Okay, so in The New York Times today Dwight Garner reviews Tyler Cowen's An Economist Gets Lunch (and, incidentally, rips it to shreds -- "It's flat, padded with filler, flecked with factual errors and swollen with a kind of reverse snobbery that's nearly as wince-inducing as anything you'll hear at the Park Slope Food Coop in Brooklyn.", etc.).
Several Cowen titles are under review at the complete review -- see, for example, Creative Destruction -- and I am curious about this book, but here would like to address something different: the frequently cited huge numbers for search results supposedly returned on Google-searches.
In his review, Garner mentions:
When I conduct a search for tofu fajitas on Google the top line tells me there are about: "About 484,000 results"; a search for "tofu fajitas" (i.e. with the quotation marks) returns -- much like Garner found -- "About 29,600 results".
Those top-line numbers are the ones that media-mentions etc. invariably go with. But look at those results more closely.
I set my Google search results to return 100 results per page. At the bottom of the first 100 I can click through to nine additional pages of search results; clicking through to the last one -- page 10 -- of the tofu fajitas search I find I don't even make it to page 10; rather I find myself on: "Page 8 of about 484,000 results". Scrolling down to the bottom of that page I find myself pretty much at the end of the road, Google announcing:
(For "tofu fajitas" -- with quotation marks -- I get a total of 779, rather than the promised 29,600.)
In fact, try to search for more (like by including those omitted results) and you'll soon reach the point where Google lets you know: "Sorry, Google does not serve more than 1000 results for any query". So much for the 484,000, or 29,600, etc. ..... Good for headlines, useless (because non-existent) in real life.
So, please, everyone in the media, when you make claims about X number of Google search results coming up for search Y: check to see how many really come up (it's really easy to do). The numbers won't be nearly as impressive, but they will be much more accurate.
(Given how I've never made any inroads complaining about the misuse of statistics re. the number of titles published annually (which, for various reasons, are also grossly inflated -- though not nearly by such orders of magnitude), I'm not holding my breath.)
Several Cowen titles are under review at the complete review -- see, for example, Creative Destruction -- and I am curious about this book, but here would like to address something different: the frequently cited huge numbers for search results supposedly returned on Google-searches.
In his review, Garner mentions:
Mr. Cowen later writes, "Google brings up over a million mentions for 'tofu fajitas.'" That sounds crazy, so you check it. It turns out that Google offers only about 30,000 mentions of "tofu fajitas"; giving it a wider search range (without quotation marks) brings it up to about 115,000. Confidence further rattled.Google results vary from person (well, computer) to person (computer) (depending on settings -- how much filtering you allow -- for example) and time to time, but in practically every media/book mention the total number is a grossly inflated one -- so also here (by both Cowen and Garner).
When I conduct a search for tofu fajitas on Google the top line tells me there are about: "About 484,000 results"; a search for "tofu fajitas" (i.e. with the quotation marks) returns -- much like Garner found -- "About 29,600 results".
Those top-line numbers are the ones that media-mentions etc. invariably go with. But look at those results more closely.
I set my Google search results to return 100 results per page. At the bottom of the first 100 I can click through to nine additional pages of search results; clicking through to the last one -- page 10 -- of the tofu fajitas search I find I don't even make it to page 10; rather I find myself on: "Page 8 of about 484,000 results". Scrolling down to the bottom of that page I find myself pretty much at the end of the road, Google announcing:
In order to show you the most relevant results, we have omitted some entries very similar to the 777 already displayed. If you like, you can repeat the search with the omitted results included.So rather than 484,000 results, Google actually only offers a total of 777. Quite a big difference, no ? Apparently the other 483,223 results are simply "very similar" to the ones already offered.
(For "tofu fajitas" -- with quotation marks -- I get a total of 779, rather than the promised 29,600.)
In fact, try to search for more (like by including those omitted results) and you'll soon reach the point where Google lets you know: "Sorry, Google does not serve more than 1000 results for any query". So much for the 484,000, or 29,600, etc. ..... Good for headlines, useless (because non-existent) in real life.
So, please, everyone in the media, when you make claims about X number of Google search results coming up for search Y: check to see how many really come up (it's really easy to do). The numbers won't be nearly as impressive, but they will be much more accurate.
(Given how I've never made any inroads complaining about the misuse of statistics re. the number of titles published annually (which, for various reasons, are also grossly inflated -- though not nearly by such orders of magnitude), I'm not holding my breath.)