Monday, November 2, 2009

Selecting means more than one choice (Ex parte Hemmat)

Ex parte Hemmat
Decided October 13, 2009
(Appeal 2009-003911; Appl. No. 10/429,615; Tech. Center 3600)

I ran across this case when reading the blog Gray on Claims. (I like this blog a lot because it covers BPAI cases in addition to Federal Circuit and District Court.) The subject matter of the application is a business method and several issues were on appeal, including 101/Bilski.  I'm going to discuss only the indefiniteness issue. In particular, I'm going to explore other ways of writing the claim  to avoid indefiniteness.

I've edited the claim to show the portion involved in the indefiniteness issue:
holding a meeting...wherein the meeting results in additions to the concept document identifying at least one proposed approach, ... ;

identifying impacted systems for each proposed approach by ...;

selecting a final selection from the at least one proposed approach with a view of the long-range plans and overall direction of an enterprise rather than to a specific problem.
The Board found that selecting requires more than one choice, and this was inconsistent with the use of "at least one", which allowed for only a single choice.

When I read this, my first thought was: "That's not inconsistent. A POSITA understands that if there's only one choice, the selection degenerates into selecting that one choice".

Give the Board some credit: the indefiniteness holding did not rest on this inconsistency alone. The Board also found that other parts of the claim were inconsistent with selecting the single choice. For further details on that, read the case summary on Gray's blog (here).

What I'm interested in is this: though here the Board did not find selecting-from-what-could-be-a-single-choice to be indefinite per se, I can definitely see an Examiner giving me such a rejection. And I wouldn't be surprised if, in another case, the Board did find indefiniteness for this reason alone.

So I'm wondering if there is a better way to claim this sort of thing, to avoid the specter of indefiniteness completely. The claim in Ex parte Hemmat is complicated, so I'm going to discuss this issue in the context of a much simpler hypo.

Here's the feature I want to claim, in plain English rather than claim language. The method performs a search which comes up with possible candidates, then chooses the best candidate as a match. The spec would explain that when the search results include only one candidate, that one candidate is the best match.

Using the style from Ex parte Hemmat, this translates to:
Approach #1
performing a search that produces at least one candidate;
selecting a best match from the at least one candidate.
But the lesson from Ex parte Hemmat  is that approach #1 is vulnerable to an indefiniteness rejection.

How about this instead?
Approach #2
performing a search that produces a plurality of candidates;
selecting a best match from the plurality of candidates.
I don't like approach #2 because no infringement occurs for those data sets that return only one candidate. While the accused infringer doesn't escape entirely — infringement still occurs whenever data sets return 2+ candidates — this does complicate damages and proof of infringement, doesn't it?

How about this?
Approach #3
performing a search that produces a group of candidates;
selecting a best match from the group of candidates
The idea behind "group" is that it allows for 1 or more without screaming "indefiniteness problem" like "at least one" does. But maybe not. Maybe it's got the same problem: since it allows for the possibility of only one, this conflicts with the verb "selecting."

Then the obvious solution is to get rid of the pesky verb "selecting," which means we can go back to "at least one". How about this:
Approach #4
performing a search that produces at least one candidate;
determining a best match from the at least one candidate.
Since determining doesn't carry the connotation of more-than-one, this one should be safe from the charge of indefiniteness. But determining isn't quite the same as selecting from a group of predetermined choices, and it's possible that you need to express this difference in order to avoid prior art.

Here's my last attempt. Does the following approach allow you to keep the determine-vs-select distinction yet avoid indefiniteness?
Approach #5
performing a search that produces a group of candidates;
determining a best match from the group of candidates by:
    if the group has more than one candidate, selecting the best match from the more than one candidate;
    if the group has exactly one candidate, appointing the exactly one candidate as the best match.
Certainly #5 is clunky, and wouldn't be my first choice. But maybe it's good to have as a backup.


  1. How about:
    A method for identifying a best match candidate, the method comprising:
    initiating a search to obtain a list of one or more (potential)candidates;
    obtaining a search result; and
    using the search result to identify the best match candidate.

  2. It looks like you avoid the at-least-one problem. That's good. But it seems you accomplish this by being vague about the quantity of "search result". I don't like vague, although some practitioners think vagueness is a good thing. [Or even a great thing.]

    Is "result" is meant to be singular or plural?
    Also, how does "result" tie to the list and the candidates?