Ed Lazowska began the class by giving a very brief overview of the class and whirlwind tour of some of the guest speakers who would be talking here. The great thing about the class is that we’ll get to hear about the history of computing from people who actually //created// a lot of the history. After some more introduction, Steve Maurer took over.
I really like listening to this guy. He talks like one writes a book. He’s articulate, he’s fluid, he keeps talking and he usually always makes sense. I think thats a rare quality for a public speaker, specially if you’ve to talk for 2-3 hours.
In general, we want a society where the value of inventions is consistently higher than the costs. That way we always increase the net value (or some metric of //happiness//) of the society. **All known incentive mechanisms are imperfect**. Basically we’re saying that there’s no one incentive that works for everyone in all situations. We need to pick the incentives depending on the //context//. Technology has improved, but society has learnt to consume more data as well. Inventions happen because a //consumer// (someone who is facing the problem) also has the technical know-how to put things together and come up with a solution. Rarely do these things occur in isolation.
Again, a recurring theme is that the incremental cost of technology has decreased, driving innovation and competition. 60 years ago, making a second computer was just as expensive as making the first one. But the ability to cheaply replicate components (think integrated circuits) is what led to the proliferation of the personal computer.
What was the value of “computing” in prehistoric times? Government uses, data for the military. Ancient libraries came up as a research tool, to enable knowledge transfer. Then we have the dark ages, where a data loving society collapses to one where no records are kept. Followed by renaissance and a revival in warfare and commerce. Tyco Brahe and Kepler got a machine built to compute trigonometric functions for their astronomical needs.
Then comes Pascal. Mathematical prodigy. Interesting anecdote: he was deformed at birth, and his father who was a tax official thought some client he didn’t serve well had bewitched their son. Pascal, as everyone would know, built the first mechanical calculator. Did you know that calculations for the atom bomb were also done on mechanical calculators? 50 “Pascalines” were built through 1652 and sold for 100 “livres”. The society at the time wasn’t “ready” to use such a technology.
In the late 19th century, the largest and most complex data processing consumers were the banks and the railways. The society’s data consumption was **exploding**. This was also the time of the rise of big businesses (precursors to today’s corporations) like Nabisco and Traveler’s Insurance. William Burroughs of the Burroughs Adding Machine Company made a lot of money out of these calculating machines. They had 58 models, one for each line of business. Fast and huge market penetration, and significant demand.
//Ex post// efficiency is that post the invention, the idea should be free for all. //Ex ante//, on the other hand, is to motivate people to invent in the first place but retain the monopoly once the invention has been made.
Punch cards were motivated by musical organs (where pins raised, so the opposite would be holes). The target market was to improve handlooms and automate the process of weaving using punch cards. Patents are not always good — toys can often lead to significant inventions and if all toys were to be patented, we would never make progress. There was also a debate between the British and the French on the subject of patents vs. prizes. Photography was born out of a prize, and at that time it was stifled by patents. The prize route ended up setting industry standards for photography technology.
Then the era of Charles Babbage and Ada, Countess of Lovelace. Babbage built the difference engine. A prototype was built after Babbage’s time. Babbage became dis-interested in the difference engine because he came up with the //analytical engine// — a steam powered, **programmable machine**. Its almost like the first Von Neumann, stored procedure kind of device.
Herman Hollerith and the age of IBM. Driven by the census challenge in 1880s. Taught at MIT, built Hollerith machine to “read” punch cards which held censor data. Interestingly, Hollerith had a //service model// with the Government for the Census. He “rented” machines for $1000/year with $10 for downtime. Also in part because they didn’t trust the technology as much. He did two census (1880 and 1890). Around the 1900 census, people start grumbling about monopoly of Herman Hollerith machines.
Then Hollerith went around selling the idea to corporations (New York Central railways). And adds functionality to the machine as demands show up (only counting was needed for the Census). He spins off the “Tabulating Company” (venture capitalism?) out of this effort, and eventually Hollerith machines end up all over the world. They start building peripheral business around it — accumulators, keypunches, sorters etc. Leads to defining of interfaces, standards, modularity and all these other concepts that we take for granted.
Hollerith institutionalized this ability to //find and discover// potential use cases and customers within his company. Later years see the decline of Hollerith’s market leadership followed by a merger. Interesting discussions are borne out of this turn of events: does monopoly fund innovation? How do mopolies and innovations stack against one another. Joseph Schumpeter came up with the notion of Schumpeterian competition where capitalism leads to monopolies, monopolies innovate, but only last until the next technological revolutions. And therefore, since in the long run we value technological progress more, monopolies are not only inevitable, they’re actually OK so long as there is competition at the innovation level.
Finally we have Vannever Bush and the Memex. Vannever Bush made the first power amp (it was mechanical!). He also built a differential analyzer, which was later improved upon by Claude Shannon. One thing to note is that throughout the past 100-200 years, a lot of the innovation came from grant money and universities. This is a key point, one that we shall have occasion to revisit in the future.
Back to Hollerith thread. The CTR company became IBM in 1924 with T.J. Watson as the CEO. Patent wars started as early as 1921. Practices such as customer lock-in and competing standards also raise their heads during this time. IBM used its patent position to force Remington-Rand into a joint monopoly pricing. We see this kind of thing again when in the 1980s IBM lets indepedent vendors build the IBM-PC.
Competition spurs a lot of R&D, primarily scientific calculators. At about the same time (1930s), HP is born. Electronic computers don’t become part of IBM’s portfolio until 1960s. Mechanical contraptions lead to electromechanical to electronic. In the early days, the computer industry was a capital intensive market. These days the dynamics have //completely// changed. Software is probably the least capital intensive industry, and therefore the most volatile. When markets are small, there is no long tail — the winner takes it all. But when markets are big, and there is a lot of choice, long tail effects come into play and being small can actually be profitable.
Interesting point just came up: why do some big companies have long term, research oriented investments (IBM, AT&T, M$) while others don’t (Cisco, Dell). One point Steve mentioned is that a lot of these companies head into these long term investments because they’re not working in a competitive economy. AT&T was insulated from all market forces. So were IBM and M$ in their hay days. Of course, a lot of competition also drives research, but even there small companies really can’t afford such investments.
Howard Aiken builds Mark 1 (Babbage’s dream comes true). $100,000 estimate price (1939). Final price tag of $200,000 (1943). This is 60 years ago!!!!