How Moore’s Law Made Google Possible
Gordon Moore’s famous calculation of the gains in power and economy that would drive chip production continues to have profound implications for every enterprise, no matter what the sector. But most of us have difficulty grasping the full impact of what Moore has laid out. Our handicap is that we are laboring under the illusion that the impossible is impossible.
But those who truly understand Moore’s Law know its corollary: the impossible is the inevitable. Right after Moore’s prescient prognostication, anyone with a slide rule — or a Texas Instruments calculator — could have easily run some numbers and determined that within a generation there would be computational gains a billionfold or more. The much more difficult task would be to believe this, let alone figuring out what it meant for rates of innovation, for businesses, and even for the human race.
The nonlinear gains that Moore predicted are so mind-bending that it is no wonder that very few were able to bend their minds around it.
But those who did would own the future.
Case in point is Larry Page. From birth in 1973, Larry Page was incubated in the growth light of Moore’s Law. His father and mother were computer scientists. He grew up on Michigan college campuses, never far from a computer center. He took for granted the dizzying gains in computation that would come. So he did not think twice about proposing schemes that exploited the effects of Moore’s Law, especially the big idea he had as a Stanford graduate student of dramatically improving search by taking advantage of the links of the World Wide Web.
When his thesis professor noted that such a task meant capturing the whole web on Stanford’s local servers, Page was unfazed. With a firm grip of how much more powerful and cheap tomorrow’s technology would be, he realized such a feat would eventually be relatively trivial; so would making the complicated mathematical analysis of those links, which would have to be done in well under a second. These would be written by Page’s partner, Sergey Brin, who shared Page’s comfort with the nonlinear effects of Moore’s Law. Both knew for the first time in history, the massive computation required to analyze all those links was within the grasp of grad students. Thus, by recognizing the “new possible,” Page and Brin were about to do what once was impossible — instantly combing through all of human knowledge to answer even the most obscure question.
In interviews, including those I conducted with him while writing In the Plex, my biography of Google, Page has outlined what might be known as his own variation on Moore:
Huge acceleration in computer power and memory
+ rapid drop in cost of same
= no excuse for pursuing wildly ambitious goals
Companies that develop products for the world in its present state are doomed for failure, he says. Successful products are created to take advantage of tools and infrastructure of the future. When Google whiteboards new products, it assumes they will be powered by technologies that don’t exist yet, or do currently exist and are prohibitively expensive. It is a safe bet that in a very short period of time, new technologies will exist and the cost of memory, computation and transit will fall dramatically. In fact, Moore’s Law (and similar phenomena in storage and fiber optics) means that you could bet the house on it. “The easiest thing is to do some incremental improvements,” says Page. “But that’s guaranteed to be obsolete over time, especially when it comes to technology.”
A clear example of this came in 2004 when Google announced Gmail, its web-based email product. It was not the first entry in the category. But the competitors offered very limited storage — the most popular product at the time, Microsoft’s Hotmail, gave users 2 megabytes of free storage. Users constantly had to pare down their inboxes. Gmail gave users a gigabyte of storage — five hundred times the industry standard. (It soon doubled the amount to 2 gigs.) At the time, it was so unusual that when Gmail was announced on April 1 of that year, many people regarded it as a prank — how can you give away a gigabyte of data? Indeed, in 2004, the outlay of such RAM storage to millions of users drained Google’s resources. Yes, it was costly. But only temporarily. As Page says, “That’s worked out pretty well for us.”
When Page takes meetings with Google’s employees, he relentlessly badgers them for not proposing more ambitious ideas.
Much worse than failure is failing to think big.
Earlier, Steve Jobs of Apple had a similar conundrum when releasing the Macintosh. The problem was that in 1984, technology was not ready for the computer his team had designed. To provide a satisfactory user experience, the Macintosh required at least a megabyte of internal memory, a hard disk drive, and a processor several times speedier than the Motorola 68000 chip that drove the original. Jobs knew that Moore’s Law would provide help soon, and wanted to sell the Macintosh initially priced at a money-losing $2000 to grab market share. But his bosses at Apple did not understand that setting a low price would only mean losses temporarily — the company would soon be paying less for much more powerful chips. Indeed, in a few years the Macintosh had all the power and storage it needed — but had lost the market momentum to Microsoft.
Ray Kurzweil, the great inventor and artificial intelligence pioneer now at Google, has a theory about those who are best suited to create groundbreaking products.
The common wisdom, he says, is that one cannot predict the future. Kurzweil insists that, because of Moore’s Law and other yardsticks of improvement, you can predict the future.
Maybe not enough to tell if a specific idea will succeed but certainly well enough to understand what resources might be available in a few years. “The world will be a very different place by the time you finish a project,” he told me in an interview a couple of years back.
The problem, explains Kurzweil, is that so few people have internalized that reality. Our brains haven’t yet evolved in synch with the reality that Moore identified. “Hard-wired in our brain are linear expectations, because that worked very well a thousand years ago, tracking an animal in the wild,” he says. “Some people, though, can readily accept this exponential perspective when you show them the evidence.” The other element, he adds, is the courage required to act on that evidence. Accepting Moore’s Law means understanding that what was once impossible is now within our grasp — and leads to ideas that may seem on first blush outlandish. So courage is required to resist that ridicule that often comes from proposing such schemes.
For the past few years, critics both in and out of Silicon Valley have been griping about what I call the “Jetson Gap.” The critique is embodied in venture capitalist Peter Thiel’s charge, “We were promised flying cars and instead what we got was 140 characters.” But flying cars are rather tame compared to the fantastic inventions we now use every day: a search engine that answers our most challenging questions in less than a second; a network of a billion people sharing personal news and pointers to news and gossip; and a palmtop computer that among other things can beam live video to the world and have a conversation with you.
Those who understood Moore’s Law had the fortitude to make those advances. And more people are catching on. A generation raised on Google thinking is now working on new inventions, new systems, new business plans. Businesses in virtually every sector are being challenged — and in some cases shut down — by young entrepreneurs applying Moore’s Law. (Call it the Uberization of everything.) It’s quite probable that on someone’s drawing board right now is a project that will change our lives and earn billions — but is a funding challenge because the pitch sounds, well, crazy.
But as Page told me in 2102: “If you’re not doing some things that are crazy, you’re doing the wrong thing,”
Moore’s Law guarantees it.
This article originally appeared in the Winter 2015 edition of Core, a Computer History Museum publication. The issue is a special edition commemorating the 50th anniversary of Moore’s Law.
Notebook photo copyright Douglas Fairbairn Photography, courtesy of Computer History Museum. Moore photo courtesy of Intel.