Computer Chronicles Revisited 4 — Singer Link and SOM

Normally, The Computer Chronicles highlighted consumer software and hardware. Stewart Cheifet often described his role as doing the legwork on behalf of the viewer so they knew what products to buy. This particular episode, however, goes in a somewhat different direction. The subject is simulator software, but aside from the opening host segment, the episode is largely devoted to proprietary software used in non-consumer applications.

Flight Simulators – Computer Game vs. Training Tool

Cheifet and Gary Kildall opened the episode by playing Microsoft Flight Simulator 1.0 on an IBM Personal Computer. Cheifet noted this was a good example of how you could use a computer to simulate real-world situations. Foreshadowing one of the episode’s guests, Cheifet added that simulation software could also be used in architecture and urban design. Cheifet asked Kildall that given everything a computer does is essentially a simulation, what do we mean by “computer simulations.” Kildall replied that it’s about generating scenes or a situation that someone wants to experience, which requires graphics and a lot of computing power. But it’s ultimately less expensive to do than the real thing. For example, it’s cheaper to run a simulation of a 747 landing than to land an actual plane.

Stewart Cheifet and Gary Kildall with an IBM PC running ‘Microsoft Flight Simulator’.

Cheifet then narrated a short feature on the use of flight simulators by the U.S. military to train pilots in air-to-ground combat. While this might resemble a computer game, Cheifet said these simulators included “random factors” to help better simulate real-life conditions. He compared the progress of computer simulation to film animation, which managed to achieve a “remarkable level of detail” during the 1940s and 1950s. Cheifet then introduced a computer architectural model where buildings “can be viewed at almost every perspective,” eliminating the need for hand-drawn blueprints.

As it turned out, both of these items were produced by the episode’s guests. The first guest was Frank Lewandowski, a senior scientist with the Link Division of the Singer Company, also known as Singer - Link. You may recognize the “Singer” name as that of a sewing machine manufacturer. It’s the same company. The Link Division developed the military flight simulation software introduced by Cheifet earlier.

Kildall opened up the discussion by noting his “second great love is aviation,” with his first obviously being computers, as that paid the bills. There was then another demo of Microsoft Flight Simulator, which was developed by Bruce Artwick. Cheifet noted this program was based on a Cessna. Kildall added the actual scenery in the game was rather choppy and not a “real time” picture. Lewandowski explained that Flight Simulator has a picture composed of 20 to 30 lines updated about twice per second. In military applications, there needs to be a “much more sophisticated scene.” A combat pilot could not accept the type of scenery seen in Flight Simulator. They require something closer to “8,000 edges for every TV frame time and to compute a new scene at each frame time,” rather than 2 or 3 times per second.

Kildall said that required more computing power than an IBM PC could achieve. Lewandowski agreed, noting that while a PC had a single integrated circuit, the Link Division’s simulators had 40,000 such circuits–quite a substantial difference.

Cheifet asked for more specifics about the types of flight simulations done at the Link Division. Lewandowski said that for commercial aircraft and airline pilots, the simulators were limited to takeoffs and landings. That was not that difficult to train. The more difficult applications were for the military, as those involved air-to-air and air-to-ground combat. This required simulating an entire environment, not just the mechanics of the flight. Lewandowski then narrated some footage from the Link Division’s military combat simulator, which showed a “real-time scene” of a Boeing AH-64 Apache attack helicopter firing a missile at a tank. Cheifet pointed out that the scenery was “not painted, but computer generated.” Lewandowski said yes, every part of the picture was generated 30 times per second, which was the “present state of the art.”

Kildall asked if the simulator used a database to represent the terrain. Lewandowski said it did. Everything in the simulated world, such as trees or mountains, were coded into an XYZ coordinate. Kildall noted this was what required so much computing power. Lewandowski added that over the next few years the goal was to develop “textures,” so that for example you could show the “foliage” on individual trees. Or as an enemy helicopter got closer, the simulator operator could see the “other pilot” and weapons officer. And of course, you could see more textures on ground buildings and equipment. Kildall closed the segment by asking how long it would be before we would see this type of detailed scene generation. Lewandowski said some people were getting close now with the use of video disc, which mixed live imagery with computer graphics. But the Link Division’s simulators relied entirely on computer graphics.

Using Proprietary Simulation Software to Stay “One Step Ahead” of the Competition

For the final segment this week, Steve Harrison joined Cheifet, Kildall, and Lewandowski. Harrison was the data processing manager at the architecture firm of Skidmore, Owings & Merrill (SOM) in San Francisco. Cheifet asked Harrison how an architect uses computer simulation. Harrison said his firm used it for “perspective drawing,” including the study of how shadows from a building affect the urban environment. Simulators could also test the effects of earthquakes on buildings, as well as the energy performance of a building over the course of a year as the sun traveled in different paths.

Cheifet asked Lewandowski how this differed from the simulators he works with at the Link Division. Lewandowski noted that architecture simulations did not have the “same constraints” as an aircraft simulator in that the former did not have to be in “real time.” That is to say, a flight simulator needs to react as the pilot operates the controls, and the scene must reflect those changes immediately. Cheifet asked Harrison to clarify that something like an earthquake simulator for a building was not in real time. Harrison said no, instead they would take a tape that was made from the “actual forces” of a prior earthquake and play it back over 4 to 5 hours of computer analysis.

Harrison then showed some perspective drawings that were made using his firm’s proprietary architecture simulator. The demonstration building was actually San Francisco’s City Hall (which was not designed by Harrison’s firm). These were wire frame drawings without much surface detail, which Harrison said were commonly used in “massing studies.” Kildall asked what type of computing power was necessary to run the simulation. Harrison said it ran on a VAX 11/780 mainframe using a Tektronix 4113 color terminal. Harrison added the software could also run on the “smaller” VAX 730 and 750 machines. Cheifet asked about the software itself. Harrison replied it was a proprietary program called “Draft” developed by SOM for perspective, 2D, and 3D drawing. Harrison then showed some simulation drawings of the Louise M. Davies Symphony Hall in San Francisco, which SOM designed. Harrison pointed to the high amount of window detail in these simulator renderings.

Kildall asked if this type of simulation software was being used throughout the architecture profession or just by SOM. Harrison said this particular software was “special to his firm,” but that other architects had systems “that begin to approach this.” He said SOM was trying to keep “one step ahead” of its competition in this regard. There were also a number of non-proprietary systems that had some aspects of SOM’s software, but his company offered an “integrated system” that applied to “all of the architectural and engineering disciplines in the firm.”

Cheifet asked if SOM’s software simulated the environmental impact of a structure. Harrison said it did, and in fact that was a major use of the system. He again referred to “massing studies,” which Cheifet asked him to define. Harrison explained that this simply referred to the mass of a building and how it fit into a “tight urban environment” like San Francisco.

Harrison then showed a final example, this time from a shadow study of the Bank of American Building in San Francisco, then known as the Bank of America Building. Specifically, this simulation showed the shadow cast each hour by the building. Cheifet asked if it was worth the cost of a computer simulation to do such studies. In other words, Was it that difficult to study a shadow manually? Harrison replied it was, especially if you are looking at making major changes to a building. Traditionally, architects build small physical models to do these type of studies. But in a city like San Francisco where there were a lot of different structures and interaction between shadows, it was easier to test a “large number of different designs” using a computer simulation.

Kildall followed up by asking if computer simulation could also be used for interior design and the placement of objects within a room. Harrison said yes, the SOM system is actually used for interior perspective as well as exteriors. He added that interior simulations were inherently more complicated, as there were a lot more edges inside of an “open office environment,” and that was difficult to show clearly using wire frame graphics.

Cheifet turned back to Frank Lewandowski to ask what he’d like to be able to do with military simulators going forward. Lewandowski reiterated the need for “much better textures,” particularly with respect to foreground objects like rocks. Such details mattered if you wanted to train someone to drive a tank, and that was difficult to do presently with the existing “geometric textures.” Kildall asked if it was possible to “digitize” real world properties to create more realistic textures. Lewandowski said they did some of that now, such as using civil engineering drawings of airports to extract certain data about those locations, such as where to place trees, shrubs, and radar towers.

Cheifet asked Lewandowski how it was possible to simulate the human factor, i.e., the “emotional response of human actors” in a computer simulation. Lewandowski said the simulator gives the impressio of sound and motion, as well as other cues that combine with the “realistic scenery” to give the pilot everything they need. Cheifet then closed the show by asking Harrison how he would like to improve his own simulator software. Harrison joked he’d liked to get one of Lewandowski’s systems from the Link Division. Short of that, Harrison agreed that more realistic scenery was the next stage.

Unfortunately, I was not able to track down anything definitive about Frank Lewandowski. I did come across a couple of obituaries that might have been for him, but nothing I could confirm. So let’s talk about Singer - Link instead. At first glance, it seems odd that a sewing machine company would have been involved in building military combat flight simulators. But during the 1960s, Singer Corporation decided to “diversify” its operations through a series of acquisitions, which included Link Aviation Devices, a company founded by Edwin Link in 1929 to market the Link Trainer (also known as the “Blue Box”), an early flight simulator he’d developed two years earlier. According a corporate history pamphlet, Edwin Link ran Link Aviation as an independent company until 1954, when it merged with General Precision Equipment Corporation (GPE). Link continued as president of the merged company until 1959. GPE then merged with Singer in 1968.

In 1981, Singer split the Link Division into two parts. The first part, Link Simulation Systems Division (later Link Tactical Simulations Division), was based in Maryland and focused on energy and industrial simulations. The second part, Link Flight Simulation Division, apparently continued to focus on flight simulators and was likely where Lewandowski worked. A few years after Lewandowski appeared on Chronicles, Canadian firm CAE Industries purchased Link Simulation Systems from Singer. By 1990, CAE had also acquired the Flight Simulation Division and consolidated the two divisions back into one unit. CAE renamed itself as CAE-Link.

Hughes Electronics Corporation purchased CAE-Link in 1995. Three years later, Raytheon purchased Hughes. Finally, in 2000, L-3 Communications acquired Raytheon. And despite all of the corprorate turnover, the Link simulation business continues in 2021 as L3Harris.

Harrison Moved from Architecture to Academia

SOM, Steve Harrison’s architecture firm, is also still in business today. Skidmore, Owings & Merill dates its own history back to 1936, when it was founded in Chicago by architects Louis Skidmore and Nathalien Owings. John O. Merrill, an engineer, joined the firm three years later. Today, SOM is known for its work on a number of famous buildings, including New York City’s One World Trade Center and the Burj Khalifa in Dubai, which is currently the world’s tallest building.

Harrison himself is no longer with SOM. He left the firm not too long after his Chronicles appearance. In 1985, he joined Xerox PARC as a member of its research staff. He stayed there until 2002, when he left to found two startup companies before entering academia. Since 2003, Harrison has been an associate professor at Virginia Tech’s Department of Computer Science and School of Visual Arts. He also serves as director of Virginia Tech’s Human-Centered Design Program and co-director of the Social Informatics Area for the Graduate School.

Notes from the Random Access File