Digital sweatshops then and now

In the early 1960s my family lived in New Delhi. In 1993, working for BYTE, I returned to India to investigate its software industry. The story I wrote, which appears below, includes this vignette:

Rolta does facilities mapping for a U.S. telephone company through its subsidiary in Huntsville, Alabama. Every night, scanned maps flow through the satellite link to Bombay. Operators running 386-based RoltaStations retrieve the maps from a Unix server, digitize them using Intergraph’s MicroStation CAD software, and relay the converted files back to Huntsville.

I didn’t describe the roomful of people sitting at those 386-based RoltaStations doing the work. It was a digital sweatshop. From the window of that air-conditioned room, though, you could watch physical laborers enacting scenes not unlike this one photographed by my dad in 1961.

I don’t have a picture of those rows of map digitizers in the city soon to be renamed Mumbai. But here’s a similar picture from a recent Times story, A.I. Is Learning From Humans. Many Humans.:

Labor-intensive data labeling powers AI. Web annotation technology is a key enabler for that labeling. I’m sure there will digital sweatshops running the software I help build. For what it’s worth, I’m doing my best to ensure there will be opportunities to elevate that work, empowering the humans in the loop to be decision-makers and exception-handlers, not just data-entry clerks.

Meanwhile, here’s that 1993 story, nearly forgotten by the web but salvaged from the NetNews group soc.culture.indian.

Small-systems thinking makes India a strategic software partner

by Jon Udell

BYTE, 09/01/1993

Recently, I saw a demonstration of a new Motif-based programmer’s tool called Sextant. It’s a source code analyzer that converts C programs into labeled graphs that you navigate interactively. The demonstration was impressive. What made it unique for me was that it took place in the offices of Softek, a software company in New Delhi, India.

It’s well known that Indian engineering talent pervades every level of the microcomputer industry. But what’s happening in India? On a recent tour of the country, I visited several high-tech companies and discovered that India is evolving rapidly from an exporter of computer engineering talent into an exporter of computer products and services. Software exports, in particular, dominate the agenda. A 1992 World Bank study of eight nations rated India as the most attractive nation for U.S., European, and Japanese companies seeking offshore software-development partners.

The World Bank study was conducted by Infotech Consulting (Parsippany, NJ). When the opportunity arose to visit India, I contacted Infotech’s president, Jack Epstein, for advice on planning my trip. He referred me to Pradeep Gupta, an entrepreneur in New Delhi who publishes two of India’s leading computer magazines, PC Quest and DataQuest. Gupta also runs market-research and conference businesses. He orchestrated a whirlwind week of visits to companies in New Delhi and Bombay and generously took time to acquaint me with the Indian high-tech scene.

A Nation of Small Systems

Even among Indians, there’s a tendency to attribute India’s emerging software prowess to the innate mathematical abilities of its people. “After all, we invented zero,” says Dewang Mehta, an internationally known computer graphics expert. He is also executive director of the National Association of Software and Service Companies (NASSCOM) in New Delhi. While this cultural stereotype may hold more than a grain of truth, it’s not the whole story. As NASSCOM’s 1992 report on the state of the Indian software industry notes, India has the world’s second-largest English-speaking technical work force. Consequently, Indian programmers are in tune with the international language of computing, as well as the language spoken in the U.S., the world’s largest market.

Furthermore, India’s data-communications infrastructure is rapidly modernizing. And the Indian government has begun an aggressive program of cutting taxes and lifting import restrictions for export-oriented Indian software businesses while simultaneously clearing the way for foreign companies to set up operations in India.

Other countries share many of these advantages, but India holds an ace. It is a nation of small systems. For U.S. and European companies that are right-sizing mainframe- and minicomputer-based information systems, the switch to PC-based client/server alternatives can be wrenching. Dumping the conceptual baggage of legacy systems isn’t a problem for India, however, because, in general, those systems simply don’t exist. “India’s mainframe era never happened,” says Gupta.

When Europe, Japan, and the U.S. were buying mainframes left and right, few Indian companies could afford their high prices, which were made even more costly by 150 percent import duties. Also, a government policy limiting foreign investors to a 40 percent equity stake in Indian manufacturing operations drove companies like IBM away, and the Indian mainframe industry never got off the ground.

What did develop was an indigenous microcomputer industry. In the early 1980s, Indian companies began to import components and to assemble and sell PC clones that ran DOS. This trend quickened in 1984, when the late Rajiv Gandhi, prime minister and an avid computer enthusiast, lifted licensing restrictions that had prevented clone makers from selling at full capacity.

In the latter half of the 1980s, a computerization initiative in the banking industry shifted the focus to Unix. Front offices would run DOS applications, but behind the scenes, a new breed of Indian-made PCs — Motorola- and Intel-based machines running Unix — would handle the processing chores. Unfortunately, that effort stalled when the banks ran afoul of the unions; even today, many of the Bank of India’s 50,000 branches aren’t linked electronically.

Nevertheless, the die was cast, and India entered the 1990s in possession of a special advantage. Indian programmers are not only well educated and English-speaking, but out of necessity they’re keenly focused on client/server or multiuser solutions for PCs running DOS (with NetWare) or Unix — just the kinds of solutions that U.S. and European companies are rushing to embrace. India finds itself uniquely positioned to help foreign partners right-size legacy applications.

The small-systems mind-set also guides India’s fledgling supercomputer industry. Denied permission by the U.S. government to import a Cray supercomputer, the Indian government’s Center for the Development of Advanced Computers built its own — very different — sort of supercomputer. Called PARAM, it gangs Inmos T800 transputers in parallel and can also harness Intel 860 processors for vector work. Related developments include a transputer-based neural-network engine intended to run process-control applications. The designers of this system impressed me with their clear grasp of the way in which inexpensive transputers can yield superior performance, scalability, modularity, and fault tolerance.

Software Products and Services

Many of the companies I visited produce comparable offerings for LAN or Unix environments. In the realm of packaged software, Oberoi Software in New Delhi sells a high-end hotel management application using Sybase 4.2 that runs on Hewlett-Packard, DEC, and Sun workstations. A low-end version uses Btrieve for DOS LANs. Softek offers 1-2-3, dBase, and WordStar work-alikes for DOS and Unix.

Shrink-wrapped products, however, aren’t India’s strong suit at the moment. PCs remain scarce and expensive commodities. According to DataQuest, fewer than 500,000 PCs can be found in this nation of 875 million people. To a U.S. software engineer, a $3000 PC might represent a month’s wages. An equivalently prosperous Indian professional would have to work a full year to pay for the same system. To put this in perspective, the average per capita wage in India is about $320, and the government caps the monthly salary of Indian corporate executives at around $1600 per month.

Software piracy is another vexing problem. “The competition for a 5000-rupee [approximately $165] Indian spreadsheet isn’t a 15,000-rupee imported copy of Lotus 1-2-3,” says Gupta, “but rather a zero-rupee pirated copy of Lotus 1-2-3.”

Painfully aware of the effect piracy has on the country’s international reputation as a software power, government and industry leaders have joined forces to combat it. The Department of Electronics (DoE), for example, has funded an anti-piracy campaign, and Lotus has a $69 amnesty program that enables users of illegal copies of 1-2-3 to come clean.

Reengineering Is a National Strength

The real action in Indian software isn’t in products. It’s in reengineering services. A typical project, for example, might involve re-creating an IBM AS/400-type application for a LAN or Unix environment. A few years ago, Indian programmers almost invariably would perform such work on location in the U.S. or Europe, a practice called “body shopping.” This was convenient for clients, but it wasn’t very beneficial to India because the tools and the knowledge spun off from reengineering projects tended to stay overseas.

More recently, the trend is to carry out such projects on Indian soil. Softek, for example, used a contract to build a law-office automation system for a Canadian firm as an opportunity to weld a number of its own products into a powerful, general-purpose client/server development toolkit. Softek engineers showed me how that toolkit supports single-source development of GUI software for DOS or Unix (in character mode) as well as Windows. They explained that client programs can connect to Softek’s own RDBMS (relational DBMS) or to servers from Gupta, Ingres, Oracle, or Sybase. That’s an impressive achievement matched by few companies anywhere in the world, and it’s one that should greatly enhance Softek’s appeal to foreign clients.

While reengineering often means right-sizing, that’s not always the case. For example, the National Indian Institution for Training, a New Delhi-based computer-training institute rapidly expanding into the realm of software products and services, has rewritten a well-known U.S. commercial word processor. Rigorous development techniques are the watchword at NIIT. “We have a passion for methodology,” says managing director Rajendra S. Pawar, whose company also distributes Excelerator, Intersolv’s CASE tool.

Other projects under way at NIIT include an X Window System interface builder, Mac and DOS tutorials to accompany the Streeter series of math textbooks (for McGraw-Hill), a simple but effective multimedia authoring tool called Imaginet, a word processor for special-needs users that exploits an NIIT-designed motion- and sound-sensitive input device, and an instructional video system.

Although services outweigh products for now, and the Indian trade press has complained that no indigenous software product has yet made a splash on the world scene, the situation could well change. Indian programmers are talented, and they’re up-to-date with database, GUI, network, and object-oriented technologies. These skills, along with wages 10 or more times less than U.S. programmers, make Indian programming a force to be reckoned with. Software development is a failure-prone endeavor; many products never see the light of day. But, as Tata Unisys (Bombay) assistant vice president Vijay Srirangan points out, “The cost of experimentation in India is low.” Of the many software experiments under way in India today, some will surely bear fruit.

A major obstacle blocking the path to commercial success is the lack of international marketing, but some help has been forthcoming. Under contract to the U.K.’s Developing Countries Trade Agency, the marketing firm Schofield Maguire (Cambridge, U.K.) is working to bring selected Indian software companies to the attention of European partners. “India does have a technological lead over other developing countries,” says managing partner Alison Maguire. “But to really capitalize on its software expertise, it must project a better image.”

Some companies have heard the message. For example, Ajay Madhok, a principal with AmSoft Systems (New Delhi), parlayed his firm’s expertise with computer graphics and digital video into a high-profile assignment at the 1992 Olympics. On a recent U.S. tour, he visited the National Association of Broadcasters show in Las Vegas. Then he flew to Atlanta for Comdex. While there, he bid for a video production job at the 1996 Olympics.

Incentives for Exporters

According to NASSCOM, in 1987, more than 90 percent of the Indian software industry’s $52 million in earnings came from “on-site services” (or body shopping). By 1991, on-site services accounted for a thinner 61 percent slice of a fatter $179 million pie. Reengineering services (and, to a lesser extent, packaged products) fueled this growth, with help from Indian and U.S. government policies.

On the U.S. side, visa restrictions have made it harder to import Indian software labor. India, meanwhile, has developed a range of incentives to stimulate the software and electronics industries. Government-sponsored technology parks in Noida (near New Delhi), Pune (near Bombay), Bangalore, Hyderabad, and several other locations support export-oriented software development. Companies that locate in these parks share common computing and telecommunications facilities (including leased-line access to satellite links), and they can import duty-free the equipment they need for software development.

The Indian government has established export processing zones in which foreign companies can set up subsidiaries that enjoy similar advantages and receive a five-year tax exemption. Outside these protected areas, companies can get comparable tax and licensing benefits by declaring themselves fully export-oriented.

Finally, the government is working to establish a number of hardware technology parks to complement the initiative in software. “We want to create many Hong Kongs and Singapores,” says N. Vittal, Secretary of the DoE and a tireless reformer of bureaucracy, alluding to the economic powerhouses of the Pacific Rim.

The Indian high-tech entrepreneurs I met all agreed that Vittal’s tenacious slashing of government red tape has blazed the trail they now follow. How serious is the problem of government red tape? When the government recently approved a joint-venture license application in four days, the action made headlines in both the general and trade press. Such matters more typically take months to grind their way through the Indian bureaucracy.

The evolution of India’s telecommunications infrastructure shows that progress has been dramatic, though uneven. In a country where only 5 percent of the homes have telephone service, high-tech companies increasingly rely on leased lines, packet-switched data networks, and satellite links. The DoE works with the Department of Telecommunication (DoT) to ensure that software export businesses get priority access to high-bandwidth services.

But the slow pace of progress at the DoT remains a major frustration. For example, faxing can be problematic in India, because the DoT expects you to apply for permission to transmit data. And despite widespread Unix literacy, only a few of the dozens of business cards I received during my tour carried Internet addresses. Why? DoT regulations have retarded what would have been the natural evolution of Unix networking in India. I did send mail home using ERNET, the educational resource network headquartered in the DoE building in New Delhi that links universities throughout the country. Unfortunately, ERNET isn’t available to India’s high-tech businesses.

Vittal recognizes the critical need to modernize India’s telecommunications. Given the scarcity of an existing telecommunications infrastructure, he boldly suggests that for many scattered population centers, the solution may be to completely pass over long-haul copper and vault directly into the satellite era. In the meantime, India remains in this area, as in so many others, a land of extreme contrasts. While most people lack basic telephone service, workers in strategic high-tech industries now take global voice and data services for granted.

Powerful Partners

When Kamal K. Singh, chairman and managing director of Rolta India, picks up his phone, Rolta’s U.S. partner, Intergraph, is just three digits away. A 64-Kbps leased line carries voice and data traffic from Rolta’s offices, located in the Santacruz Electronics Export Processing Zone (SEEPZ) near Bombay, to an earth station in the city’s center. Thence, such traffic travels via satellite and T1 lines in the U.S. to Intergraph’s offices in Huntsville, Alabama.

Rolta builds Intel- and RISC-based Intergraph workstations for sale in India; I saw employees doing everything from surface-mount to over-the-network software installation. At the same time, Rolta does facilities mapping for a U.S. telephone company through its subsidiary in Huntsville. Every night, scanned maps flow through the satellite link to Bombay. Operators running 386-based RoltaStations retrieve the maps from a Unix server, digitize them using Intergraph’s MicroStation CAD software, and relay the converted files back to Huntsville.

Many Indian companies have partnerships with U.S. firms. India’s top computer company, HCL, joined forces with Hewlett-Packard to form HCL-HP. HCL’s roots were in multiprocessor Unix. “Our fine-grained multiprocessing implementation of Unix System V has been used since 1988 by companies such as Pyramid and NCR,” says director Arjun Malhotra.

HCL’s joint venture enables it to build and sell HP workstations and PCs in India. “People appreciate HP quality,” says marketing chief Ajai Chowdhry. But since Vectra PCs are premium products in the price-sensitive Indian market, HCL-HP also plans to leverage its newly acquired HP design and manufacturing technology to build indigenous PCs that deliver “good value for money,” according to Malhotra.

Pertech Computers, a system maker in New Delhi, recently struck a $50 million deal to supply Dell Computer with 240,000 motherboards. Currently, trade regulations generally prohibit the import of certain items, such as finished PCs. However, exporters can use up to 25 percent of the foreign exchange they earn to import and sell such items. Pertech director Bikram Dasgupta plans to use his “forex” money to buy Dell systems for resale in India and to buy surface-mount equipment so that the company can build work-alikes.

IBM returned to India last year, after leaving in 1978, to join forces with the Tatas, a family of Indian industrialists. The joint venture, Tata Information Systems, will manufacture PS/2 and PS/VP systems and develop software exports.

Citicorp Overseas Software, a Citicorp subsidiary, typifies a growing trend to locate software-development units in India. “Our charter is first and foremost to meet Citicorp’s internal requirements,” says CEO S. Viswanathan, “but we are a profit center and can market our services and products.” On a tour of its SEEPZ facility in Bombay, I saw MVS, Unix, VMS, and Windows programmers at work on a variety of projects. In addition to reengineering work for Citicorp and other clients, the company markets banking products called Finware and MicroBanker.

ITC (Bangalore) supplements its Oracle, Ingres, and AS/400 consulting work by selling the full range of Lotus products. “Because we have the rights to manufacture Lotus software locally,” says vice president Shyamal Desai, “1-2-3 release 2.4 was available here within a week of its U.S. release.” Other distributors of foreign software include Onward Computer Technologies in Bombay (NetWare) and Bombay-based Mastek (Ingres).

India’s ambitious goal is to quadruple software exports from $225 million in 1992 to $1 billion in 1996. To achieve that, everything will have to fall into place. It would be a just reward. India gave much to the international microcomputer industry in the 1980s. In the 1990s, the industry just might return the favor.

Posted in .

3 thoughts on “Digital sweatshops then and now

  1. Thankyou so much for sharing this, both the digital sweatshop observation but as always the nostalgia and reminisces of Byte magazine. Boy that article takes me back. Inmos Transputers, A/S400, Intergraph all those old names and technologies. And also the hopeful promise each brand name, technology brought with it. It was quite wide open, diverse industry. One more thing that article on India reminded me of, apart from India not having a Mainframe legacy to deal with, but wired telecom legacy either. I dare say that helped usher in the cell phone far and wide, as it has done in other countries too. If you don’t have twisted pair copper infrastructure, you don’t need to wait to build it. Just wait and leap frog it altogether when the new, NEW thing comes along.

    1. > not having a Mainframe legacy to deal with, but wired telecom legacy either


      Btw, the 64Kbps leased line to the satellite that’s mentioned in that article wasn’t just for that once company, it was for the entire “software industrial park” in which the company was located.

Leave a Reply