How many forward citations does the Microsoft touchscreen patent US8077153 have? And why does this matter?

最終更新日

While preparing a recent analysis of leading Microsoft (and Google and Apple) patents we came across US patent 8077153, for Precise selection techniques for multi-touch screens.

According to our analysis, this was one of the most important Microsoft patents with an AmberScore value of 8.3, or 8.3 times the average value for US patents filed in the last 20 years. This patents sits in the middle of a crowded network, with 76 forward citation patents, as shown below (please click on this image to see an interactive version of this network).

US8077153-network.jpg

But if you were select this patent in the popular seach tool Google patent, a funny thing happens – only 4 forward citations are shown:

US8077153-in-Google-patent.jpg

 

As the developer of AmberScope, we were a little concerned about this discrepency, so we looked up the same patent in Espacenet:

Espacnet-forward-citations.jpg

 

Thankfully, this showed 76 forward citations as well. 

But if we say look up the same patent at the USPTO, you will end up with a forward citation count of 1.

And this varying forward citation count is not limited to these websites. Patentbuddy gives a forward citation count of 1, Patentlens gives a forward citation count of 4, and RPX gives a forward citation count of 1.

 

So what is going on here?

After a bit of detective work, the answer became clearer. US8077153 has been published twice, firstly as patent application US20070247435 in October 2007, and then as the granted version US8077153 in December 2001. And if we look at this patent application, it indeed has as foward citation count of 82, which is quite close to 76 (if not exactly the same – for reasons that will become clear shortly).

US2007247435-forward-citations.jpg

 

And why was the application cited so much and not the granted patent? We suspect two reasons:

  • Because the application has been published for over 7 years now, while the granted patent has only been published for almost 3 years
  • And maybe because examiners and patent applicants for later and similar patents got so used to citing the patent application number they just kept doing so, rather than checking to see if the patent had been granted when they referred to this patent

So the explanation for the varying forward citation count is as follows:

  • Espacenet (and AmberScope) treat a patent application and granted patent as being the same record . There is a good reason for this, and for example in Europe and many other countries both the patent application and granted patent only ever have one number between them. And why not – it is the same invention right?
  • But Google patent and others treat the patent application and granted patent as being two separate documents. This could also explain the sum of the forward citations for the application (82) and granted forms of the Microsoft patent (4) is higher than 76 – because there might be some double counting back to these seperate documents, and double counting from other separate applications and granted patents.

 

Why does this matter?

Forward citation count is used as a key factor many patent quality prediction indices, including (indirectly) in the AmberScore metric supplied by AmberScope. For example, one of the aformentioned patent search sites likes to supply a patent citation ranking in the form of % this patent is cited more than, within patent classification classes. With a forward citation count of 1, this patent ends up with this percentile ranking of 65.3%, which is nothing special and a position shared or beaten by over 21,600 other patents in this patent class.

But if this same site used what we believe to be the correct figure of 76 forward citations, this patent would have a percentile ranking of 99.2%, a position that would place it in the top 10 out of over 63,000 patents in its class. Which is a very different ranking indeed.

 

Implications

If you are using forward citation count as an imput into your patent quality metrics, or even just using this figure as a patent quality rule of thumb, be very careful about what figures you are using. Because as this simple example as shown, different patent search website can give you very different forward citation counts, in turn leading to very different quality estimates. 

If for example you were preparing a report on this patent for your clients or boss, would you base your analysis on a forward citation count of 76 (which considers how the patent application was cited) or a forward citation count of 1, as per the USPTO and some other websites?

(For the record, in our earlier Microsoft patent ranking US8077153 ended in 12th position out of the over 19,000 US Microsoft patents we reviewed, although we did not specifically publish this ranking at the time).

 

Read More