The complaint against Google (NASDAQ:GOOG) et al that was filed on the 15th of September at the Eastern District of Virginia, Alexandria division, by I/P Engine Inc, a company that was bought by Vringo (VRNG) this year, gives a good overview of what precisely is the issue behind the trial. I have always found that when discussing lawsuits, one tends to get confused by the barrage of witnesses, documents in evidence, expert testimonies, and other legal paraphernalia, unless one takes a chronological outlook to the whole matter and goes back to the origins, which is generally the first complaint filing. So let's take a look at the Vringo complaint, of which I have a copy, with some footnotes, elaborations, and commentaries.
How Will A Study Like This Help Investors?
A study like this has immediate and long-term predictive value. If you understand just what is at issue, you will be in a position very similar to where the jury is now, and you will be able to make an informed call about how the case will be decided by the jury. This has immediate predictive value for both Google and Vringo investors, both Long and Short position holders or those interested in initiating any position. This case, as has been discussed by me and others already, will have tremendous impact on how Google conducts business in the future, since, as we will see, it directly relates to its principal source of income, Adwords. This case will also have tremendous impact on Vringo because for this small company this is an all-out war for survival.
As for longer term predictive value, understanding the nitty-gritty of this case will help investors make informed calls on future cases that Vringo might bring against other parties, or Google might face for any other similar patent infringement. If you understand how and why Google decided to move with the trial, then, once you know this one's outcome in the next few days, and the next time Google faces a case like this, you will know whether the company should proceed to trial, or not.
The Defendants in the Lawsuit and Vringo's Strategy
The complaint was filed against 5 companies:
- AOL.com (NYSE:AOL), for infringing on Vringo's patents in its advertising.com ad placements, as well as while using Google's search advertising system.
- Google, for infringing on Vringo's patents by using the Lang/Kosak Relevancy Filtering Technology in developing "Quality Scores" to filter advertisements.
- IAC's (IACI) ask.com search engine, for infringing the Lang/Kosak Relevancy Filtering Technology directly, and indirectly by using Google's ad system.
- Target (NYSE:TGT), an ecommerce website that uses Google's ad system to place relevant ads besides product descriptions on its website search results pages.
- Gannett (NYSE:GCI), a publishing business, for similarly using Google's ad system.
Note that although there are 5 named defendants, the issue at the core -- except in the case of AOL's and IAC's direct infringement -- is Google, and Google's ad placement system and its infringement of the Lang/Kosak Relevancy Filtering Technology through its "Quality Score" method. Why did Vringo's litigation team take this approach? There are literally millions of websites and companies that use Google's adwords/adsense models, so why target AOL and the other 3? I think the strategy is twofold; one, to actually try to bolster their case by using the other defendants as so many "examples" of the scope of the infringement and the volume and profitability of the business; and two, to enforce licenses from a slew of similar businesses if the trial is won by Vringo. In the first part, Vringo is basically telling the jury, "Here, take a look at all these big businesses using our patents indirectly through Google. This should be counted in trillions of dollars of money generated from our patents; so, please remember that when awarding damages." In the second, Vringo is telling those hundreds and thousands of other companies out there that use Google adwords, "Watch how we won against AOL, or IAC, or Gannett, or Target - companies just like yours.Wonder what will happen if we take you to court?"
Ken Lang and Donald Kosak's Involvement in the Technology
Andrew Kennedy Lang was a doctoral student at Carnegie Mellon in 1995, where he was writing his dissertation on adaptive filtering and recommendation system technologies. His Professor Michael Mauldin was the founder of Lycos, at that time the largest "portal" on the Internet. A portal was the precursor to the modern search engine. It was a directory of webpages sorted by sectors like travel, industry, news etc. Since the number of websites at that time could be counted only in tens of thousands, it was relatively easy to manually sort them into directories, where users could come and look up a particular site.
However, as the Internet grew to billions of webpages, it became impossible to sort them manually into portal directories. An associated problem was placing advertisements on portals that were relevant to the webpage a user was viewing. For example, suppose you are viewing a travel page on Vietnam; an advertisement for your local dentist would hardly be relevant here. However, in manual sorting, it was difficult to manually link each algorithmic search query to relevant ads.
A Related Issue: Related to the advertisement issue above, a company called goto.com (later named overture.com) launched the "Pay-per-click" (NYSE:PPC) system in 1998. In this system, advertisers only paid when an user clicked on their ads (links) placed on goto.com's search results pages. The ad placements were ordered according to the highest bidders, so if you bid more, your ad was on top. However, the problem with PPC was that the ad placement was not done relevant to search queries; so an advertiser could potentially bid his way to the top of a search query that was irrelevant to his ad. So, an user would see the ad - thereby letting the advertiser gain publicity - but it would be free since the user would not click the ad (it being irrelevant) and goto.com would not make any money.
All of the problems discussed above: a) relevance based sorting of webpages as search engine query results, b) placement of relevant ads besides these results, and c) PPC manipulation through lack of relevancy, could be addressed by technologies developed by Ken Lang and Donald Kosak.
The "420" and "664" Patents
Both patents are directed to (that is lawyer lingo for "related to") the so-called Lang/Kosak Relevance Filtering Technology which produces more relevant search results for user queries. This technology aims to
- filter items for content relevancy to a search query, so if you search for "Vietnam travel" it does not produce pages about "local dentist."
- provide feedback information from prior users, and in filtering the items. So it tells you if the algorithm was not working for some query.
- combine the provided feedback information with the content relevancy to determine whether (or where) an item should be "ranked" in a search results response to the query.
To quote from the actual patent (this is a set of patents, starting with US Patent 5,867,799) the process of relevancy determination:
The integrated filter system compares received INFORMONS to the individual user's query profile data, combined with collaborative data, and ranks, in order of value, informons found to be relevant. The system maintains the ranked informons in a stored list from which the individual user can select any listed informon for consideration As the system continues to feed the individual wire the stored relevant informon list typically due to factors including a return of new and more informons adjustments in the user's query feedback evaluations by the user for considered informons and in collaborative feedback data. (US Patent 6314420)
Google's Quality Score concept, extensively used in its search engines, Adwords and targeted ad systems, is a broad development of this concept, according to the complaint. To understand how quality score follows from the Lang/Kosak Relevance Filtering Technology, let me quote from Google's Help Pages:
Quality Score is an estimate of how relevant your ads, keywords, and landing page are to a person seeing your ad. Having a high Quality Score means that our systems think your ad, keyword, and landing page are all relevant and useful to someone looking at your ad. Having a low Quality Score, on the other hand, means that your ads, keywords, and landing page probably aren't as relevant and useful to someone looking at your ad.How we calculate Quality Score
Every time someone does a search that triggers your ad, we calculate a Quality Score. To calculate this Quality Score, we look at a number of different things related to your account, like the following:
As we can see, what Google calls Quality Score appears to be what Lang and Kosak called relevance in their patents. There are some modifications, especially in the user feedback system that Lang-Kosak developed; Google makes this feedback system automated with account history and past CTR. However, the basic ideas: that 1) relevance is the most important factor in producing search query results, 2) relevance can be determined using certain algorithms and signals, and 3) that relevance is to be used to sort query results, including ad placements - these ideas are common to the two models, and since Lang-Kosak is patented prior art, (some of) Google's concepts derive from it.
Disclosure: I have no positions in any stocks mentioned, and no plans to initiate any positions within the next 72 hours. I wrote this article myself, and it expresses my own opinions. I am not receiving compensation for it (other than from Seeking Alpha). I have no business relationship with any company whose stock is mentioned in this article.