Another day in the office with AI

<Technical post, probably belongs in linkedin. Posting here regardless> 

It's a new world of coding with AI. I have been pair-programming with ChatGPT (and to a lesser extent Google Gemini) for a while now. My emotions run the full gamut and it is quite a roller-coaster :-). One minute I am full of awe and the next minute I am like "I can't believe you're that stupid". I will describe my experience with a couple of projects one big and one small. If you want to TLDR the project descriptions, move to the bottom to the Morals of the Story section.

Mobile App with Backend

I am starting on a project this year which is the first one where we are coding from scratch after the recent AI revolution. I wanted to provide the team that I am leading with some starter code for the backend (Python and the mobile app frontend (Flutter/Dart code).

• Backend data model: I had written a document describing the entities and their fields and their relationship with other entities. I fed this document to ChatGPT and asked it to generate the model class definitions for a model-first Object-Relational-Mapping approach using Flask/SQLAlchemy ORM. This saved a lot time writing code for the classes. ChatGPT did 90% of the work and I could make adjustments to get it to 100%.
• Backend data generation APIs: I gave detailed prompts asking ChatGPT to write code for a REST API to generate the data walking the object model to package all the nested objects in the object hierarchy and return the data as JSON. ChatGPT's code worked quite well here and again the 90%-10% split above worked out.
• Client App code - I had mixed results here with LLMs:
• I asked ChatGPT to write the Flutter/Dart code for the App to consume the JSON sent by the server API. I needed to persist the objects on the client on Hive local storage for offline availability. ChatGPT's first effort only persisted the top-level objects and not the objects embedded within it. When I pointed this out, ChatGPT added the additional code to persist embedded objects. But the code had one mistake that I was able to debug easily and fix. But I still ran into issues with Hive that I could not fix with additional prompts to ChatGPT. I ran into my session limit and ChatGPT asked me to come back after a few hours to continue.
• I then moved to Gemini which did slightly better in some respects but worse than ChatGPT in others. It had some silly mistakes like a missed dependency in the project dependencies YAML file, which wasted some of my time. Gemini's advice to fix the issue was not useful, I ended giving up the advice to Gemini, which it graciously seemed to accept and promised not to make the mistake After these were fixed, somehow the Hive issue tht happened with Hive did not happen and I was able to finish the work. The generated code will essentially will serve as the Proof-of-concept of how our design would work with the backend date model, APIs for generating data in JSON, and client code to consume and persist in local cache.

Text-to-Speech

I was trying to integrate Google's text-to-speech converter to get to speak Indian/Sanskrit words and had to do some coding on top to get it to work. I needed to provide custom pronunciations for the Indian words that the text-to-speech would by default mispronounce. There was some gnarly handling required when the words with custom pronunciations are in possessive form with an aphostrope (say "Durga's") and Google TTS wasn't up to the task without additional coding. And I enlisted ChatGPT's help. V1 of the code didn't quite work. I had to do some debugging on my end and I realized what might be happening and gave ChatGPT a hint; it duly used the hint and tried to fix it with V2. Still didn't work. I now just said, nope, try again. Now it seemed to figure out itself what the problem might be and produced V3. That didn't work either. I gave up on this particular thread after V5 or so.

While I could not get that last thing (handling apostrophes gracefully) working, thanks to AI, I was still able to try out 3 different Text-to-speech solutions (AWS, Google, Elevenlabs) and came up with potentially a reasonable V1 approach figured out.

Morals of the story

1. At the time of writing, LLMs like ChatGPT are good enough to generate a lot of the boiler plate code that needs to be written. This can save a lot of time.
2. Often the 80-20 rule applies; you get 80% of the lines of code from the LLM and you finish the last 20; it is 90-10 if you are lucky. And if you are really lucky, the code would be bug-free. That gets us to the next point.
3. A big caveat is that the code generated is very likely to have bugs and sometime serious gaffes and you need to be able to understand the logic completely so that you can fix the bugs to get it to work; at times, you can keep iterating (telling the AI that it is not working, try again) but you are better off finding the bugs, or at least do some partial troubleshooting so that you can frame your incremental prompts with helpful information to the AI. If you run into a situation where you are not able to figure out the fix, you may start losing the time-saving advantage of getting the boiler-plate code; but then again if you had written the code from scratch, your code may have the same (or higher) number of bugs.
4. It is super important to invest in good English documentation (for e.g. the data model) of your high level design so that you can give detailed requirements in English in your prompts. That can server your entire project to generate both server and client code. This might finally be the forcing function to get your developers to write good documents. This is a bit worrisome because in my experience often developers (especially in India) hate to or are not good at (or both) documenting their designs, requirements and algorithms. But this is going to become more important in the new world.

R. Balaji

Comic relief in Climate conversations (or more generally all left vs. right conversations)?

 We attended the Hyderabad Literary Fest which had a track on Climate Conversations and one of the best sessions was the one titled with this blog's title (not the part in parantheses). It was funny and thought provoking. 

The standup comic Raghava Mandava who specializes in bringing humor to the highly charged conversations of Climate Change talked about how he was a"nature lover" and not an "environmentalist". Because "environmentalist" has become associated with radical lefties whose cries of outrage (according to him) have only served to alienate the people who they want to convince. Or how imperfect sustainability solutions drive people away. Case in point, the turtle-killing plastic straw and its lame recent substitute, the paper straw which seems to dissolve in the drink you are sipping before you are half way done. I can relate to this personally. One of his lines was something like how a normal person who has nothing against turtles tries to drink using the paper straw and is driven to outrage ('**** the turtles'). 

The larger point is how climate change warriors perhaps should rethink how they communicate, how not come across as scolds etc. I do think the right suffers from a bit of persecution complex (e.g. the whole anti-woke thing); after all they are supposed to be the tough guys, are their feelings so delicate that they are hurt by bleeding heart liberals? Or do they have a point, has the radical left gone too far?

Another sentiment I heard in the session was how a single experience with nature can be more effective than 100 exhorting messages and lectures on how we should be reducing the carbon footprint. 

While the above was in the context of just one topic that is close to the heart of those on the left, the same arguments can be made for all hot-button cultural/political issues. After all, "the condescencion of  liberal elites" is a standard refrain you hear that mobilizes and even energizes the opposite side. 

Considing how the left got its collective backsides kicked in November (it was Inauguration week and the left is indeed licking its wounds, including yours truly), I am going to say, perhaps it is the latter. So time for some collective soul searching and how the messaging can be improved. Because while I continue to believe the left's heart is often in the right place, how they frame the conversation and the messaging is clearly due for some serious improvement.

R. Balaji

Chola Temples of Darasuram/Kumbakonam

We recently visited the Airavatesvarar Temple at Darasuram (near Kumbakonam); it is one of the three Great Living Chola Temples (as anointed by UNESCO in its World Heritage citation), along with the Brihadisvara Temples at Thanjavur and Gangai Konda Cholapuram. The Airavatesvarar temple is not as big as the other two. While it lacks in size and majesty, it more than makes up for this in the quality and richness of its sculptures and mandapas/pillars with ornate carvings. 

Raja Raja Chola and Rajendra Chola are perhaps the greatest father/son duos to ever rule anywhere in India. They took the Chola kingdom to the peak of its 400 year rule in the south both in terms of territory expansion and cultural and architectural pursuits. They built the great Shiva temples at Thanjavur and Gangai Konda Cholapuram respectively. Rajendra Chola in a reverential nod to his father, made his vimanam (the tower on top of the sanctum santorum - garba griha) a bit shorter than his father's architectural magnum opus. It was during the reign of Raja Raja II who ruled almost 150 years later, that the Airavatesvarar temple was built. The vimanam is similar in style to the bigger Chola temples, but even shorter than the one in Gangai Konda Cholapuram.

The decline of the Chola empire is said to have started in Raja Raja II's  period. Nevertheless, he has added another jewel to the extensive list of Shaivite temples of the Chola era. Airavatam refers to Indra's mount, the white elephant that lost its complexion thanks to a curse from sage Durvasa. Airavatam is said to have sought redemption from Lord Shiva in this temple who duly got Airavatam back to its pearl-white splendour. 

The Mukhamandapa also known as the Rajagambhiran Mandapam  features ornate pillars and ceilings with a truly spectacular horse-ridden chariot motif featured on either side of the steps that go up to the great hall. 

The large wheel is reminiscent of the stone chariot in Hampi's Vitthala temple. While the chariot in Hampi is monolithic, what is remarkable about the horse and the wheel here is that they are pieced together like jigsaw puzzles by carving the  rectangular stones that make up the mandapa's plinth structure behind the wheel and the horse; i.e. the horse and the wheel are not separate from the super-structure of the temple, which in its ingenuity epitomizes the creativity and skill of the Chola artisans. 

The pillars on the mandapam all feature a gaja yaali at the bottom which is a hybrid of at least 5 animals, the eyes and mane of a lion, an elephant's trunk, goat's horns, a boar's ears and a bull's tail which you can see on the back side of the pillar. The way the tail curves on itself differs in each pillar. The yaalis' trunk is curved to point in the direction of the inner sanctum leading some people to speculate that it serves as a sign-post. 

The pillars on the mandapam have carvings of scenes from Shiva Purana and Skandha Purana. As a sample, the following shows the birth of Lord Muruga or Karthikeya as 6 babies before they get fused to a single 6-faced baby god by Parvathi. You can see Lord Shiva holding 3 of the babies, while Parvathi is nursing another child. Shiva and Parvathi are surrounded by the Krittikas who helped raise the baby Karthikeya. 

On one of the other balustrades, a running elephant is depicted, with great anatomic accuracy,  capturing the legs in running stride and a swinging bell indicating the motion. The jewelry and decorations on the elephant and the afore-mentioned horse are exquisitely carved. 

On the outer walls of the inner sanctum, you can see key moments from the lives of the 63 Shaivite Nayanmars carved as miniatures. You would really need a heritage expert or a very good guide to interpret each one of these. One of the panels shows Kannappar, a hunter  known for his innocent devotion to Lord Shiva. Like every one of his ardent devotees, Shiva tests Kannappar by showing a bleeding eye on the linga. Kannappa first removes one of his eyes and puts it on the lord; and then the other eye on the linga starts bleeding. Without hesitation, Kannappar offers his second eye, this time making sure to mark the spot of the lord's eye with his foot as he is scooping out his eye (!). 

And then there is the bull and the elephant panel called the rishabha gunjara, a double entendre in stone, if you will. The head in the middle can be either that of an elephant (body on the right) or of a bull (body on the left).  

We also visited Nageswaraswamy Temple at Kumbakonam, which is considered one of the earliest Chola temples. Early Chola temples are much smaller in size and grandeur than later day temples. However they have a sculptural innovation of their own, in the form of miniature relief panels depicting scenes from the Ramayana. These panels narrate the entire story in about 50 panels. I will talk about just one sample here where a bhootha is coming out of a yagna pyre bearing the amirtha (or payasam) which Dasaratha takes and distributes to his three wives which leads to the birth of Lord Rama and his brothers. The panel is half the size of  an A4 page and still the artist packs so much detail. Prof. Sivaramakrishnan's excellent talk (part of a series on Chola Art and Architecture organized by Tamil Heritage Trust) describes this panel in great detail wih all the intricacies; it is staggering to see the level of artistic thought, such as framing of the characters, where they are looking at (Dasaratha and his wives are all looking at the urn, while the 2 rishis are looking at each other with satisfaction after the success of the yagna), the curvature of the yagna flame in the middle and how Vashista's body is aligned with this curve. You have to listen to to the talk to know more, I can barely do justice.  

Each of these temples have so much to see and a short article like this can only be a teaser at best; the best way to experience them is in person with a well-informed guide. 

- R. Balaji

Keezhadi Museum

 Around 2015, the Archaeological Survey of India (ASI) made an exciting new archaeological discovery south-east of Madurai in the Keezhadi village. What looked like urban settlements with houses built of standardized bricks and tiled roofs were found at a depth of about 2.5m. More significantly, they were dated between the 6th and 3rd century BCE! Pottery, beads, iron tools and weapons, teracotta tools and toys have been found. To put this in context, the only other urban settlement in India that is older than this is the Indus Valley (or Harappan) Civilization which dates back to the 3rd millennium BCE. 

Some of the pottery found in the site had inscriptions that seem to be Tamil Brahmi providing tantalizing circumstantial evidence that links this settlement with the Sangam poetry corpus. That is, if you are willing to take a leap of faith. Until now, the Tamil Sangam poetry itself has been dated to a much later period: 1st and 2nd century CE. In my humble opinion, the fact that there is Tamil Brahmi writing suggests the Keezhadi people were potentially Tamil speakers, but does not by itself mean their language was evolved enough to produce the poetry of the richness and sophistication of Sangam Tamil. On the other hand, I have always felt the current dating of Sangam poetry also highly suspect.  Compressing that vast corpus to 200 years seems like a rather tight fit. In any case, I am hoping findings like Keezhadi make historians revisit the dating of Sangam poetry.

There is also some controversy around the age of the pottery with inscriptions (as to which end of the 6th century BCE - 2nd century BCE time frame they belong to). The controversy is that, if they turn out to be older than the 3rd century BCE, then that pre-dates Ashokan Brahmi, the script that was used in Ashokan edicts found throughout India. However Tamil Brahmi has until now been considered an evolution of (Ashokan or Mauryan) Brahmi and not the other way around.

We cannot access the archeological sites themselves, where excavations are still going on. However an excellent museum has been put together that houses many of the artifacts and some model recreations from the Keezhadi site. We visited this museum recently and this post is inspired by that visit. 

As you enter the museum there is a small auditorium where they show a 15 minute video providing the historical context and background which is best seen before seeing the museum itself. 

Let me just give a brief background on stages of growth of a particular human culture or civilization before we see where Keezhadi fits in; human populations go though various stages in their development as a culture starting from hunter gatherers and three peiods of Stone Age, namely Paleolithic (Old), Mesolithic (Middle) and Neolithic (New) Stone Ages. In the Paleolithic period, large stone tools with  sharpened edges are usually found. In latter ages, smaller, more specialized tools and more polished stone tools are used. Some of these human cultures discover iron and/or copper and gold go through the Iron Age developing metallic tools, coins and weapons. 

As they go through these stages, humans move from a nomadic hunter gatherer life style to settled societies, with farming and agrictulture, producing food that can be saved across seasons. They start producing pots for storage. As the society becomes more sophisticated, it produces goods for leisure such as jewels and ornaments using beads and metals and toys and games. Coins from other remote civilizations can be found suggesting trade links to those societies. If they develop writing, they start making inscriptions or other signages on rocks, caves and pots. 

The Keezhadi settlement is clearly an Iron Age settlement which shows all the stages of development listed above:

• Walls made of standard sized bricks, tiles, ringed wells
• Terracotta pipes and brick-made channels for water
• Black and red pottery including Black Polished ware 
• Seals, Roman coins and arretine pottery shards suggest trades to remote civilizations
• Tamil Brahmi script in pot shards
• Many iron tools such as ladles, sickles and knives
• Beads of different kinds and small gold ornaments
• Toys like cube and cuboid dice made of ivory and terracotta

The museum includes references to Sangam Literature (especially Natrinai, Purananuru and Agananuru) that indirectly refer to some of the artifacts. The juxtaposition of Sangam poetry references next to artifacts might suggest a direct link but I found these connections quite tenuous.  For instance, next to an antimony rod that was evidently used for decorating a woman's eyelash, there is a citation of a line from a Natrinai poem that talks about a women's hazel eys with decorated eyelashes. 

On less controversial matters, I found the museum very well organized. Local design elements such as Athankudi tiles and wood work that is reminiscent of Chettinadu have been used tastefully in the multiple buildings that make up the museum. The sheer number and variety of artifacts on display for a settlement that was discovered only 10 years ago is quite remarkable. We went on a week day and the crowds were sparse; I hope the museum gets added to everyone's must-see list for Madurai. 

R. Balaji

On the New Test Cricket - part 1

 It is difficult to believe I haven't written more on cricket - the avid fan that I am, especially of the Test format. This is likely to be the first of a series of posts, so bear with me while I ramble and rant a bit on T-20 and then the pleasures of Test cricket. 

Test cricket remains the purest and most balanced battle of bat vs. ball. I realize I might antagonize T-20 fans, but I have always believed T-20 tilts the balance heavily in favor of bat over ball. When there is no fear of getting all out in a 20 over span, batsmen hit with impunity and scamper home in 20 overs with 5-6 wickets lost. So yes, lots of 4's and 6's to go wild over; but there is rampant run inflation (i.e. the value of a run, a 4 or a 6 has come down).  I wonder what will happen if they tweak T-20 to say 5 wickets down and you are all out, with 5 players who only bat and 5 who only bowl. 

The 50 overs ODI is a reasonable middle ground where the threat of getting all out remains. So I believe predictions of the demise of the ODI format are premature, but I could very well be proven wrong on this one, given the huge success of IPL (I recently learnt it is one of the top 10 leagues in the world across all sports -  measured in terms of the money in it). It remains to be seen whether 3 formats can survive, given the demands it puts on players and the cricketing nations to squeeze in local T-20 leagues, along with international competition.

So anyways, as a Test cricket fan, I never miss out on Indian test cricket and also watch the Ashes with great interest . There is something magical about the sheer visual image - players sporting pure whites on the lush green fields of England. The ball swings like crazy and 250 is a par score for a Test inning. Tennis fans can perhaps relate to this visual delight when they tune into the first week of Wimbledon every year; players in pure whites on pristine verdant turf.

When HD coverage started happening, for the first time, you could see the ball's movement in the air even before it pitched - yes there is movement before it hits the pitch and then movement after it pitches. And in slow motion, if the bowler has perfect seam position you can see which side of the ball is shiny. Which in turn determines the swing direction - towards the rough side in traditional swing and the other way when the ball gets older and starts 'reversing'. We could have an entire post on the art and physics of swing bowling.

If you grew up watching test cricket in the 80s and 90s, India always had great batsmen (Gavaskar, Vishwanath, Vengsarkar and Azharuddin), but we were never a fast bowling powerhouse, with Kapil Dev standing out as the only great seam bowler (and he only bowled medium pace) from that era. Overall we were a middling team. We would lose most test series against the dominant teams of that time, viz. West Indies, Australia and England. 

In the 2000s, the batting stayed strong, if not stronger with a formidable batting unit that consisted of Sehwag, Tendulkar, Dravid and VVS Laxman; not to mention Ganguly and a strong batting wicketkeeper in MS Dhoni. We continued to be strong in spin bowling with great spinning duos such as Kumble/Harbhajan Singh and more recently Ashwin/Jadeja. All of this resulted in some really solid performances at home where we could pulverize the opposition in spin-friendly pitches, but our record abroad was still mediocre, especially against Australia, England and Sout Africa. Zaheer Khan was a great swing bowler for a while, but fast bowling remained an area of weakness. 

Into the 2010s, we finally started producing genuine fast bowlers who could move the ball both ways.  This brings us to who we have now - Jasprit Bumrah, Mohammed Shami and to a lesser extent Mohammed Siraj. These guys bowl regularly in the high 130s kph, and Bumrah can bowl in the low 140s kph with great accuracy. Bumrah is currently ranked the no. 1 bowler in ICC Test Rankings, a first for an Indian pacer. More on Bumrah later.

The batting dominance continued after the aforementioned unit retired. Kohli came in and has become arguably the greatest multi-format player of all time. As a Test batsman, his form has declined a bit in recent years, but still ranks easily among the top 10 around the world. Pujara took over "Wall" duties from Dravid and others like KL Rahul and Rahane have chimed in. May be there aren't as many now that will end up in the all-time-great list (Kohli being the only one). But at any given point in time, we have enough in-form batsmen to field two world-class batting sides. Youngsters like Shubman Gill and Yashaswi Jaiswal show great promise.

Thanks to a stronger and more balanced bowling unit (as in both pace and spin), India has beome a cricketing powerhouse in the last 15 years or so. Nowadays we almost always win series at home against anyone and have started winning abroad as well, with back to back series wins in Australia being the highlight. South Africa is the only place where we are yet to win a series. India is often ranked no. 1 or 2 in the World Test rankings. In the new World Test Championship format (which is a 2 year cycle), India is the only team that reached the finals in both instances of so far. True, we lost both the finals to NZ and Australia respectively, but reaching both finals is a testament to the team's consistency over this period.

I wrote all of the above without looking at any numbers, I then looked it up and thankfully the numbers as produced by Google Gemini confirmed my impressions*; India has steadily gotten better as a Test cricket team over the last 4 decades. The following table is reproduced verbatim from Gemini's answer to my question "India's Win loss draw record in Test cricket matches in the last 50 years by decade". 

Decade	Played	Won	Lost	Drawn	Win%	Note
2020s (2020-2023)	54	30	13	11	55.56	Current period
2010s (2010-2019)	82	41	24	17	50.00	Kohli Era
2000s (2000-2009)	81	34	28	19	42.53	Tendulkar, Sehwag & Dravid era
1990s (1990-1999)	80	27	34	19	33.75	Azharuddin & Tendulkar era
1980s (1980-1989)	81	20	31	30	24.69	Gavaskar & Kapil Dev era

So overall, it is an exciting time to be a cricket fan in India. And then Test cricket itself is evolving. We have a new breed of young cricketers (aforementioned Jaiswal, Shubman Gill and the swashbuckling Rishabh Pant) who defy conventional wisdom and bat with supreme confidence and aggression. There is this thing called Bazball that England has been playing in the last few years which has brought a new fearless (or reckless?) and aggressive approach to Test batting. Their record over this period is not bad at all (14-6). But is it because of Bazball? Would they perhaps have an even better record if they did not play Bazball? 

More on Bazball and other recent trends in Test cricket in the next post.

Balaji 

* I checked ChatGPT as well and it returned different numbers which puts both the ChatGPT and Gemini numbers in question; however both sets of numbers confirm the same trend - that India's win percentage in Tests has been improving every decade since the 90s, till midway through the 2020s. 

 

  

   

Cultural and religious connections between South India and South East Asia

Recently we traveled to Bali which piqued my interest in this topic and has been reinforced by a recent lecture I attended. As we learned while preparing for the Bali trip, while Indonesia is predominantly Islamic, the population of Bali is mostly Hindu, Balinese Hindu to be specific. 

We grew up learning about Rajendra Chola's military expedition to South East Asia. As per K. A. Nilakanta Sastri's treatise on South Indian History, Rajendra's naval expedition resulted in victory over the Sri Vijaya kingdom, forcing the Sri Vijayan king, Sangrama Vijayatungavarman to accept Chola suzerainty. The Sri Vijaya kingdom at the time, spanned the Malayan peninsula and Sumatra (the largest Indonesian island).  It is well known that Rajendra Chola earned the title Kadaram Kondan after this expedition. Kadaram is the Tamil name of modern day Kedah, a port in modern-day northern Malaysia. 

It might be tempting to say that the connection to South East Asia was initiated by this expedition. That would be inaccurate and simplistic to assume that cultural connections happen solely through or as a result of warfare. The connections between India - especially South India -  and South East Asia had clearly started 100s of years earlier. After all, the rulers of the Sri Vijayan dynasty (modern day Indonesia/Malaysia) and the Khmer Dynasty (modern day Cambodia) bore Hindu/Sanskrit names*. Nilakanta Sastri states that diplomatic relations existed between the Chola and Sri Vijaya kingdoms since at least Rajaraja Chola's (Rajendra Chola's father) time. 

The more immediate impetus for this post is from a very interesting and insightful lecture  I had the privilege to attend, by  Prof. Vasudha Narayanan, (Dept. of Religion, University of Florida) on Vishnu Temples and Iconography in Cambodia. Dr. Vasudha covered multiple examples of how deities and stories from Hindu mythology have been interpreted and represented sculpturally by Cambodian sculptors including at the world-famous Angkor Wat temple. Following are some excerpts and my take-aways from this talk.  

The story of the churning of the Ocean of Milk (பாற்கடல்) by the Asuras and Devas has for some reason attained a lot of significance in Cambodia and is a common motif across the country. It is a fascinating story of how Lord Vishnu takes multiple forms - first the Koorma (tortoise) Avataram that supports the Mandara Mountain which is used to churn the ocean to extract the amirtam (the nectar of immortality) and later takes on the form of Mohini to oversee the distribution of the amirtam to the devas and asuras.  Dr. Vasudha talked about the massive 49-metre bas relief panel on the eastern gallery of the outer wall of Angkor Wat that depicts the churning scene **.  

There is a curious character near one end of Vasuki the snake's tail (the snake is used as the rope to churn) on the panel. It is a robust looking monkey figure bigger than the churners next to it. This character has been previously identified as either Hanuman or Sugriva by scholars. 

Photo courtesy of Bahnfrend licensed under the Creative Commons Attribution-Share Alike 4.0 International license.

However, Dr. Vasudha has made a compelling case that this is none other than Vali. Perhaps the most compelling evidence she cites is from Kambaramayanam, where Vali's participation in the churning effort is mentioned. It should be noted that in Valmiki's Ramayana there is no mention of Vali's involvement. Kamban however mentions this in multiple instances including one where Vali is mollifying his wife Tara saying he - whose strength helped the Devas themselves in churning the ocean - can handle his brother Sugriva in battle. There is also a 13th century Telugu text called Ranganatha Ramayana that mention's Vali's involvement in the churning of the ocean. Dr. Vasudha also showed a relief panel of the churning scene in the Virupaksha temple at Pattadakkal in which a monkey figure is shown (which happens to be an amazing heritage site in its own right as a Badami Chalukya temple from the 8th century). 

Dr. Vasudha also showed instances where the Cambodian representation of stories from Hindu texts is different from how they are typically seen in India. 

• There are sculptures representing the story of Vishnu protecting his devotee Gajendra the elephant from the crocodile. In Cambodia, the crocodile is above Lord Vishnu which is very different from how it is portrayed in India. 
• The reclined posture of Lord Vishnu on Adhiseshan the multi-headed serpent is usually depicted with Vishnu's head on the left. However in Cambodia, it is more commonly depicted with his head on the right. There are rare instances of this in India as well which Dr. Vasudha has identified; she has come up with a correlation of the direction the scultpture is facing to which side the head is positioned.

It is as if Cambodian sculptors had access to the same texts, but perhaps not the Indian iconography and came up with their own. Or as Dr. Vasudha stresses, there is the possibility that they did not necessarily have access to the texts, but had knowledge of the same stories that are narrated in those texts. 

R. Balaji

P.S.

* It is also interesting to note the an suffix of the male rulers' names which bears similarity to Tamil (and Malayalam?) names - as opposed to other Indian languages which do not have the an suffix; e.g. Bheeman in Tamil vs. Bheema in Telugu/Kannada or Bheem in Hindi).

** See this WSJ article for a nice description of the panel at Angkor Wat. 

An objective measure of how majestic a mountain is

Ever since our days in Seattle, I have been endlessly fascinated with mountains like Rainier and Denali - mountains that stand much taller than their neighbouring mountains thereby offering dramatic and spectacular views, often from long distances. It turns out there is an objective measure that measures this quality: topographical prominence (thanks to a Windows screen saver photo of Mt. Kilimanjaro which led me to this Wikipedia page). Some really high peaks (like K2) don't have as much topological prominence as they are surrounded by mountain masses that are also very high.

By Muhammad Mahdi Karim - Own work, GFDL 1.2, Link

The definition of topographic prominence is a bit tricky but makes intuitive sense if you think about it. Wikipedia defines it as " the height of a mountain or hill's summit relative to the lowest contour line encircling it but containing no higher summit within it". Basically:

1. Find the maximal land area surrounding the peak which contains no higher peak. 
2. Find the lowest point in that area. 
3. Topographic prominence is the height of the peak from that point. 

Now, which mountain do you think is #1 in terms of topological prominence? As per another helpful Wikipedia page on the Ranking of Mountains by Topological prominence, it is indeed Mt. Everest. Well, no surprise there, you might say. But I was actually surprised. I knew that you need to get to a very high altitude before you even see Everest. The base camp for a mountaineer making an attempt at Mt. Everest is at a whopping 17000 ft. So I would have thought Mt. Everest's topological prominence would be around 12000 ft, not that high. I thought even Mt. Rainier might beat it, Denali for sure. You can see Denali from 100 miles. 

But then if you go back and read the definition again, Mt. Everest gets the top nod based on a technicality. You see, as per step 1, you need to find the maximal area surrounding the peak which contains no higher peak. Since there is no higher peak than Everest, the maximal area is the Earth's entire landmass (or at least Eurasia if you want to be very precise). And the lowest point in that area (step 2) is sea level. Hence the topological prominence of Everest is its entire height from sea level, all 29000ft of it. So the measure actually doesn't really work that well for the tallest mountain in the world because it is no. 1 by definition, but it works well for others. So none of the high peaks in the Himalayas other than Mt. Everest  come up in the top 10, because they are surrounded by really high contiguous landmasses that includes higher peaks  and they don't stand out as much. 

Not surprisingly, Mt. Aconcagua and Mt. Denali the tallest mountains in South and North America respectively are 2nd and 3rd in the list. And Mt. Kilimanjaro, the highest mountain in Africa is fourth.  And my favourite neighbourhood Mt. Rainier comes in at a respectable 21st  in the list. 

- Balaji