Dr Tan Tin Wee has had a huge impact on Internet development in Singapore. While practising as an academic at the National University of Singapore (NUS) since the Internet’s emergence as a global network, he has introduced the Internet to the masses, minority and disability communities, and the government.
He has worked to bridge the digital divide, built high-speed next-generation links between Singapore and the US, and pioneered Internet services and applications to showcase the incredible potential of the Internet through multilingual websites, e-commerce applications, Java applets, virtual reality VRML, video delivery, satellite Internet, remote rapid prototyping (the precursor of 3D printing), IPv6, and more. Tin Wee continues to impact Internet operations in high-performance computing (HPC), building green data centres housing petascale supercomputers, and laying the groundwork to support future quantum-safe communications between supercomputers in Singapore. In this post, he shares the details of this 30+ year journey.
What’s in a name?
My name is Tan Tin Wee. Tin Wee is my given name; Tan is my family name and is a transliteration of the Hokkien Chinese dialect pronunciation of my name. So you might be confused, like Vint Cerf at the Internet Hall of Fame Inaugural Induction ceremony (YouTube), when he referenced me as ‘Tin Tan Wee’ in his acceptance speech. According to my linguist friends, Tin Tan Wee sounds much better linguistically because it is a phenomenon of ablaut reduplication, in which high-sounding vowels always sound better preceding low vowels in the ears of Westerners. Think ‘tick-tock’ rather than ‘tock-tick’, or ‘King Kong’, rather than ‘Kong King’ — that’s ablaut reduplication at work.
In Eastern languages, we have less of this linguistic phenomenon, as you see in this picture of my Chinese name (Figure 1).
The transliteration took place at a time before Singapore, where I was born and bred, had achieved independence. We used to be a British colony, so the bureaucratic administrators of the time adopted whatever phonetic transliteration was convenient to them, as opposed to standardized transliterations like those adopted in Taiwan, such as the Wade-Giles standard, or in mainland China, where they have the standard Hanyu Pinyin romanization system.
Name order is another issue. For Chinese, Japanese, Korean, Vietnamese, Southern Indian, and some French names, family names are conventionally written first and are followed by given names, which can be generational names. In Singapore, the original Chinese name order is retained, but for Japanese and mainland Chinese names, upon romanization, the name order is also reversed for ordinary people. For example, easily recognizable Chinese names like Lee Kuan Yew or Deng Xiao Ping retain this name order. However, people in China would probably render an ordinary person with the reversed order of Xiaoping Deng in romanized form. For Yao Ming, the international basketball star, some are left wondering which is the family name and which is the given name. Famous cellist Yo-Yo Ma has his family name moved to the end, like Western names, but in Chinese, he would be known as ‘Ma You You — 馬友友’.
Why am I giving you a lesson in linguistics? The answer is simple — Internationalized Domain Names, or IDNs.
The Internet is for everyone, everywhere, and in their own language
In the early days of the World Wide Web (WWW), I took it as a personal challenge to deal with such issues, particularly with the internationalization of web content in 1995. In 1998, we began work on internationalizing the Domain Name System (DNS), especially for the Singapore context. This work went beyond just name order or transliteration issues and delved deeper into the underlying encoding systems, and the conflicts they create when there is insufficient standardization.
The rest is history. Today IDN standards are universally adopted for web content and domain names using the Unicode consortium’s encoding system. It’s not yet a perfect system and still has some technical issues but, by and large, they work for most people.
Biological inspiration
Currently, I’m an Associate Professor in the Department of Biochemistry at NUS, and Chief Executive at the National Supercomputing Centre (NSCC) Singapore. Studying biochemistry and molecular biology may not be an obvious segway to Internet technologies but it was for me. I’ll explain.
In 1985, I found it necessary to write a Fortran program for my final year Biochemistry project at Cambridge University to augment the findings of my experimental research.
My last research project a few years back worked on a fish vaccine that enabled a local fish farming company to prevent mass fish death due to a bacterial infection. That Singapore company, which was founded by an army reserve buddy of mine, survived and recently IPO-ed on the Oslo Stock Exchange.
My last PhD student whom I supervised graduated in 2021 after working on a project about the rapid identification of disease pathogens using machine learning and Artificial Intelligence (AI) techniques.
So, I have been involved in molecular biology and computing for more than three and a half decades!
It was natural for me to get caught up in computing during the beginnings of the human genome project, and the emergence of biocomputing — now known as computational biology and bioinformatics — when I returned to Singapore from Edinburgh after my PhD in recombinant DNA vaccines. In the late 1980s and early 1990s, and still, 30 years later, I was concerned about the need for biologists to access biomedical data.
The biggest developments at that time were computer networks beyond Local Area Networks (LANs) and Wide Area Networks (WANs) to global networks such as BITNET and the nascent Internet. By 1992, I was able to demonstrate biomedical databases I had created using WAIS and Gopher at the Geneva Palexpo to an excited medical informatics audience, during the MEDINFO’92 World Congress. At that time, explosive Internet growth had come from Tim Berners-Lee’s WWW. This growth was driven by the graphical user interface of the web browser and provided the visual impact needed to move the Internet beyond a largely text-based system. By September 1993, I launched biomed.nus.sg as the co-first WWW site for Singapore focusing on biological data and information online.
Back then, in an article I wrote, I had predicted that one day at the click of a button, a researcher would be able to read all kinds of scientific papers without visiting the library journal stacks and download and analyse scientific data. Today this is a given.
The birth of IDNs
Because I was known as the biggest user of the Internet for research, I was eventually asked in 1993 to help run the Technet Unit, Singapore’s first Internet Service Provider (ISP) for the research and scientific community, until it was bought by Sembawang Media in 1995. It was commercialized with a much sought-after ISP licence to operate Singapore’s second commercial Internet service (after SingNet from Singapore Telecoms). I managed to upgrade the link from 256kbps to 512kbps and expand the customer base to offer graphical user interfaces for access to websites.
This was the first time an IT grant project became commercially viable in Singapore. My entire Technet team was assimilated into the new company called Pacific Internet. The network room, servers, and the entire customer base were transferred to them in an instant. I was asked by my then boss, Dr Thio, to create the Internet Research and Development Unit (IRDU) funded by the remaining grant and/or proceeds from the sale. One of the small research groups in the IRDU specialized in the multilingualization of the Internet.
Singapore has four official languages: Malay, Chinese, Tamil, and English. It was a major challenge in 1995 to offer official web content in all four languages. The lack of standardization in encoding systems was a serious problem as Unicode was still in its infancy at the time. As the problem of rendering multilingual characters on the web interface was then progressively resolved with web browsers adopting code page support, I turned to the issue of multilingual domain names. By 1998, we had derived a working solution. It was presented to the international participants of the Green Paper, who met in Singapore to formulate the formation of what is known today as ICANN, the regulatory body overseeing Internet names and numbers.
This prototype IDN solution was spun off with an eight-figure, multimillion-dollar, investment from a private investment company, General Atlantic, to become i-DNS.net International. I continued to focus on building the international consensus around IDNs and promoting the concept. To do this, I proposed the formation of the Multilingual Internet Names Consortium (MINC) to promote the support of domain names in all languages.
It should be clear by now that my passion for the Internet and supporting technology is not driven by profit. Both ISP and domain name businesses were huge back then, as they are today, but I chose the path towards invention, innovation, research, development, implementation, and promotion of technology.
Pushing IDNs uphill, and bridging the digital divide
The early days of IDNs were an uphill task. Trying to introduce new technology to highly experienced and sceptical Internet professionals is much more difficult for people with no official training in computer science or network engineering. The early introduction and demonstration of a working IDN prototype in 1998 was overshadowed by the introduction of the significant Green Paper proposal (formally titled “A Proposal to Improve the Technical Management of Internet Names and Addresses”), in Singapore, by Ira Magaziner.
It was only after the formation of ICANN in the late 1990s, and the tumultuous period of the dot com crash, that IDNs slowly floated up through the engineers’ discussion forum and finally onto the radar of policymakers. Official adoption of IDNs in the form of IDNA only took place in 2003 (RFC 3492). More than ten years later, the first IDN country-code Top-Level Domains (ccTLDs) were inserted into the DNS root zone. By then, .com had launched mixed character-set IDNs, which may have played an important part in cybersecurity problems whereby many scammers made use of lookalike characters in other languages to spoof domain names. The deplorable situation of domain name spoofing, including IDNs, is still rife today.
Bridging the digital divide everywhere we encounter gaps
So, what was my motivation for doggedly driving IDNs? Since the 1990s, I’ve had an enduring concern for the digital have-nots. While I was initially motivated by digital opportunities for biomedical scientists, I became increasingly concerned the digital divide would continue to expand if the Internet did not support local language email addresses and fonts. Because of this, I demonstrated the technical and operational feasibility of the multilingualization of the DNS and promoted the global effort to standardize IDNs.
Today, that concern is unfounded as almost everyone is familiar with English characters as it has become the lingua franca of technology, science, business, and commerce. Perhaps sometime in the future, in an increasingly polarized world, the language divide may become a big problem again. If it does, however, I am heartened that the Internet has already been future-proofed, at least at the web and the DNS level.
One area that I feel is still under-served is the Internet for people with disabilities. In 1998, I co-initiated the Enable2000 effort to introduce the Internet as an empowerment tool for disability communities. This was launched off the back of an earlier success in 1995 when I introduced Internet Relay Chat (IRC) chat and the web to primary school kids at the Singapore School for the Deaf (SSD). I installed a modem in the principal’s office, wired up their computer room with 10BASE-T cables, and installed network cards in all of their 286 computers. Dreams of introducing the kids at the Singapore School for the Visually Handicapped (SSVH) to the already digitized literature of Project Gutenberg is still a challenge for me today, nearly 30 years on.
Another area of concern is the rapidly ageing society in Singapore. I volunteer at the Caregivers Alliance to train caregivers and give them support in better managing their loved ones afflicted with dementia. I constantly look for ways on how technology can improve the care offered by caregivers and healthcare professionals. For a start, NSCC has signed project agreements with SingHealth, Singapore’s largest healthcare cluster, and the National University Hospital System (NUHS), to empower their clinician-scientists with edge supercomputers on-site within their campuses to allow healthcare professionals better low latency access to fast and powerful GPUs for their AI projects as well as modelling and simulation for their genome informatics projects.
Supercomputing origins of NSFNET, the backbone of the early Internet
From the 2010s until today, I have been involved in driving high-performance computing adoption and the democratization of supercomputing in Singapore. This endeavour came about from my work in 1996 as part of the nascent Asia Pacific Advanced Network (APAN) to connect Asia to America with high-speed next-generation links.
It started out with the linking of Singapore’s emerging National Research and Education Network (NREN), now called SingAREN, to the USA’s STARTAP in Chicago (now called Starlight) in 1997. It would be a quarter century this November 2022 since we established that trans-oceanic link. I was only able to continue this effort in the 2010s, this time, promoting Asia-Connects-America (ACA) 100Gbps, a far cry from the 512kbps I had attained during the Technet days, and later, achieving a routable InfiniBand Ring-around-the-World through the InfiniCortex project. As if history was repeating itself, just as the NSFNET in the 1990s connected supercomputing centres in the USA, we pioneered this experimental InfiniBand network to connect supercomputing centres, which we demonstrated successfully in successive Supercomputing conferences, from SC14 to SC16.
Based on this successful global-range InfiniBand network, from 2016 we operated Singapore’s first petascale supercomputer over a metropolitan-range InfiniBand network on a production basis continuously for six years without any hiccups. This was to demonstrate that the network infrastructure today can support more than just TCP/IP as a communications protocol. It can support InfiniBand as a data transport protocol for global distances despite its popularity as an in-rack or intra-data centre protocol for supercomputing centres.
Great power, greater responsibility
Just as I brought powerful Internet-enabling technologies to the masses, my current goal is to bring the power of high-performance computing and networking to ordinary scientists. In 2015, we built the first petascale supercomputer facility, the Advanced Supercomputer for Petascale Innovation, Research and Enterprise (ASPIRE 1) in South East Asia at the innovation precinct of Singapore’s Agency for Science, Technology and Research (A*STAR) based on long-range InfiniBand interconnecting login nodes in multiple data centres throughout Singapore city.
ASPIRE 1 will soon make way for ASPIRE 2A as part of Phase 2 of the NSCC infrastructure project, which is funded by the National Research Foundation of Singapore. In operating these supercomputers, my key concern was the environmental impact of power-hungry supercomputers, a point that I elaborated on in a TEDxSingapore talk in 2015 on what it takes for responsible operators of data centres to be green and sustainable. The new ASPIRE 2A supercomputer will be housed in a hot tropical data centre hall with no air conditioning systems and tapping on dehumidified ambient air with direct-to-chip cooling, and zero water evaporative loss using dry coolers.
Few people realise that every bit moved over distance costs energy, and every byte stored on spinning disks over decades requires enormous amounts of energy. Every calculation on a supercomputer consumes energy, which then needs even more energy for cooling. I hope to raise awareness of how all data centre operators should explore all possible ways to reduce the energy burden and operate ever more energy-efficient systems.
More recently, the National Quantum Computing Hub (NQCH) Singapore, which includes NSCC as a partner, was established. My next task is to provide the necessary supercomputing infrastructure to support the drive toward quantum computing. To protect future communication links, I intend to build the basis for a quantum Internet, initially using Quantum Key Distribution (QKD) technologies under the framework of the National Quantum-Safe Network (NQSN).
The priority for 21st-century society
Growing the local supercomputing community and the means to protect future computing networks and communications are not the most pressing issues we face. In a world gripped by pandemics and polarization, anti-globalization, and war — all of which have been driven in part by the irrational frenzy of human foibles and amplified globally on digital and social media, Internet spam, scams, fake news, and deepfakes — what can be more important than for mankind to figure out how to manage and handle technology, and to control our basic instincts at the individual level right up to the political arena of international discourse and diplomacy? What we have right now is a networked world that resembles a ‘Wild Wild West’, without sheriffs.
Our generation, the Internet pioneers of the world, made available a wonderfully powerful technology — unprecedented in the history of mankind. The Internet has brought great good and benefits but has also brought the uncontrolled proliferation of bad actors who continue to exploit the Internet with nefarious activities for greed or power — or worse — simply because they can do it with impunity. Perhaps this should be our priority.
Internet Hall of Fame inaugural inductee (2012), Tan Tin Wee is an associate professor at the Department of Biochemistry at the National University of Singapore and Chief Executive of the National Supercomputing Centre (NSCC) Singapore.
The views expressed by the authors of this blog are their own and do not necessarily reflect the views of APNIC. Please note a Code of Conduct applies to this blog.