One Google query = one Apollo program's worth of computing

Here's a thought:

"It takes about the same amount of computing to answer one Google Search query as all the computing done — in flight and on the ground — for the entire Apollo program."

(Quote from Seb Schmoller’s "Learning technology – a backward and forward look," attributed to Peter Norvig and Udi Mepher of Google on hearing of the death of Neil Armstrong)

I remember hearing that the processor in a singing greeting card had more capacity than all the electronic computers on Earth at the time of Sputnik's launch, though I can't find a cite for it at the moment. Exponential processor improvements are pretty wild.

Learning technology – a backward and forward look (PDF)

(via Memex 1.1)


  1. I’m not sure if thats amassing or terrifying. Mechanical Engineering with a TI-83 and the net is scary enough, cant think what it must have been like with a slide rule and a pocket protector.

  2. Thought question: Is it a good thing or a bad thing that the amount of technology that once supported a fantastic scientific achievement is now relegated to something as mundane as a Google search?
    Was it worth the lifetime efforts of two generations of systems developers and programmers so you can look up ______ (insert waste of time here)?

    1. The point isn’t about “technology relegated to stupid searches”, because the advanced technology is still available to science (look at Curiosity and Mars). This article is just about how computational power is growing and growing every day, at the point that you can now execute simple tasks in a very efficient way, using the same amount of technology used 44 years ago to go to the Moon. :)

      1. I know this.  I was programming in the 1970s in Hex and spent half a career as a Mainframer.  My point was that computing power started out doing amazing stuff like breaking codes and helping to win a war, then the Space Program(s).  Computing power is growing today but it’s growing sideways, taking over ordinary existing functions and doing them in a different way.  I spent my day squinting at a laptop helping tune the optics on a 7kW industrial laser.  This used to be done with mirrors, gratings and photo paper.  Now you just plug in and log on to the computer in the laser and bring up a Logitech camera that views the beam.  Balance the Dell on top of the resonator cavity and tweak parameters.  For a while computers changed what could be done.  I’m not sure so much of that is still happening.  Progress may just be allowing us to do things faster.

        1. For a while computers changed what could be done.  I’m not sure so much of that is still happening.

          Ok then, why don’t you try searching the internet with mirrors and photo paper

        2. Doing things faster *is* doing more.  Spectrometers work better and cheaper than before, which means they can be used for more, which means we know things we wouldn’t have known otherwise – sure, from the perspective of the technology the spectrometer isn’t that different, but from the perspective of someone who wants to know about the chemical properties of an item the availability and accessibility of that tool is the difference between knowing and not knowing.  Doing things faster means we have a transcription of human DNA now and not sometime further in the future, and we continue to know more about all sorts of other organisms.  Is knowing the DNA sequence of a worm new and different?  If you study worms it certainly is!

          And if anything, it’s not that the novel technology has become any less common, or progressed any less, it’s just that the practical or even mundane application of technology that’s one or two generations behind the cutting edge has become so relevant and important that it seems to dominate the conversation.  But really there’s just a lot more conversation about technology.

        3. Ross, I’m really lost on your point here. Do you think there is some moral imperative for all computing power to be dedicated to grand projects? Is anyone with a grand project complaining about the lack of computing power because it is being used up by google searches?

          It has simply become a ubiquitous resource and has been applied as such. There’s no reason to see it as a finite resource being wasted on our empty decadent western lifestyles. There are plenty of them to go around already.

        4. Human Genome Project, high-frequency trading, Google Earth, SETI@home FightAIDS@home etc., Wikipedia, BitCoins, phones for the poor, self-driving cars, telekinesis, kickstarter, chatroulette…
          Do any of those count as having changed what could be done?

    2. I think, (in as much as Google is a portal to the internet), search engines make the slow processes of learning and research much more efficient.  A Google search is not mundane.  And it did spawn a valuable company, for what it’s worth.

    3.  “Was it worth the lifetime efforts of two generations…”  Absolutely YES it was worth it.  Cause we/they/somebody went to the frikkin MOON.  :)

  3. Apollo 11 also used 5,625,000 pounds of rocket fuel. Without that, a Google search won’t help you get to the Moon.

  4. My interpretation of this is that it is a major testament to the computational successes of the Apollo program. A Google search is actually a phenomenally complex task performed with a high-end hardware infrastructure. The idea that the Apollo program could do similar computing with the resources of the 1960s blows me away.

  5. “I remember hearing that the processor in a singing greeting card had more capacity than all the electronic computers on Earth at the time of Sputnik’s launch, though I can’t find a cite for it at the moment. Exponential processor improvements are pretty wild.”

    It comes from Ray Kurzweil. I can’t remember whether it’s from The Singularity is Near or The Age of Spiritual Machines, but it’s definitely from one of those. That’s also not the exact fact, although I don’t have my copy of the books in front of me to look it up.

    1. The closest I can find is this quote from Dr. Michio Kaku speaking at a conference in Abu Dhabi (at which Kurzweil also spoke): “The chip that sings ‘happy birthday’ in a novelty greetings card contains more technology than was available to the entire allied forces in 1945, and what do we do with it? We throw it in the trash.” Thank you, Google. Source:

      1. Contains “more” technology? How does one in a meaningful way quantify that? More advanced? More in one place? The first version of the quote, attributed to Kurzweil, made more sense, though it sounded a bit unlikely.

  6. O Hell Noes did we evar goe to teh Moon. Howe we evar get to teh moon usin such crappy compooters??!?!

    Too bad what I just said isn’t the joke it should be. I’ve actually heard moon landing deniers say this. 

  7. So perhaps it’s time to update the old kvetch, “If we can send a man to the Moon, why can’t we __________?” to “If we can execute one Google query…” et cetera.

    Doesn’t carry quite the same punch somehow, does it?

  8. The fact that the same amount of computing power that enabled man to land on the moon (the entire program, start to finish) is used to execute a Google search is extremely abstract and really says nothing at all. Something more quantifiable, say, like, “processor power in a smartphone” or “modern laptop” would say much more, because we really have no idea how much processing actually goes into a Google search. Think about it. Add up all the processors used to do that search. From the searcher’s machine, through routers and ISPs, up to Google’s servers, then back, and rendered in the browser. It actually sounds like a lot of processors processing a lot of data done really quickly.

    1. Good point. I have no idea what exactly goes on with a Google search, but if its anything like a MySQL query, filtering through a bajillion words of text to find matches seems like quite a few shit-tons of computational power. In fact it seems likely that its equivalent to all the CPU clock cycles that happened globally up until 1980.

  9. One of these days  … one of these days … Pow! I’d like to know how much computing power, in Apollo-Programs, does it take to watch Ralph Kramden threaten Alice on Youtube.

    1. They don’t show their work!  So it’s hard to know what’s being compared here.  They mention 5 10 MHz IBM 360 mainframes, and appear to be claiming that they were running continuously for 11 years.   That would be 5 computers * 10M cycles/sec * 30M sec/year * 11 years = 16 * 10^15 cycles (16 quadrillion).   It is very hard to believe that it takes that many CPU cycles to answer a query.   If each Google machine runs at 1 GHz and the query takes 1 second to answer, there are 16 million machines involved!  Someone dropped some zeros.

      I could believe that Google has 16M processors total, so perhaps the claim is really that Google consumes in 1 second the CPU time used for all of Apollo.

Comments are closed.