This is a discussion of a Microsoft patent granted today that may not have been implemented, and may never be. It’s unclearly written, but worth discussing…
When you perform a search at a search engine, the page that shows the results of your query is often referred to as a search results page.
Search engines don’t like to show a link to the same page more than once in their search results pages – at least in the unpaid Web search part of their pages. But, most search engines also show advertisements on many search results pages, which look similar to the Web search results.
It’s also possible that a search query using multiple terms, each of which an advertiser may be bidding upon, may cause a page to show up in paid results more than once.
And it’s not completely unusual to see a link to the same page in the paid results and the Web results on a search results page. But what if a search engine decided to show the paid-for result, and remove the Web result?
A newly granted patent from Microsoft describes what happens when links to the same page might appear more than once upon a search results page in Web search (usually one version gets filtered). It also discusses what might happen when the same displayed URL may show up more than once in the paid results listings. Part of it also discusses the instance where the same link appears in both paid results and web results.
Duplicate Pages in Web Search Results
Sometimes the same web page may be indexed more than once by a search engine under different web addresses (URLs). Here’s an example of how that might happen:
- http://www.example.com
- http://www.example.com/home.html
The page “home.html” is the default file in the root directory for the “example.com” domain and the search engine may have recognized (it doesn’t always) that they are the same page. If the search engine finds that both of these pages are relevant for a search that someone has performed, it may filter one of the versions, and show the other one.
The summary of the patent tells us:
The present invention provides a system and method for removing unnecessary multiple references to a common resource such as removing certain redundant listed Uniform Resource Locators (URLs) that reference the same display URLs (and thus the same Web page) as another listed URL.
Duplicate Pages in Paid Results
There are times when the owner of a page may pay to have the page show up in the paid search listings section of a page.
Normally, a paid search result is tied to a keyword being bid upon, and someone advertising their site may choose to bid upon more than one keyword term or phrase. What should a search engine do when a searcher uses a query which uses more than one of those bid-upon terms or phrases?
Should it show the same page twice? Should it show the most relevant? Frankly, it’s a little hard to tell what this patent is suggesting here when there is more than one URL showing up in paid results.
What’s worse is what it seems to imply when there’s a page that shows up in both paid and unpaid results.
Microsoft’s Granted Patent
Systems and methods for removing duplicate search engine results
Invented by Navin Martin Joy and Sally Salas
Assigned to Microsoft
US Patent 7,185,088
Granted February 27, 2007
Filed March 31, 2003
Abstract
The present invention is directed toward efficiently locating desired information and, more specifically, to providing a system and method for removing unnecessary multiple references to a common resource such as redundant listed Uniform Resource Locators (URLs) that reference the same display URLs (and thus the same Web page) as another listed URL.
Consequently, in circumstances where only a smaller, finite number of listed results are immediately used (such as displaying only the twenty most relevant results on the first page presented to a search engine end-user), the finite number of listed results may correspond to a greater number of unique display URLs than would otherwise occur absent this form of filtering.
Some Controversy: Removal of Web Results
Keep in mind that not every patent that is granted gets implemented. And some that are granted maybe shouldn’t be implemented. Some patents also aren’t very clear as to what they are presenting either.
I’m quoting the following section out of context, and it probably needs to be in context to make more sense – but I’m struggling with it reading in context, too. It appears to say that in some instances, when a link to a page appears both in paid listings and in Web results, that the unpaid listing might be removed from results.
On the other hand, if any of the listed URLs is a paying URL, at step 458 the system determines if any of the paying URLs are immune URLs (“URLs”)–that is if any of the paying URLs are immune from elimination. (The present embodiment presumes that some paying URLs may be immune from elimination and that only paying URLs may be immune; of course, other presumptions are possible and the present invention is in no way limited to this particular presumption as described herein).
If there are no immune URLs (again, paying URLs that are immune from elimination) then, at step 460, the system determines if any of the non-paying URLs (“npURLs”)–that is, any listed URL that is not a paying URL–has a higher relevance than the one paying URL having the highest relevance among paying URLs. If the most relevant paying URL is also the most relevant listed URL in the group, then at step 456 the system eliminates all but the most relevant listed URL (which, by default, is a paying URL).
However, if there is at least one non-paying URL with a higher relevance than the most relevant paying URL (“MaxPURL”), then at step 462 the most relevant paying URL and the most relevant non-paying URL (“MaxnpURL”) swap their respective relevances (including any relevance content) and then, at step 456, all of the listed URLs are eliminated except for the most relevant listed URL (which, because of the swap, is the most relevant paying URL).
Conclusion
Some of my uncertainty comes from the use of the phrase “immune from elimination.” This patent doesn’t describe how a paid listing becomes “immune.” Is it something that an advertiser would be able to decide upon, with full disclosure that a web result might not be shown if the paid result was? Does it mean something else?
Also, some parts of this patent could be used without other parts being implemented. The parts of this application that could cause duplicates of Web results, and possibly duplicates of paid results from being shown, could be adopted without the part that might cause a choice to be made about whether or not to remove a Web result when a paid result was also showing.
I’ve seen links to the same page in Google listed in both paid and web results on a single results page. Microsoft’s search may do the same thing.
What do you think?
I’ll tell you what really irks me about this patent….that it was granted. The US Patent office grants a patent on showing each URL only once on a page or in side-by-side search results? That’s not an invention. It’s a technique, and an obvious one at that. I could program that in my sleep. This country really needs to revamp its patent laws to exclude this type of application. I might as well file a patent on the shape I twist a paper clip into to clean my fingernails.
Don’t tell anyone, but that thought crosses my mind sometimes when I read some of the patents that I come across.
Most patent filings I see aren’t earth shaking and revolutionary, but are instead small incremental steps away from things that people might be doing already.
I can’t say that I’ve seen any patent filings that discuss this topic directly, and talk about the relationship between Web results and paid results in quite this manner. When I saw the title, I thought it was going to be mostly about duplicate content issues in organic results, but it really opens the door to get us thinking about paid listings in some new ways.
I can’t recall ever seeing a search results page (from any search engine) where the top two paid results point to the same page, and then the top organic result points to it, too. Maybe I wasn’t paying attention. Or maybe there’s something else going on.
Hi Miriam,
I think a lot of folks would make that choice, too.
Gosh, it would seem to me that MS would be shooting themselves in the foot with this. I would assume they’d like people to spend as much as possible on advertising, but if that were to interfere with their organic rank…well, what’s the web business owner going to choose? A ‘free’ top organic rank, or one they have to pay for? If this patent were to be implemented and MS forced website owners to make this choice…well, I know what I’d choose!
Miriam
Maybe re-reading the patent on Dispersing search engine results by using page category information can help to understand the motivation behind this one. Bill had a post about it last year, as far as I remember, and it came up again in January. That patent was about
“…a search engine that displays search results related to various topics or categories on a single page of search results independent of conventional rankings. By displaying such dispersed search results, the user is able to view a variety of results on the first page of results.” (0006)
Imagine the ~10 positions on a results page are not simply shared according to normal ranking, but split among the possible meanings of the query first. E.g. on the first results page for “Jaguar”, only five positions are used for the car, three for the cat, two for other purposes. If you only consider the five positions available for car-related sites, the impact of a duplicate in this group, or between this group and the ads, on the experience of a user who is interested in car-related sites is considerably higher than if you take the “normal” standpoint and ask for the percentage of duplicates among all positions on a results page.
Good point, Chris.
A reasonable results diversification does seem like a goal that Microsoft is trying to persue. Does that desire, and should that goal cross over from organic results to paid search results? I don’t know.
Microsoft has mapped this process out with the patent, but we don’t know if they will ever use it.
From a web marketing point of view the exclusion of an organic result in favor of paid search is most likely a step in the wrong direction.
In fact those web sites capable of achieving both organic and paid listings on the SERPs enjoy greater CTRs when compared to those with only a paid or organic listings, and the difference is rather considerable according to the results presented at the RIMC Conference in Iceland in November of last year on the panel Doing Business on the Internet.
Also questionable if a technique can be raised to the status of a patent – but that’s another issue …
Sante
Hi Sante,
Thanks for your comment. You do raise some great issues.
The benefit of such advertising does appear to favor an advertiser. But, the patent focuses upon the income of the search engine itself. Given a choice between clicking on the ad or the organic result, which would a searcher be more likely to do if both appeared on the same results page?
I don’t know.
Is Microsoft doing this? Any of the other search engines? I don’t know that either. I hadn’t considered that they even might think about doing something like this without the patent. The patent describes how this might be done, and describes a little about why – but not in a detailed exploration of whether or not it really might be beneficial to them or to advertisers.