That proved it. There was a memory flaw in MAXC—only not where everyone (including Thacker) had been searching for it. What they had overlooked was that MAXC actually had two ports into memory, one via the central processor and the other through the disk controller. Because Thacker's test program ran from inside the machine, it surveyed only the processor port, which worked fine. But Munger, like the ARPANET link that had stymied Metcalfe, ran as an external program from the disk—where the memory port was indeed broken. "My fucking program found the bug," Metcalfe recalled, "and Chuck never forgave me."
"I believe firmly to this day that Metcalfe misread Chuck," Taylor said later. "Thacker liked to have bugs pointed out, because he loved to fix them." On the other hand, not many of Thacker’s colleagues displayed Metcalfe's pure delight in rubbing it in. He even had a rubber stamp made up echoing a catchphrase from the movie Love Story. It read, "Reliability is never having to say you're sorry."
"I used to stamp that on all the memos I was writing," Metcalfe said many years later, still grinning at the thought. "Chuck hated it. Poor Thacker!"
After finally getting MAXC hooked up to the ARPANET, Metcalfe moved on to the challenge that was to bring him and Boggs together. This involved finding a simple and reliable way to connect PARC's Altos to each other. The local network was the sine qua non of interactive distributed computing, Taylor believed: He was after more than the symbiosis of one man and one machine, but rather the unique energy sure to issue from joining togedier a multitude of people and machines all as one.
Unfortunately, none of the network architectures then in use suited PARC's specifications. The ARPANET was too large-scale and required too much extra hardware to link computers together in discrete local networks at a reasonable cost. IBM and several other computer manufacturers had developed their own proprietary systems, but they were specifically tailored to their own machines and difficult to adapt to others. They also tended to break down when the local loop got too large. The POLOS group's adaptation of a network technology provided by Data General for its Nova minicomputers underscored these shortcomings.
The networks maximum capacity was fifteen computers. POLOS's attempt to double the number had produced a multi-tentacled horror of cable and hardware. "We were able to network up to 29 Novas, but that was the limit," Metcalfe recalled. "The ultimate 29-Nova daisy chain had twenty-eight 40-conductor cables, sixty 40-conductor connectors, and the nasty habit of crashing if any one of these fragile devices was disturbed." The basement room at PARC where all these cables came together was aptly labeled the "rats nest."
Obviously this would not do for Taylor, who envisioned a system linking hundreds of Altos. His other specifications were similarly stringent. The network had to be cheap—no more than 5 percent of the cost of the computers it was connecting. It had to be simple, without any fussy new hardware, in order to promote long-term reliability. It had to be easily expandable—unlike POLOS, where adding a Nova meant taking down the network and splicing a new line into the rat s nest. And it had to be fast, because it would be feeding files to Gary Starkweathers swift laser printer and would need to keep up.
When Metcalfe first got to PARC he found several networking schemes already percolating on CSL's back burner, none of them to his liking. One was a local version of the ARPANET (but 1,000 times faster) devised by Charles Simonyi, who had finally rejoined his old Berkeley Computer colleagues at PARC. Simonyi’s design was nicknamed SIGnet, which stood for "Simonyis Infinitely Glorious Network." Metcalfe studied the specifications for about a week before rejecting it for having "too many moving parts for a local network."
He started to look elsewhere while a deadline loomed. Thacker’s Alto schematics, which were coming together around the end of 1972, left a blank space where the network controller was supposed to fit. If Metcalfe could not come up with something to fill the blank, the matter would be taken out of his hands—which would be not only a challenge to his intellectual authority as the network guy, but a blow to his pride.
That dismal outcome was averted when he suddenly recalled a concept he had first encountered months earlier. Back in June, while visiting Washington on ARPANET business, he had lodged on the guest room sofa-bed of his friend Steve Crocker, an ARPA program manager. Late that night he pulled down from a handy bookshelf a heavy volume of papers from an obscure technical conference, "a sure cure for jet-lag sleeplessness," and lumbered his way through one written by a University of Hawaii professor named Norman Abramson.
Abramson's paper described ALOHAnet, a radio network designed to allow computers to communicate with one another along the Hawaiian archipelago. ALOHAnet was loosely derived from the ARPANET, as could be seen from the nickname of its central control computer: Menehune, a mythical Hawaiian "imp." Metcalfe was annoyed by the pun but intrigued by the scheme. ALOHAnet messages were transmitted in discrete digital packets through the atmosphere. Because air is a passive medium (in contrast to, say, an electrically charged phone line), that feature made the system fetchingly simple. Abramson further described the networks clever means of handling the interference that occurred whenever two or more stations tried to transmit simultaneously. If they failed to hear an acknowledgment from the receiving station indicating that their messages had arrived safely, they retransmitted after waiting a random interval so the messages would not collide a second time. This, Metcalfe perceived, would be a highly useful feature in a local network where scores of computers might be trying to send messages on the same line.
The main limitation of ALOHAnet appeared to be its tendency toward gridlock. The paper suggested that the channel could be loaded up to only 17 per cent of its capacity before breaking down into a incoherent jabber of retransmitted and recolliding messages.
"That can't be right," Metcalfe said to himself, propped up on Crocker's sofa-bed. It had not escaped his notice that Abramson's figures were not based on experience—the existing ALOHAnet linked only seven computers—but on theory, and misapplied theory at that. He realized Abramson had made two impossible assumptions: That the number of users was infinite, and that each one kept mindlessly typing even after the acknowledgments stopped coming. No wonder the model filled up with messages and retransmissions until it crashed like an overloaded blimp. "Totally unacceptable," Metcalfe thought.
But suppose one imposed a couple of real-life assumptions on Abramson's model? Such as that the system had a finite number of terminals— thirty, forty, even a hundred—and that users stopped transmitting if the system stopped responding. In that case, Metcalfe calculated, the system should remain stable even at 90 per cent of capacity.
Cheap, simple, and capacious: Back at PARC, he realized that ALOHAnet possessed most of the qualities the lab sought in a local network. Over the next few months Metcalfe worked to adapt it to the centers high-volume, high-performance specifications. He junked the central control computer, Menehune, because each Alto would control its own transmission rate. He designed a scheme by which each station would listen to the line and stop transmitting the instant it heard any interference, instead of continuing to chatter. And rather than transmit via radio, he proposed joining the Altos by some sort of physical line.
The key element was that the medium had to be inert. Metcalfe understood that if the line had to carry an electrical current to aid transmission, like a phone line, Murphy's Law would take over. The line voltage would become the component most vulnerable to failure. But if there was no power on the line, Murphy would be defeated. It was possible and much better, he reasoned, to send messages into a passive medium, like the " 'luminiferous aether' once thought to pervade the universe as the medium for the propagation of light."
On May 22, 1973, he drafted his first memo describing the concept for PARC's patent attorneys. Subject: "The ETHER Network." Soon after that, he met David Boggs for the first time.
Meanwhile, Boggs had found his own separate way to PARC— escorted, as had been so many
others, by Alan Kay.
Like all of Stanford's grad students in electrical engineering, Boggs had been sentenced to snooze through a weekly one-hour seminar in the department's largest lecture hall featuring a talk from some person prominent in industry. The point was to inoculate the ripening "double-E s" with the excitement of engineering in the real world. For the most part, however, the sessions merged into a single soporific ten- week drone.
That is, until Kay showed up shortly before Christmas 1972. He started speaking in general terms about the interesting work taking place at the research center Xerox had opened up across the street from the campus. Then he put up a series of slides of a machine he and his colleagues had built to mimic the PDP-10. Boggs shook off his torpor and sat upright. High-performance computing was his field. He understood that any group that could build a PDP-10 from scratch was something special. When the seminar ended at five and everyone was free to leave, he bolted down front and subjected Kay to close questioning.
The latter, who was always on the lookout for potential recruits with what he called "special stars in their eyes," noticed the telltale stellar glow in Boggs s. Knowing that Novas were starting to arrive at the rate of two or three a week for the POLOS team and that they needed someone to assemble them to make sure nothing was dead on arrival, he forwarded Boggs’s resume to Bill English, who hired him to work part-time through the end of the Stanford school year. Boggs was duly anointed keeper of the test stand, which was a steel rack erected in the basement of Building 34. It held a perfectly functional Nova, the cover removed and the parts arranged to be easily accessible in case they needed to be swapped with those of a balky machine to determine which piece was causing the glitch. To the rest of the lab Boggs seemed rather a solitary figure in his basement lair. But he was available for kibitzing the day Bob Metcalfe stumbled by, hauling his bale of yellow co-ax.
Metcalfe had sketched Ethernet out in a series of memos with a fair amount of input from Thacker and Lampson—"inventing the network in real time, working out bits and pieces of the idea," as Boggs later recalled. But until then he had made no effort to determine if the parts would work together in the real world as well as they did on paper. The most critical question concerned the cable—the passive ether itself. An electrical pulse, Metcalfe understood, becomes attenuated, or stretched, as it travels along a wire. The longer the distance, the worse the resulting dilution and the more difficult for a receiver to recapture the original data. As he and Boggs soldered the test apparatus together, he explained that this was the reason he needed to fire pulses down the cable and read what came out at the other end.
"I have to know how bad it is," he said.
After that day they did not encounter each other for a couple of months, until Metcalfe reappeared in the basement one afternoon, this time holding a small piece of hardware he had designed to connect the POLOS Novas to the ARPANET.
"Can I smoke test this on your rack?" he asked Boggs. (The allusion was to a procedure that works exactly as it sounds: You shoot a voltage surge through a circuit to test whether some hidden fault will make it burst into flame.)
They spent a week or two testing the circuit together for a few hours each day. Debugging a complex electronic device being almost as powerful a bonding experience as, say, serving on a submarine in wartime, Metcalfe learned a lot about his partner: That he was a digital whiz, accomplished at wielding the oscilloscope, and, most interesting, underemployed in his POLOS work. Presently the pair showed up at Bill English's office door, figuratively holding paintbrushes and a bucket of whitewash. "Metcalfe wants me to work on something with him for a while," Boggs said. "Is that okay with you?"
Having secured English's acquiescence they walked on down the hall to Metcalfe's office, where Metcalfe raked together a thick wad of memos comprising the Ethernet invention record he had assembled for Xerox's patent department. "Go read this," he said.
"That," Boggs recalled, "was pretty much the last time SSL got any work out of me. For the next twelve months at least I spent every working day with Metcalfe."
They slept when they were exhausted and the rest of the time they worked, as unconscious of alarm clocks or the sun as casino players on a roll. "There was no chip on the Ethernet board that both of us didn't know about," Metcalfe recalled. "There was no line of my microcode that Boggs did not understand. We worked on the whole thing together, every minute, every piece of it." Boggs was placed on the payroll full-time for the summer and stayed even after the school year resumed, placing his Stanford Ph.D. studies on hold. He did not finish his doctorate for another nine years.
Metcalfe, by contrast, resubmitted his doctoral dissertation to Harvard, fattened up with a properly theoretical digression covering the ALOHAnet. In June 1973 his thesis, entitled "Packet Communication," was finally accepted ("without enthusiasm," he later groused).
As a working system Ethernet differed from other PARC inventions in one crucial detail: It was explicitly designed to be imperfect. Metcalfe labeled the network a "best efforts" system—that is, the computers were instructed not to rely on everything working perfectly. This ensured that the system would not crash in the event of a single minor glitch (or even a torrent), of the sort certain to crop up in a network of bug-prone experimental computers. "I loved it," said Kay, one of its earliest fans. "It was one of the great finesses of all time, an object lesson in how to make something work when you don't know how to make it work well."
Ethernet's basic procedure resembled getting somebody's attention in a crowded library by the most efficient, if crude, method: by shouting. The ether—that is, the coaxial cable connecting the Altos—was usually silent. When a machine was ready to transmit a message, it shot a wakeup bit onto the ether, alerting every other machine that something was about to happen. Then it sent a packet comprising, consecutively, an eight-bit destination address (the digital tag of the Alto for which the message was intended); its own address; the message itself; and a string of verification bits known as a "checksum." Receiving stations would check the destination address to see if the message was intended for them. If so, they would copy the whole packet into memory; if not, back to sleep.
Meanwhile, the transmitting station would listen for any sign that its packet had collided with another machine's. If it detected interference, it would instantly stop sending, count off a random delay (as would the transmitter of the conflicting message), and send again. The listen-and- retransmit process could be repeated as many as fifteen times before the machines would give up.
As much an enemy of "biggerism" as Thacker, Metcalfe implemented these complicated electronics on the single circuit board the Alto design allotted to Ethernet by stripping the system down to its bare essentials. The original Ethernet board did not even have a timer of its own, relying instead on the Alto's internal clock for the critical duty of synchronizing transmissions.
Toward the end of the design phase, however, Boggs insisted on adding one feature he deemed crucial. This was the "checksum," a bit sequence that would enable the receiving station to verify that a message had not been subtly garbled in transmission.
"Sure, David," Metcalfe said. "If you can find room on the board to fit the checksum logic, you can add it."
This struck Boggs as a little cynical. A checksum system would require at least eight integrated circuits, or chips. Of the sixty chip positions on the boards they were using, fifty-nine were already occupied. Then he noticed that just enough space remained around the margins to wedge in a few more chips. By the time he was done there was scarcely a millimeter of unused room. Some chips literally hung off the edge of the board, like refugees clinging to a packed lifeboat. But Ethernet got its checksum.
With that, Metcalfe and Boggs s invention proved as facile and forgiving as they had hoped. Adding new machines, or "nodes," to the system without interrupting service for even a split second was a cinch: One punched a tiny hole in the main co-ax and, using a simple piece of cable TV hardware
called a "Jerrold tap," plugged the needle-like end of a branch cable into it. (This stratagem was suggested by David Liddle, a POLOS engineer and a basketball-playing crony of Metcalfe's, whose familiarity with Jerrold taps dated from his college job as a cable TV installer.) The network proved almost infinitely expandable while remaining emphatically simple, not much more than a cable terminated at both ends that anyone could tap into as easily as a water line.
Yet the Alto's first users were disconcertingly slow to get on the Ethernet bandwagon. Because the network connection was a costly $500 budget option on the first machines, many PARC engineers chose to dispense with it altogether. This was especially true as long as the network appeared to be useful mainly for sending files between computers—a superfluous function, since the Altos were equipped with removable disks that could easily be transferred from one machine to another. "Ethernet was up against 'sneakernet' from the very start," Metcalfe recalled.
All that changed overnight in 1975 with the advent of SLOT, Starkweather's laser printer. The virtues of the combined system called "EARS"—the Ethernet, the Alto, the research character generator, and SLOT—-were too powerful to ignore. One could now write a memo, letter, article, or dissertation and with the push of a button see it printed in professional-quality type. ("Before that, you had to have an article accepted for publication to see your words rendered so beautifully," Liddle mordantly observed much later. "Now it could be complete rubbish, and still look beautiful.")
Dealers of Lightning Page 23