Friday, April 30, 2010

Answer: How to find faces of people

The big problem here is to limit your search to just content that you care about. 

To start with... You might try searching just on Syrian websites.  The simplest way to do this is to limit your search results JUST to .SY   (which is the top-level domain for Syria).  

For example, the query:  [ woman ] will find web pages that use the word "woman" only on Syrian websites.  

However... you might like to search Syrian websites in Arabic, since that's the language there.  So you might try going to Google Translate and convert the word "woman" into Arabic, which Google Translate says is:  امرأة

So... now try this search:  [ امرأة ]  -- which will look for pictures of women ONLY from Syria web sites.  Of course, many of the websites in Syria have celebrities and such, so this isn't a great way to sample for Syrian women, per se. 

SO... maybe you'd like to just look on Syrian newspapers.  I found a list of Syrian newspapers ( by doing the fairly obvious search for [ Syrian newspapers ] and so now we can combine the SITE: trick with this list of newspapers.  Example:

The query is now... [ امرأة ]  

And if you do this search in Images, and then click on the Advanced Search tab, you can filter by the Faces option... and that will give you a good number of faces of Syrian women.  

And of course... Another approach would be to search over image collections--think about doing a search like [ Syria OR syrian woman OR women ] on or or  

The key point here is that you should be thinking in the mind set of those who are creating the results you want.  And that's why you might want to search in Arabic.  

Another idea to keep in mind is how and why to scope a search (for example, by using SITE: to limit your search to just a particular newspaper site).  Newspapers often have photos of faces and people that might just be what you're looking for.  (Or not.. but if you can limit your search to collections of things you care about, this might be just the ticket.) 

Search on! 

Wednesday, April 28, 2010

Wednesday Search Challenge (April 28, 2010): How to find faces of people in Syria?

A question that recently appeared in the Google Help Forum (for Search) asked: 

     For a research project I'm doing, I'd like to find a bunch of images of Syrian women.
     But everything I try doesn't seem to work.  Help!  

So today's Search Challenge is this:  How WOULD you find a large collection of Syrian (or any other country) people?  

Can you quickly find more than a couple dozen images?  

Monday, April 26, 2010

Misreading texts

A while ago I walked down the sidewalk on a brilliant, clear day, my coffee in hand.  I’m rested, happy and generally enjoying life. 

At the long newspapers rack I looked down the line, and at one stand I saw  a headline blurb about the upcoming “Princess & The Frog” Disney movie.  I read the caption:  “Princess is no royal film treat” 

WHAT??  It’s NOT a treat?  I’m supposed to take my daughter to see it tomorrow.  ZOMG. Really?  That SO contradicts what I’ve heard about the movie that I do a doubletake. 
Re-reading, more carefully this time I see:  “Princess is a royal film treat.”  Now I’m really confused.  How could I have possibly misread that so badly?  And, if I hadn’t gone back to re-read something that contradicted what I *thought* I knew… 

Here’s the whole headline, more-or-less the way I saw it that morning: 

Clearly, as I walked down the walkway, I scanned the “no” from the subheadline and somehow inserted it into my reading of the headline.  It was a simple misread, somehow caused by a subtle blip in the way my eyes flitted over the text. 

This brought up a question that I do not know the answer to:  “How often do such misreads occur in everyday life… and how often are they corrected by the detection of contradictions and careful re-reading?”  I read the text and noticed the contradiction with what I thought was true, so I stopped and carefully re-read this snippet of text. 

Or, to be blunt about it, how often have you misread something, then NOT noticed the contradiction and just incorporated that bit of knowledge into your world-view? 

To extend this even farther, how often do any of us mis-HEAR something and get a bogus interpretation?  Unlike written-out texts, you can’t just backup time and re-hear the spoken phrase.  This is what leads to the creation of modegreens—misheard lyrics that become well-known, as in the case of “ ‘Scuse me while I kiss this guy.”  (from a lyric in the song "Purple Haze", by Jimi Hendrix which is actually "' ‘Scuse me while I kiss the sky.")

Perhaps this is just our normal condition—misreading, mishearing, misunderstanding—as we went our way through life.  A big part of the trick is to pick up the missed-interpretations as quickly as possible so the mis-information doesn’t stick in your mind and become part of your assumptions about the world. 

As Steven Pinker says, “You'll be amazed at the number of things you remember that never happened, at the number of facts you were certain of that are plainly false. Everyday conversation, even among educated people, is largely grounded in urban legends and misremembered half-truths. It makes you wonder about the soundness of conventional wisdom and democratic decision-making — and whether the increasing availability of fact-checking on demand might improve them. “    (

When I read that, I thought he was, in Pinkerian style, exaggerating the case in order to make a point.  Now, I’m not so sure. 

For the past month, ever since I noticed that first misreading of the newspaper headline, I’ve been keeping track of how often I notice myself re-scanning a text because something seemed inconsistent or just plain wrong.  Much to my surprise, it seems to happen more-or-less constantly, on the average of  2 – 4 times / day! 

I don’t think I’m especially sloppy in my scanning, so this makes me wonder—is this really a common condition? 

I did a little searching (because that’s what I do…), trying to find out Here are a couple of pull quotes that are suggestive:

“Misreading is normal and expected; it can be checked and corrected. Even good readers are likely to make quite conspicuous misreadings, but they will not self—correct unless the misreading makes a difference to meaning. This is the way fluent readers read.”  ( Smith , 1978:34)

According to Traugott & Pratt (1990:341) “readers often tend to read carelessly and stereotypically, that is, they often notice only a few features of the language they read without paying attention to what particular variety has been chosen , or to how it is represented.”  
So my hunch seems right.  People misread all the time.  The question is, how often, and how bad does it get?

I have two immediate responses: 

My first reaction is to be an engineer about it:  How can we set up texts so that misreads happen less often?  Is there some way to write the text, lay it out or otherwise present it in a way to minimize misreading?  For instance, some fonts seem to lend themselves to misreads.  Consider how some sans fonts lend no support for seeing the letters 1 l I  (esp. Gill Sans, shown here) as different letters (that is: the number 1, the miniscule L, and the majuscule I).  That’s an easy, small example.  What about line breaks that mislead, or paragraph lengths that intimidate?   

(As you’ve probably noticed by now, I tend to write with many small paragraphs rather than large, intimidating academic-seeming blocks of prose.  This is largely because I dislike reading deep into forbidding textual forests.  I find that smaller chunks seem to improve my reading and understanding.) 

So, what can one do set up the physical conditions for superior reading?  It’s going to be lots of typographic design, but design WRT scanning/readability, not necessarily grace-and-beauty (although I’d be surprised if those two goals were far apart). 

Or can I, as a writer, write in such a way as to encourage your close attention?  Perhaps one way to do this is by adding in unexpected ideas and attractors (sex & violence go here) into the middle of the text to encourage close examination. 

My second reaction is my teacherly response:  How can I teach people to read more carefully, or, barring that, to read in a way that reduces the effect of misreading? 

Teachers teach many skills—but I don’t know how often the skill of *careful* reading is taught.   In particular, I don’t recall any teacher trying to point out that as you read you need to constantly be on the lookout for things that don’t line up, contradictions and misalignments.  (Maybe I just missed that day.) 

My point remains.  Reading isn’t a simple act of visual ingestion, but a constant struggle for reconciliation between what you know, what you’ve already read, and whether or not it all makes sense together.  When it doesn’t, you have to go figure out why. 

In particular, that last comment rings especially true… You HAVE to go figure out why…  usually that just means backing up and re-reading a few words. 

We don’t just live with garden path sentences (“The old man the boat.”), but we actively reparse them until it makes sense.  In the same way, good readers look for fixes to inconsistencies within the text, or with respect to their world knowledge. 

And, as Groucho Marx said in a paraprosdokian, "Outside of a dog, a book is man's best friend. Inside of a dog, it's too dark to read." 

That’s funny because it causes you to re-interpret the first clause (“Outside of a dog..”) in a new way when you read the second sentence.  In effect, it’s a garden path paragraph that needs a re-interpretation going back one entire sentence. 

That’s what happens to good readers, or at least to me, this supposedly good reader.  I reparsed the conflicting evidence of the headline until I found the offense, my misscan of the page. 

Of course, I’m high on the “Need for Cognition” scale, so I pay attention to things like this.  I wonder how many people do…

Or, as a question for you:  Could you take note of how often you find yourself misreading a text during the next week and let me know?  (Details encouraged!)   I’m dying of curiosity to see how wide-spread this is. 

As an alternate question, and for extra credit, how would you design an experiment to get at this phenomenon?   Any ideas?

Traugott, Elizabeth Closs & Mary Louise Pratt (1990). Linguistics for Students of Literature. Harcourt Brace Jovanovich, Inc.

Smith, Frank (1978). Reading. Cambridge UniversityPress

Friday, April 23, 2010

Answer: When did jockeys start riding differently?

As with many research questions, I first noticed this difference in the body pose of jockeys more-or-less by chance.  I actually first noticed the "standing upright" pose in a old lithograph in a hotel lobby and thought to myself, "how odd is that?"

Then, when I started looking at images of horse riders over time, it became clear that something changed around 1900--riders before that time rode their mounts in a near vertical pose, while after 1900 the jockeys would crouch closely to the horse.  When you see videos of great horse runs (e.g., YouTube video of horse races ), the jockey's body keeps a steady position, almost floating above the horse.

On my first attempt to figure this out, I went looking for a historical series of jockey images--from 1800 up to 2000.  Alas, my attempts to find such an album in both Books and Images just didn't pan out.  I gave the attempts roughly 10 minutes each, looking down to pages 3 and 4 of the results.

But it just wasn't working!

I next tried to search [ horse jockey pose ] and found an article in Time on this very topic!  ("Secrets of Jockeying:  Why horses go fast" by Jeffrey Kluger, July 21, 2009)  As surprised as I was, I followed the suggestion in that article to find the Science Magazine article with the query:

[ Science magazine racehorse speed jockey ]

Why "Science magazine"?  Because I knew that just the word "Science" would be much to general, but the two-word phrase "Science magazine" would probably find articles in the journal "Science."

Why did I use "racehorse"?  Because the article in Time said "...the greatest single increase in racehorse speed..."  and I figured that was probably a direct quote.  What's more, the word "racehorse" is a little unusual, and I was hoping that the Science article would have used it.

And sure enough, the Science article is in the top 5 of the search results page!  Fantastic!

Unfortunately, if you don't subscribe to Science, it'll cost you a bundle to get the article.

So I did a search on the title of their article ("Modern Riding Style Improves Horse Racing Times") and found that lots of other reputable sources were quoting it extensively.  Ah ha!  This was good news because it suggested that the article was both widely read AND quoted by other reputable sites.

My strategy at this point was just to find another article that quoted the Science paper extensively.

This wasn't hard to do.  The article "Jockey 'monkey crouch' helped improve horse speed" published in the HorseTalk web site confirmed my impression.  Around 1897 stirrup length changed from long (with a near-vertical riding posture) to short, with the newer "monkey crouch" style of riding.  With a little more reading, you'll find that Tod Sloan introduced the monkey crouch to US horse racing in 1897.

As the article goes on to say, the newer style is much easier on the horse.. .but harder work for the rider.  Not only do you have to be small and strong, but also very cardio-fit to be a great jockey!

And... as Ach444 commented yesterday--another good query is [ jockey posture ] which has a high quality science blog as the first result.  (Wish I'd done that one first!)

There are many ways to solve this particular search challenge--but in all cases, be sure to look for multiple links to the same resource so you can see who's writing about this topic.  You want to know if you can believe them!

Search on!

Wednesday, April 21, 2010

Wednesday Search Challenge (April 21, 2010): When did jockeys start riding differently?

I've been thinking a good deal about innovation this week--how and when do new ways of doing things take over from the old, traditional ways.  

I'm especially interested in major innovations that have to do with sports.  Dick Fosbury's radically different way of going backwards over the high jump leaps to mind as a major track & field innovation that caused a real discontinuity in high jump records.  In the same way, cross-country skiing had a huge innovation when skating passed the diagonal stride as the best way to move yourself down the snowy trail. 

But this week when I was looking through some old books I noticed that even something as tradition-bound as horse racing seems to have gone through a major change in performance.  

Here's an image of a jockey racing a horse from around 100 years ago.  Notice the upright posture of the jockey while the horse is at full run. 

But look at a more modern jockey: 

Notice the crouch?  

So... my big question for today is this:  When did jockeys switch from the upright posture to the more compact form we see today?  

For extra credit, how much of a difference does it make?  Is the change just a stylistic change, or does it really make a difference in a horse race?? 

Answers tomorrow! Search on!

Thursday, April 15, 2010

Answer: A tough song to find

When my friend sent me that puzzler he commented that he'd been searching for quite a while and hadn't had any luck. He wrote:

    What's the name of the jazz standard song, recorded by Billie Holiday,
    Frank Sinatra, and others, that includes the lyric "It was just one of those things",
    but is not Cole Porter's "Just one of Those Things"?

His question gave me a couple of important clues.  First, I know he's a smart guy, so the first, obvious query wasn't going to work.  (Otherwise he wouldn't have said anything to me.)

Second, he gave me a clue that the Cole Porter song "Just one of those things" was going to be invading my search results.  So if I'd need to exclude those results somehow.

And THAT's a big clue about what to do.

Whenever you find unwanted results "invading" your search results page with hits that are close, but not really right, that's a call to use the minus operator as a way to remove unwanted results.

So my first query was:

    [ "it was just one of those things" billie holiday -porter ] 

Note that MINUS sign in front of the term "porter."  I knew I didn't want to see anything from the song with the title "Just one of those things" and I knew there was no way to remove any results that had song TITLE as "Just one of those things."  The only thing left to do was to move Cole Porter, somehow, from the results set.

I chose to do -porter because I didn't know if all the songs would have "Cole" in the attribution (as opposed to saying "C. Porter").

The big point here is that I chose a relatively uncommon term ("porter") that would knock out any of the songs he'd written, but leave everything else.

My results were pretty good.  All of the top ten hits were exactly the thing I was looking for.   "Come Rain or Come Shine" by Harold Arlen and Johnny Mercer.

My search time was around 10 seconds, which included 5 seconds of clicking through the first result to verify that it was a real song...

Tip of the day:  Use minus to remove "invading results" from your SERP.  When you pick the term to exclude, try to pick something that's pretty obscure (but always a part of the thing you want to exclude--you don't want to OVER-exclude things that might actually have the answer.  

Search on!  

Wednesday, April 14, 2010

Wednesday Search Challenge (April 14, 2010): a tough song to find

A friend asked me a hard question the other day that is today's Search Challenge.   

     What's the name of the jazz standard song, recorded by
     Billie Holiday, Frank Sinatra, and others, that includes the lyric 
     "It was just one of those things", but is NOT Cole Porter's 
     tune "Just one of Those Things"?

Give this a try!  Write down your starting time (hh:mm:ss) and then work on it.  See how long it takes you to solve this little puzzle.  Post a comment with your time and how you managed to solve it.  

Masterful performance:  less-than 15 seconds 
Good job:  15 seconds - 1 minute 
Okay job:  1 minute - 2 minutes 
Need to read SearchResearch blog more often:  > 2 minutes 

My answer tomorrow.  

Search on! 

Monday, April 12, 2010

Figuring out the differences between fishes

Sorry about being away the past few days... I was on a SCUBA diving trip with a few friends in the Bahamas.  (I know, I know... life is hard.)

But the experience made me appreciate the value of the biologist's nomenclature, the Genus/species name for everything.

You probably know that humans like you and I are Homo sapiens.  And you probably know that the "scientific" name for a living creature is something like that.  This way of giving a name to an organism is called "binomial nomenclature" and just means "two names naming system."

The capitalized term (Homo, which just means "man") is the genus name.  Think of it as the family of related organisms.  And the second term, sapiens, (meaning "wise") is the specific species name.

Why go to all this trouble?

Because it gives just one name to an organism, rather than a whole bunch of names that vary from place to place.

Now, why do I bring this up?

Because in the Bahamas there's a fish called (locally) the "zebrafish."  It's black and white (mostly) and has obvious stripes. Here's a picture I took this past week:

So it makes sense. Looks a bit like a zebra.

UNFORTUNATELY.. that's also the name of a freshwater fish that lives in Australia (and has no connection with this fish).

If you look up [ zebrafish ] you'll see all kinds of fish with this name.  That's where the binomial name comes in handy.  The fish above is really a Pterois volitans -- which is also called the "red lionfish" and the "scorpion fish" in different places.  

What does this have to do with search?  

Using the binomial name for the fish clears up the ambiguity.  And a search for that name: 

 [ Pterois volitans

will give you information ONLY about this particular fish.  This is generally true for many animal and plant names.  A "daddy longlegs" in the US is a kind of spider.  In the UK it's a kind of crane fly, which is a very different insect.  

In general, when you know the specific name of something you're looking for, that's often the best way to search for it.  

But be careful that you really know that this is the right name.  Using the wrong search term when you think it's the right term can be a recipe for un-success.  

Search on! 

Tuesday, April 6, 2010

All the magazines in Google books

One of the great, underused resources at Google are all the magazines that are scanned and put into the Google Books index.  In there you can find back issues of Life magazine--marvelous for getting a sense for what daily life was like between 1936 and 1972... check out this 1968 story about Georgia O'Keefe, complete with advertisements that ran in the magazine (such as the advertisement to eat more rice, from the Rice Council), or the fashions of the day:

But even better... Now you can look at the entire list of magazines available in Google Books.


Saturday, April 3, 2010

Show "Images from the page" -- when and why you want to use this

A friend just told me that he enjoyed reading my blog, but that "sometimes your posts are kind of long..."  

Sorry about that.  I even know that shorter writing is more easily read; it's just that my thoughts are sometimes in a longer format than is common for blog posts. 

But let me make today's post short and sweet.  

Images from the page:  Today I want to just point out one feature of Google search--"Images from the page"--which is a handy tool, if you know about it and know when to use it.  

Suppose someone told you that there's a sphinx (an actual, carved stone body-of-lion-with-head-of-person sculpture) somewhere in Palo Alto.  Is that true?  How would you find it?

You might plausibly do a search like this: 

[ palo alto sphinx ] 

And you'd get some results, and spend a lot of time trolling through them trying to figure out which was real, and which ones were just related pages without actually answering the question. 

You could then click on the Images tab to switch to Google Image search.  But it turns out that's not very useful either.  It still doesn't show you anything that looks sphinxy (!) or tell you where you might find one in Palo Alto. 

HOWEVER.. if you open up the options panel (by clicking on the "Show Options..." link)... 

.. you'll see the option to "Show images from page..." 

Once you do this, it will change the display on the SERP to show you images that actually are on the page... like this... 

And, near the bottom of the page, you'll see a real, stone-carved sphinx statue that's on the Stanford campus in Palo Alto.  

With this "Images from the page" option, you can quickly scan for items that you can recognize (or discriminate from false hits) very quickly.  

Consider using this option next time you need to find something that you (a) don't know quite how to describe, but can rapidly recognize when you see it, or (b) when you need to visually filter out all of the true hits from the noise that inevitably creeps in.  

Search on!  

Friday, April 2, 2010

A little more on quotations--why attribution is tricky

The introduction to Ralph Keyes book (The Quote Verifier: Who said what, where and when) turns out to be a marvelous discussion on why quotations are so difficult to attribute.  I'd like to recap some of his arguments here because they're so interesting... 

Misquotation and missed attributions happen all the time.  ALL the time.  Even people you think would get it right--say, JFK, who was a prolific re-quoter of others and had a speech-writing staff to boot--often got the attributions wrong.  

But misquoting happens for some good reasons.  

The primary reason is that when using quotes, the reference we're most likely to consult is our memory. This is a hazardous form of research. Our memory wants quotations to be better than they usuall were, and said by the person we want to have said them. (p xi)

It's an excellent point that's supported by all the cognitive science memory research.  Memory is fickle and errorful in specific ways.  We tend to remember the most common version of a story, rather than what actually happened, and we tend to remember things in comfortable ways, with excess baggage trimmed for recollection.  

When it comes to quotations, memory is too much the servant of aspirations, not enough the apostle of accuracy.  That is why misremembered quotations so often improve on real ones.  Memory may be a terrible librarian, but it's a great editor.  Excess words are pruned in recollection and better ones added.  The essence of a good remark is preserved, but its cadence is improved.  Churchill's "blood, toil, tears, and sweat" becomes "blood, sweat, and tears."  Durocher's "The nice guys are all over there.  In seventh place" morphs into "Nice guys finish last."  Gordon Gekko's "Greed, for lack of a better word, is good" ends up as "Greed is good."   (p xi)

In essence, quotations get improved over time with retelling, the rough edges smoothed and the cadence made right.  

They ALSO shift location from who might have really said something to someone who might-have-plausibly-said-something.  

The misattribution process is not random.  Patterns can be discerned.  If a comment is saintly, it must have been made by Ghandi (or Mother Teresa).  If it's about honesty, Lincoln most likely said it (or Washington), about fame, Andy Warhol... about courage, John Kennedy...about winnings, Vince Lombardi... malaprops by Yogi Berra.  If witty, a quip must have been Twain's concoction, or Wilde's, or Shaw's, or Dorothy Parker's. (p xii)

We remember things because they fit into a larger pattern, that is, they make sense in a particular context and don't fit in so well to some other.  

A quotation often attributed to Nelson Mandela takes this form:  "Our deepest fear is not that we are inadequate.  Our deepest fear is that we are powerful beyond measure.  Is it our Light, not our Darkness that most frightens us."  When any source is given at all, this is said to be from an inaugural speech by South Africa's two-term president.  Aside from the fact that these words don't even sound like him, they do not appear in either inaugural address given by Mandela.  On the other hand, those sentences can be found in the 1992 book, A Return to Love by pop theologian Marianne Williamson. (p xiv) 

The elevating and uplifting tone of the quote makes more of an impact on the listener/reader if it comes with the moral authority of Mandela.  It's less effective (and less memorable) if it's from Williamson.  And so we tend to remember what makes sense together, rather than what actually is.  

As a consequence, we remember the common and what we want to be true.  And in an age of fast and facile internet search for quotation sources, it's easy to be susceptible to confirmation bias--you find something you expect to be true, something that's easily relinking, copied and pasted or retweeted, and you think it's true.  

But more on this in another post.  In the meantime, remember that just because something has been repeated a million times on the web doesn't make it true.  It just makes it repeated, and we all know how easy that is to do.  

Thursday, April 1, 2010

Answer: Quotes / misquotes / mondegreens

Finding out who said what turns out to be trickier than you'd think. 

The obvious web search looks like it gives good answers.  But I'm interested in more than just what's commonly believed, I'm really interested in what's true.  

Let's look at the obvious query: 

Writing is easy; all you do is sit staring at the blank paper until the drops of blood form on your forehead 

This is a bit long for a query, especially if (as is likely with such a long query about a quotation) you've probably got a few parts of it wrong.  So let's change this query to be just the unlikely parts.  Something like:

[ writing is easy   staring at the blank   drops of blood  ]   

(Note that I didn't use double quotes here.  I could have, but it doesn't matter much.) 

Both searches work reasonably well and give more-or-less the same results.  However, the second (shorter) query is much more likely to work, as it's much LESS likely to have an error that will send you off on a goosechase search for some other pithy expression.  

As you'll see by looking at the SERP, the answer seems to be Gene Fowler, the journalist and writer (1890 - 1960). 

But I'd like to hold us to a higher standard than just what appears in the snippets on the search results page as they're automatically extracted from the original text on the landing page.  

And, as it turns out, quotes are VERY often misattributed or misremembered.  This starts the problem.  

A good quotation attribution is actually often tough to find.  What I'm really looking for is something that the supposed author actually wrote (that is, with their byline), or was reported by someone who heard it firsthand (and then actually wrote their recollection).  

When I did this search/research, I kept finding attributions to Gene Fowler.  But oddly, for someone who was so prolific in his writing, I could NOT find that phrase (or anything like it) in his writings.  

That struck me as odd.  

I mean, I checked Google Books (which has a number of Fowler's books in it), and I checked Google News Archives--again, I can find many articles by Fowler, but nothing with this quote in it.  

So I started checking collected books of quotations, hoping to find a good reference for the source.  But it turns out that many quotation books just give the attribution, but not the reference. 

This was getting to be disappointing.

Finally, after browsing through many books of quotations, I finally discovered one that gave the best source references I've found yet. 

The Quote Verifier:  Who said what, where and when by Ralph Keyes (2006)   This book is a great compendium of pithy quotes WITH the references for each given in great detail.  As I read through the book, I began to believe that this was the resource for me.  
I went to Google Books, found Keyes' book, and did a Search in this book for [ Fowler ] 

And there, on page 257 I found his entry about this quote.  

(Click on the image to make it larger.) 

This little analysis by Ralph Keyes points out what really happens with quotations.  They're often muddy in origin, frequently based on an earlier saying and evolve towards the more succinct and more memorable.  

In his book I Never Metaphor I Didn't Like, Mardy Grothe, also attributes this quote to Walter (Red) Smith.

Both books come across as fairly carefully researched (especially Keyes book, which goes to a great deal of trouble to track down origins and variants of phrases).  

So, as Keyes says "Verdict:  Credit Red Smith..."  

I still haven't found a direct quotation for either Smith OR Fowler, but I have found confirmation in two separate books, both of which are well-written and carefully researched.  

And I trust these sources over endless mindless repetitions of attributions without sources.  
In truth, that's what a lot of this boils down to--an evaluation of which source you're going to trust.  

In this case, I trust carefully researched books (which have an excellent reference list in the back) over one-line attributions on random web pages.