Randomized Controlled Trials for Microsoft Copilot for Security with James Bono, Sida Peng, Roberto Rodriguez, and Sandra Ho. updated March 29, 2024.

Randomized Controlled Trials for Microsoft Copilot for Security. SSRN Working Paper 4648700. With James Bono, Sida Peng, Roberto Rodriguez, and Sandra Ho.

We conducted randomized controlled trials (RCTs) to measure the efficiency gains from using Security Copilot, including speed and quality improvements. External experimental subjects logged into a M365 Defender instance created for this experiment and performed four tasks: Incident Summarization, Script Analyzer, Incident Report, and Guided Response. We found that Security Copilot delivered large improvements on both speed and accuracy. Copilot brought improvements for both novices and security professionals.

(Also summarized in What Can Copilot’s Earliest Users Teach Us About Generative AI at Work? at “Role-specific pain points and opportunities: Security.” Also summarized in AI and Productivity Report at “M365 Defender Security Copilot study.”)

Aspira Networks Charging Merchants for Traffic That’s Otherwise Free

Affiliate marketing is supposed to be low-risk for merchants: in theory, merchants only pay affiliates when a user makes a purchase. Specifically, an affiliate should earn a commission only if 1) the user browses the affiliate’s site, 2) the user clicks the affiliate’s specially-coded link to the merchant, and 3) the user makes a purchase from the merchant. But rogue affiliates find ways to bypass these requirements — be it cookie-stuffing, adware popups, typosquatting, or network-based traffic interception. In this piece, I show Delaware-based Aspira Networks approach: configuring its partners’ networks so that if a user makes a purchase from a targeted merchant’s site, the merchant has to pay Aspira an affiliate commission — even though Aspira did nothing to cause or encourage the user’s purchase.

The method is straightforward. Aspira partners with network operators to monitor and reroute users’ browsing. If a user requests a targeted merchant, Aspira intercedes — redirecting the user to an Aspiralabs.com “finder” page, through an affiliate click link, then back to the merchant. The affiliate redirect causes merchants to conclude, mistakenly, that Aspira caused users’ subsequent purchases.

See examples of Aspira’s redirects in action, targeting the following merchants:
   GoDaddy – video, packet log
   Home Depot – video, packet log
   Travelocity – video, packet log

In testing, I found that Aspira targets roughly a quarter of large mainstream US affiliate merchants. I found Aspira systematically using CJ publisher account 3965551. In testing, I did not find Aspira currently targeting merchants using LinkShare, though an early Aspira investor pitch (dated August 2009 in metadata) indicates that Aspira has worked with LinkShare as well as Google Affiliate Network.

Merchants have no reason to pay for Aspira’s traffic. Though affiliate network reports may attribute sales to Aspira, Aspira does not actuallycause additional or incremental purchases. Rather, these are purchases that merchants would have received anyway. Aspira tells prospective merchants that they will “sell more products” by working with Aspira, but I see no evidence to support that claim. Quite the contrary, in fact; in my testing, Aspira did not genuinely promote any merchants. Aspira claimed commission only after a user had already reached a merchant’s site.

Notably, Aspira’s networks violate longstanding and broadly-applicable network policies. For example, the Commission Junction Publisher Service Agreement indicates that commission is only due for “clicks through links” (provision 3.a). Aspira’s automatic redirects entail no genuine user “click” on any link; there’s an automatic redirect but not an actual click. Other CJ rules disallow “transactions … not in good faith” including all manner of automatic and nonincremental leads (provision 1.d.ii). Here too, Aspira falls short.

In August 2011, Zhang et al. uncovered Paxfire similarly redirecting users through affiliate links. Under litigation pressure and media scrutiny, Paxfire found itself banned from Commission Junction, LinkShare, and Google Affiliate Network. I suggest the same resolution for Aspira.

Aspira’s site focuses on public installations (coffee-shops and the like) where, in principle, one might imagine that network access was made available by payments from Aspira. But I found Aspira redirecting traffic from an ordinary office served by Indiana ISP Smithville Communications. So far as I know, Smithville never notified its customers that it would be monitoring their communications, redirecting them through affiliate links, or sharing their browsing activity with Aspira. (Smithville’s privacy policy makes no mention of Aspira or sending users’ browsing to third parties.) Nor did Smithville offer customers a discount for allowing Smithville and Aspira to monitor and redirect their browsing. Other users report (and criticize) similar redirects when using standard residential and commercial ISPs Access Media 3, Arvig, and OnShore Networks.

Aspira’s site gives little detail about its revenue from redirecting users through affiliate links. Partner Cash4trafik says “typical operators have realized historically …. about $1.00 – $10.00 per subscriber per year.” Notice that revenue must be shared among affiliate networks, Aspira, Cash4traffik, and ISPs — so if an ISP gets $5, the others might take another $15 along the way. Indeed, early Aspira financial projections — posted to the web and readily found by web search — indicate that Aspira planned to retain a 40% to 67% share of affiliate commissions.

Meanwhile, Aspira’s financial projections show a particularly brazen attempt to claim payments despite minimal effort. As of 2010, Aspira projected 2013 revenue climbing to $63 million with expense reportedly to stay below $7 million. Business plans are often overly optimistic, but an 89% profit margin is difficult to reconcile with genuine efforts to find incremental customers. In contrast, it’s wholly consistent with the practice I observed in which Aspira claims commission without any expense to find or attract users.

Whatever the benefits to Aspira and its partners, the effect on merchants is clearly negative: Aspira causes extra advertising expense without providing incremental purchases. Affiliate merchants should reject Aspira’s approach, save their marketing budgets for publishers with genuine incremental value, and encourage Aspira to shift to other activities.

Thanks to Thomas Rice for bringing this practice to my attention and facilitating my collection of proof of Aspira’s practices.

Facebook Leaks Usernames, User IDs, and Personal Details to Advertisers updated May 26, 2010

Browse Facebook, and you wouldn’t expect Facebook’s advertisers to learn who you are. After all, Facebook’s privacy policy and blog posts promise not to share user data with advertisers except when users grant specific permission. For example, on April 6, 2010 Facebook’s Barry Schnitt promised: “We don’t share your information with advertisers unless you tell us to (e.g. to get a sample, hear more, or enter a contest). Any assertion to the contrary is false. Period.”

My findings are exactly the contrary: Merely clicking an advertiser’s ad reveals to the advertiser the user’s Facebook username or user ID. With default privacy settings, the advertiser can then see almost all of a user’s activity on Facebook, including name, photos, friends, and more.

In this article, I show examples of Facebook’s data leaks. I compare these leaks to Facebook’s privacy promises, and I point out that Facebook has been on notice of this problem for at least eight months. I conclude with specific suggestions for Facebook to fix this problem and prevent its reoccurrence.

Details of the Data Leak

Facebook’s data leak is straightforward: Consider a user who clicks a Facebook advertisement while viewing her own Facebook profile, or while viewing a page linked from her profile (e.g. a friend’s profile or a photo). Upon such a click, Facebook provides the advertiser with the user’s Facebook username or user ID.

Facebook leaks usernames and user IDs to advertisers because Facebook embeds usernames and user IDs in URLs which are passed to advertisers through the HTTP Referer header. For example, my Facebook profile URL is http://www.facebook.com/bedelman. Notice my username (yellow).

Of course, it would be incorrect to assume that a person looking at a given profile is in fact the owner of that profile. A request for a given profile might reflect that user looking at her own profile, but it might instead be some other user looking at the user’s profile. However, when a user views her own profile page, Facebook automatically embeds a “profile” tag (green) in the URL:

http://www.facebook.com/bedelman?ref=profile

Furthermore, when a user clicks from her profile page to another page, the resulting URL still bears the user’s own user ID or username, along with the details of the later-requested page. For example, when I view a friend’s profile, the resulting URL is as shown below. Notice the continued reference to my username (yellow) and the fact that this is indeed my profile (green), along with an appendage naming the user whose page I am now viewing (blue).

http://www.facebook.com/bedelman?ref=profile#!/pacoles

Each of these URLs is passed to advertisers whenever a user clicks an ad on Facebook. For example, when I clicked a Livingsocial ad on my own profile page, Facebook redirected me to the advertiser, yielding the following traffic to the advertiser’s server. Notice the transmission in the Referer header (red) of my username (yellow) and the fact that I was viewing my own profile page (green).

GET /deals/socialads_reflector?do_not_redirect=1&preferred_city=152&ref=AUTO_LOWE_Deals_ 1273608790_uniq_bt1_b100_oci123_gM_a21-99 HTTP/1.1
Accept: */*
Referer: http://www.facebook.com/bedelman?ref=profile

Host: livingsocial.com

The same transmission occurs when a user clicks from her profile page to a friend’s page. For example, I clicked through to a friend’s profile, http://www.facebook.com/bedelman?ref=profile#!/pacoles, where I clicked another Livingsocial ad. Again, Facebook’s redirect caused my browser to transmit in its Referer header (red) my username (yellow), the fact that that username reflects my personal profile (green). Interestingly, my friend’s username was omitted from the transmission because it occurred after a pound sign, causing it to be automatically removed from Referer transmission.

GET /deals/socialads_reflector?do_not_redirect=1&preferred_city=152&ref=AUTO_LOWE_Deals_ 1273608790_uniq_bt1_b100_oci123_gM_a21-99 HTTP/1.1
Accept: */*
Referer: http://www.facebook.com/bedelman?ref=profile

Host: livingsocial.com

In further testing, I confirmed that the same transmission occurs when a user clicks from her profile page to a photo page, or to any of various other pages linked form a user’s profile.

With a Facebook member’s username or user ID, current Facebook defaults allow an advertiser (and anyone else) to obtain a user’s name, gender, other profile data, picture, friends, networks, wall posts, photos, and likes. Furthermore, the advertiser already knows the user’s basic demographics, since the advertiser knows the user fits the profile the advertiser had requested from Facebook. For example, in grey highlighting above, the advertiser learned from Facebook my age, gender, and geographic location.

Facebook’s Contrary Statements about User Privacy vis-a-vis Advertisers

Facebook has made specific promises as to what information it will share with advertisers. For one, Facebook’s privacy policy promises “we do not share your information with advertisers without your consent” (section 5). Then, in section 7, Facebook lists eleven specific circumstances in which it may share information with others — but none of these circumstances applies to the transmission detailed above.

Facebook’s recent blog postings also deny that Facebook shares users’ identities with advertisers. In an April 6, 2010 post, Facebook promised: “We don’t share your information with advertisers unless you tell us to (e.g. to get a sample, hear more, or enter a contest). Any assertion to the contrary is false. Period.” Facebook’s prior postings were similar. July 1, 2009: “Facebook does not share personal information with advertisers except under the direction and control of a user. … You can feel confident that Facebook will not share your personal information with advertisers unless and until you want to share that information.” December 9, 2009: “Facebook never shares personal information with advertisers except under your direction and control.” As to all these claims, I disagree. Sharing a username or user ID upon a single click, without any disclosure or indication that such information will be shared, is not at a user’s direction and control.

Facebook Has Been on Notice of This Problem for Eight Months

AT&T Labs researcher Balachander Krishnamurthy and Worcester Polytechnic Instituteprofessor Craig Wills previously identified the general problem of social networks leaking user information to advertisers, including leakage through the Referer headers detailed above. In August 2009, their On the Leakage of Personally Identifiable Information Via Online Social Networks was posted to the web and presented at the Workshop on Online Social Networks (WOSN).

Through Krishnamurthy and Wills’ research, Facebook eight months ago received actual notice of the data leakage at issue. A September 2009 MediaPost article confirms Facebook’s knowledge through it spokesperson’s response. However, Facebook spokesperson Simon Axten severely understated the severity of the data leak: Axten commented “The average Facebook user views a number of different profile pages over the course of a session …. It’s thus difficult for a tracking website to know whether the identifier belongs to the person being tracked, or whether it instead belongs to a friend or someone else whose profile that person is viewing.” I emphatically disagree. As shown above, when a user views her own profile, or a page linked from her own profile, the “?ref=profile” tag is added to the URL — exactly confirming the identity of the profile owner.

What Facebook Should Do

Since receiving actual notice of these data leaks, Facebook has implemented scores of new features for advertising, monetization, information-sharing, and reorganization. Inexplicably, Facebook has failed to address leakage of user information to advertisers. That’s ill-advised and short-sighted: Users don’t expect ad clicks to reveal their names and details, and Facebook’s privacy policy and blog posts promise to honor that expectation. So Facebook needs to adjust its actual practices to meet its promises.

Preventing advertisers from receiving usernames and user IDs is strikingly straightforward: A modified redirect can mask referring URLs. Currently, Facebook uses a simple HTTP 301 redirect, which preserves referring URLs — exactly creating the problem detailed above. But a FORM POST redirect, META REFRESH redirect, or JavaScript redirect could conceal referring URLs — preventing advertisers from receiving username or user ID information.

Instead, Facebook has partially implemented the pound sign method described above — putting some, but not all, sensitive information after a pound sign, with the result that sometimes this information is not transmitted as a Referer. If fully implemented across the Facebook site, this approach might prevent the data leakage I uncovered. However, in my testing, numerous within-Facebook links bypass the pound sign masking. In any event, an improved redirect would be much simpler to implement — requiring only a single adjustment to the ad click-redirect script, rather than requiring changes to URL formats across the Facebook site.

Finally, Facebook should inform users of what has occurred. Facebook should apologize to users, explain why it didn’t live up to its explicit privacy commitments, and establish procedures — at least robust testing, if not full external review — to assure that users’ privacy is correctly protected in the future.

Update – May 26, 2010

On May 20, 2010, the Wall Street Journal reported the problem detailed above. On or about that same day, Facebook removed the ref=profile tags that were the crux of the data leak.

I yesterday spoke with Arturo Bejar, a Facebook engineer who investigated this problem. Arturo told me that after Krishnamurthy and Wills’ article, he reviewed relevant Facebook systems in search of leakage of user information. At that time, he found none, in that Facebook revealed the URLs users were browsing when they clicked ads, but did not indicate whether the user clicking a given ad was in fact the owner of the profile showing that ad. However, in a subsequent Facebook redesign, beginning in February 2010, Facebook user home pages received a new “profile” button which carried the ref=profile URL tags I analyze above. Because this tag was added without a further privacy review, Arturo tells me that he and others at Facebook did not consider the interaction between this tag and the problem I describe above. Arturo says that’s why this problem occurred despite the prior Krishnamurthy and Wills article.

Arturo also pointed out that the problem I describe did not affect advertisers whose landing pages were pages on Facebook (rather than advertisers’ own external sites).

Meanwhile, Facebook’s May 24 “Protecting Privacy with Referrers” presents Facebook’s view of the problem in greater detail. Facebook’s posting offers a fine analysis of the various methods of redirects and Facebook’s choice among them. It’s worth a read.

After discussing the problem with Arturo and reading Facebook’s new post, I reached a more favorable impression of Facebook’s response. But my view is tempered by Facebook’s ill-advised attempts to downplay the breach.

  • Rather than affirmatively describing the specific design flaw, Facebook’s post describes what “could” “potentially” occur. Facebook’s post never gives a clear affirmative statement of the problem.
  • Facebook says advertisers would need to “infer” a user’s username/ID. But usernames and IDs are sent directly, in clear and unambiguous URLs, hardly requiring complex analysis
  • Facebook claims that the breach affected only “one case … if a user takes a specific route on the site” (WSJ quote). Facebook also calls the problem “a rarely occurring case” (posting). I dispute these characterizations. It is hardly “rare” for a user to view her own profile. To view her own profile and click an ad? There’s no reason to think that’s any less frequent than clicking an ad elsewhere. To view her own profile, click through to another page, and then click an ad? That’s perfectly standard. Furthermore, although Facebook told the Journal there is “one case” in which data is leaked improperly, in fact I’ve found many such cases including clicking from profile to ad, from profile to friend’s page to ad, and from profile to photo page to ad, to name three.
  • Through transmission in HTTP Referer headers, usernames and IDs appears reach advertisers’ web servers in a manner such that default server log files would store this data indefinitely, and default analytics would tabulate it accordingly. Facebook says it has “no reason to believe that any advertisers were exploiting” the data breach I reported, but the fact is, this data ends up in a place where advertisers could (and, as to historic data, still can) access it easily, using standard tools, and at their convenience.
  • Although Facebook’s post says the problem is “potential,” I found that a user’s username/ID is sent with each and every click in the affected circumstances.

So the problem was substantial, real, and immediate. Facebook errs in suggesting the contrary.

Sony’s Crackle: Invisible Traffic Galore

Advertisers buying display ads from Sony’s Crackle.com rightly and reasonably expect that users can see the ads. After all, a visible ad is a basic and crucial condition for effective display advertising: If a user can’t see ad, then the impression is wasted, as is the associated spending. Nonetheless, in a surprising series of incidents, numerous Crackle partners are loading the Crackle site invisibly — thereby overcharging advertisers for worthless invisible impressions.

Below, I present three recent examples of Crackle partners loading the Crackle site invisibly, largely via 1×1 IFRAMEs. I then tabulate observations preserved by my automation, demonstrating that Crackle’s tainted traffic has continued for more than a year. I conclude by flagging implications for traffic measurement and ad pricing, and by suggesting what Crackle should do to clean up this mess.

Example 1: Yahoo Right Media, Adjuggler Invisible (1×1) IFRAME Loads Crackle Invisibly

In testing of April 24, 2010, my Automatic Spyware Advertising Tester browsed a series of ad URLs I had previously observed to be loaded by various spyware (installed through security exploits without user consent). One such URL embedded Bcserving tags for a 160×600 IFRAME, which passed traffic through Yahoo Right Media (yellow) to Adjuggler (green). Crucially, Adjuggler responded with an invisible 1×1 IFRAME (red) loading a URL on the Crackle site (blue). Meanwhile, another 1×1 IFRAME loaded competing video site Buddytv (grey).

GET /servlet/ajrotator/875404/0/vj?z=pdn&dim=753182&pos=1&pv=1292398882782181&nc=26008239 HTTP/1.1
Accept: */*
Referer: http://ad.yieldmanager.com/iframe3?AAAAAJ57DABJF0gAAAAAAABnEwAAAAAAAgAcAAoAAAAAAP8AAAAHGBe4GAAAAAA
ANXYaAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAVewYAAAAAAAIAAgAAAAAAXI.C9Shczz
9cj8L1KFzPP2ZmZmZmZtY.ZmZmZmZm1j8AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAADJXq7VaJccC
LvQxO2LYSYeejv1pj-PofdqHVgeAAAAAA==,,…,4256ee3e-5018-11df-ace3-001e6837e93f
Accept-Language: en-us
Accept-Encoding: gzip, deflate
User-Agent: Mozilla/4.0 (compatible; MSIE 6.0; Windows NT 5.1)
Host: rotator.adjuggler.com
Connection: Keep-Alive
Cookie: optin=Aa; ajess1_185B9A7222F0F2E13794DA5C=a; ajcmp=2023xkW0101Em0039kn

HTTP/1.1 200 OK
Server: JBird/1.0b
Connection: close
Date: Sun, 25 Apr 2010 03:11:36 GMT
Pragma: no-cache
Cache-Control: private, max-age=0, no-cache, no-store
Expires: Tue, 01 Jan 2000 00:00:00 GMT
P3P: policyref=”http://rotator.adjuggler.com:80/p3p/RotatorPolicyRef.xml”, CP=”NOI DSP COR CURa DEVa TAIa OUR SAMa NOR STP NAV STA LOC”
Content-Type: application/x-javascript

document.write(“<“+”IFRAME FRAMEBORDER=0 MARGINWIDTH=0 MARGINHEIGHT=0 SCROLLING=NO WIDTH=1 HEIGHT=1 SRC= “http://search.dailygamingupdates.com/ioq1wEIC6YxRSnWiIC9BHpdX0b1i.html”><“+”/IFRAME><“+”br><“+”/br>n”);
document.write(“n”);
document.write(“n”);
document.write(“<“+”IFRAME FRAMEBORDER=0 MARGINWIDTH=0 MARGINHEIGHT=0 SCROLLING=NO WIDTH=1 HEIGHT=1 SRC=”http://crackle.com/c/A_River_Runs_Through_It/?cmpid=762“><“+”/IFRAME><“+”br><“+”/br>n”);
document.write(“n”);
document.write(“<“+”iframe src=”http://www.buddytv.com/home2/american-idol-home2.aspxwidth=”1″ height=”1″ scrolling=”no” frameborder=”0″ marginheight=”0″ marginwidth=”0″><“+”/iframe>”);

The net effect was to load the Crackle site completely invisibly. The page-load also embedded tracking tags for comScore/ScorecardResearch (yellow). Unless comScore takes special steps to recognize and discount these invisible loads of the Crackle site, the presence of these tags would cause comScore services to overstate Crackle’s popularity.

<script type=”text/javascript”>
document.write(unescape(“%3Cscript src='” + (document.location.protocol == “https:” ? “https://sb” : “http://b”) + “.scorecardresearch.com/beacon.js’ %3E%3C/script%3E”));
</script>

<script type=”text/javascript”>
COMSCORE.beacon({
c1: 2,
c2: 6035898,
c3: “”,
c4: “Crackle.com”,
c5: “030224”,
c6: “”,
c15: “”
});
</script>

Example 2: Yahoo Right Media, Adjuggler, Media Javelin Overflowing IFRAME (1280×800 inside 160×600) Loads Crackle Invisibly

In testing of April 14, 2010, my tester browsed another publisher passing traffic through Yahoo Right Media (yellow) to an ad on Adjuggler (green) with a HTML comment referencing Media Javelin (pink). The placement was purportedly a 160×600 (grey), yet the response included a further 160×600 ad (completely filling the available space) (grey) followed by three 1280×800 IFRAMEs (red). The third of these IFRAMEs passed traffic to an Adspeed URL (orange) which passed traffic to Crackle (blue).

GET /servlet/ajrotator/875404/0/vj?z=pdn&dim=753182&pos=1&pv=9001545836619459&nc=92606617 HTTP/1.1
Accept: */*
Referer: http://ad.yieldmanager.com/iframe3?AAAAAJ57DABJF0gAAAAAAABnEwAAAAAAAgDkAAoAAAAAAP8AAAAEAh e4GAAAAAAANXYaAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAVewYAAAAAAAIA AgAAAAAAXI.C9Shczz9cj8L1KFzPP2ZmZmZmZtY.ZmZmZmZm1j8AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA AAAAAAAAAAAAAAAAAT0W5xeDoOCKeWj8s538mkN2SqqfrSTmyoa.sbAAAAAA==,,…
Accept-Language: en-us
Accept-Encoding: gzip, deflate
User-Agent: Mozilla/4.0 (compatible; MSIE 6.0; Windows NT 5.1)
Host: rotator.adjuggler.com
Connection: Keep-Alive
Cookie: optin=Aa; ajess1_…=a; ajcmp=…

HTTP/1.1 200 OK
Server: JBird/1.0b
Connection: close
Date: Wed, 14 Apr 2010 05:43:20 GMT
Pragma: no-cache
Cache-Control: private, max-age=0, no-cache, no-store
Expires: Tue, 01 Jan 2000 00:00:00 GMT
P3P: policyref=”http://rotator.adjuggler.com:80/p3p/RotatorPolicyRef.xml”, CP=”NOI DSP COR CURa DEVa TAIa OUR SAMa NOR STP NAV STA LOC”
Content-Type: application/x-javascript
Set-Cookie: ajcmp=…;Max-Age=315360000;expires=Sat, 11 Apr 2020 05:43:20 GMT;Path=/

document.write(“<“+”!– BEGIN STANDARD TAG – 160 x 600Mediajavelin.com: Run-of-site – DO NOT MODIFY –>n”);
document.write(“<“+”IFRAME FRAMEBORDER=0 MARGINWIDTH=0 MARGINHEIGHT=0 SCROLLING=NO WIDTH=160 HEIGHT=600 SRC=”http://ad.yieldmanager.com/st?ad_type=iframe&ad_size=1024×800&section=820955“><“+”/IFRAME>n”);
document.write(“<“+”!– END TAG –>n”);
document.write(“n”);
document.write(“n”);
document.write(“<“+”!– AdSpeed.com Serving Code 7.9.4 for [Ad] Buddy_home_cpc 1280×800 –><“+”iframe width=”1280″ height=”800″ src=”http://g.adspeed.net/ad.php?do=html&aid=82609&wd=1280&ht=800&target=_top” frameborder=”0″ scrolling=”no” allowtransparency=”true” hspace=”0″ vspace=”0″><“+”img style=”border:0px;” src=”http://g.adspeed.net/ad.php?do=img&aid=82609&wd=1280&ht=800&pair=as” width=”1280″ height=”800″/><“+”/iframe><“+”!– AdSpeed.com End –>n”);
document.write(“n”);
document.write(“<“+”!– AdSpeed.com Serving Code 7.9.4 for [Ad] Buddy_idol_cpc 1280×800 –><“+”iframe width=”1280″ height=”800″ src=”http://g.adspeed.net/ad.php?do=html&aid=82610&wd=1280&ht=800&target=_top” frameborder=”0″ scrolling=”no” allowtransparency=”true” hspace=”0″ vspace=”0″><“+”img style=”border:0px;” src=”http://g.adspeed.net/ad.php?do=img&aid=82610&wd=1280&ht=800&pair=as” width=”1280″ height=”800″/><“+”/iframe><“+”!– AdSpeed.com End –>n”);
document.write(“n”);
document.write(“n”);
document.write(“n”);
document.write(“<“+”!– AdSpeed.com Serving Code 7.9.4 for [Ad] Crackle_blood_cpc 1280×800 –><“+”iframe width=”1280″ height=”800″ src=”http://g.adspeed.net/ad.php?do=html&aid=82611&wd=1280&ht=800&target=_top” frameborder=”0″ scrolling=”no” allowtransparency=”true” hspace=”0″ vspace=”0″><“+”img style=”border:0px;” src=”http://g.adspeed.net/ad.php?do=img&aid=82611&wd=1280&ht=800&pair=as” width=”1280″ height=”800″/><“+”/iframe><“+”!– AdSpeed.com End –>n”);
document.write(“”);

GET /ad.php?do=html&aid=82611&wd=1280&ht=800&target=_top HTTP/1.1
Accept: image/gif, image/x-xbitmap, image/jpeg, image/pjpeg, application/x-shockwave-flash, */*
Referer: http://ad.yieldmanager.com/iframe3?AAAAAJ57DABJF0gAAAAAAABnEwAAAAAAAgDkAAoAAAAAAP8AAAAE Ahe4GAAAAAAANXYaAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAVewYAAAAA AAIAAgAAAAAAXI.C9Shczz9cj8L1KFzPP2ZmZmZmZtY.ZmZmZmZm1j8AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA AAAAAAAAAAAAAAAAAAAAAAAT0W5xeDoOCKeWj8s538mkN2SqqfrSTmyoa.sbAAAAAA==,,…
Accept-Language: en-us
Accept-Encoding: gzip, deflate
User-Agent: Mozilla/4.0 (compatible; MSIE 6.0; Windows NT 5.1)
Host: g.adspeed.net
Connection: Keep-Alive

HTTP/1.1 200 OK
P3P: policyref=”http://g.adspeed.net/w3c/p3p.xml”, CP=”NOI CUR ADM OUR NOR STA NID”
Expires: Sat, 01 Jan 2000 00:00:00 GMT
Pragma: no-cache
Cache-Control: private, max-age=0, no-cache, no-store, must-revalidate
Content-type: text/html
Connection: close
Transfer-Encoding: chunked
Date: Wed, 14 Apr 2010 05:43:19 GMT
Server: AdSpeed/s10

<html><head><title>Advertisement</title></head><body leftmargin=0 topmargin=0 marginwidth=0 marginheight=0 style=”background-color:transparent”><SCRIPT language=”JavaScript”>
<!–
window.location=”http://crackle.com/c/Blood/?cmpid=763“;
//–>
</SCRIPT><div style=”position:absolute;left:0px;top:0px;visibility:hidden;”><img src=”http://g.adspeed.net/ad.php?do=imp&zid=0&aid=82611&auth=0DB7FD0BC9&wd=1280&ht=800&cb=1271223799″ alt=”i” width=”1″ height=”1″ /></div></body></html>

The net effect was to load the Crackle site completely invisibly. Here too, the Crackle site returned comScore and ScorecardResearch tags as detailed in example 1.

Example 3: Yahoo Right Media, Extreme-sportsonline, Hotbizguide Double Invisible (1×1) IFRAMEs Load Crackle Invisibly

In testing of April 13, 2010, my tester browsed another publisher passing traffic through Yahoo Right Media (yellow) to Extreme-sportsonline (green) which included a 1×1 IFRAME (grey) passing traffic to Hotbizguide (pink). Hotbizguide returned two separate 1×1 IFRAMEs (red) loading Crackle (blue)

GET /BhkG9KftZrBezVPLOKuGW5pr4LRA.html HTTP/1.1 …
Referer: http://ad.yieldmanager.com/iframe3?AAAAAJ57DABJGEgAAAAAAARoEwAAAAAAAgA0AAIAAAAAAP8AAA ADChe4GAAAAAAASncaAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAVewYA AAAAAAIAAgAAAAAAXI.C9Shczz9cj8L1KFzPP2ZmZmZmZtY.ZmZmZmZm1j8AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA AAAAAAAAAAAAAAAAAAAAAAAAAAAAAHyzysEFgNCMHLNiu.ziQAnw9Ws9ezWVJH53CoAAAAAA==,,…
Host: search.extreme-sportsonline.com …

HTTP/1.1 200 OK
Server: nginx/0.7.62
Date: Tue, 13 Apr 2010 13:37:58 GMT

<!DOCTYPE html PUBLIC “-//W3C//DTD XHTML 1.0 Transitional//EN” “http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd”>
<html ; }
function trigger() { step++; if(step==max_steps){ load_content(); } }
</script>
<div id=”inifrcode”></div>
<iframe src=”http://search.hotbizguide.com/66682d5b6048f47cf70e56deb9989bb1.php” marginwidth=”0″ marginheight=”0″ hspace=”0″ vspace=”0″ frameborder=”0″ scrolling=”no” width=”1″ height=”1″ onload=”trigger();”></iframe><!–inifrcode–>
<div id=”stuff”></div>
<div id=”stuff1″></div>
<div id=”innercode”></div>
</body>
</html>

GET /MTNkyUL9SGHUGOPKuJJ7uj3AOWuN.php HTTP/1.1
Accept: image/gif, image/x-xbitmap, image/jpeg, image/pjpeg, application/x-shockwave-flash, */*
Referer: http://search.mobilegamesearch.com/NgVbJ4eWaW7bMt57RcZVGMggRW9t.html
Accept-Language: en-us
Accept-Encoding: gzip, deflate
User-Agent: Mozilla/4.0 (compatible; MSIE 6.0; Windows NT 5.1)
Host: search.hotbizguide.com
Connection: Keep-Alive
Cookie: PHPSESSID=a8qbtgec2k79fd8c6onqp5k3q6

HTTP/1.1 200 OK
Server: nginx/0.7.62
Date: Tue, 13 Apr 2010 21:45:49 GMT
Content-Type: text/html
Transfer-Encoding: chunked
Connection: keep-alive
X-Powered-By: PHP/5.2.11
Expires: Thu, 19 Nov 1981 08:52:00 GMT
Cache-Control: no-store, no-cache, must-revalidate, post-check=0, pre-check=0
Pragma: no-cache
Set-Cookie: trkid=…; expires=Fri, 16-Apr-2010 21:45:49 GMT

<!DOCTYPE html PUBLIC “-//W3C//DTD XHTML 1.0 Transitional//EN” “http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd”>
<html >iframe src=”http://crackle.com/c/A_River_Runs_Through_It/?cmpid=728width=”1″ height=”1″ scrolling=”no” frameborder=”0″ marginheight=”0″ marginwidth=”0″></iframe>
<iframe src=”http://crackle.com/c/The_Beast/?cmpid=692width=”1″ height=”1″ scrolling=”no” frameborder=”0″ marginheight=”0″ marginwidth=”0″></iframe>
<!–customcode–>

The net effect was to load the Crackle site completely invisibly, twice. Here too, the Crackle site returned comScore and ScorecardResearch tags as detailed in example 1.

A Long-Term Problem

To confirm the scope of Crackle’s invisible and forced-visit traffic and to evaluate trends over time, I searched logs preserved by my Automatic Spyware Advertising Tester. My tester records the URLs of pages loaded by the spyware-infected virtual computers, but my tester never intentionally visited the Crackle site, nor did my tester ever prompt or provoke traffic to Crackle in any way. So if my tester observes traffic to Crackle, that traffic must have occurred unrequested — invisibly or, at best, through a spyware popup or popunder.

The following table summarizes my tester’s observations of traffic to Crackle:

Month Number of observations of
traffic to Crackle.com
Q1 2009
15
Q2 2009
57
Q3 2009
95
Q4 2009
31
Q1 2010
51
April 2010 (month to date)
74

Although my testing methods have changed over time (e.g. to test new spyware), my testing intensity has remained constant. I have no specific reason to expect that adjustments to my methods would be more or less effective at uncovering Crackle incidents. I therefore tend to attribute month-to-month changes to changes in the intensity with which Crackle’s partners purchase invisible traffic and other tainted traffic on Crackle’s behalf. In my experience, most traffic-buyers run on-and-off campaigns — buying traffic for a time, then taking a break. If Crackle’s traffic buying followed a similar approach, spikes would be expected.

In any event, invisible and forced-visit traffic to Crackle cannot be written off as a short-term anomaly. Quite the contrary, my tester’s observations confirm that these problems have persisted for many months.

Implications and Next Steps

Advertisers buying placements on Crackle reasonably expect high-quality traffic. For example, the first clause of Crackle’s “About” page boasts that Crackle is owned by Sony, and Sony’s size and wealth provide a level of accountability that smaller video sites cannot offer. Nonetheless, my observations confirm that some placements on Crackle are entirely worthless.

Crackle may blame its tainted and invisible traffic on traffic brokers, affiliates, and other external forces. But traffic-buying inevitably invites exactly these shenanigans. If Crackle intends to buy traffic, it should better vet its partners — including confirming partners’ bona fides, overseeing partners’ specific methods, and developing procedures to hold partners accountable for any shortfalls. I’ve seen no sign of any such efforts. If Crackle can’t buy traffic with the requisite skill, perhaps Crackle would do better by ceasing to buy traffic at all.

Crackle traffic rank from Alexa - April 2010Three years ago, I posted How Spyware-Driven Forced Visits Inflate Web Site Traffic Counts, pointing out that cheap spyware and popup traffic can increase measurements of web site popularity. Users at least see those spyware popups, giving some nugget of rationale for resulting traffic figures. But users cannot even see the Crackle site when it is loaded invisibly as detailed above. Nonetheless, invisible site-loads are still likely to inflate measurements of Crackle’s traffic. For example, during the first few days of April 2010, my tester observed rampant fake traffic to Crackle — and Alexa simultaneously reported a major spike in Crackle’s traffic. (See image at right.) Furthermore, with comScore tags embedded in each Crackle impression, comScore systems receive a report each time Crackle’s site is loaded. Indeed, such reports come from a large number of distinct IP addresses — seemingly confirming that the traffic is legitimate. Can comScore successfully recognize and discount invisible loads of the Crackle site? If not, comScore will join Alexa in overstating Crackle’s popularity.

My bottom line? Buying traffic is a dirty business — so many sellers who provide worthless traffic, and so many buyers who use too little care in selecting and assessing the traffic they buy. As it stands, advertisers and networks doing business with Crackle risk paying for ads users cannot see. Plenty of small-time video sites play these games, but it’s disappointing to see a Sony site stoop to that level.

Measuring Typosquatting Perpetrators and Funders

Moore, Tyler, and Benjamin Edelman. Measuring Typosquatting Perpetrators and Funders. Light Blue Touchpaper. February 17, 2010.

Reprinted at CircleID.

Introduction to Moore, Tyler, and Benjamin Edelman. “Measuring the Perpetrators and Funders of Typosquatting.” Lecture Notes in Computer Science. Springer-Verlag. Financial Cryptography and Data Security: Proceedings of the International Conference 6052 (2010).

Google Click Fraud Inflates Conversion Rates and Tricks Advertisers into Overpaying

I’ve repeatedly reported improper placements of Google ads. In most of my write-ups, the impropriety occurs in ad placement — Google PPC ads shown in spyware popups (1, 2, 3, 4), in typosquatting sites (1, 2), or in improperly-installed and/or deceptive toolbars (1, 2). This article is different: Here, the impropriety includes a fake click — click fraud — charging an advertiser for a PPC click, when in fact the user never actually clicked.

But this is no ordinary click fraud. Here, spyware on a user’s PC monitors the user’s browsing to determine the user’s likely purchase intent. Then the spyware fakes a click on a Google PPC ad promoting the exact merchant the user was already visiting. If the user proceeds to make a purchase — reasonably likely for a user already intentionally requesting the merchant’s site — the merchant will naturally credit Google for the sale. Furthermore, a standard ad optimization strategy will lead the merchant to increase its Google PPC bid for this keyword on the reasonable (albeit mistaken) view that Google is successfully finding new customers. But in fact Google and its partners are merely taking credit for customers the merchant had already reached by other methods.

In this piece, I show the details of the spyware that tracks user browsing and fakes Google PPC ad clicks, and I identify the numerous intermediaries that perpetrate these improper charges. I then criticize Google’s decision to continue placing ads through InfoSpace, the traffic broker that connected Google to this click fraud chain. I consider this practice in light of Google’s advice to advertisers and favored arguments that click fraud problems are small and manageable. Finally, I propose specific actions Google should take to satisfy to prevent these scams and to satisfy Google’s obligations to advertisers.

Introducing the Problem: A Reader’s Analogy

Reading a prior article on my site, a Register discussion forum participant offered a useful analogy:

Let’s say a restaurant decides [it] wants someone to hand out fliers … so they offer this guy $0.10 a flier to print some and distribute them.

The guy they hire just stands at the front door and hand the fliers to anyone already walking through the door.

Restaurant pays lots of money and gains zero customers.

Guy handing out the fliers tells the owner how many fliers were printed and compares that to how many people bring the fliers into his restaurant.

The owner thinks the fliers are very successful and now offers $0.20 for each one.

It’s easy to see how the restaurant owner could be tricked. Such scams are especially easy in online advertising — where distance, undisclosed partnerships, and general opacity make it far harder for advertisers to figure out where and how Google and its partners present advertisers’ offers.

Google and Its Partners Covering Advertisers’ Sites with Spyware-Delivered Click-Fraud Popups

PPC advertisers (e.g. Finish Line)
money viewers
   Google   
money viewers
InfoSpace
money viewers
Cheapstuff
money viewers
Adfirmative
money viewers
dSide Marketing
money viewers
Netaxle
money viewers
eWoss
money viewers
AdOn Network
money viewers
Trafficsolar

The money trail – how funds flow from advertisers to Google to Trafficsolar spyware.

In testing of December 31, 2009, my Automatic Spyware Advertising Tester browsed Finishline.com, a popular online shoe store, on a virtual computer infected with Trafficsolar spyware (among other advertising software, all installed through security exploits without user consent). Trafficsolar opened a full-screen unlabeled popup, which ultimately redirected back to Finish Line via a fake Google PPC click (i.e., click fraud).

My AutoTester preserved screenshots, video, and packet log of this occurrence. The full sequence of redirects:

Trafficsolar opens a full-screen popup window loading from urtbk.com, a redirect server for AdOn Network. (AdOn, of Tempe, Arizona, first caught my eye when it boasted of relationships with 180solutions/Zango and Direct Revenue. NYAG documents later revealed that AdOn distributed more than 130,000 copies of Direct Revenue spyware. More recently, I’ve repeatedly reported AdOn facilitating affiliate fraud, inflating sites’ traffic stats, and showing unrequested sexually-explicit images.)

AdOn redirects to eWoss. (eWoss, of Overland Park, Kansas, has appeared in scores of spyware popups recorded by my testing systems.)

eWoss redirects to Netaxle. (NetAxle, of Prairie Village, Kansas, has also appeared in numerous popups — typically, as here, brokering traffic from eWoss.)

Netaxle redirects to dSide Marketing. (dSide Marketing, of Montreal, Canada, says it provides full-service SEO and SEM services.)

dSide Marketing redirects to Adfirmative. (Adfirmative, of Austin, Texas, promises “click-fraud protected, targeted advertising” and “advanced click-fraud prevention.”)

Adfirmative redirects to Cheapstuff. (Cheapstuff fails to provide an address on its web site or in Whois, though its posted phone number is in Santa Monica, California. Cheapstuff’s web site shows a variety of commercial offers with a large number of advertisements.)

Cheapstuff redirects to InfoSpace. (InfoSpace, of Bellevue, Washington, is discussed further in the next section.)

InfoSpace redirects to Google, which redirects through DoubleClick and onwards back to Finish Line — the same site my tester had been browsing in the first place.

This placement is a bad deal for Finish Line for at least two reasons. First, Google charges Finish Line a fee to access a user already at Finish Line’s site. But that’s more of a shake-down then genuine advertising: an advertiser should not have to pay to reach a user already at its site. Furthermore, Google styles its advertising as “pay per click”, promising advertisers that “You’re charged only if someone clicks your ad.” But here, the video and packet log clearly confirm that the Google click link was invoked without a user even seeing a Google ad link, not to mention clicking it. Advertisers paying high Google prices deserve high-quality ad placements, not spyware popups and click fraud.

Finally, the popup lacks the labeling specifically required by FTC precedent. Consistent with FTC’s settlement in its Direct Revenue and Zango cases, every spyware/adware popup must be labeled with the name of the program that caused the popup, along with uninstall instructions. Furthermore, the FTC has taken an appropriately dim view of advertising software installed on users’ computers without user consent. But every single Trafficsolar installation I’ve ever seen has arrived on my test computers through security exploits, without consent. For these reasons, this Trafficsolar-Google popup clearly falls afoul of applicable FTC requirements.

Critiquing InfoSpace’s role

As shown in the prior section and diagram, traffic flows through a remarkable seven intermediaries en route from Trafficsolar spyware to the victim Google advertiser. Looking at such a lengthy chain, the problem may seem intractable: How could Google effectively supervise a partner’s partner’s partner’s partner’s partner’s partner’s partner’s partner? That insurmountable challenge is exactly why Google should never have gone down this path. Instead, Google should place ads only through the companies with which Google has direct relationships.

In this instance, when traffic finally gets to Google, it comes through a predictable source: InfoSpace. It was InfoSpace, and InfoSpace alone, that distributed Google ads into the morass of subsyndicators and redistributors detailed above.

Flipping through my records of prior InfoSpace observations, I was struck by the half-decade of bad behavior. Consider:

June 2005: I showed InfoSpace placing Google ads into the IBIS Toolbar which, I demonstrated in multiple screen-capture videos, was arriving on users’ computers through security exploits (without user consent). The packet log revealed that traffic flowed from IBIS directly to InfoSpace’s Go2net.com — suggesting that InfoSpace had a direct relationship with IBIS and paid IBIS directly, not via any intermediary.

August 2005: I showed InfoSpace placing ads through notorious spyware vendor Direct Revenue (covering advertisers’ sites with unlabeled popups presenting their own PPC ads). The packet log revealed that traffic flowed from Direct Revenue directly to InfoSpace — suggesting that InfoSpace had a direct relationship with Direct Revenue and paid Direct Revenue directly, not via any intermediary.

August 2005: I showed InfoSpace placing ads through notorious spyware vendor 180solutions/Zango. The packet log revealed that traffic flowed from 180solutions directly to InfoSpace — suggesting that InfoSpace had a direct relationship with 180solutions and paid 180solutions directly, not via any intermediary.

February 2009: I showed InfoSpace placing Google ads into WhenU popups that covered advertisers’ sites with their own PPC ads.

May 2009: Again, I showed InfoSpace using WhenU to cover advertisers’ sites with their own PPC ads, through partners nearly identical to the February report.

January 2010 (last week): I showed InfoSpace’s still placing Google ads into WhenU popups and still covering advertisers’ sites with their own PPC ads.

And those are just placements I happened to write up on my public site! Combine this pattern of behavior with InfoSpace’s well-documented accounting fraud, and InfoSpace hardly appears a sensible partner for Google and the advertisers who entrust Google to manage their spending.

Nor can InfoSpace defend this placement by claiming Cheapstuff looked like a suitable place to show ads. The Cheapstuff site features no mailing address or indication of the location of corporate headquarters. WHOIS lists a “privacy protection” service in lieu of a street address or genuine email address. These omissions are highly unusual for a legitimate advertising broker. They should have put InfoSpace and Google on notice that Cheapstuff was up to no good.

This Click Fraud Undercuts Google’s Favorite Defense to Click Fraud Complaints

When an advertiser buys a pay-per-click ad and subsequently makes a sale, it’s natural to assume that sale resulted primarily from the PPC vendor’s efforts on the advertiser’s behalf. But the click fraud detailed in this article takes advantage of this assumption by faking clicks to target purchases that would have happened anyway. Then, when advertisers evaluate the PPC traffic they bought, they overvalue this “conversion inflation” traffic — leading advertisers to overbid and overpay.

Indeed, advertisers’ following Google’s own instructions will fall into the overbidding trap. Discussing “traffic quality” (i.e. click fraud and similar schemes),Google tells advertisers to “track campaign performance” for “ROI monitoring.” That is, when an advertiser sees a Google ad click followed by a sale, the advertiser is supposed to conclude that ads are working well and delivering value, and that click fraud is not a problem. Google’s detailed “Click Fraud: Anecdotes from the Front Line” features a similar approach, advising that “ROI is king,” again assuming that clicks that precede purchases must be valuable clicks.

Google’s advice reflects an overly optimistic view of click fraud. Google assumes click fraudsters will send random, untargeted traffic. But click-frauders can monitoring user activities to identify the user’s likely future purchases, just as Trafficsolar does in this example. Such a fraudster can fake the right PPC clicks to get credit for traffic that appears to be legitimate and valuable — even though in fact the traffic is just as worthless as other click fraud.

What Google Should Do

Google’s best first step remains as in my posting last week: Fire InfoSpace. Google doesn’t need InfoSpace: high-quality partners know to approach Google directly, and Google does not need InfoSpace to add further subpartners of its own.

Google also needs to pay restitution to affected advertisers. Every time Google charges an advertiser for a click that comes from InfoSpace, Google relies on InfoSpace’s promise that the click was legitimate, genuine, and lawfully obtained. But there is ample reason to doubt these promises. Google should refund advertisers for corresponding charges — for all InfoSpace traffic if Google cannot reliably determine which InfoSpace traffic is legitimate. These refunds should apply immediately and across-the-board — not just to advertisers who know how to complain or who manage to assemble exceptional documentation of the infraction.

More generally, Google must live up to the responsibility of spending other people’s money. Through its Search Network, Google takes control of advertisers’ budgets and decides, unilaterally, where to place advertisers’ ads. (Indeed, for Search Network purchases, Google to this day fails to tell advertisers what sites show their ads. Nor does Google allow opt-outs on a site-by-site basis — policies that also ought to change.) Spending others’ money, wisely and responsibly, is a weighty undertaking. Google should approach this task with significantly greater diligence and care than current partnerships indicate. Amending its AdWords Terms and Conditions is a necessary step in this process: Not only should Google do better, but contracts should confirm Google’s obligation to offer refunds when Google falls short.

I’m disappointed by Google’s repeated refusal to take the necessary precautions to prevent these scams. InfoSpace’s shortcomings are well-known, longstanding, and abundantly documented. What will it take get Google to eject InfoSpace and protect its advertisers’ budgets?

Google Still Charging Advertisers for Conversion-Inflation Traffic from WhenU Spyware updated January 7, 2010

When an advertiser buys a pay-per-click ad and subsequently makes a sale, it’s natural to assume that sale resulted primarily from the PPC vendor’s efforts on the advertiser’s behalf. But tricky PPC platforms take advantage of this assumption by referring purchases that would have happened anyway. Then, when advertisers evaluate the PPC traffic they bought, they overvalue this “conversion inflation” traffic — leading advertisers to overbid and overpay.

In this piece, I show Google and its partners still covering popular sites with PPC advertisements promoting those same sites. I present the role of InfoSpace, the Google partner at the core of these misplacements, and I argue that Google should long ago have severed its ties to InfoSpace. I cite specific Google promises that these placements violate, and I critique Google’s contractual disclaimers that claim advertisers must pay for these bogus placements. Finally, I propose specific actions Google should take to satisfy to its obligations to advertisers.

Google and Its Partners Still Covering Advertisers’ Sites with Spyware-Delivered Popups

WhenU covers Continental with its own Google ads -- charging ad fees for traffic Continental would otherwise receive for free
WhenU covers Continental with its own Google ads — chargingad fees for traffic Continental would otherwise receive for free

As shown in the thumbnail at right and detailed in screenshots, video, and packet log, WhenU continues to cover web sites with PPC popups. Crucially, those popups show Google ads — often promoting the very same sites users are already browsing.

In the example shown at right, I browsed the Continental Airlines site. WhenU opened the popup shown at right — covering the Continental site with a list of Google ads, putting a prominent Continental ad front-and-center. Thus, Google charges Continental a fee to access a user already at Continental’s site. That’s a rotten deal for Continental: For one, an advertiser should not have to pay to reach a user already at its site. Furthermore, advertisers paying high Google prices deserve high-quality ad placements, not spyware popups.

The details of the Continental ad, as shown in the WhenU-Google popup, further entice users to click. The ad promises a “low fare guarantee” — suggesting that users who book some other way (without clicking the ad) may not enjoy that guarantee. And the ad promises to take users to the “official site” — suggesting that users who don’t click the ad will book through a site that is less than official. In fact both suggestions are inaccurate, but a reasonable user would naturally reach these conclusions based on the wording of the advertisement and the context of its appearance.

The WhenU-Google popup lacks the labeling specifically required by FTC policy. In particular, all sponsored search ads are to be labeled as such, pursuant to the FTC ‘s 2002 instructions. But look closely at the popup screenshot. On my ordinary 800×600 screen, no such label appears. I gather the required label would ordinarily appear on a sufficiently large screen, but the FTC’s policies make no exceptions for users with small to midsized screens. Indeed, as netbooks gain popularity, small screens are increasingly common.

The diagram below (left) confirms the specific intermediaries passing traffic from WhenU to Google in this instance.

The money trail: how funds flow from advertisers to Google to WhenU
(three examples persisting over ten months)
December 2009

PPC advertisers
(e.g. Continental)
money viewers
   Google   
money viewers
InfoSpace
money viewers
LocalPages
money viewers
(unknown company*)
money viewers
WhenU

PPC advertisers
(e.g. RCN)
money viewers
   Google   
money viewers
InfoSpace
money viewers

*  LocalPages
money viewers
Nbcsearch
money viewers
LocalPages

money viewers
WhenU

PPC advertisers
(e.g. Verizon)
money viewers
   Google   
money viewers
InfoSpace
money viewers
LocalPages
money viewers
WhenU

This observation marks the third sequence by which I have observed Google paying WhenU to cover advertisers’ sites with the advertisers’ own Google ads. The center and right diagrams (above) show the intermediaries in my May 2009 and February 2009 observations of similar placements.

The Impropriety of Google’s Relationship with InfoSpace

In all three instances I reported (as summarized in the diagram above), Google’s closest link is to InfoSpace. That is, Google pays InfoSpace, and InfoSpace pays the various entities that follow. In my view, Google’s relationship with InfoSpace is ill-advised for at least three reasons:

First, InfoSpace has a track record of improper placements of Google ads. InfoSpace is implicated in all three of the placements detailed above — misplacements that have continued over a lengthy period despite ample notice and opportunity for correction. Furthermore, I have personally observed other improper placements by InfoSpace. (Perhaps I’ll post more in a futher piece.) Google need not continue to do business with a distributor with such a poor track record.

Second, Google does not need a distributor whose business model entails farming out ad placements to subdistributors. If InfoSpace’s subdistributors seek to distribute Google ads, and to be paid for doing so, let them apply directly to Google and undergo Google’s ordinary quality control and oversight. Inserting InfoSpace as an additional intermediary serves only to lessen accountability.

Third, InfoSpace’s corporate history undermines any request for lenience or forgiveness. The Seattle Times chronicles InfoSpace’s accounting fraud in a three-part investigative report, “Dot-Con Job“, presenting 12,000+ words of analysis as well as primary source documents and even voicemail recordings. The Seattle Times byline summarizes their findings: “Investors were cashing out millions, and faithful investors were left with pennies.” Hardly a mark of trustworthiness!

These Ads Violate Google’s Promises to Users

These ad placements fall short of Google’s promises to users. By paying spyware vendors to show advertisements, Google both enlarges and prolongs the spyware problem. In particular, Google’s funding supports software that users struggle to remove from their computers. Google’s payments make it more profitable for vendors to sneak such software onto users’ computers in the first place.

Furthermore, Google’s Software Principles specifically disallow WhenU’s practices. Google’s “installation” and “upfront disclosure” principles disallow deceptive and nonconsensual WhenU installations. (I have video proof on file showing nonconsensual WhenU installations.) Google’s prohibition on “snooping” prohibits certain WhenU privacy practices, including WhenU’s historic violation of its own privacy policy (transmitting full page URLs despite a privacy policy promising “As the user surfs the Internet, URLS visited by the user … are NOT transmitted to WhenU.com or any third party server”).

Crucially, Google’s partnership with WhenU directly contradicts Google’s call for software makers and advertising intermediaries to “keep[] good company” by supervising partners. Despite that commitment, present on Google’s site for 4+ years, Google inexplicably continues its relationship with WhenU.

These Ads Violate Google’s Promises to Advertisers

These ad placements also fall short of Google’s obligations to advertisers. For example, when Google describes its Search Network, Google promises:

Ads are targeted based on a user’s search terms.   (emphasis added)

But here, the user performed no search — so there was no proper cause to display a Search Network ad or charge an advertiser a high Search Network price.

Google confirms:

On the Search Network, ads are shown … on … the search results pages of … Google’s search partners … within the Search Network. On our search partners, your ads may appear alongside or above search results, as part of a results page as a user navigates through a site’s directory, or on other relevant search pages.   (emphasis added)

A placement through a spyware popup does not meet these criteria: A spyware popup is not a “page.” Furthermore, a user browsing an ordinary web site (like the Continental site shown above) is neither “search[ing]” nor navigating a “directory,” contrary to Google’s promise that search ads are shown to users at search engines and directories.

Despite these clear promises, Google’s AdWords Terms and Conditions purport to allow these placements and any others Google might choose to foist on unsuspecting advertisers. Google requires advertisers to accept the following form contract provisions:

Customer understands and agrees that ads may be placed on … (z) any other content or property provided by a third party (‘Partner’) upon which Google places ads (‘Partner Property’).   (emphasis added)

That’s circular, uninformative, and a rotten deal. Advertisers should demand better. Nor should Google’s fine print claim the right to impose such bogus charges. Google should amend its contract to disavow charges from spyware, adware, conversion-inflation, and other schemes contrary to Google’s affirmative promises.

What Google Should Do

Google’s first step is easy: Fire InfoSpace. Google doesn’t need InfoSpace, and there’s zero reason for this relationship to continue in light of InfoSpace’s repeated failings.

Google also needs to pay restitution to affected advertisers. Every time Google charges an advertiser for a click that comes from InfoSpace, Google relies on InfoSpace’s promise that the click was legitimate, genuine, and lawfully obtained. But there is ample reason to doubt these promises. Google should refund advertisers for corresponding charges — for all InfoSpace traffic if Google cannot reliably determine which InfoSpace traffic is legitimate. These refunds should apply immediately and across-the-board — not just to advertisers who know how to complain or who manage to assemble exceptional documentation of the infraction. (Indeed, in response to my May 2009 report, I know Google provided a credit to RCN — the specific advertiser whose targeting I happened to feature in my example. But I gather Google failed to provide automatic credits to all affected advertisers, even though Google’s billing records provide ample documentation of which advertisers faced charges from which Google partners. And I understand that Google denied requests for refunds or credits from other affected advertisers.)

More generally, Google must live up to the responsibility of spending other people’s money. Through its Search Network offering, Google takes control of advertisers’ budgets and decides, unilaterally, where to place advertisers’ ads. (Indeed, for Search Network purchases, Google to this day fails to tell advertisers what sites show their ads. Nor does Google allow opt-outs on a site-by-site basis — policies that also ought to change.) Spending others’ money, wisely and responsibly, is a weighty undertaking. Google should approach this task with significantly greater diligence and care than current partnerships indicate. Amending its AdWords T&C’s is a necessary step in this process: Not only should Google do better, but contracts should confirm Google’s obligation to offer refunds when Google falls short.

I’m disappointed by how little has changed since my year-ago reports of these same practices. In a conference presentation in February 2009, I demonstrated substantially similar WhenU placements, with Google’s Rose Hagan (Senior Trademark Counsel) present in the audience. In May 2009 I wrote up these WhenU placements on my web site in great detail. Yet ten months later, the problem continues unabated. Indeed, the other misplacements I identified in May 2009 also continue: Google continues partnering with IAC SmileyCentral (deceptive browser plug-ins that induce searches when users attempt navigations), placing ads on typosquatting sites (including sites that show a company’s own ads when users mistype that company’s domain name), and, through Google Chrome, inviting users to search (and click prominent top-of-page ads) when direct navigation would better satisfy users’ requests and avoid unnecessary advertising costs for advertisers. I’m disappointed by the lack of progress when, in each instance, the improper charges are clear and well-documented. Google’s intransigence confirms the need for the Bill of Rights for Online Advertisers I proposed this fall.

Measuring the Perpetrators and Funders of Typosquatting

Moore, Tyler, and Benjamin Edelman. “Measuring the Perpetrators and Funders of Typosquatting.” Lecture Notes in Computer Science. Springer-Verlag. Financial Cryptography and Data Security: Proceedings of the International Conference 6052 (2010). (Introduction, Web appendix.)

We describe a method for identifying “typosquatting”, the intentional registration of misspellings of popular website addresses. We estimate that at least 938,000 typosquatting domains target the top 3,264 .com sites, and we crawl more than 285,000 of these domains to analyze their revenue sources. We find that 80% are supported by pay-per-click ads, often advertising the correctly spelled domain and its competitors. Another 20% include static redirection to other sites. We present an automated technique that uncovered 75 otherwise legitimate websites which benefited from direct links from thousands of misspellings of competing websites. Using regression analysis, we find that websites in categories with higher pay-per-click ad prices face more typosquatting registrations, indicating that ad platforms such as Google AdWords exacerbate typosquatting. However, our investigations also confirm the feasibility of significantly reducing typosquatting. We find that typosquatting is highly concentrated: of typo domains showing Google ads, 63% use one of five advertising IDs, and some large name servers host typosquatting domains as much as four times as often as the web as a whole.

The Dark Underbelly of Online Advertising

Edelman, Benjamin. “The Dark Underbelly of Online Advertising.” HBR Now. (November 17, 2009).

The Internet is sold to advertisers as a highly measurable medium that is the most efficient way to target exactly the right customers. But online advertising is also easily subverted–letting fraudsters claim advertising fees for work they did not actually do. The trickiest frauds deceive advertisers so effectively that measurements of ad effectiveness report the fraudsters as exceptionally productive and high quality, rather than revealing that their traffic was actually worthless. This is a quiet scandal. In a time of tightening ad budgets, losses to advertising fraud come straight from the bottom line–but savings can be equally dramatic. Here’s a look behind the veil–an explanation of ad practices that have cheated even the Web’s largest advertisers. Advertising scams take plenty of victims, both witting and not, but I offer strategies to help determined marketers protect themselves.