Month: May 2012

The Feeling of Space

We have a new website.  If you take a look at it, you will see metal and cloth.

But it is what you will feel when you see it that I want to talk about.

About fifteen years ago, we brought home a computer that could run Windows 3.1, and that was the first time I came into contact with the ‘desktop metaphor’.  It was on that day that the desktop became a big part of my life.  (To all the Mac fans out there, I’m sorry, but I never saw a Mac in my life till I landed in North Carolina in the year 2000).

I spend most of my time from 9 to 5 looking at a two dimensional space decorated with the tools and artifacts of my trade (software).  These tools and the desktop are now so deeply burnt into my consciousness that I believe the slider in my screen to be real.  Any time I see a slider or a button, my mind believes that it can be touched and manipulated, even if it is only a projection on a wall.  I might not believe that you (dear reader) exist.  But I have no doubts whatsoever about the existence of the slider on the edge and I know it will scroll just so.

This is not an experience of reality that is uniquely mine.  Anyone who spends all their waking hours before a computer screen would be just too familiar with the feeling.

One day, I started working on building a three dimensional user interface based on the ‘real’ world.  I designed it with my files stored in a folder in a cabinet against the back wall of the room, and planned to add in user avatars that would walk across and talk to you.  I gave up a few years later.  I just wasn’t good enough at graphics to be able to build such a UI.  I remember having read somewhere that either Apple or MS did build such a system, but that it didn’t take off.  Virtual worlds that seamlessly incorporate your desktop are difficult to do.  The closest I have seen someone come to it is Coverflow, up until now.

The Wikipedia has this to say about Coverflow: ‘Cover Flow is an animated, three dimensional graphical user interface integrated within iTunes, the Macintosh Finder, and other Apple Inc.products for visually flipping through snapshots of documents, website bookmarks, album artwork, or photographs.

Cover Flow was conceived by artist Andrew Coulter Enright and originally implemented by an independent Macintosh developer, Jonathan del Strother. Enright later named the interaction style fliptych to distinguish it from the particular Cover Flow implementation.

Cover Flow was purchased by Apple Inc. in 2006, and its technology was integrated into its music application, iTunes 7.0, which was released September 12, 2006.  The name was previously “CoverFlow” without a space.’

Our website now tries to create a 3D UI metaphor.

If you take a look at our website, you will see metal and cloth.

You will see a thick edge of heavy brushed metal.  It is a protective metal enclosure.

The cloth is inside the metal enclosure, but you can touch it.

Scroll to the bottom of the scroll of cloth and click on “Publications”.

A translucent and crunchy acrylic card pops up with information for you on it.

That is all that it takes to give you the feeling of space.

Because now you are looking into a box.  A box contains 3D space.

It is a feeling of space that we programmers probably all yearn for with or without our knowing.

Significance

Metal and cloth both indicate passion.  Steel is associated with bravery, strength and conquest (with a soldier).  A ‘man of the cloth’ on the other hand is a man who embarks on the path of religion, who goes on a quest to understand how and why we exist, and what is right and wrong.  AI is in many ways nothing but a quest to build machines that can learn and think (that can live).  It is a quest to conquer knowledge and create life.

Acknowledgement

The craftsman is a gentleman by the name of Venkatesh.  The thing that struck us all about Venkatesh’s work was how easily he could communicate a sense of brooding moodiness through a design.  That touched a chord.  I mean, India is all about a certain moodiness.  Look around you.  One moment it rains.  The next, it is dry.  The cycle of seasons goes from the burning yellow heat of summer in March to the grey thunder and rain of June to the cool blue silence of winter in November, and then back to summer again, if all goes well.  The designer put in the grey of the Indian monsoon, the shocking yellow of the summer (and the red gulmohur trees), and some winter blue (just a little).

Visit the Aiaioo Labs website to see a lot of Indian grey, some slices of yellow, and those healing touches of blue.

Postscript:  A friend pointed out that this style of design might be said to tend towards Skeuomorphism (and there’s a lot of debate about it) http://tobiasahlin.com/blog/skeumorphism-and-storytelling/.  Skeuomorphism is relevant to what we’ve done and it was fun reading about it.  However, there’s more to this design than the facade, lines, colours, and semblances.  What we’re doing is creating more dimensions that a 2D screen really has, and capturing a mood in the process.

Advertisements

Machine Learning and Intuition

What is intuition?  The Wikipedia entry says “Intuition is the ability to acquire knowledge without inference or the use of reason.”

Intuition is something that enables you act without using inference (or knowledge gained from inference) to help you decide upon the right course of action.

So, an intuitive course of action is determined by your feelings – those proverbial feelings in the pit of one’s stomach – and not by conscious reasoning.

However, intuition is not the only tool that decisions come to be based on.

Instead of relying on intuition, a human being can base his or her decisions on logic.

Logic-based (or you might call it rule-based) decision-making is a method of proceeding from a number of premises and eventually arriving at a decision/conclusion (or a course of action) through inference.

So, there seems to be a dichotomy in decision making:  intuitive and logic-based.

In AI, there seems to be a similar dichotomy.  One area of research in AI deals with planners and inference engines which are clearly logic/rule based.

Then there is an area of research called Machine Learning (ML) which deals with the study of systems that can be trained to recognize patterns.

You first need to train ML systems on examples of decisions made under certain circumstances and the rewards obtained as a result.  After sufficient training, the ML systems ought to by themselves be able to take decisions when presented with a set of circumstances, even if it is a set of circumstances they have never encountered before.

Some ML systems are rule-based.  Decision tree classifiers are a good example.

But there are some ML systems that seem to have a method of operation that resembles intuition.

One example is ANNs (Artificial Neural Networks).

Another is a family of ML systems that are statistical in nature and are known as Discriminative Models.

There are also systems that are a combination of both.  They are called Probabilistic Graphical Models.  These seem to use what closely resembles a combination of intuition and inference.

Kashgar to Merovingian France

In a previous post, I talked about a group of people who might have migrated from Parthia (North East Iran) all the way to Southern Vietnam, leaving cultural and linguistic evidence of their passage all the way.

Today, I am going to talk about another group of people whose story I find extremely interesting.  This migration had a lot to do with a very famous King of India – Kanishka.

We all learn in our history textbooks that Kanishka was a Kushan King.  But very few of us in India know the origin of the word Kushan.  It comes from the word Guishuang (貴霜) in Chinese.

The Guishuang (Kushan) were one of the five tribes of the Yuezhi.  Sometimes called the Da Yuezhi (大月氏) or Big Yuezhi.  The Yuezhi were a tribe of people who lived near Kashgar in China (called Kashi in Chinese – Kashgar in Chinese is Kāshí (喀什) which is a shortened form of Kāshígé’ěr (喀什噶爾)).

They came into conflict with a nomadic tribe who lived to the north of China called the Xiongnu 匈奴 (supposedly also pronounced Hunnu – and therefore possibly the Huns – the etymology is interesting – apparently the first character 匈 was once also pronounced ‘hong’), and were forced first West and then South.

In their first migration to India a few hundred years after Alexander, they passed through Scythia, which is the name of a vast expanse of land north of Parthia (Northern Iran) and came to be called Scythians (Saka) in Persia and India.  The origin of the word Saka itself is thought to be related to an early Iranian word for nomad.  From the Wikipedia article on Scythians: ‘Saka, on the other hand, Szemerényi relates to an Iranian verbal root, sak-, “go, roam”.’

By the time of the second Yuezhi migration, at around 100 AD, the Kushan tribe had became the most powerful tribe in the Yuezhi confederation and had come to dominate it.

Apparently, the Kushans continued to be referred to as Da Yuezhi in China hundreds of years later, and I quote a line from a book called the Sanguozhi (三國志, Chap. 3) “The king of the Da Yuezhi, Bodiao 波調 (Vasudeva I) [Vasudeva became Bazodeo in the Kushan language], sent his envoy to present tribute, and His Majesty (Emperor Cao Rui) granted him the title of King of the Da Yuezhi Intimate with the Wei (魏) (Ch: 親魏大月氏王, Qīn Wèi Dà Yuèzhī Wáng).”

Quick digression:  There is a story about this Kushan King having returned the remains of Thomas (one of the apostles) from where he was buried in South India to Edessa.  Any evidence of a connection is based on guesses about nomenclature and is very shaky.  But if it is true, it would explain (to me) why I saw no body whatsoever when I took a peek into the crypt supposed to contain the relics of St. Thomas (I happened to visit the Mount church when it was closed for restoration and no one was around).

Many ethnic groups consider themselves as descended from Scythian tribes like the Yuezhi.  They include the Pashtun in Afghanistan, the Jats of India, and possibly some people from Sindh (there was a Scythian tribe called the Sindi that came from a city on the Black Sea called Sinda – but whether they had anything to do with Sindh no one really knows).

But one of them is a tribe very far from Iran and India.  It is the French.  There is one story that the Franks were in some way related to the Scythians.  Here is what the Wikipedia has to say about it:  ‘The Carolingian kings of the Franks traced Merovingian ancestry to the Germanic tribe of the SicambriGregory of Tours documents in his History of the Franks that when Clovis was baptised, he was referred to as a Sicamber with the words “Mitis depone colla, Sicamber, adora quod incendisti, incendi quod adorasti.”‘. The Chronicle of Fredegar in turn reveals that the Franks believed the Sicambri to be a tribe of Scythian or Cimmerian descent, who had changed their name to Franks in honour of their chieftain Franco in 11 BC.’

Tulu to Kusunda

The young engineer who joined our firm a few months ago is from the coast, and one of the languages she speaks is Tulu.  One day, on the way to a business meeting, she taught me a few sentences of Tulu.

The way I like to learn languages is this.  I choose simple sentences in the language and then create variations of these sentences till an understanding of the grammar begins to emerge.  So I asked her, “How do you say, ‘What is your name’ in Tulu?”  Her answer amazed me.  In Tulu, you say “Irna podar dada?”

The answer was amazing because it contained function words that I’ve never come across before.  In Kannada, you would say something like: ‘Ninna hesaru enu?’  In Hindi, it would be “Tumhara naam kya hai?”  Tulu is the first language I’ve encountered in South India that has function words that don’t resemble those in Kannada or Hindi.

But it turns out that there are more interesting and unique languages in the subcontinent, and that some are disappearing.  On the BBC website today is an article: ‘Nepal’s mystery language on the verge of extinction‘ that talks about a language called Kusunda in Nepal.  I quote:

The unknown origins and mysterious sentence structures of Kusunda have long baffled linguists.   … Professor Pokharel describes Kusunda as a “language isolate”, not related to any common language of the world. “There are about 20 language families in the world,” he said, “among them are the Indo-European, Sino-Tibetan and Austro-Asiatic group of languages.  Kusunda stands out because it is not phonologically, morphologically, syntactically and lexically related to any other languages of the world.”

It turns out that there are only two speakers of the Kusunda language left, one of them a very old woman named Gyani Maiya Sen and the other a woman named Kamala Khatri who had ‘left the country in search of a job’.

Parthia to Panduranga

Subtitle:  Alternative History

I am writing today about the story of one thread of migrants who settled in India two thousand years ago and tied together a number of countries, cultures and peoples.

Three hundred years after Alexander, a group of people from Parthia in Central Asia (Northern Iran) seem to have, in a series of migrations, settled in India between Sindh and Kashmir, and on the Konkan coast.  Upon entering India in 100 AD, they established a kingdom extending from Sindh to Kashmir.  Historians call this group of Parthians the Indo-Parthians.

Two hundred years later, they were forced south by another group of migrants called the Indo-Scythians (the Kushans) who then came to rule the land between Sindh and Kashmir.

The Kushans are thought to have come from the area around Kashgar in Xinjiang province in China.  One of the names for Kashgar in Chinese is Kāshí (喀什) which is a shortened form of Kāshígé’ěr (喀什噶爾).  The Kushans were also known by the Chinese name Yuezhi (月氏).

What happened to the Indo-Parthians after they were driven South?  Inscriptions indicate that they went South all the way to the Konkan between 100 AD and 300 AD.

In 400 AD, the Pallava dynasty came to power in Tamil Nadu.  It is thought by some that the Pallava Kings were Indo-Parthians (assuming the name Pallava was a corruption of Pahlava meaning Parthian).  What is interesting is that the first inscriptions of the Pallava Kings were in Sanskrit and Pali, and it was only later in the Pallava period that their inscriptions changed to Tamil.

What is also very interesting is that their influence went East possibly all the way to Vietnam.  The Pallava Kings after 275 AD had names like Simhavarman and Mahendravarman, and the ‘varman’ suffix in their nomenclature was one that no other South Indian dynasty used.

When I visited Ayutthaya (the city of Siam in Thailand), I met an amateur historian who told me that the Khmer Kings were influenced by the Pallavas.  The Khmer Kings of Cambodia had names like Jayavarman (the name means Victorious Shield).

To celebrate the independence of Khambuja (Cambodia) from Java, Jayavarman 2nd had a ritual performed on Mahendraparvata, now known as Phnom Kulen, declaring him Kamraten jagad ta Raja in Cambodian.

From the Wikipedia: ‘An inscription from the Sdok Kak Thom temple recounts that on the top of the Kulen Hills, Jayavarman instructed a Brahman priest named Hiranhadama to conduct a religious ritual which placed him as a chakravartin, universal monarch.’

The Angkor Wat was built by someone known as Suryavarman (the name means Sun’s Shield).

Recently, I read an account by a mariner on Zheng He’s expedition to the southern seas.  He mentions that the King of Thailand looked South Indian.  He also talks of a stop at a city called Panduranga in a country called Champa (Champa was the name of South Vietnam, and Champa remained a distinct political entity till the mid nineteenth century).

I looked up the Kings of Champa.  Their names were Manorathavarman, Devavarman, Vijayavarman.  But could the suffix ‘varman’ have been transferred from India to Vietnam, and did the Pallavas have anything to do with it?

Two of the earliest known Kingdoms in South East Asia were Tarumanagara and Salakanagara in Java. They came into existence at around the same time as Champa in Vietman.

According to their respective legends, these three kingdoms (Tarumanagara, Salakanagara and Champa) were established by Indian traders or priests who arrived and married local princesses.

The story of Tarumanagara is particularly detailed. A man called Jayasingawarman is said to have founded the Tarumanagara kingdom in 358 AD. Jayasingawarman hailed from the kingdom of the Salankayanas in India. He married the daughter of King Dewawarman VIII of Salakanagara in Java and established the Kingdom of Tarumanagara.

The Wikipedia has this to say about the Salankayana kingdom: ‘The Salankayanas succeeded the Andhra Ikshvaku dynasty and were vassals of the Pallava kings of southern India.’

The earliest inscriptions in Java were by the Kings of Tarumanagara. These inscriptions were in Wengi script (a Pallava period script) and in the Sanskrit language.

So, three South East Asian kingdoms had very similar origin stories, and the same royal suffix, and were established at about the same time, and in at least one case, could have had a connection to the Pallavas.

If these name suffixes had indeed come from the Pallavas, as the amateur historian in Siam informed me, and if the Pallavas were related to the Indo-Parthians, that’s a long way for the immigrants from northern Iran to have gone.

Some day, I’ll probably also tell you the story of the Yuezhi, which, trust me, is no less interesting.

Caveat Emptor:  A lot of what I have talked about in this article is based on conjecture. I read or heard all that I’ve written, but they were often no more than guesses by historians.