One of the best interviews I ever got came over a decade ago, with then Red Hat CEO Jim Whitehurst.
Whitehurst, who like me is a Rice University alum (class of ’89) put open source into context for me.
He compared it to the standardized nuts, bolts, and washers that emerged in the early 19th century, after the Industrial Revolution had been rolling for about 50 years. Standards let you mass produce, and specify, bolts, washers and fittings. Standards made planes, trains, and automobiles possible. The revolution became something you could engineer.
Open source is now standard in cloud computing. We don’t talk about operating systems anymore. The complexity is buried until simple, standard, operating interfaces.
It’s true that cloud companies have abused open source. They failed to pay their fair share of the maintenance, putting money only into their own additions to the code base. But the standard systems still run, and by now are bulletproof.
But planes, trains, and automobiles didn’t happen all at once. It took time, experimentation, and fundamental breakthroughs like the Bessemer process, to build the 20th century world.
It will take time to build the post-industrial world, too. Meanwhile, in the early 19th century, there were engineers, like Isambard Kingdom Brunel, building vast constructions that didn’t always work. They pushed the technology beyond where it could go.
But clouds remain wondrous. Clouds mean you can build software and services without worrying about the bottom of the stack, just as you can specify a screw or a washer when designing a fry pan. Everything in the clouds is based on standards that became standards because they were free to use.
AI Is a Railroad
We don’t think about cloth anymore. We think of clothes. By the same token we don’t think of machine parts anymore. We think of machines. That’s because we take what’s on the bottom of the industrial stack for granted and build from there.
Clouds did that for computing. They’re the platform on which AI will be built.
It took 60 years for the Industrial Revolution to reach this point, from Watt’s steam engine to standards that made the Stephenson locomotive possible. It took computing about the same time, from the first IBM mainframe to the clouds. It then took another 40 years for the Industrial Revolution to transform the world, with steamships reliably crossing the oceans and railroads making cross-country travel possible.
The computing revolution, in other words, is right on schedule.
Railroads were all the rage in England 200 years ago, much like Generative AI is today. GenAI promises to compress intellectual time as the railroads did travel time. But it took time to make a reliable locomotive, and the railroad stocks that looked so good 200 years ago turned into failures once the roads started operating. Along the way there were bubbles, called manias, that took down fortunes.
Back when I talked with Jim, I wrote that software was the only good remaining that was produced by hand. You still typed out software programs, then tested them, then edited them and tested them again. It was like weavers at their hand looms, or travelers on horse-drawn carriages. The Creative Class has made itself redundant.
Time Loves a Hero
This is the social revolution of our time. We ask each other, as in the John Henry story, whether our minds will become useless, as people asked that of their bodies then. The difference was that John Henry in the legend was a proletarian, a former slave. Today it’s the global bourgeoisie that feels threatened by AI.
But we know what happened then. The same will happen now. It will just take time, trial and error, and lots of failures. We wonder who will create the breakthroughs, and who will reap the benefits? Will it be only the factory owners, or will other people get their share?
It took a century of war, strife, and struggle to answer the questions that emerged in John Henry’s time. Can’t we learn from that, for the sake of our grandchildren?
