Singapore

My plane arrived after 1 A.M. in the morning at Changi Airport and I steeled myself for waiting in long lines while customs officials woke up and ambled over. To my amazement, from the time I came off the plane to the time I got into a taxi, including clearing customs and getting my bags, a total of seven minutes passed.

My taxi hurtled into town on a wide, immaculate freeway lined with flowers and trees. We passed a Pizza Hut and a bowling alley. This is Asia? Cars even pulled over to the side of the road to let an ambulance through.

Getting to the hotel, I found that the five copies of Stacks sent over by Interop had not arrived. I placed a call to David Brandin, the Vice President for Programs, who assured me they had been sent over a week ago by Federal Express.

The next morning, we found out that the books had in fact ar rived five days earlier, but had been intercepted by the Undesirable Propagation Unit, one of the government censors. I knew that pornography and communism were not welcome, but I hadn't figured my descriptions of SMDS would offend anyone other than Frame Relay manufacturers. After pointing out that the books were engineering manuals, Federal Express eventually sprang them loose.

I spent the morning at the National University of Singapore (NUS) talking to Dr. Thio Hoe Tong, director of the computer center. NUS, with 17,000 students, is the largest university in Singapore. The university had just installed a 64 kbps link to JVNCnet in Princeton and joined the Internet.

The campus internet had recently caught the attention of the National Science and Technology Board (NSTB), which decided that the experiment should be expanded. Dr. Thio's deputy, Tommi Chen, had been delegated to develop TECHnet, a project with the aim of developing a national research IP network.

Technically, TECHnet was simple. For SG $300 per month, an organization got a port on the NUS router. Politically, however, it was a big step forward. TECHnet involved both commercial and educational users. As long as the NSTB determined that a company is a "bona fide" research and development group, it could join the network.

This may seem a bit pale to some readers, but the idea of multiple users sharing a single leased line (which is what TCP/IP does), runs contrary to a long-established tenet of international communications, the infamous Recommendation D.1.

Recommendation D.l of the CCITT established the principle that multiplexing multiple users over a single international line should be the province of the "network provider," a code word signifying the single monopoly, the PTT. In this parlance, a "value-added net work" was anybody who was forced to buy services from the telephone company. By defining multiplexing over international lines as the proper function of the network, not the value-added network, competition never even got out of the gates.

Recommendation D.l has been defanged for most of the world. When you think about it, D.l and similar barriers to entry run counter to the Treaty of Rome, the GATT, and any other principles of fair trade and antitrust ever promulgated in international law. While most organizations and some countries ignore it, there are a few places where D.l still rears its ugly head. Although the recommendation itself is rarely cited specifically, it has certainly established itself as part of the telephone company mentality.

Rumors were that Singapore Telecom was not pleased by the idea of TECHnet. After all, if a company uses the Internet to communicate with the U.S., the reasoning went, they wouldn't be placing phone calls.

This reasoning had been applied before to networks. In Singapore, BITNET was not available to undergraduates. The rumor was that Singapore Telecom objected because if undergraduates could use electronic mail to reach brothers and sisters studying in the U.S., they wouldn't place international voice calls.

In the case of TECHnet, the government had exerted enough pressure so that the system moved forward despite the objections of more conservative elements. However, as a safety check, CEOs of companies joining TECHnet had to sign a certification that the employees of a company would only use the network for R&D, strenuously avoiding commercialistic or even (save us) personal purposes.

After describing TECHnet, Dr. Thio explained the rather unusual approach to campus networking at NUS. A mere 18 months before, several thousand PCs and Macintoshes had been isolated islands with no connectivity. Rather than phase in networks over a long period of time, NU.S. adopted a "big bang" theory of network design.

An RFP was issued which was won by a consortium that included Hewlett Packard and Fibronics. In two stages, six months apart, 3,000 machines went on the network. Most of the network was based on HP9OOOs and PS/2s acting as LAN Manager servers. An FDDI backbone and Ethernet subnets tied it all together.

Needless to say, with several thousand users going live at the same time, training and support became, as Dr. Thio put it, "an issue."

That night, I was scheduled to give a talk to the local IEEE chapter. I gave my usual rambling discourse on "our friend, the Internet." Afterwards, I was presented with an attractive pewter dish and we went out for a dinner of hairy crabs and garlic prawns.


I had agreed to be a guest lecturer Friday morning at the Information Communication Institute of Singapore (ICIS). ICIS, a cooperative program between AT&T and the Singapore government, offered a postgraduate degree in telecommunications.

I was picked up by Don Stanwyck, an ICIS faculty member and AT&T employee. Don had been one of AT&Ts representatives to various CCITT and ANSI study groups and had served as editor for several key ISDN recommendations.

Don took me down to Little India, where we had a breakfast of roti prata, Indian fried bread which you dip into a bowl of spicy mutton curry. Don explained how he had ended up on the expatriate circuit in Singapore. As a CCITT delegate, he had logged well over a million frequent flier miles, traveling 70 to 80 percent of the time. He finally decided "enough is enough," and told AT&T he wanted to live in Singapore. In addition to Singapore, Don had lived in Japan, and speaks proficient Mandarin and UUCP.

With a love-hate relationship to Singapore, he was typical of many of the expats who live in Asia. He knew more about Singapore than many locals, yet remained very much a visitor here. After two years, he had kept an intellectual detachment and alternated between wild praise and cynical criticism.

At ICIS, I lectured to a packed room with several classes meeting together. It was a typical lecture in some ways. The students laughed at my jokes and seemed to enjoy the talk. Yet, when I was done and asked for questions, not a single person raised their hand.

I had experienced this in the past, so was not as disconcerted as I might have been. In many Asian countries, it is considered bad manners to ask a question. To do so implies either that the teacher did a bad job explaining or that the student wasn't paying attention. Risking either implication would be a loss of face for somebody.

Instead, questions are best asked in a more informal setting. After the lecture, sitting in the lounge, I spent a half-hour answering detailed questions on everything from namespace administration to exhaustion of the 32-bit address space.


I spent Friday afternoon with Singapore Telecom. Singapore Telecom is one of the classic PTTs, controlling every aspect of telecommunications, as well as the postal system. For many years, local telephone service in Singapore had been essentially free, with a flat annual equipment rental of SG $190 for homes and SG $250 for businesses. (The Singapore dollar was trading at 1.694 to the U.S. dollar at the time.)

When I arrived in Singapore, the flat rate was about to be abolished in favor of a straight usage-based system of 1.4 cents per minute during peak periods and 0.7 cents per minute during off-peak periods. "Rentals" would decrease to SG $100 for homes and SG $150 for businesses.

This change was causing such an uproar that the Prime Minister, traveling in Zimbabwe, had called a press conference to discuss the subject, and his comments were front page news in the Singapore Straits Times. For Singapore, this was a major political flap. A newspaper columnist felt compelled to write a piece debunking the theory that the entire social fabric would disintegrate because children would place fewer calls to their mothers. Singapore Telecom issued a study saying that 66 percent of bills would go down and that the scheme would be revenue-neutral. Of course, recent year end results by Singapore Telecom of profits of SG $1.2 billion didn't help matters.

In order for a residential bill to go down, a residential customer had to make less than 107 hours per year of peak, or 214 of off-peak, calls. It boiled down to a situation where families with teenagers and modems were going to face higher bills.

Singapore Telecom contended that the previous flat rate policies had made it difficult to sell some new services. ISDN was a key example. Singapore was quite proud in 1989 of being the first country to have nation-wide availability of ISDN.

ISDN cost 10 cents per minute in Singapore. With variable pric ing on regular phone lines, Singapore Telecom hoped to increase the demand for data services. I asked Chew Mun Chun, manager of the Network Technology Group at Singapore Telecom how successful ISDN had been. He smiled bravely and said, with all the positive spin he could muster, "today we have close to 300 subscribers." Needless to say, most businesses had found it simpler to use high speed modems on dialup lines.

While ISDN had not been a raging success, Singapore Telecom had been more successful in other areas. It teamed up with British Telecom and the Norwegian Telecom to offer Skyphone, a service that allowed airplanes anywhere in the world to place calls. Each of the three telephone companies provides a ground station that covered a certain geographic area: Norway covered the Indian Ocean, Singapore the Pacific, and British Telecom the Atlantic.

Skyphone is built on top of a network of eight satellites run by INMARSAT, a consortium headquartered in London and sponsored by 60 member countries. INMARSAT provides a generic satellite service which members then use to build value-added services such as Skyphone. An airborne user communicating with an INMARSAT satellite indicates which service she is trying to use. When a call reaches one of the earth stations, such as the one that Singapore Telecom operates on Sentosa island, it is routed through the local telephone network and then out to the International Direct Dial (IDD) links. Since Skyphone is a telephone link like any other (albeit, an expensive one at U.S. $6.70 per minute), it can be used for fax and data as well as voice traffic. It is even possible to place ground-to-air calls, although the caller would need to know the air craft ID in order to do so.

The first customer for Skyphone was the Sultan of Brunei. Singapore Air was the first commercial airline to install the service. Other airlines quickly followed suit.


Another interesting Singapore Telecom project I saw was Teleview, a videotex system that went into field trials in 1988 and went public in 1990. The government had very ambitious plans for the system, but by late 1991 there were only 6,500 subscribers.

Teleview was an unusual videotex system in that it supported high-quality photographic images. Commands to Teleview went in via normal asynchronous dialup lines. Photos returned at 2,400 bps would not be very effective, however, so a return data path was provided via a television channel dedicated to Teleview.

Each frame on the television channel had a subscriber ID on it, directing it to the appropriate system. The signal had a 4.5 Mbps data throughput rate, allowing frames for lots of users to share the single television channel.

Your average PC is not set up to handle either photographic quality displays or to receive RF signals. To provide these capabilities, the original Teleview terminal was a specially developed device which could be seen in many shopping centers and public facilities.

Dedicated terminals, especially fairly expensive, single-purpose terminals, were not an ideal way to penetrate the home market. PC users were able to access Teleview by adding an RF card for receiving data, a modem for sending commands, and a photo display card to drive a monitor at 480 by 280 pixels with 18 bits per pixel for a total of 256,000 simultaneous colors.

While photographic display was a unique feature of Teleview, it was possible to bypass the feature and access the system simply with a modem for an 80 by 40 character display. It appeared that many, if not most, of the users were content with the character-only displays and used the system for things like stock quotes.

As videotex systems went, Teleview had a fairly good selection of information available. In addition to the obligatory stock prices, it also had the Reuters news service, airline information, real estate listings, and even record and book shops.

Some of the more unusual applications reflected the influence of a strong government involvement. Mandarin tutorials, for example, proved quite popular as an instructional tool. Always sensitive to the cultural makeup of the island, the government added Malay les sons soon after the Chinese.

The initial architecture for the system was a single proprietary computer system with X.25 links to service providers and asynchronous and RF links to users. The next generation system was an at tempt to vastly expand the functionality (or at least the capacity). Teleview officials were still in the middle of the procurement process and could not discuss details, but interviews with an anonymous source established that the contract for the system had been won by Digital.

The front-end to the new network was going to be a series of modem racks, presumably connected to VAXen, which are in turn connected to a series of front-end Ethernets. Each front-end Ethernet would then be connected, via a bridge, to an FDDI backbone. Other bridges would connect to a back-end Ethernet. On the back end Ethernet would be another series of VAXen, which would contain a cache of recently accessed frames.

The cache VAXen would be connected by a mystery net (presumably Ethernet and LAT) to a series of VAX Clusters, which would serve as data repositories. Other VAXen would act as X.25 links to other information sources.

As one of my anonymous sources said, "they're throwing VAXes around like water." Singapore had, as of October 1990, invested SG $50 million in the system and it was likely that current investment was an order of magnitude higher.

The government placed great importance in Teleview. When it was deployed commercially, Yeo Ning Hong, the Minister for Communications and Information said "beginning today, Teleview will bring the Information Age into every home and workplace." Like France, Singapore saw videotex as the way to bring widespread computing into every home.


Saturday morning, I went back to Science Park and entered yet an other ultra-modern office building, this one housing the offices of Singapore Network Services, developers of the Tradenet EDI system.

I was met by two individuals wearing the international marketing uniform of business suits, business cards, and business-like smiles. They whisked me into the board room and sat me down at a table the size of a golf course. They sat down far away on the other side of the table, yet still close enough that I could see them.

I knew I was in trouble when the slide projector came out. The junior of the team, bearing the title of "Executive Marketing" got up and started smiling broadly and talking about "increasing productivity" and similar concepts that we could all agree on as being positive.

Singapore Network Services (SNS) was formed with very strong government support as a joint venture between the Civil Aviation Authority of Singapore, the Port of Singapore Authority, the telephone company, and the government.

Although Tradenet, designed to clear cargo through customs by using EDI, was the first application, SNS was branching out into other EDI applications. The slide presentation was meant to illustrate a grand vision of the world, and words like "network hub" and "information services" kept getting tossed out.

A bit of inquiry established the fact that the "network hub" was in fact an IBM 9000 which traders accessed via dialup or leased lines. Not being familiar with the IBM 9000, I naively asked what operating system was being used.

"That is a very technical question," my presenter responded. "We were not told you would want such technical information."

It is the bane of marketing the world over that successful projects, such as a massive application of EDI in Tradenet, gets papered over with superficial nonsense like "information services for productivity." Instead of simply describing the success and how it came to be, which would certainly be very impressive, marketing feels compelled to invent silly catch phrases.

Tradenet had in fact been quite a successful system. Traders used the EDIFACT standard to format documents which were uploaded to the IBM 9000. SNS didn't use X.400 to transfer the document, although Executive Marketing agreed with me that X.400 was a good thing. Once the files were uploaded OfficeVision mail was used to move the documents between different groups.

The system linked traders in Singapore to customs, the Trade Development Board (TDB), and to what was euphemistically referred to as "other controlling agencies." An example of the latter was the Undesirable Propagation Unit, the government agency that had held my books for inspection.

Typical transactions were import/export declarations, certificates of origin, and other documents. Many of these operations, such as customs clearance for a shipment, were semi-automated. An answer from customs on whether clearance is automatically granted or an inspection is needed, could be returned in a few hours instead of the few days required in a manual system.

Tradenet was launched in 1989 and at the time the government had hoped that by 1993, 90 percent of the traders would be using the system, at which time it would become mandatory. By 1990, 90 percent of traders were already on the system and it became mandatory in 1991.

After Tradenet, Singapore Network Services started branching out into other EDI applications. Medinet was used for automated processing of medical claims. $Link was an electronic funds transfer application. OrderLink was a system developed to allow the government to issue Requests for Quotation and receive bids from suppliers.

Executive Marketing finished his slides and I escaped the plush confines of the Singapore Network Services board room into the refreshing humidity of the street. I flagged down a taxi which managed to decelerate from 50 mph to 0 in a little under 1 second. I jumped in with a bit of trepidation and the tires squealed as he took off.

"Where you from?" the cabbie asked.

"Colorado, U.S.A.," I replied.

"Ah, Chicago!" he replied. He seemed to find this hilarious and laughed furiously. After laughing for a few minutes, he would quiet down, then repeat the word "Chicago" again and start laughing. I occupied myself with pressing my foot against an imaginary brake pedal.


Saturday night, Michael Yap from the National Computer Board picked me up for drinks. After a couple of beers in a generic pub in one of Singapore's huge entertainment developments, we went for a ride and ended up in Chinatown.

There, we went into a traditional tea house, Tea Chapter. After taking off our shoes, we walked up narrow stairs two stories into a large room filled with low tables and pillows.

"This is a Singapore Chinese Yuppie joint," Michael explained. It had been started a few years ago by a few friends as a place people could sit and relax. Around the room, people were sitting at tables. At a corner, a young couple sat on cushions on the floor while their two young children bounced off the walls.

Tea Chapter was part of a growing trend in Singapore to rediscover ethnic roots and to allow ethnic diversity. Originally, the country pursued an aggressive melting pot strategy, but it had since taken an about face. Malays, Indians, and Chinese all have distinct ethnic communities in Singapore.

Tea arrived with an elaborate array of devices. A big pot of boiling water was used to feed another tiny pot where the tea steeped for a few minutes. The tea from the first batch was thrown away and more water added. Then, a few batches from the tiny pot were used to fill a medium-sized serving pot. From there, the tea was poured into teacups so small they held only a swallow. The tea was accompanied by eggs which had been preserved in green tea leaves, a delicious snack.

Along with the teapots was a container filled with mysterious implements: miniature shovels, tongs, picks and the like. I asked Michael to explain the purpose of all these devices.

"I'm not quite sure," he confessed sheepishly.

Michael Yap is typical of the technocrats who run Singapore. His education at the University of Maryland was paid for by the government as part of an elite scholars program for fast-trackers. In return, he is bonded for eight years of government service, although it is likely that he will stay much longer.

Michael is passionately devoted to the joint causes of technology and Singapore. He continually steers the conversation back to one of the two topics (and preferably both at once), always accentuating the positive aspects of Singapore society.

Michael's job is Programme Manager for IT2000, Singapore's effort to build an "intelligent island" by the year 2000. The scope of IT2000 is vast, one of the most encompassing visions ever formulated by a government to build a network infrastructure for a society.

First and foremost, IT2000 is a symbolic commitment by the government to use technology. It is an effort to find a way to maintain the stunning economic growth that Singapore has managed in the last 20 years through shipping, finance, and hi-tech manufacturing.

As a vision, it starts with high-level goals, such as increasing productivity and well-being. At the next level down, it moves on to the applications that can help achieve these goals. Only at the very end is specific technology considered.

Thus, unlike many national plans, fiber to the home or use of ISDN are not the goals. Instead, the vision of IT2000 is to maintain productivity increases of 3 to 4 percent per year for a population of 2.65 million people for a sustainable period of time.

Singapore is one of the few countries that could seriously advance such a goal. Known as Singapore, Inc. to the financial press, the country is run much like a big corporation. It is not unusual for senior management (i.e., government ministers) to give direction to corporations, citizens, and the civil service.

Throughout its history, Singapore has been run with this combination of broad consensus and top-down direction. IT2000 is not the first large scale plan. In 1984, there was "Vision 1999," in 1986, the "Agenda for Action." In 1990, the government issued a beautiful coffee-table book produced by Times Editions called Singapore‹ the Next Lap. The book contained a series of broad goals, such as the following:

Pictures of smiling children and happy families were accompanied by specific goals, in this case a target population of 4 million people. These goals were then used to guide government policies, such as tax policies, or even housing assignments.

In the area of information technology, which the Singapore government refers to as IT, policy was spearheaded by the National Computer Board (NCB). The NCB coordinates government computerization efforts, and is also charged with creating a "National IT Plan." In consensual Singapore, the NCB shuns the term "create," preferring instead to "lead to the realization" of a National IT Plan.

In the civil service computerization area, NCB had deployed over 800 of its staffers into government offices. An example of a system developed by the NCB is the immigration system, which claims to be able to check passports for foreigners in 30 seconds and residents in 15. My own experience at the airport, even though in the middle of the night, certainly bore this out.

Other examples of government automation efforts are an auto mated card catalogue for the national library, a computerized registry of companies, and automated production of birth certificates. No one of these applications was particularly unique; what was interesting was the breadth and focus of the computerization effort.

NCB had been slowly trying to tie these different systems together to allow data sharing among government bodies, an effort known as IDNet. I met with employees of NCB to learn more about IDNet. The meeting took place in the vast NCB headquarters located in Science Park. On the walls were posters of people looking very modern, sporting slogans like "Success Through Partnership" and "Fulfillment Through People Orientation."

I was met by an NCB public relations officer and ushered into a conference room. "Help yourself to coffee," she said, pointing to a device called the Café Bar and leaving to get the people I would be meeting with.

The strange looking contraption had different buttons you pushed to dispense various powders (coffee, milk, soup) into a cup. You then pressed the water dispensing bar to put hot water into your own personal mix of soluble powders. After a few false starts, I figured out the apparatus and made myself a cup of fresh, steaming Nescafe.

Being the curious type, I had the cover off the Café Bar and my head buried inside when the door opened and a group of NCBians (as the posters downstairs referred to employees) were ushered in by the PR officer. I slammed the cover on my finger and sat down to learn about IDNet.

The network is fairly simple. It consists of an IBM 3090 at the Ministry of Finance acting as the hub for a multidomain SNA network. The basic service allows a user on one net to send and receive 3270 data streams to another net. A few custom applications are also available.

To use cross domain services requires a special ID. One thousand IDs had been granted, each tied to a user profile detailing which services that particular user could access. IDNet has 31 sites, 90 machines, and 40 applications.

The IDNet architects didn't seem particularly adventuresome technically, as is often the case with people whose job is to keep a network running. They grudgingly acknowledged that token ring networks might play a role in the future, but more for terminal access based on PCs than for peer-to-peer networks. Protocols like TCP/IP and OSI were acknowledged, but with a notable lack of enthusiasm.

In a distinct twist from many large organizations, NCB was centralizing its servers into four very large data hubs, each devoted to a particular type of data. Centralizing the location of computers would, hopefully, save on overhead and permit better control over operations. The four centers would have data about land, people, legal establishments, and finance. While each data center might have many computers, the NCB staffers didn't rule out the possibility of replacing many systems with a single mega 3090 running DB2.

The guiding principle for computerization of government was the slogan "one stop, non stop." This meant that a citizen ought to be able to go anyplace and do all government business at once (and quickly). As one official pointed out, no matter how comfortable you make the chairs in a government waiting room, people will still want to leave. Of course, it is worth pointing out that reality does not always meet the lofty goals of slogans, but it was encouraging to see that Singapore policy makers had a policy they were trying to communicate.

Data sharing among government agencies was one aspect of one stop, non stop. The other was delivery of services. Systems like Teleview videotex were envisioned as the delivery vehicle for many government services. Some services were already available on Teleview, such as listing of government tender notices and awards.

Other systems were not on Teleview but did show the effects of data sharing. For example, the census bureau tapped into existing databases and preprinted census forms, allowing people to verify data instead of reentering it.

One stop, non stop was part of the broader vision of IT2000. Singapore had had a national IT plan for several years, but in 1990, Dr. Tay Eng Son, the Senior Minister of State for Education unveiled the IT2000 concept which greatly increased the scope (and the stakes).

IT2000 is being called a "national masterplan for an information infrastructure." When I talked to Michael Yap in late October 1991, the NCB had just finished the first phase of feasibility studies. Based on that first phase, NCB had received significant government funds to develop an architecture to make it all happen

The feasibility phase divided the economy into 11 sectors, ranging from construction to health care to transportation. Each sectoral study group was chaired by a "captain of industry."

NCB staffers assisted each study group. The staffers led the groups through top-down analyses such as trends analysis and wish lists and through bottom-up planning procedures such as dependency graphs and functional decomposition.

The goal was a set of strategic applications: things that people should be able to do on an intelligent island. Many of the projects that fell out of this planning are truly ambitious, ranging from a national multi-function smart card and an intelligent vehicle highway system, to a queueless, counterless government services system.

Altogether, the planning process involved 200 industry leaders and 50 NCB staffers. This resulted in the goals for IT2000. The next step will be to begin defining a series of target applications. Underlying the applications will be a set of common network services (e.g., authentication) which in turn use the national telecommunication network. Tying all this together is a series of technical standards and a legal framework.

Defining each of these pieces is the challenge. When Michael Yap talks about a legal framework, for example, he includes such difficult issues as copyright for intellectual property. When he talks of common network services he means everything from transport connections to "look and feel" to application tool kits.

Will it work? When NCB decides an item is a priority, it can achieve impressive results. For example, it developed a ship planning system for the Port of Singapore Authority that reduces the planning time for each ship by half. Together with the massive use of EDI in TradeNet, there is no doubt that the port is a highly efficient, automated enterprise.

The real challenges will be the common network infrastructure and the legal framework. In the U.S., such issues are still in the very early conceptual stages. The Corporation for National Research Initiatives has attempted to address both issues in their Knowbot architecture, but Knowbots are still at the early prototype stage.

Michael and I finished our tea and he dropped me off at my hotel. I was off the next day to Dublin to learn more about a key U.S. infrastructure, the NSFNET, which had come into being because of an Irishman.

The next morning, I stood in long lines at the Singapore airport and read the Singapore Straits Times. I noticed a story that had an interesting mix of old Asia and high-tech. In the Taoist religion, when people die, paper models of important things in their life are burned in a ceremony, allowing the newly departed to take the good things they enjoyed in life with them. Typically, these paper models are things like houses and cars. On Sin Ming Drive, however, paper models of portable phones, VCRs, and even karaoke sets were available. No ISDN handsets were on sale yet, though.

<<<>>>