I was reading this article in Wired when I came upon the claim that “Google: Accounts for almost four out of five internet searches (which includes sites that license Google’s search technology), and 75 percent of all referrals to websites.” No references are offered for these figures. The rest of the piece is filled with other supposed facts without one link to or mention of a source.
Having followed the search engine market for a while the numbers in the quote above sound suspicious to me. I have never seen figures suggesting that Google (with or without affiliates) accounts for 80 percent of all searches. I contacted the author for his sources. To his credit, he got back to me very promptly. However, he did not point me to a source that can verify the information. (I do not quote from personal communication in public unless I indicated to the author that I would – which I did not – so I will not give you his exact words, but there is no source with the above figure that I can pass on to you or a collection of sources whose aggregated information leads to the above number.)
Newspaper and magazine articles do not require citations so unless the source is mentioned in the text as part of the article (e.g. “a study by the Pew Internet and American Life Project found…”) then the reader has no way of verifying the information (unless the reader decides to contact the author and the author responds). In academic writing, it is well understood that you have to cite your sources whether you are referencing ideas or specific facts. I realize that this may be tedious to do on the limited pages of newspapers and magazines. However, it seems that in online publications there should be less of a constraint to cite sources. If the reporter did his or her job and looked up relevant references for an article then why not link to them? Sure, if these are proprietary sources then that may be difficult. But I am sure that is not always the case. Yet we rarely see references to original sources in traditional newspaper and magazine pieces.
Now that the above article has appeared in Wired with the mentioned numbers stated as supposed fact, future writers (of blogs, newspaper articles, academic papers or what have you) can simply cite the Wired piece as the source of these figures and be done with it. And then we will have an unverified (and highly unlikely) figure taking on a life of its own.
PS. It is a whole other issue to figure out what it really means that a search engine accounts for x% of all searches. That may still just mean y% of all users (where y is a much smaller number than x). You can read more about this here. It would take a whole other post to get into why this may also be relevant here. I’ll leave that for another time.
{ 1 trackback }
{ 17 comments }
Nat Whilk 04.21.05 at 4:06 pm
Old data?
“Before Yahoo dumped Google, some 80% of all Internet searches were through Google technology.” (_USA Today_, 4/30/2004, p. 5b.)
“Google’s technology is used to power searches on other sites, such as Yahoo! and AOL (though Yahoo! plans to use its own technology soon). Taking this into account makes Google responsible for around 80% of all internet searches.” (_The Economist_, 1/31/2004, Special Section 2)
Brett Bellmore 04.21.05 at 4:12 pm
“If the reporter did his or her job and looked up relevant references for an article then why not link to them?”
Well, duh! Because the relevant references may be obviously lacking in credibilty, (A hotbutton topic, and you got all your info from one side’s lobbying organizations…) or actually contradict the message you’re trying to deliver.
And even if, in a specific instance, or even in most instances, you could supply that cite, you really do not want to get your readers in the habit of expecting it, because they might get suspicious in those cases where, for the reasons above, you don’t dare let them know the source.
Jane Galt 04.21.05 at 4:12 pm
The answer is twofold: the print side and the online side generally aren’t that close, and it takes quite a bit of time to enter all of those sources, since the online side takes its copy from the print output. In order to link sources, you would first have to get the reporter to take hours out of their day to highlight each source, dig up exactly where they found the number or item, and then somehow append a footnote. On the other end, you would need someone on the online side to manually enter each of those sources, including dealing with the inevitable typos. This is considerable cost even for a monthly magazine. I already spend at least an hour with each of my articles getting the sources ready for the fact checker; to have to footnote my sources extensively would take me at least another hour that I could be writing new copy. On the dotcom side, adding all of those hyperlinks would take at least another couple of hours per article. Given that the online sides generally don’t make money under any reasonable accounting standard, no one is going to add the equivalent of at least one or two full time jobs in order to provide sources.
eszter 04.21.05 at 4:24 pm
Nat – Definitely old data, Yahoo! doesn’t use Google anymore and given that it comes up as second most popular (e.g. Nielsen//NetRatings data I cite) with a considerable share it’s likely to affect the figures.
Anita Hendersen 04.21.05 at 4:36 pm
I just hope this doesn’t turn into one of those CrookedTimber threads where people say that only smart people (e.g. us) use Google, while the unwashed masses use whatever search function their computer fools them into using. Because that isn’t true.
I don’t know the correct percentage of Google’s market share or where to find this information. It is certainly less than 80% since Yahoo dropped them. And I don’t think anyone ever claimed it was that high outside the United States.
Nat Whilk 04.21.05 at 4:47 pm
Eszter:
Perhaps I was being too cryptic. Yes, the data I posted was old. It acknowledges its own oldness with the phrase: “Before Yahoo dumped Google”. My “old data?” question was whether that is the simple explanation for the _Wired_ writer’s mistake. And when you said that you had “never seen figures suggesting that Google (with or without affiliates) accounts for 80 percent of all searches”, I was wondering if “never” meant, well, never.
eszter 04.21.05 at 5:15 pm
Anita – Care to cite your sources re earlier CT content such as “only smart people use Google”? By the way, Google seems to have larger penetration in some countries other than the US (so for certain countries the 80% figure may be closer to the mark). The Wired quote doesn’t specify countries included in the figure cited.
Nat – Interestingly, you didn’t cite the sentence following the one you quoted from the USA Today article: “Now, it’s down to 49.7%, says ComScore Media Metrix.” They cite comScore, but it would be nice to see how they come up with these figures. The Economist does a better job of explaining how it came up with it. Finally, if that was the explanation for the Wired author’s comment then I would have mentioned it as the source to which he had referred me when responding to my query.
CG 04.21.05 at 5:38 pm
By coincidence, I was reading an article linked to on Slashdot, about the influence of the PR industry on the news media, and it discussed the same issue:
http://www.paulgraham.com/submarine.html
John Emerson 04.21.05 at 5:41 pm
To me, the fact that e-publication potentially, and sometimes actually, offers instant source-checking (through embedded links) means that the internet is a superior medium. A paper reference requires a trip to the library to check, and in many cases the use of interlibrary loan, and some authors (Ann Coulter!) depend on readers’ NOT checking footnotes.
My own goddamn site gets way too many of its links from msn and aol searches.I don’t know what algorithm they use but a majority of the searches are useless and off-topic.
Lance Knobel 04.21.05 at 6:35 pm
One minor point. Wired News, where the article you cite appears, no longer has an editorial relationship (name apart) with Wired Magazine. So you should distinguish between “appeared in Wired” and “appeared in Wired News”. Confusing, I know.
My sense is that Wired Magazine has maintained higher standards than Wired News. It’s of course easier to be rigorous on a monthly magazine than a news site that publishes on a reasonably constant basis.
Tom T. 04.21.05 at 6:41 pm
Here’s a BBC article that, near the bottom, says that Google is shrinking from 80% down to an expected market share of 50%. The sources of those figures are cited only as “experts.”
Jeremy Leader 04.21.05 at 7:07 pm
Around here, the information about Google paying sysadmins $35K in the Bay Area excited almost as much comment as the market share “fact”.
There was speculation that they were using a different definition of “systems administrator” than the rest of the industry, or perhaps that was a pre-IPO salary which was accompanied by a nice chunk of stock, now worth $millions.
Seth Finkelstein 04.21.05 at 7:10 pm
“75 percent of all referrals to websites.â€
Hmmm … it really depends. I do track this for my website.
Google is certainly the greatest single named source of referrals. Though there are a lot of hits where the referral can’t be determined. Aggregators and crawlers take up a lot of the traffic for my site (as well as blog-spammers).
I’d say Google has definitely lost market share from what it once was. But quantifying the exact amount isn’t trivial.
Chris 04.21.05 at 10:21 pm
Funny that you just wrote this. I just got out of my magazine article writing class, and in that spirit, I’ll attempt to offer some justification on behalf of magazine folk. Let me play devil’s advocate here.
One common activity for those writing and selling articles is to repurpose and repackage the same article for different publications. In some respects, the collection of sources and research is proprietary information, and it possesses a value itself independent of a single article. If I’m writing a magazine article and it gets published, as a writer, I want to be able to take my content and try and sell different versions to others. If a complete bibliography is posted with my article, I’m giving up a competitive advantage that I have created for myself by filtering sources to arrive at a useful set. Another writer may simply take my sources and use those to sell a similar article to a publication that I may have been targeting.
Of course, this is the only reason I know for keeping all sources confidential. (The writer in this case may simply be lazy. Who knows?) Given the different nature of the magazine article market than that of the academic “market”, different values and considerations prevail. In the magazine world, the reader is not expected to fact-check the writer (and editors), so lists of citations have less value than in the academic world. In the academic realm, a little more responsibility is placed on the reader.
I share in your concern of an unverified fact taking a life of its own. I’m honestly surprised that it might be acceptable for an academic paper could cite a magazine article as a source. Those writing magazine articles and those writing journal articles are motivated by very different factors, and it’s useful to have some sort of virtual firewall between the two.
Nat Whilk 04.22.05 at 10:14 am
Eszter wrote: “Nat – Interestingly, you didn’t cite the sentence following the one you quoted from the USA Today article: “Now, it’s down to 49.7%, says ComScore Media Metrix.—
I’m missing the interesting part of me not citing that. The sentence that I did cite specifically stated that the 80% figure was for the period before the loss of their major affiliate.
Eszter 04.22.05 at 10:27 am
Nat, as I understood it, you cited that bit as a possible source of confusion for the author of the Wired News piece. But how could the author have been confused by that information when the next sentence specifically states that their popularity is down considerably? I found it interesting that you decided to exclude a piece of info that was located right next to the fragment you cited and was quite relevant to the conversation at hand.
Nat Whilk 04.22.05 at 1:24 pm
Eszter wrote: “Nat, as I understood it, you cited that bit as a possible source of confusion for the author of the Wired News piece.”
No. I cited it as evidence that until a little over a year ago, before the loss of Yahoo, Google was alleged to account for 80% of Internet searches, the exact same figure you said you’d “never” heard associated with Google’s share of Internet searches.
“But how could the author have been confused by that information when the next sentence specifically states that their popularity is down considerably?”
I don’t think, and didn’t think, that the specific _USA Today_ article cited was the _Wired_ writer’s source. That source could have been any of a number of other articles written before Yahoo dumped Google that mentioned the 80% figure. For example, it could have been the article in _USA Today_’s 2/1/2004 issue that said: “Companies pay because about 80 percent of Internet searches are performed with Google technology, says online newsletter Search Engine Watch.”
Of course, if the _Wired_ writer had been confused by the 4/30/2004 _USA Today_ article, it would have been no stranger than you reading my original post (#1) that specifically mentioned that “Yahoo dumped Google” and turning right around (in post #4) and informing me that “Yahoo! doesn’t use Google anymore”!
Comments on this entry are closed.