Measuring Claim Changes - Patent Law Blog (Patently-O)

by Dennis Crouch

Most of the time, patent applicants amend their claims at some point during the patent prosecution process. But, I wanted identify some measure of the amount of amendment. ?Amendments during prosecution are important for legal reasons because because they impact claim scope, claim interpretation, application of the doctrine of equivalents, and potential for written description failure. From the business side, claim amendments are a form of re-work that cost money and potentially delay patent issuance. ?

Yesterday I created a comparison database that includes all of the patents issued on February 12, 2013. For each patent, I also found the corresponding published application (if any). Using a minimum-edit-distance algorithm, I then compared the first claim of the patent with the first claim of the published application. The minimum-edit-distance approach is to find the minimum number of steps needed to transform the application-claim into the patented claim. Once we find the minimum number of changes, we then simply count the number of words deleted from the original application and the number of words inserted to form the new claim. (Here, words also include numbers and punctuation.)?

As an example, Claim 1 of U.S. Patent No. 8,370,968 looks something like the following, with each deletion shown in strikeout and each insertion underlined.:

1. a cervical spine protection device adapted for use with a users head protection device , the cervical protection device comprising ,

a lower portion engaged with a users shoulders , the said lower portion having an opening at a the forward facing side of a the users body ;

an upper portion designed for engagement with a the users head protection device when an axial force is applied to a top portion of the users head protection device , the said upper portion having an opening at the forward facing side of the users body and a rearward facing side sides of a users body ;

and a plurality of compression elements contained within the said upper and lower portions of said cervical spine protection device , wherein the whereby axial force applied to the a top portion of the users head protection device is directed away from a the cervical spine of the user in an oblique direction and wherein the via said plurality of compression elements comprise a first horizontal portion within and adjacent to the upper portion of the cervical spine protection device and a tapered lower portion designed to follow the contours of a users neck wherein a force applied to the first horizontal portion is directed along the tapered lower portion away from the users cervical spine in an oblique direction.

By my calculations here, the minimum-edit-distance for the '968 involves 36 words being inserted and 87 words being deleted. You can see that the inserted words are, for the most part, related directly to elements found in the claim original applied-for claim. To take that into account, I added an additional measure that counted the number of new words inserted into the claim, where "new words" are defined as words that were not found in the original applied-for claim.

The two findings surprise me most: (1) In over 45% of the cases, the only amendments to claim one involved deleting or reshuffling terms (i.e., no new words were added that were not in the original claim); and (2) In the majority of cases the patented claim is shorter than the applied-for claim. The difference in application-patent claim length is about 30% on average. Taken together, these results run contrary to the conventional wisdom that the usual amendment practice is to simply add language from a dependent claim to create a new independent claim. ?Rather, the suggestion here is that a substantial number of patentees are expanding the scope of their claims during prosecution rather than contracting.?

Bi-Modal Distribution of Claim Strategy: It turns out that speaking about the "average" changes probably does not make much sense in this situation because there is someting of a bi-modal distribution (two modes) and anything-but a proper bell curve. ?Since longer claims have the potential of having more inserts and deletes, I normalized the change amount to a 0-100% scale. ?A 0% change represents absolutely no change to the claim. ?A 100% change means that the claim was entirely rewritten using none of the same words. ?The histogram chart below shows the tendency of applicants to choose one of those two extremes.?

As usual with patent issues, there is significant differences if we divide the data according to technology center. Applications in TC 2800 were most likely to have changes of <5% while those in TC 1600 and 3700 were the most likely to see changes of >95%. ?TC 1600 is also the only technology center where the averge claim length increases during prosecution.?

?

Across the board, dramatic changes are more likely when the application is a continuation or divisional rather than an originial non-provisional. Japanese filers are more likely to have very small changes; German filers are more likely to have big changes. In my small sample, I found that Google & Apple almost always make substantial changes to their applications, while Honda & Monsanto were the most likely to get issued patent claims that were virtually identical to the originally filed claims. Finally, applications examined by an assistant examiner were much less likely to be patented with only very small amendments to the claims as compared with those seen by the primary examiner.?

In a second go-round I will improve my algorithm a bit to give relatively less weight to changes such as "the?said" and relatively more weight to more major insertions.?

?

?

Source: http://www.patentlyo.com/patent/2013/03/measuring-claim-changes.html

nikki minaj album of the year grammy red carpet grammy award winners the band perry grammy awards whitney houston autopsy