the potential for a much greater life of an electric car and the ability to have greater design freedom may mean you'll buy a car for 15 or 20 years and remodel it when you need to along the way.  Some of the concepts are very clever indeed.

⁴ The highly peaked power curve of an IC engine and its torque characteristics demand the use of a multispeed transmission.  People on bikes have similarly narrow characteristics.  Electric motors have much broader ranges where they are efficient and usely can get by with a single speed gearbox.

 

bicycles are not made of cheese and frosting

The real reason for Thanksgiving is the gathering of friends and family in gratitude for the food and company.  It may be the purest and most perfect holiday we have.

 

Last Monday evening we were lucky enough to attend a fine Thanksgiving dinner with friends in the city.  The food was excellent  and the careful orchestration in the kitchen and service caused my mind to wander...

 

What if food could be used to describe 3d printing?

 

Presentations could be mind-bogglingly complicated and would be perfect down to the millimeter.  Chefs could describe any presentation they could imagine, exchange their presentations with other chefs around the world possibly making a bit of money on the side.  Home cooks could have machines in their kitchens and reproduce anything they could afford.  The amount of preparation time would be well-known making it easier to bring all of the dishes in a dinner together at the same time.  Experimenters could create designs on their computers and know how they would turn out in advance.  Neat, but there is a nagging problem ...

 

Materials

 

 
Three dimension printers can only deal with a limited palate of materials.  To make the analogy with the current world of 3d printers imagine machines where one model could only deal with a four varieties of cheese.  It might be very good with cheddar and less good with three other hard varieties, but no other cheese would work with it.  Another printer might be very good working with ice while still another model does a fine job with frosting.  The cheese model allows you to print with colored cheeses and that feature will be added to the frosting model sometime next quarter.  There is a home model planned based on a hobbyist kit that can print spam (the stuff from cans rather than the email variety).  

 

Materials and how they are treated during manufacture are really important!¹

 

Most of us don’t think about it much, but consider the thousands of fabrics used in clothing, the complexity of materials that go into food, the hundreds of different woods used in building, the thousands of materials at an architect’s command when a building is designed, the exotic materials of a turbine blade in a jet engine or that make up one of the billions of transistors in your phone.  

 

A good designer needs a fundamental understanding of the materials used in their domain.  

 

3d printing is still very primitive.  It has enormous value fabricating prototypes, but there isn’t much flexibility in material choice.  It is like living in a culinary world limited to spam, icing, hard cheese and ice. 

 

Some disciplines push material use and design more than others.  Bicycles and airplanes are among the most interesting as the final vehicles should be strong and lightweight.  The bicycle is particularly fascinating as it is both elegant and simple.  It is still possible for a single designer to make most of the choices alone rather than relying on a large design and engineering team.  Airplanes were once that simple - when two bicycle mechanics from Ohio created the revolution, but the sport now requires large teams.

 

At about three to four times as efficient as walking, bicycles are the most efficient way to move a person using only muscle power.   There isn’t a lot of material, but a frame can be very strong often support fifty or more times its weight.  The shock of uneven roads is absorbed without losing much efficiency while the bike is rigid at the same time.  The drive train can transmit as much as 98% of the power you supply to the wheel and simple brakes that weigh a few ounces can dissipate the energy of a racer flying downhill at 50 miles per hour in a few seconds.  All of this is done by a machine that can be beautiful.  A bike is one of those places where art and technology come together.  

 

Twenty years ago most bicycles had steel frames. The type of steel and construction technique used were a function of the cost and purpose of the bike and high end bikes like my old Raleigh Professional were made of hand brazed exotic steels like Reynolds 531.  This made them light for the day and imparted a certain signature quality to the ride of the bike.

 

About then bikes with aluminum frames started becoming practical.  Aluminum has very different characteristics than steel and was difficult to manufacture, but Trek and others started using hydroforming techniques and prices dropped.  The bikes could be lighter than equivalent steel frames, but the ride quality was different.  Some of us preferred high quality steel, but there was a choice.

 

China builds an enormous number of bikes and began to export them.  A decision was made that bike technology is strategic for China.  Learning how to make aluminum bikes was seen as a way for China to dominate the international bike market as well as learn how to world with leading edge aluminum fabrication techniques.  What works for bikes also works for airplanes and thousands of other products.  

 

The first Chinese models were awful. but they stuck with it through partnerships and got to the point where they pioneered some techniques of their own.  in large quantities a low end bike frame from China now sells for about $4 - probably less than the cost of the electricity used to make the aluminum in many parts of the world.  China may be losing on the frames, but they own the market and make enough other high value parts to end up winning.

 

Now they are working on reducing the price of carbon fiber.  

 

Carbon fiber is found in most racing bikes.  I’m not fond of its ride quality, but it is stiff and very light.  It is possible to make a frame that weighs under a pound with a final, albeit exotic, bike tipping the scale at under ten pounds.  The Union Cycliste Internationale responded to this by requiring bikes to weigh at least 6.8 kilograms - about 15 pounds.   The bikes are lighter, but carry some electronics now.  Racing is an arms race in many dimensions.

 

Good carbon fiber frames are difficult to make with a large amount of labor and high material prices.  A few years ago a reasonable frame might start at three thousand dollars and unless your name was Lance Armstrong, fabricating one to exact dimensions was usually not practical. Now the price has fallen by about an order of magnitude as, once again, China has determined that carbon fiber fabrication techniques are a strategic need and a good are to gain focus and experience is the bicycle industry.  Such techniques have great application in aviation, wind turbines and will become common in the automobile industry during this decade.  it is possible the scale of the carbon fiber industry dwarfs the automotive battery industry.

 

But back to the basic bicycle.  It is possible to make bikes from other materials.  Titanium has desirable characteristics, but is expensive and difficult to work with.  Bamboo can be used by itself or as a composite material and can create beautiful bikes with a ride quality some people find wonderful.  And cranky purists like Dave and I have a variety of steels and techniques to chose from.

 

About two years ago a special bike was created for Colleen - a good friend who happens to be approximately tall.  Her body happened to beyond the limits of the CAD tools common to bike design and a careful bespoke design was called for.  It was decided to base the design on an upright commuting bike rather than a racer and ride quality was important.  The issue was the frame was so tall that to get the right shock absorption, rigidity and “feel” the steel tubes used had to be carefully tailored.  The project required carefully measuring her with the precision normally reserved for racers.  This made sense as it was an entirely new design.  The steel was Reynolds 953, an exotic stainless steel requiring unique manufacturing techniques.  There is only one company on the planet that can make this stuff The wall thickness on the tubing is very thin - less than a half mm and as little as about a third of a mm - with tubes being carefully matched to produce the right ride quality.  

 

Jony Ive at Apple showed the world that consumer products can become art when mixed with great design and an informed materials choice.  My MacBook Pro has a body machined (!) from a single piece of aluminum stock.  An expensive process, but the quality shows and laptops from other manufacturers feel like toys to me even before the machine comes to life.

 

We live in a material world that is not made of hard cheese, frosting, ice and spam.

 

As it is Thanksgiving eve it makes sense to finish with a recipe that celebrates the enormous amount of material and fabrication choice available to the cook along with a bit of simplicity.  

 

On Monday we had some excellent roasted asparagus as part of an early Thanksgiving dinner in the city with friends.   There were asparagus dreams that night andI  had to make some on Tuesday.

 

I usually make asparagus when it is in season, but roasting gives you a bit of latitude and if you are willing to sacrifice a bit of the produce, you can still get great results.  I bought about a pound of ok looking stalks yesterday and fired up the oven.  The warning is the measurements are guesses as I cook mostly by throwing things together.  (baking is where I’m careful)

 

• Pre-heat oven to 400° F

• Snap off the bottoms of the stems at the natural breaking place (I used these for asparagus soup later, so they aren't wasted).  These weren't great asparagus, so I peeled the bottom inch or so 

• Put on a baking sheet, drizzle on some olive oil.  Add a bit of a good sea salt (Maldon for example), grind some pepper and toss.  Now arrange on the pan so the stalks aren't touching.

• Bake for 20 minutes  and plate

• Melt a half ounce of butter and add 3 or 4 chopped cloves of garlic.  Heat until about the time the butter starts to brown.  Drizzle/spoon or whatever over the asparagus  (note - the heart healthy way would be to use olive oil here.  I strongly prefer butter and rationalize it with the idea that each of us is only getting a quarter ounce.

• Sprinkle with chopped pecans or walnuts  

• Squeeze a bit of lemon juice on top or a drizzle good aged balsamic vinegar.

 

Happy Thanksgiving!

 

_____________

¹ I note that 2d printing is just cutting flat objects and technologies abound that allow a great choice of materials.  We will see, and are seeing, dramatic change here well before 3d printing becomes important outside of a few niche markets like prototyping.

 

clothing and mass customization

For an early post I'll touch on something that is a nice intersection of technology and sociology and anthropology...

 

fashion

No one has successfully accused me of being fashionable, but I'm interested in apparel and fashion for a variety of reasons.  

In the mid 90s I found myself consulting for an Army project designed to lead to better fitting uniforms.  It turns out sizing is an artifact of mass production that is anything but perfect.  It would be nice to create individual made to measure uniforms, but to mass produce them. Our approach was to build a fast body scanner that created a 3d model of a person that was then used to create a custom two dimension digital pattern.  This pattern would be used to drive automated cutting machines and the result would be sewed together (the last step probably by people). 

The military scanned a large number of soldiers and created an excellent anthropometric database.  For a variety of reasons the idea of using scanners for measurement never caught on, but several firms have experimented with the idea of practical paths to semi mass produced made to measure clothing.

I mostly forgot about this work until a few years ago when a very tall young woman became a friend of the family.  Colleen's height is something more than five standard deviations out from the female mean*.  The practical result is many manufactured items that need to fit don't.  Watching her frustration with mass produced items started me down the path of thinking about the problem again.  In another blog I discuss the general problem in two posts.

There are several interesting possible apparel industry disruptions.  Not only is better fit possible along with some customization, but the idea of what a design is becomes more fluid.  A curious thing about practical design is that copyright doesn't apply - at least not in the US and functionally not in the EU.  One reason the fashion industry has survived in its current form is fashion is dynamic and the time to manufacture has traditionally been long.  This is changing and there is some worry in the industry - shades of another industry that obsesses about copyright.

More important is the idea that design might be distributed and design mashups are possible.  This begins with customization, but it can go much beyond that.  Few of us are Jean Paul Gaultiers, but there are thousands of great street fashion designers who can modify existing designs into something that you might find more to your taste.  At some level you can also participate - perhaps choosing to customize from a safe palate of choices.

This may not happen with traditional labels, although there are models where they might participate.  It also has the ability to break the current byzantine and inefficient distribution and sales model.  You might select a pattern from a great designer and modify it yourself or use the modifications of others.  It may be possible for you to specify materials as well as the manufacturer and the resulting garment would be shipped directly to you.

The technology and business models haven't emerged yet, but the industry is huge (larger than entertainment and sports) and massive change seems likely as there is need for something better - perhaps starting with fit.

Currently the social component of fashion and design fascinates me.  This wasn't on my screen fifteen years ago, but it may be a significant driver of the disruptions that may confront the industry.  Those with serious platforms - Amazon, Facebook, Apple, Google and a few others - may be in an interesting position to participate.  It is also possible that new players might emerge.

These days many people are interested in 3d printing.  A very neat technology  that has completely changed prototyping,  but there are serious scaling and material issues for final products that move this well beyond the current time horizon.  I'll have more to say on mass customization in the future and particularly the low hanging fruit of apparel - I generally refer to mass customization of apparel as two plus epsilon dimension mass customization.   stay tuned.

 

I spend a lot of time talking with and observing people in their twenties - a time of life when there is considerable experimentation and it happens to be the group where change often emerges.  A friend I consult with is Jheri - another difficult to fit young woman in her mid twenties who lives in Copenhagen. Our conversations caused her to write a piece on the history of sizing and she has graciously allowed me to quote it here.  It was originally targeted at difficult to fit women to provide a bit of background on why the current fashion industry can't help, but it also shows why even average people can have a difficult time with fit.

so thank you Jheri!

I’m curious about fashion and have spent a bit of time studying the industry. There are basically three ways clothing is made: ready to wear, made to measure and bespoke.  

 

Most clothing is ready to wear.  A design is made and turned into a pattern which can be made larger and smaller.  It is up the the manufacturer to define the "fit" to a given size.  In fact the concept of size is quite arbitrary.  In North America the standard size, which turns out to be very different from the average size, is a size 8.  The designs are made into patterns that will fit a size 8 fit model very well and then they are adjusted incrementally up and down to make other sizes.  Adjustments are made for talls and petites and a different type of sizing usually happens around size 14 or so.  A brand may have different lines each with their own fit.  So even if you are a perfect size 8 fit model for one line another line from the same manufacturer may have a very different fit. 

 

The problem is women are curvy and the shape of our curves vary a lot.  For me I am very flat on top, but I have a nice curve from my waist to my hips.  The length and shape of that curve can vary a lot as well as how big my butt is or what shape it is and a hundred other things.  We don't need perfectly fitting clothing, but the specification of what a size 8 is only gets you in the ballpark.

 

Average sized women can shop and try on different pieces of clothing until something fits.  Even there they don't get exactly what they might like.  A friend of mine is an 8 and maybe about 5'5, which is average height for a woman. She tells me about half of the clothing she tries are terrible fits and she knows what lines are likely to fit her shape.  But there are so many clothes that she can usually find something she likes and sometimes there are good sales.  She will tell you shopping can be a bit frustrating, but I think she is in paradise.

 

If you are a manufacturer you want to be able to sell everything you make.  This means you need a good design - that is up to the designer, but you need to know the distribution of women's sizes.  You don't want to make as many size 0s as size 8s or 10s.  If 20% of your customers are size 8 you should make 20% of your line in size 8.  If 5% are size 4, you make 5% in size 4 and so on.   At some point it doesn't make sense even stocking sizes where there will be few customers  On the small side you rarely see many size 0s or 2s in stores.  It gets worse when you add tall sizes and that usually means something like 5'8 to many manufacturers.

 

You have probably seen the bell shaped curve in your math class.  These are often called gaussian distributions and many things in nature, including human height, come very close to following them.  I won't go into the math, but we are at the right end of the scale.  Depending on how you count, only about 1 in 20,000 women are my height or taller in the US. This sort of makes sense when I travel around.  I have two very tall friends.  One is 6'7 and the other 6'8 and at that height we are talking about one in more than a million.  It does not make financial sense making manufactured clothing even at my height. But you may be able to mix and match pieces that fit here and there.  My 6’7 friend can get by with short sleeved tshirts and jeans are ok as long as she wears them as cutoffs.  But dresses, nice slacks, and jeans that really fit simply don’t exist in stores.

 

But it gets worse.

 

Most women in America are a bit on the heavy side.  About two thirds are considered overweight and a third obese.  This means that plus sizes are the common size.  Why is it so many women who are size 14 and up can't find clothing that fits if the average woman requires these sizes?  A lot of women blame the manufacturers and models for an image that is super thin and unrealistic.  This isn't true.  There are marketing reasons why models are tall and thin, but not reason for a manufacturer to leave money on the table by not offering clothing to a large number of women who want to hand over cash for something pretty.

 

The problem goes back to how clothing is made.  I mentioned ready to wear.  That started to become popular in the early 20th century.  It became very cheap to mass produce large number of clothes in sizes that would sort of fit the majority of women.  Before then you had to sew or use the town dress maker and clothing, for most women, was expensive and lacked style.  Few dressmakers had the skill to make beautiful togs.  With ready to wear quality improved and the price dropped.  It caught on. 

 

Before mass produced and sized clothing caught on there were two basic ways to make it.  Bespoke and made to measure.  Bespoke is what it sounds like - you "speak" for a piece of fabric and, and in the hands of an expert,  a piece is made to fit perfectly.  There is no pattern and no concept of sizing.  It is still done today and in the hands of masters is the summit of clothing.  It is also mostly reserved for mens suits; suits that start around $5,000.  There are a few who will craft women’s clothing using this technique.

 

Made to measure is where the fabric and design is known but your measurements are taken.  The fit is much closer than ready to wear, but a pattern is involved.  If you are wealthy and go to a designer's Salon you can often have designs made to your measurements.  It usually takes a month or two of waiting along with a boatload of money.  Much more than I have.  

 

For ready to wear it was necessary to find a way to scale patterns.  Remember our size 8 fit model?  For hundreds of years - going back to the 16th century in Italy it had been known that women of average weight varied in a predictable way.  If a certain amount came off the bust, there was another amount that came off the hips and still another than came off the waist.   As long as these jumps were not huge you could make patterns, which can be very complex and hard to make, that would scale easily.  In made to measure adjustments could easily be made and you may have done this yourself if you have done dressmaking from patterns.  Additionally there are other parts of a design that can be altered, usually by a tailor, for the final fit.  But we are still waiting for something that fits heavier women, right?

 

The first real attempt for mass produced larger sized clothing was pioneered by Lenna Bryant in about 1910.  She had a dress shop and a client asked her for a maternity dress that could be worn on the street.  It was a big success, but it was considered unfashionable and no one would run ads.  Bryant and her husband decided this was a real business and started a mail order catalog.  Mail order was becoming popular sort of like the Internet of its day. It turned out many, if not most, of their customers were full-figured rather than pregnant.

 

Bryant's husband was an engineer who designed railroads.  He knew how to design curves in the tracks and used this knowledge with measurement statistics from thousands of life insurance records to define three or four basic full figured body types.  This made it easier to get a dress that mostly fit a woman of average stature and that was in a day when women could sew and alter or lived close to someone who could.  It was a pretty big advance. The Bryants refined their designs and became millionaires when a million was a huge amount of money.

 

But there is another problem.  Why aren't we just scaling from size 8?  It turns out the fuller figured women have very different curves and the range of those curves varies hugely.  There may be a size 16 fit model, but rather than fitting half of size 16s, she might only cover 10%.  The concept of size doesn't work anymore because the curves themselves are not being properly fitted. There is far too much variation.  Plus sized manufacturers try to do as best they can and try to find ranges of fit models that give them a better chance at making money, but for the majority of plus sized women the sad truth is they can't get the off the rack fit that their smaller sized sisters get.  The manufacturers would love to solve this as they could mint money.  There are even some good ways to do it, but that would be getting ahead of ourselves.

 

Of course none of us are of average height.  We have curves, some of us more than others.  But we also have legs, arms and torsos that tend to be longer than what "tall" sizes account for.  We have a few speciality manufacturers.  They are faced with the same problem a regular manufacturer has but it is worse. Not only do they have to guess the right size range if they want to sell enough to stay in business, but they have to know what percentage are.  5'10 to 6'0, 6'0-6'2 and so on...  You get a sheet of paper and draw a horizontal line with that bell shaped curve for size distribution. Depending on your line maybe most are size 6 through 12s.  Now you draw a vertical line and for each one of those sizes you try to figure out how many will be in each height range.  The space for 6'3 and size 0-2 has such a small number in it that you decide not to make anything specifically for that woman.  At 5'10 and size 8 or 10 there may be a large enough number that you want to have several designs.  For the fuller figured women it is very tricky as now as the type of curve becomes a more important variable.  And there is long and short waisted, etc. This is really really difficult and risky business.  You will please a few customers, but many won't have great fits and there can't be a big selection.

 

Ok, so far I've mostly given problems and things may sound hopeless. But they aren't.  Most of us manage by putting together something even if it doesn't fit great or have the fabrics and designs we would love to have.  If one of the speciality companies has a line that fits, support them!  None of these places makes a lot of money as the costs are high for low volumes. We have to acknowledge that clothing is going to be more expensive, so buy it and take good care of it.  If the fit isn't fabulous get to know a good tailor.  I have a great one.  She is about 70 and I bribe her with cake and chocolate even if I don't have something for her to work on.  Be creative and look at vintage and used clothing and work with your tailor.  Sometimes you can put together very interesting outfits for not a lot of money and it is much more your style.  Perhaps you can find a dressmaker or even learn yourself. One of my extra tall friends was complementing her neighbor on a beautiful dress and learned it was homemade.  Now that neighbor is making a made to measure dress for her and is making a little money on the side too.

 

Made to measure is the holy grail assuming the patterns can be matched well enough.  It is beginning to happen.  You may have seen shops that take a lot of measurements and turn them into clothing.  Indi is one that does semi-custom jeans.  What they do is create a pattern in the computer that is unique to you and then that information is used to run a machine that cuts the fabric.  People sew it together and if that is done right and the measurements are good, it should fit well.  These places offer a bit of customization.  Now your height and shape are much less important.  It is expensive now, after all it is competing against designs made in the third world by women who only made a few dollars per day at most.  

 

This is a developing area and there are problems, but it will probably get much better.  In the mean time develop an eye for style, find a good tailor or dressmaker and learn a bit of sewing.  

______________

* 5.6 sd is very rare and a good illustration that the distribution of human height is only approximated by a gaussian.  Specifically it implies that she is taller than all by one in about ninety three million women in this age range.  Looking at women's basketball teams this clearly isn't the case and a more realistic number based on other work is probably between three and ten million women with six million being a good guess.

A male her height is much more common at 3.6 standard deviations from the mean - about one in sixty three hundred men in this age range.  The male curve is only slightly non-gasssian that far out if you use the largest data sets.  

It turns out most of the error in the tails in induced by sampling.  The study of rare events demands extra care with normal techniques and tools often breaking down unless unless you have a good fundamental understanding of what is going on under the hood.

If you want to probe human height more deeply an excellent non-technical article appeared in The New Yorker in 2004 - The Height Gap by Burkhard Bilger (fortunately not behind their paywall).  Having a tall friend opened my curiosity on the subject and I have spoken with some of the researchers mentioned and then branched out a bit.  It turns out the height of a population (not individual height!) is a good proxy for health.  In the US female height has been falling in the past few decades and male height is relatively stagnant.