flak rss random www

some more books 1

There was a lot happening here. The summaries started getting really unwieldy. Just listing all the relevant names in some cases would stretch to several paragraphs. I’ve tried to streamline things a bit, but tons of great material has been cut out.

soul

Data General, founded by some renegade DEC engineers, notably Edson de Castro who DEC does not acknowledge as the designer of the PDP-8, makes piles of money. The big computing giant of the time was IBM, in the midst of a decade long antitrust trial, but they sell mainframes, not minicomputers. After ten years in business, at the end of 1978 Data General landed in the Fortune 500. They’d survived one disaster in 1973 when a parts shortage meant they weren’t shipping anything, putting several customers out of business.

Tom West is in charge of developing the Eagle, Data General’s 32 bit minicomputer. Data General’s rivals are starting to sell 32 bit minis, including the DEC VAX 11/780. It’s ok not to be first, but it’s bad to fall too far behind. West finds a friendly VAX customer who lets him in to the server room to take it apart. He’s not too impressed. The VAX is a complicated machine.

Stepping back, Data General had been trying to make a 32 bit mini for a while. In 1976 West had taken over the Eclipse group, the successor to DG’s initial Nova system, as the original architects moved on to the Fountainhead Project (FHP), the 32 bit system. As a result of some political gamesmanship involving tax breaks, Data General had just opened a new office in North Carolina, to which FHP was relocated. de Castro then made a number of public statements about the NC office (“that’s where the talent is“) which didn’t really inspire the home office.

Some of the employees who remained decided to work on EGO (the HAL to FHP’s IBM), but de Castro cancelled it after internal arguments. West regroups and considers simply extending the Eclipse to 32 bit operation. This would offer software compatibility, which customers like. IBM had done this, at great expense, with the 360 line. Customers have large investments in their software. Switching systems means throwing all that away. Now called Eagle, this project receives internal support, but the messaging is split. To other divisions in the company, West is telling them it’s just a modest improvement over the Eclipse until FHP takes off. To the engineers working on it, he tells them it’s the machine of the future.

Com. It’s not clear if the FHP was Eclipse compatible or not. That was a major point in Eagle’s favor, so why would the flagship project omit it?

showstopper

Dave Cutler was a football star before breaking his leg in a college game. After graduating, he turned down a job programming computers at GM. There’s no creativity in being forced to follow some machine’s rules. So he went to DuPont, but ends up using the computer anyway. He goes to IBM programming school for a week, and comes back with a changed attitude. He gets to control the program. He wants to be a great programmer.

The idea for a stored program, in which the operation of the computers is stored within its own memory, originated with John von Neumann in 1944. Grace Murray Hopper invented the compiler to make it easier to enter ones and zeroes. Languages like Fortran and Cobol were developed, but software was still often tied to a particular machine. In the early 60s, IBM developed the 360 line to allow any software program to run on any hardware system.

Cutler gains experience working with DEC PDPs at DuPoint, then jumps to DEC to work directly with the machines. He quickly earns a reputation as a great coder and a great shouter. He catches the eye of Gordon Bell. In 1975, the PDP line is falling behind, and Bell has the answer. The VAX, running VMS. It will be backwards compatible with the PDP, and upwards scalable. Cutler is picked to lead development on the VMS operating system. Bell formally launches the project on April 1, 1975. (I always knew VMS was a joke.)

The VAX ships in 1977, and everyone praises Cutler’s accomplishment. “To his distress, he could no longer direct the VMS project with his customary authority. The program was now so important to the company’s future that a bevy of managers wished to improve it.” He threatens to quit, but Bell convinces him to stay by offering to let him lead his own team and develop something entirely new. In 1981 he moves to an office in Seattle, far away from headquarters. But after Bell leaves in 1983, his support wanes. In 1985 he barely gets approval for an entirely new computer design, Prism, running a new operating system, Mica. In 1988, DEC cancels the project.

Bill Gates was not a football star. When the Altair comes out, Gates and Paul Allen, write a Basic compiler for it. The Altair dies, but Intel’s chips do not, and the Microsoft Basic compiler works for all the systems using it. Later, they (re)sell DOS to IBM, but keep the rights, selling them to clones. By 1988, Gates and Microsoft are filthy rich, IBM is tiring of competitors, and they start work on OS/2. But customers aren’t taking to it, in part because it lacks compatibility.

Nathan Myrhvold wrote a clone of IBM’s TopView program, which Microsoft bought because Gates didn’t want IBM getting ahead, although IBM then cancelled TopView, leaving Myrhvold at Microsoft. He warns Gates of two coming trends. RISC processors that won’t run DOS, and Unix. Gates learns that Cutler is unhappy at DEC, and offers him a job building the next generation of system at Microsoft. Cutler arrives October 31, 1988.

Com. It’s unclear to me what Cutler was doing between 1977 and 1988. Four years unhappy with the VAX project, then seven whole years in Seattle? That seems like an awful lot of time.

troublemakers

Bob Taylor works at the Pentagon, running the Information Processing Techniques Office (IPTO) at Advanced Research Projects Agency (ARPA). He once refused to use a punch card computer, doing the calculations by hand because it was still faster. But then he reads a paper “Man-Computer Symbiosis” by J. C. R. Licklider which inspires him. Computers need to be interactive. Two years later they meet, while Taylor is at NASA and Licklider is running IPTO. In 1965, he joins IPTO as assistant to Ivan Sutherland, Licklider’s successor. ARPA had funded the initial dedicated computer science departments at universities across the country, with a focus on timesharing machines which allowed multiple interactive users.

Taylor travels around the country, visiting ARPA sites and talking to the funded graduate students. He learns that many of them met friends and colleagues via timesharing computers by logging in and seeing what others had done. What if we can connect more computers, connecting more people?

At the time, many incompatible systems prevailed. In order to use a program written by others, it often had to be rewritten. If the computers were networked, then everybody would have access to all the programs and data.

Taylor however doesn’t understand computers at a sufficiently low level to lead such a project himself. He needs someone to run the project at a technical level for him. The ideal candidate, Lawrence Roberts at MIT, doesn’t want to work for Taylor, so Taylor blackmails MIT by threatening to cut their funding unless Roberts takes the job. And initially the universities that were supposed to develop the network were reluctant, since a network would mean sharing their own CPU cycles with outsiders, so again funding cuts were threatened. With things underway, Taylor and Licklider outline their vision for the network in a paper, “The Computer as a Communication Device”.

Taylor, incidentally was also a big supporter of Doug Engelbart, even before ARPA. In October 1967, he attended a demo of a presentation in which information was presented via screen sharing. At the time, few people were aware of what Engelbart was up to. The Mother of All Demos would not take place until December 1968. Taylor grows tired of the increasing military focus of ARPA, as it becomes DARPA, and leaves.

Al Acorn is a student at Berkeley in 1969 and works at Ampex, the company that made the first audiotape recorder in 1948, and then the first videotape recorder. Among other people who worked at Ampex at some point include Larry Ellison and Ray Dolby. His office mates are Ted Dabney and Nolan Bushnell. Acorn went back to Berkeley for a semester to avoid being drafted. When he returns to Ampex, Bushnell is gone and Dabney is thinking of leaving. They are going to make a video game.

At the same time, 12 year old Fawn Alvarez is stuffing envelopes at ROLM, where her mother works. Vineta Alvarez started work at Lockheed, which trained her in soldering and related skills. She then went to work at Fairchild Semiconductor, then Sylvania. She also did some side projects on the kitchen table for extra money because state law prohibited women from working overtime. Several former Sylvania engineers founded ROLM, and they want Vineta to be in charge of production. ROLM’s plan is to take Data General minicomputers and harden them, selling ruggedized versions to the military.

At the same time, Mike Markkula is meeting his former boss Jack Gifford. Gifford convinced Markkula to come work at Fairchild Semiconductor in 1966. Now he’s founding a new company, Advanced Micro Devices, and wants Mike to join. Lots of people are leaving Fairchild. The previous year, Robert Noyce and Gordon Moore founded Intel. Before that, Charlie Sporck led an exodus to National Semiconductor. But Markkula is happy to stay put, since he is advancing rapidly where he is. But a year later, in 1970, Fairchild is not as exciting as it once was, so Markkula accepts an offer at Intel. After negotiating for a pile of stock options, 20x the normal amount.

At Stanford, Niels Reimers works in university contract administration office. At the time, practice was for inventions discovered by university researchers to be sent to Research Corporation in New York, which would patent them, then distribute royalties to the university. Between 1954 and 1967 a total of $2944.91 had been paid out in this manner. Not much. Research Corporation itself wasn’t much motived to license its patents. Reimers wants to take a more active approach, marketing research to industry for development.

Com. For a guy without much computer education, Bob Taylor certainly got a lot done. Perhaps it was his distance from the field that let him envision more applications for computers than simple number crunching. So far the other players haven’t done much, but here they are.

dealers

We start with Bob Taylor’s retirement party in 1996. Doug Engelbart is here. So is Charles Simonyi. But many people are not. “The common knowledge was that for every guest who owed a career to the guest of honor there existed not a few individuals who had felt the sting of Taylor’s rivalry and damned him as one of the most arrogant, elitist, and unprincipled persons on the planet.” And then another recap of his career at ARPA IPTO.

Late 1968, Xerox CEO Peter McColough has hired Jack Goldman to be chief scientist. He will be responsible for opening a new pure research facility. McColough also announces that Xerox will expand its business to “the office of the future” and provide “the architecture of information.” Perfect CEO speak. Xerox had previously tried to purchase DEC, but now they’re stuck buying SDS.

George Pake is the provost of Washington University, St. Louis, but he’s tired of college administration. He gets a call from Jack Goldman, with whom he’d worked at Westinghouse Research during WWII. Pake is wanted to run the new research facility. He’s skeptical it will maintain funding without producing results. In 1944 Haloid (as Xerox was then known) supported Chester Carlson’s research even though the first copier, the Model 914, didn’t launch until 1959.

Pake is in charge of staffing a computer research facility, but he doesn’t know much about computers. He does, however, know Bob Taylor. There’s a wrinkle though that Taylor doesn’t have a PhD which we’re going to overlook for now.

In late 1970, Xerox has missed its profit targets, and executives are looking to cut back. Why not cut off PARC, this new thing that hasn’t produced anything yet? Finally, board member John Bardeen, only two time Nobel prize for physics winner, says cancelling PARC would be a mistake. It survives.

Through Taylor, PARC raids the SRI office where Engelbart works, hiring away many of his assistants.

Com. A few fortuitous results here. What if Xerox had bought DEC? Xerox hiring Goldman hiring Pake hiring Taylor, all as they were ready to move on from previous positions.

zeroday

2010. Sergey Ulasen works at the small Belarusan antivirus firm Virus-BlokAda, reviewing samples when he notices one in particular from Iran is very complicated. It’s using a zero day .LNK exploit to spread. They received the sample when an Iranian customer reported their computer was stuck in a boot loop. Additionally, the driver installed by the exploit appears to be signed by a legit certificate from RealTek. On July 12, he discloses the discovery, and soon after a similar malicious driver is found, but it’s signed by JMicron instead. And then other variants of stuxnet, as it is now known, are found, with differing compilation dates. Some of them date to 2009.

Liam O’Murchu works at Symantec. He was only going to give stuxnet a quick examination, but realizes it’s much larger and more complicated than typical malware he sees. Stuxnet goes to great lengths hide its presence on a system. He asks DNS providers to redirect the command and control domains to his server, and records the traffic. Strangely, of the 38000 infections, 22000 were from Iran, a pattern not seen before.

Com. Stuxnet came pretty close to slipping by without notice. I guess maybe AV is good for something?

fin

Like reading the same story multiple times. DEC figures prominently. Bob Taylor is everywhere. Doug Engelbart gets plenty of attention. I skipped recounting a lot of smaller stories, but the same names weave in and out of the story.

We all know the story of the mother of all demos, but the private demo he gave earlier to a smaller audience is possibly more impressive. Every participant sat at their own terminal, watching on their own screen. He wasn’t just showing off interactive computing, but interactive communication.

We’re beginning to see signs of software eating the world. At first the major expense with computers was the hardware, but over time the software accumulated as well. The IBM 360 was a major development, although it remains in the periphery of our stories here.

Human nature is to detect patterns where there are none, but nevertheless some of these intertwined stories capture the imagination. If this or that project had worked out differently, or this guy had gone there instead of here, how would things be different? In some ways, some of these seem like ideas whose time had come, making them almost inevitable, but one can also imagine a timeline in which timesharing and networking had been pushed back many years. Where would be today in that scenario?

1979 article about the scope of the IBM trial. IBM trial enters 2nd decase.

The two papers by Licklider, one coauthored with Taylor. In Memoriam: J. C. R. Licklider. I found the second especially interesting. Among the great quips at the end is a prediction that computers will eventually eliminate all unemployment as everyone turns to debugging.

Contrasting accounts of DEC management. Tom West thinks DEC is very bureaucratic when he looks at a VAX, but the absence of management is what initially attracted Cutler to DEC. DG vs DEC versus DEC vs IBM.

Posted 24 Aug 2019 20:51 by tedu Updated: 24 Aug 2019 20:51
Tagged: bookreview