Wednesday, September 05, 2012

"OpenCL" Not Merely Descriptive of API Component of Operating Software, Says TTAB

In a puzzling decision, the TTAB (Trademark Trial and Appeal Board) reversed three Section 2(e)(1) refusals of the mark OpenCL, in standard character and design forms, finding the term not merely descriptive of "Application programming interface computer software for use in developing applications for execution on central processing units (CPU) or graphic processor units (GPU), sold as an integral component of computer operating software." However, as to the two design marks, the Board agreed with the PTO that Apple's specimens of use were unacceptable. In re Apple, Inc., Serial Nos. 77616247, 77844718, and 77844736 (August 28, 2012) [not precedential].

Mere Descriptiveness: The PTO asserted that OPENCL "immediately identifies the common or generic name of an industry standard language and application programming interface," and therefore cannot be a trademark. Or, put another way, "because OpenCL is the name of a computing language, it is merely descriptive of applications programming interface software that permits software developers to use the language." [Sound like a good argument to me - ed.]

Apple asserted that "the fact that OPENCL identifies a programming language does not preclude recognition of OPENCL as a trademark for software that implements the language." [Why not? - ed.]

The Board posed the question this way: "for OpenCL to be merely descriptive, we have to find the following:

  1. OpenCL is an abbreviation for 'Open Computing Language';
  2. 'Open Computing Language' is merely descriptive of products in the [trademark] application; and
  3. A relevant consumer viewing OpenCL in connection with applicant's products would recognize it as an abbreviation of the term 'Open Computing Language.'"

[Query: didn't Apple concede that OpenCL "identifies a programming language"? - ed.]

The Board found that items 1 and 2 were satisfied, but item 3 was not. Nine of 25 articles cited by the PTO used the term OpenCL followed by the parenthetical term (Open Computing Language), indicating "that these companies and authors believed that they needed to spell out the connection between OpenCL and 'Open Computing Language' because the readers would not immediately understand that the mark OpenCl means 'Open Computing Language.'" [What about the other 16 articles? And how does the Board know why these companies and authors used both terms? - ed.].

Although some consumers might figure out that OpenCL means "Open Computing Language," the process requires some thought, "and that is the essence of a suggestive mark."

The Board therefore reversed the mere descriptiveness refusal.

[I don't understand how, if the term OpenCL identifies a programming language, it is not descriptive of a least a feature or function or characteristic of an API that implements that language. Is the word "French" not descriptive of a software program that teaches French? - ed.]

Specimens of use: In Apple's specimen of use for its design marks, the term OpenCL was used to identify the product as a programming language instead of as an API: "OpenCL in Snow Leopard is a technology that makes it possible ...." Thus that specimen of use was not acceptable.

As to the standard character mark, the specimen of use consisted of a screen shot showing the mark used to identify an API, and that specimen was okay.

And so the Board reversed both refusals as to the standard character mark OpenCL, but affirmed the refusals to register the design marks due to faulty specimens of use.

TTABlog comment: I just don't understand this decision. I spelled out why in my comments above.

Text Copyright John L. Welch 2012.


At 3:30 PM, Anonymous Anonymous said...

I think the biggest mistake in this is that Opencl was promoted as an "industry standard" language and API. And the record was full of evidence of recognition as an industry standard. If opencl is the name of an industry standard (or a standard specification for the way to do something) such as .html, mp3 or CAT5, there is no way that it can also be a sourc-indicator for their own particular goods. Their specimen only identifies one implementation of the standard that must be freely implemented by all.

At 9:01 PM, Blogger Pamela Chestek said...

The Board didn’t address the character of the letters “OpenCL” in relation to a language, but only said that a language is not a good in trade and therefore it cannot have a trademark. So its analysis of what the letters “OpenCL” mean in relation to software, a type of good that can have a trademark, started with a blank slate. Maybe an analogy is a title of a single book, which also has a bit of a sui generis rule that it can’t have a trademark. I can therefore have a book “Purple Dresses” that says nothing about purple dresses, so if it was a mark it would be arbitrary, but I can’t register a trademark for single book titles. If I then have a store that sells only the book “Purple Dresses,” I don’t think that you would consider the store name descriptive for retail store services because the contents of the store happened to be only the book “Purple Dresses.” The descriptiveness of the mark for retail store services is just a separate question altogether.

But I differ with the Board’s reversal of the refusal of the specimen. The goods are “application programming interface software.” An “application programming interface” (API) is an abstraction. The TTAB defines it in the decision as “the interface through which one program can communicate with another.” So I don’t believe an API can have a trademark either, and Apple wasn’t trying to register a trademark for APIs. Rather, it was trying to register a mark for application programming interface software. But when I look at the specimen in the opinion (a clearer version is an Aug. 24, 2011 entry in the TDR record for SN 77616247), I don’t see software that bears a mark “OpenCL,” I see software that, at best, allows one to read the documentation for APIs that are called “OpenCL.” The traditional places we look in software for the mark, like a title bar, say nothing about “OpenCL.” The software here is titled “Organizer - Documentation,” not “OpenCL,” and the banner at the top that says “OpenCL Hello World Example” is simply the name of the file that is open, not the name of the software. The Board claims that the left-hand column shows use of the mark, but these are simply file names (relating to APIs), not a trademark use for software.

The Board chided the Examining Attorney for arguing that the specimen was for APIs, not software, because the Examining Attorney hadn’t objected to the identification of goods earlier. Of course he couldn’t; he wouldn’t know if the identification of goods was correct until he saw the specimen. If I had a description of goods that said “mules” and I provided a specimen that showed a horse, the Examining Attorney can’t say “that’s a horse, not a mule” until he or she has the specimen. Had this specimen said “OpenCL” in the title bar, then it would have been an acceptable specimen.

So I don’t disagree with the Board that a trademark for software related to a language is not necessarily descriptive if it has the same name as the language, but I don’t see that Apple has actually used the mark “OpenCL” for the goods described in its application - that is, software, not the APIs per se.


Post a Comment

<< Home