The Daemon, the GNU and the Penguin - Excursus: Hardware & Ch. 9 - by Dr. Peter H. Salus

by Pamela Jones

May 19 2005

Here is "Excursus: Hardware" and Chapter 9 of Dr. Peter H. Salus' book, "The Daemon, the GNU and the Penguin," which he is publishing in installments on Groklaw under the Creative Commons license, 2.0, attribution, noncommercial, noderivatives.

Here are the earlier chapters:


The Daemon, the GNU and the Penguin

~ by Peter H. Salus

Excursus: Hardware

J. C. R. Licklider wrote that computers were communication devices, not calculating devices. Tomlinson's creation of email (1970) was a step demonstrating that. The continued expansion of the Internet provided the medium. UUCP and Usenet provided further impetus.

In January 1976, there were 63 hosts on the Internet. Five years later, there were just over 200. In August 1982, there were 235. For nearly 20 years thereafter, the number of Internet host sites doubled yearly: 562 in August 1983; 1024 in October 1984; 2308 in February 1986; 28,174 in December 1987; 727,000 in January 1992. But the slope of that curve has flattened. We no longer talk about hosts; we talk about users. But that's hard to estimate. How many people own a domain ... or several domains? And how many users are "on" some major domains? And what's the average? So it's largely guesswork. Around 2002, there appeared to be about 300 to 500 million users of the Internet. Right now, perhaps a sixth of mankind has access.

Two factors drove this (in my opinion): the development of the modem and the affordable personal computer. [This Excursus is not intended to be a complete history; my aim is to provide background so that subsequent chapters will be more intelligible.]

A modem (modulator-demodulator) sends and receives data between two computers. The first commercial modem, the Bell 103, was built by AT&T in 1962. It had full duplex transmission, frequency-shift keying and operated at 300 bits per second. Things got a bit speedier over time: 1200 bps, 9600 bps, 14,400bps. Robert Lucky invented the "automatic adaptive equalizer" at Bell Labs in 1965. Brent Townshend, a Quebec inventor, created the 56K pulse-code-modulated modem in 1996. Things had gotten a lot better.

Personal workstations and personal computers are not new, either. I can recall the LINC of 1963, for example. But these machines, as well as others, though of historical importance, had little impact on where we are now.

But we do need to look back at the 1976 Apple 1, running at 1MHz with 8kB RAM (max. 32KB); the Apple ][ (1977), the ][e (1983), and the Macintosh SE (1987). And they had their beginnings in the Altair. (Aspects of all these were developed a decade earlier, many by Doug Engelbart.)

In 1973 and 1974 a small company in Albuquerque, NM, called MITS (Micro Instrumentation Telemetry Systems), which had been producing inexpensive calculators, was seeking a new product. Texas Instruments had just taken over its calculator market. Ed Roberts, with the help of Les Solomon, decided to build a computer kit. Assisted by two hardware engineers, William Yates and Jim Bybee, they developed the MITS Altair 8800, which was featured on the cover of the January 1975 Popular Electronics. 1 The magazine called the Altair the "World's First Minicomputer Kit to Rival Commercial Models." It shipped for $400, but the purchaser had to assemble it, get it to work, and write the necessary software. It wasn't all fun, but it sold.

Among others, it sold to a Harvard freshman, Bill Gates, and his friend Paul Allen. They compiled a version of BASIC to run on the Altair. (Roberts offered Allen the post of Director of Software at MITS -- he was the only person in the department. Gates joined him upon leaving Harvard.)

The Apple ][e ran on a SynerTek 6502 board. The SE was 8MHz, 256kB, ran on the Motorola 68000, and had a serial port into which a modem could be plugged. It sold for just under $3000. (While I was Executive Director of USENIX in 1987, Telebit gave me a QBlazer and I connected the SE from home to the office using Red Rider.) Real power. And there were nearly 300 groups on Usenet.

But in 1975, IBM had tried to enter the "small machine" market with its 5100 -- a 50-pound "portable" computer, priced at $9000 to $20,000. It was a dismal flop. ("There is no reason anyone would want a computer in their home," Ken Olsen, founder of DEC, said in 1977.)

IBM then contemplated buying up Atari, but instead set up an "independent business unit" in Boca Raton, FL, to build the "Acorn." 2 The team of a dozen engineers was headed by William C. Lowe. They made a number of unusual decisions, the most notable of which was that the PC would have open, rather than proprietary architecture. They also decided to save time by purchasing an operating system, as well as making hardware components open to competitive bids. Finally, the PCs were to be sold through retail channels.

Released in August 1981, the original PC ran on a 4.77MHz Intel 8088. It had 16kB of memory, expandable to 256kB. It was priced at $1565 and launched through a brilliant advertising campaign featuring a Charlie Chaplin look-alike. By 1985, IBM's sales had overtaken Apple's and IBM had 40% PC market share.

However, the same open architecture that made the IBM PC a success, led to its decline. Open architecture meant that others could clone it, and the first of these was Compaq, coming out with an 80386-based machine in 1986. Others followed. By 1995, IBM's market share had dropped to 7.3% and in 2003 it was 6%.

In 1981, the Osborne 1, the first true portable had been released. It had 64K RAM; twin 5.25", 91K drives; and ran on a Zilog Z80 at 4MHz. Though the company eventually failed, the machines were definite landmarks. 3

Smaller size; lower price; access via the ubiquitous telephone. That did it.

Chapter 9. MINIX

Like Richard Stallman, Andy Tanenbaum was born in New York. After graduating from high school in White Plains (just north of New York City), he went to MIT, and subsequently received his doctorate from Berkeley. Since 1971 he has lived in the Netherlands (his wife is Dutch) where he's Professor of Computer Science at the Vrije Universiteit [Free University, VU].

After AT&T's 1979 announcement of the V7 licensing restrictions, precluding the use of the code in classrooms, Andy decided that the solution lay in his "helping himself."

Together with Sape Mullender, of the CWI [Center for Mathematics and Computer Science], Andy had originated the Amoeba project. Amoeba was one of the earliest attempts at a distributed operating system, contemporary with Roger Needham's work in Cambridge, and preceeding LOCUS, CHORUS, V, and Mach. [Tanenbaum & Mullender, in Operating Systems Review 13(2): 26-32 (1981)] That same year, Andy's valuable Computer Networks (Prentice-Hall) appeared.

And, while doing research, teaching classes, supervising graduate students, and writing, Andy worked on his own system. "It took about two years," he told me. He called it MINIX.

MINIX was a micro-kernel UNIX clone. While it emulated UNIX, it contained no AT&T code -- not in the kernel, not in the compiler, not in the utilities. It was 1986. The next year, 1987, Operating Systems: Design and Implementation came out, the book's title reflecting the VU course that Andy was teaching. At first, the code (v1.1) was only available on diskettes from Prentice-Hall. Soon it was available without the book.

MINIX 1.3 was on five 1.2M floppies and cost $60; MINIX 1.5 (1991) came on 12 720K diskettes at $169. 1.5 contained the accumulated bug fixes; it was V7 system call compatible; it would run on the IBM PC, PC AT, PC XT, PS/2, and 286/386 as well being "available" for the Atari ST, the Macintosh, the Amiga, and the Sun SparcStation 1, 1+, or IPC. 1.5 had a K&R compatible C compiler, a Bourne-like shell, five editors (ed, ex, an Emacs subset, a vi clone, and "a simple screen editor"), and a great deal of other goodies.

MINIX was intended as a teaching tool, and it was far from freely redistributable. It was under copyright by Prentice-Hall, but with rather liberal copy and revise/extend restrictions. However, it was certainly not free. As I understand it, the Prentice-Hall lawyers were to blame here, not Andy. MINIX now is free: the license has been redone and made retroactive. MINIX 2 is freely redistributable software.

1The Altair was named by Solomon's 12-year-old daughter, after a Star Trek episode.

2 See Pugh, Building IBM, pp. 313f., and P.E. Ceruzzi, A History of Modern Computing, 2nd Ed. (MIT Press, 2003), pp. 268-273.

3Adam Osborne (1939-2003) was a fascinating individual, but a poor businessman. See his Hypergrowth: The Rise and Fall of the Osborne Computer Corporation (1984).

Dr. Salus is the author of "A Quarter Century of UNIX" [ ] and several other books [ ], including "HPL: Little Languages and Tools", "Big Book of Ipv6 Addressing Rfcs", "Handbook of Programming Languages (HPL): Imperative Programming Languages", "Casting the Net: From ARPANET to INTERNET and Beyond", and "The Handbook of Programming Languages (HPL): Functional, Concurrent and Logic Programming Languages". There is an interview with him, audio and video,"codebytes: A History of UNIX and UNIX Licences" [ ] which was done in 2001 at a USENIX conference. Dr. Salus has served as Executive Director of the USENIX Association.

This work is licensed under the Creative Commons Attribution-NonCommercial-NoDerivs License. To view a copy of this license, visit or send a letter to Creative Commons, 559 Nathan Abbott Way, Stanford, California 94305, USA.

09:33 AM EDT

Copyright 2005 -