Why Google Doesn’t Render Memory ObsoletePosted: July 10, 2012
I’ve heard many people over the past couple years, people that I admire, say something along the lines that if you can Google it, it is not worth committing to memory. In our world of just-in-time learning, it’s a fashionable thing to say.
But just-in-time learning doesn’t release us from the need to remember. Rather, it forces us to reconsider what is worth remembering. This is why I’ve been pushing this idea of “touchstone values” in my statistical literacy course.
Here’s an example from this morning, something that was tweeted around:
Are there 6 billion mobile phone subscribers? Are they 87% of the population?
Now let me ask — are you going to Google this before you retweet it? (Do you spend your entire day researching tweets?).
Here’s the problem — in a networked world the benefits of just-in-time learning are offset by the fact that, as Clay Shirky puts it, your filter is broken. I’d say the one biggest change for the modern information consumer is that they have to learn to ignore more things (and ignore them more quickly) than any previous generation. You can’t do that by Google searching every single thing that comes your way. You have to build the quick intuition that only comes from having some broad statistical touchstones.
This stat looks wrong to me for two reasons. The first reason is this “bottom billion” thing. I remember hearing somewhere that the poorest billion people in the world survive on less than $1 a day. Something like that. That may have changed, or be a little off, but even accounting for a range of possible values around that (less than $2 a day? 800,000 people at $1.50 a day?) there’s probably a billion people that are not buying cell phone subscriptions because they barely can afford to eat.
If that is true, and that bottom billion does not have cell phones, then every other person in the world needs to have a cell phone subscription to get you to your six billion subscribers, and that’s demonstrably not the case.
The second warning flag is not so much a stat as a general fact — the world’s population can be seen as a pyramid, with a large number of young people on the bottom, less middle age people, and a few centenarians up at the peak. Because these ages run roughly from 1 to 100, and because worldwide we are undergoing population growth, in general the amount of people under a certain age will be larger than that age as a percent. So, for example, we can say the number of people under the age of 13 will likely exceed 13% of the entire world population.
But wait a second — it’s the same problem as before. We’ve found a portion of the population that has ridiculously low cell phone subscription use (maybe your 13 year old has a phone, but does your 8 year old? Does a 13 year-old in Pakistan?). And we’re back to the problem of needing 100% saturation in the remainder again.
In reality, of course, what is probably going on here is that “subscribers” are being confused with “subscriptions”. I don’t know how they are counting, but I’m guessing that if I get two Tracphones in a year because I lost one then that’s two subscriptions, that if I have a work and a home cell, that’s two subscriptions. If I have my iPad 3G on one cell company because of provider limitations and my cell on another, that’s two subscriptions. It may even be that my Google Voice number is a subscription. And so on. (We deal extensively with this issue of paying strict attention to what is be counted in our Making Fair Comparisons chapter on Defining Terms.)
But here’s the thing — that entire process above where I decided the number was not credible? It took seconds. Those two facts that made my mind reject the validity of the number — the percentage of youth and the bottom billion? They came to me automatically, since memory is highly contextual.
When you internalize facts, you begin to internalize a conception of the world. You begin to try to fit new facts into a network of old facts. You begin to operate in a liminal space that taps both conscious processing and broad intuitions. You start to accrue the benefits of Kahneman and Tversky’s System 1, and you make it much more likely you will engage System 2, since there’s less start-up cost — the appropriate numbers are triggered by the context, and you’re ready to go.
You can’t get that if you need to search Google every time. What’s more, you won’t get that, because it will never occur to you the number is improbable in the first place. Your mind won’t rebel against the number, because it will not be attempting to weave it into a previous conception of the universe.
Does this mean we should be teaching facts? Doing memory drills? That’s a different sort of question. But we should at the very least understand that the need for memory is not going away because we have the network. If anything, the filtering demands of the network require a larger number of working models of the world through which to view the firehose of data we drink from every day. And it’s the kids who master those models who will succeed.