Moore’s Law, as amended

I was listening to Peter Day’s wonderful BBC business show as he interviewed Gordon Moore, who said that his first version of his law had transistors doubling on chips in a year but the public version made that two years. He noted that it’s often said his law calls for doubling every 18 months but that’s not true; he didn’t say that.

Sure enough, that error meme popped up the next day as I read John Markoff’s story about advances in computer storage and he said that Moore’s Law “decrees that the number of transistors on a silicon chip doubles roughly every 18 months.”

Just for the hell of it, I checked with that supposedly fatally flawed Wikipedia, and it said: “Moore’s Law describes an important trend in the history of computer hardware: that the number of transistors that can be inexpensively placed on an integrated circuit is increasing exponentially, doubling approximately every two years.” But the accompanying graphic gives the impression that 18 months is still the number.

I say this not to nya-nya the Times. Search at Google for “Gordon Moore” and “18 months” and you’ll find lots of erroneous statements of the law. Still, the Times’ story is one more expression of a mistake, which many will read and spread yet farther.

So I raise again the question of how we can better map content and corrections. How does Moore assure there is a definitive statement of his law? How do we know it comes from him? Once it’s acknowledged as correct, how do we notify those who got it wrong so the can correct it and start spreading the right meme? Truth is a game of wack-a-mole.

  • Somebody should probably go and tell Moore’s company, Intel, what the period should be, because the company is the one that goes around telling everybody that it’s 18 months:

    “There’s even a law for predicting this. It’s called Moore’s Law, named after Gordon Moore, one of the founders of Intel. Moore predicted in 1965 that the number of transistors on a chip would double about every 18 months.” – from Intel site at:

    Moore was introduced at an Intel Developers Forum keynote with some cooked-up story about 18 fish. He didn’t complain then, either.

    The problem is Moore’s Law doesn’t actually describe chip density versus time: the original article by Moore talks about chip density versus chip-production process generations – they simply moved at the rate of one every four years in 1965 (giving a doubling every two). It then reduced to two years. Then went up to three; it’s running at about three right now. There is no “definitive statement of his law” that involves time. But are you going to put that in a correction – the conventional quotation of the law is close enough for a lay audience.

    Are you seriously going to ask newspapers to explain the concept of process generations to those readers simply to be arse-grindingly accurate? Especially when the net result of a doubling roughly every 18 months is what they see?

    If people want to find out what Moore’s Law (which really should be called the Moore Plot if you want accuracy) is about, the source material is available via the Internet. Maybe the NYT can link to his paper, but that is about as far as it need go in this case. It’s a poor example of what online news sources should do with corrections.

  • Did you see yesterday’s “Frazz” strip? It fits this discussion perfectly.

  • Pingback: Make Them Accountable / Media()

  • Truth and Whack-a-Mole? Probably so. I’d like to think that it is fungible. It isn’t static. It is a malleable thing. Perhaps that is why a wikipedia, with a back-channel trackability, can keep up with the fluid motion of truth?

  • Pingback: Scott Rosenberg’s Wordyard » Blog Archive » Moore’s Law, once more with feeling()