Getting to the bottom of Texas' Porn Row

Last week we dug to the bottom of an online mystery: How did the first line of T.S. Eliot's "The Waste Land" become one of Google's fastest-rising search terms? As it turned out this was no literary renaissance but rather legions of "Who Wants to Be a Millionaire Viewers" who wanted to know the answer to the show's $25,000 question -- and weren't going to wait for host Meredith Vieira to tell them.

This week we have another, more lurid mystery. There's no tidy TV ending this time, but our investigation allowed us a glimpse behind the curtain, where vice and commerce meet in that obscure and lawless way particular to the Internet.

NPR editor Tricia McKinney, who last week played Holmes to my Watson, had been noticing a strange pattern on Google's Hot Trends list: a few days a week, very early in the morning, it appeared that a whole bunch of people in Texas were searching for the same very specific lewd phrases.

Texans searching for porn in the wee hours. If it doesn't sound surprising, remember that in order for "The Waste Land" to make this hot list, it took as many simultaneous Google searches as a popular daytime game show can generate -- certainly hundreds, probably thousands -- at the same time.

And yet somehow phrases like "hardcore latina luv," "hot teen luvin," and even "young dark skinned desperate girls," were making the same list. It's true that these phrases tended to appear on the list around 3 a.m. Pacific time, when most Americans are asleep, and so searches inspired by things like TV shows or the news are infrequent. But it's no small feat to get on the list at any time -- there are 300 million people in this country, and judging from the volume of sleeping pill commercials, more than a few of us are still Googling when the lights are out.

For a company whose goal is to allow easy access to all of the world's information, Google is surprisingly prone to secrecy. The details of the history-changing algorithm upon which its search engine is based are as jealously guarded as the recipe for Coca-Cola. For members of the media too, communicating with the company tends to be a one-way affair: When it has something to tell you, it'll let you know. So it's refreshingly out of character that the company has offered a tool like Hot Trends, which gives users a window not only into what people are searching, but when, and from where.

In this case, the where was more specific than just the Lone Star State. Weeks' worth of the mysterious early morning porn searches were coming from three adjacent towns at the bottom of Texas -- just a few minutes north of the Rio Grande. Heard of McAllen? San Juan? Alamo? (Not the Alamo, the mission in San Antonio, just Alamo.)

I made a map of the three cities -- I called it "Texas Porn Row" -- and then every day for a week I wrote down the terms that would pop up like clockwork on the Hot Trends list. Whoever was searching had an array of prurient interests that, if you could say nothing else about it, did not discriminate on the basis of age, race or gender.

But now that I had the location nailed, other clues were in short supply. After hours of research, I found nothing in the news, on the blogs, on Wikipedia or in city police reports that showed anything abnormal about the area.

And when facts are few, the imagination moves in to help. Perhaps, I thought, the queries were coming from some sort of giant underground Internet cafe for reprobates, who every morning convened at dawn to search, en masse, for the dirty genre of the day. Or maybe it was a perverted version of the chocolate bar unwrapping scene in "Willie Wonka & the Chocolate Factory" -- in which some mastermind was employing an army of workers to seek some elusive anti-prize hidden somewhere in the dregs of the Internet. Yes, and every 20 minutes, a bell rings and . . . .

Bing. It was my e-mail notification sound. I'd received a communiqué from the intrepid Det. McKinney. A new unexplained search term had materialized on Hot Trends, she said, and it was even weirder than the previous ones. "Zxbfwwr," was in the Trends' top five and, even though it wasn't exactly salacious (as far as we could tell, anyway), it was most certainly from Texas. The city of Bryan was only halfway across the state from Porn Row, and even better: McKinney had noticed that people searching "zxbfwwr" were also searching "irl." Also known as Texas A&M's Internet Research Lab, which is located in Bryan-College Station. IRL, its website said, was studying "Internet traffic measurement." Bing.

What better way to study online traffic patterns than by tracking the viral traction of nasty terms, and reverse engineering Hot Trends to figure out exactly how many searches it takes to make the list?

This was a brilliant solution to the mystery. But alas, it was no closer to the truth than the Willy Wonka scenario.

I contacted Dr. Dmitri Loguinov, a professor at IRL. He would not comment on the nature of his research or whether his lab was responsible for the "zxbfwwr" searches, but he assured me his lab had nothing at all to do with the porn searches.

Loguinov did, however, have a theory about what might be going on down there.

Spam.

Bing bing bing.

While most of us weren't looking -- or rather, couldn't see -- spam has evolved far beyond junk e-mail and advertisements parked at mis-typed Web addresses (laitmes.com). What we have now is more like a whole writhing spam ecosystem, one layer of which is devoted to exploiting the weaknesses of one of the Internet's main arteries: Web search.

Let's invent a few details to fill in Dr. Loguinov's hypothesis. Say there's a Web porn business -- call it Lone Star Porn -- with a network of computers along Porn Row. These mindless "bots" do nothing but search Google for, say, the phrase "bikini party," then dig down in the results until it can find Lone Star Porn and "click" on it to open the site. Every time it does so, the search engine thinks a user looking for a "bikini party" has chosen Lone Star Porn from among the millions of other results, signaling Google that next time, it might consider moving Lone Star up in the results because people like it. Have your bot repeat this for a few weeks, and maybe Lone Star Porn gets some more traction. Meanwhile, all those searches have had the side effect of landing "bikini party" on the Hot Trends list. Whee!

It was just an odd bit of serendipity that our alleged porn czar happened to be located within a day's wagon ride of an Internet research lab devoted to monitoring similar kinds of activity. But thanks to Loguinov, the Texas porn mystery -- while not exactly solved -- made for an instructive reminder that our Internet experience is highly sanitized. If it weren't for anomalies such as these few renegade Texan spammers -- still at large -- you might mistake your Internet experience for one where the dark sea of spam and porn was absent, rather than just invisible.

Comments

[Daniel]

Great look into some of the weird crap on Google Trends!

[Jonathan Richman]

First, love this report. I too have blog dedicated entirely to figuring out these mysterious terms that show up in Google Hot Trends, everyonebutyou.com

Great detective work on the "hot teen lovin" and "zxbfwwr." That's further than I got before I gave up. There's only one problem with your theory though. You said:

"Let's invent a few details to fill in Dr. Loguinov's hypothesis. Say there's a Web porn business -- call it Lone Star Porn -- with a network of computers along Porn Row. These mindless "bots" do nothing but search Google for, say, the phrase "bikini party," then dig down in the results until it can find Lone Star Porn and "click" on it to open the site. Every time it does so, the search engine thinks a user looking for a "bikini party" has chosen Lone Star Porn from among the millions of other results, signaling Google that next time, it might consider moving Lone Star up in the results because people like it."

Unfortunately, this isn't the answer. This isn't how Google does rankings. This would be a REALLY simple way to game the system that everyone would be doing. If it were simple enough that a porn outfit in southern Texas could do it, then companies with far more computing power and resources would be doing it too. If this did work, it would be theoretically possible for Yahoo! to have yahoo.com show up as the top search result for every single keyword. It would just take some time and all their computing power. Search engines are smarter than this.

Simply finding a search term and clicking on it will not increase your rankings. Searching for a term over and over would put it on Hot Trends, but that's about it. Google is pretty protective of bots and other scripts automatically doing things. It'll block access of any IP address that it suspects is doing this. Try this for yourself. Go to Google Blogs (it seems to shut you down quicker here versus the main Google search). Search for any term you want. When you get to the results page, just keep clicking refresh over and over as soon as the page loads. After about 40-50 (sometimes more), you'll get a 403 forbidden error and won't be able to use blog search for about 2 hours. Google thinks you're a bot doing something you should be doing, so it shuts you down.

Keep digging...

[Brent D. Payne]

Great post and excellent comment by Jonathan Richman!

Automating the query would need to be done by a professional that can change IP addresses and dozen other things to get around Google's safety nets. It won't change a site's rank (significantly, but some state it does have some impact--I haven't tested it).

I'd keep digging because you are about 90% of the way there. I'd love to hear what the answer is . . . look me up in TribLink. ;-)

Brent D. Payne

[The=Mond]

Webscout - I previously pointed out the connection zxbfwwr had with Texas A&M's Internet Research Lab back on May 16th at your blog about it here:
http://latimesblogs.latimes.com/webscout/2008/05/todays-google-1.html#comment-114917758

..and it was also posted to McKinney and Laura Conway on the same day at the NPR's Bryant Park blog here:
http://www.npr.org/blogs/bryantpark/2008/05/linkfest_is_your_office_making_1.html?ft=1

yet one more place the connection was pointed out previously on May 16th:
http://bzdzb.com/?p=8#comment-105

Someone even started a Wikipedia page for zxbfwwr (but it was removed and only a Google cache version remains):
http://64.233.167.104/search?q=cache:weKNncqMbGQJ:en.wikipedia.org/wiki/Zxbfwwr+zxbfwwr+wiki&hl=en&ct=clnk&cd=1&gl=us

My guess is now that the semester is over, you won't be seeing any search terms on the Hot Trends list attributed to Bryan, TX (A&M's IRL). . . that is until the next semester starts in the fall. I wonder who will be the next school/group to try and 'game the system'.

[Brad Altemeyer]

And I thought everyone in Hollywood was weird- The Real South Texas has nothing more to do with abnormal behavior than any other part of the country. And b.t.w. unless your Wagon is doing 70mph Bryan/ College Station is more than a day's ride away. Sort of like trying to go from San Diego to whatever is 350 miles North (would that be San Luis Obispo or San Francisco, or ? somewhere up there in far far away)

I refer to the Rio Grande Valley as the Real South Texas because someone who apparently failed Geography calls San Antonio South Texas (just 4 or 5 hour drive North)