Search Patentiability: Chapter 2

White Space Patenting Chapter 2



Click to download this chapter of White Space Patenting (pdf includes illustrations)

How is it that good patent counsel can often devise white space claims in a few hours, whereas it may have taken the inventor months or years to come up with the invention? Why is it that the most effective patent counsel is often someone who is not an expert in the particular field of the invention? The answer is that a good patent attorney or agent (A) understands what constitutes a patentable invention, (B) performs a decent prior art search, and (C) employs a lot of creativity and brainstorming during the searching process to find and fill the white space. This chapter shows how each of those steps is done.

    1. A)Understanding What Constitutes A Patentable Invention.

To secure a patent from the U.S. Patent Office, one must be able to describe an invention that is (1) new, (2) useful and (3) non-obvious.

Chapter 2 White Space PatentingThe “new” part is usually fairly easy to satisfy. As long as neither the inventor nor anyone else has ever put the claimed product on the market (or used the steps of a method), and the concept was never disclosed or suggested in a book, magazine, or anywhere else, then the inventor probably has a novel idea. In the past, an inventor used to have a one-year window (referred to as a grace period) after first disclosure or commercialization to get his U.S. patent application on file. But starting March 16, 2013, that grace period is effectively all but gone. The U.S. is now on a “first to file” system, which basically means a race to the patent office.  Yes, it is possible to obtain patent rights when someone else files first, but proving that the other person stole the idea, or otherwise improperly filed, is a hard sell.

The main things to watch out for in searching novelty are: (1) that the idea is novel throughout the world; and (2) that the idea was novel at all previous times. If an idea was publically disclosed in a patent application in Germany,

it is not novel in the U.S. under the new AIA rules. If the idea was disclosed in an old patent that has now expired, it is not novel. Even if the idea was disclosed but never patented, it is not novel. With some exceptions, the proper test for determining novelty is a comparison of the claims against what is already known anywhere in the world at any time. The test is definitely not comparing the claims against what has been patented.

As with other aspects of patent law, the Patent Beast™ comic strip ( captures some of the more humorous aspects of novelty.

The “useful” part of determining patentability is also usually quite easy to satisfy. If a new idea were not useful, one wouldn’t be spending the time or money trying to get a patent. Indeed, the standard of usefulness is so low it doesn’t have anything to do with commercializability. What, for example, is the commercial benefit of patenting a comb-over to cover a bald spot, (see US4022227 to Frank and Donald Smith, 1977), or a method of swinging on a swing, (see US6368227 to Steven Olson, 2002)? Such patents could never be enforced. These and many other essentially worthless patents can be found at the Totally Absurd Archives.

Just about the only time applicants get rejections on the grounds of non-usefulness is when the claimed invention appears to be inconsistent with the accepted laws of physics. Perpetual motion machines, for example, are almost always rejected, as are cold-fusion related inventions.

The kicker is usually obviousness of the invention. A ballpoint pen with a single red dot on the barrel may be novel, but it is merely an obvious design choice. By contrast, the use of a sodium nitride ball in a pen was definitely inventive, and patentable, as would be pens that spool out writing paper.

There will be quite a bit more to say about obviousness later on in the book in the discussion on claim drafting. Suffice it to say at this stage that to get a decent patent, one needs to come up with something that is non-obvious to a Person of Ordinary Skill in the Art (a POSITA). And that usually means claiming something that is more than a mere design choice.

      1. (1)“The Invention” Is Not The Same Thing As A “Preferred Embodiment”

Inventors often get into trouble by confusing their “preferred embodiments” with their “inventions”. Preferred embodiments are merely examples of the invention. This is a critical distinction.

Focusing on preferred embodiments rather than inventions has two very undesirable consequences. First, inventors tend to focus their patentability searches too narrowly, and end up wasting a lot of money on patent applications that should never have been filed. Second, claiming preferred embodiments, rather than the underlying concepts, often produces claims that are essentially unenforceable. It is said that 80% of patents are never commercialized at all, and perhaps 95% of patent applicants never make back the money they paid for filing and prosecution of the applications. Those horrible statistics result directly from the fact that many patent claims are way too narrow.

If one is going to get a commercially significant patent, one needs to figure out the core inventive concept, and then figure out the choke points for applying that concept in the marketplace. That means doing the heavy mental lifting of figuring out all of the commercially viable alternatives, and finding a common thread that passes through all of them. Thus, if one invents a new type of serration for a steel knife, one should claim the concept broadly enough to encompass any place the market might have use for the new type of serration. That would include plastic and even wooden knives, as well as squared cutting surfaces that might not even be considered knives.

Some inventors are quite good at figuring out what the core concept is, but more often than not. They are “tree” people rather than “forest” people, and tend to get stuck on the details. And a great many inventors are just terrible at marketing. Most inventors should seek guidance of patent counsel who not only understands the technology, but who is also good at seeing the big picture.

    1. A) Performing A Good Prior Art Search

Effective prior art searching is absolutely critical to white space patenting. Indeed, this provides one very quick way to separate the better patent counsel from the mediocre or poor counsel. Patent counsel who thinks they can write good claims without searching the prior art, should probably be avoided.

The fact is that until one understands what is known in the field, one cannot realistically figure out where the white space is. A good searching strategy involves conscious decisions about when to search, what kind of search to perform, and how deep the search should go. Those decisions involve many different factors, from susceptibility of the field to keyword searching, to the impact of the search on research and development efforts.

In searching the white space, one should be careful about using outside searching services. Even if the service does a thorough job, it is the searcher who grasps the overall landscape of the available white space. It is the searcher who develops new ideas for improving the invention, and understands the right-to-practice issues. By and large, the person writing the patent claims should be the person conducting the search.

There are two important caveats.  First, some consumer products have so many prior art hits, that it is not cost-effective for a patent practioner to run the search. Second, in some technologies there is a lot of critical prior art that can only be found in foreign languages. In either of those cases, it can be very desirable to use an outside searching service.  One of the services that we have found to produce good results, and be cost-effective, is KAnalysis.

      1. (1)You Don’t Know What’s In The Prior Art

It is impossible for any inventor to know everything of relevance (i.e., all of the “prior art”) in the field of the invention. Reviewing patents on Google, and surfing the web for products, are just not adequate search strategies. Nor is experience gleaned from spending many years in the technology space. This is especially true under the new AIA version of 35 U.S.C. § 102, in which prior art includes anything “described in a printed publication, or in public use, on sale, or otherwise available to the public before the effective filing date of the claimed invention”, with very few exceptions.

As noted above, a great deal of information is available only in foreign patents and applications, and in journals and other non-patent literature. In addition, both U.S. and foreign patent applications are generally held in confidence by the patent office for 18 months, so that a patentability search likely won’t find a competitor’s disclosure that was filed only a year ago. Still another reason that an inventor can’t know everything of relevance is that what counts as prior art against an applicant is determined by what is claimed in that application. Since the claims almost invariable amended during drafting of the application, and again during prosecution before the USPTO, neither an inventor nor his or her patent counsel, nor even the world’s best searcher, can possiblypp know everything that is prior art.

What is prior art anyway? For most purposes after March 2013, prior art is the following, modified from 35 U.S.C. § 102 (a):

Prior art is anything that is patented, described in a printed publication, or in public use, on sale, or otherwise available to the public before the effective filing date of the claimed invention.

That’s the law in the United States, and for the most part that’s also the law in other countries. The text looks simple enough, but appearances here are deceiving. What, for example, is meant by “printed publication“, “public use”, “on sale”, “otherwise available” or even “the invention”? The complexity is such that the Patent Office devotes numerous sections (§2121 – § 2138) of the Manual of Patent Examining Procedure (MPEP) trying to explain to patent examiners how to apply the law.

There are three big exceptions to the general prior art rule of AIA § 102(a).

First, under AIA § 102(b)(1)(A), an inventor theoretically has one year from the time he discloses an invention within which to file his patent application. But inventors should not rely on that window. Under USPTO’s Proposed Rules For The “First-to-File” System, published in the July 26, 2012 issue of the Federal Register, an inventor can only prevail against an earlier filer if the subject matters of their respective disclosures is substantially identical. And that, of course, is very unlikely. The U.S. system now embodies “race to the patent office”.

The second exception is more reliable. A child of an earlier filed application can usually take as its priority date the filing date of the parent, grandparent, great-grandparent, etc. For CIP (Continuation-In-Part) applications, in which additional matter is added in the child application, there are two priority dates; the earlier filing date , and the filing date that the additional material was added.

The third exception is that under the America Invents Act Of 2011 (“AIA”) § 102(b)(2), patent application disclosures are not prior art if either the subject matter was obtained directly from the inventor or a joint inventor, or there is common ownership or a joint research agreement.

Inventors sometimes think that U.S. patents are only considered prior art as of their issue date. That contention used to be valid, but the situation changed several years ago. U.S. published patent applications and issued patents are now considered prior art effect as of the filing date of the application. Under the AIA, PCT and foreign applications, whether filed in English or another language, are also given prior art effect as of the filing date of the application.

There used to be a distinction between anticipation  (35 U.S.C. § 102 prior art) and obviousness (35 U.S.C. § 103 prior art). That distinction was also eliminated by the AIA, §3.

One often needs to resolve whether or not information constitutes a printed publication. For example, is a Ph.D. thesis in a major university considered a printed publication for purposes of 35 U.S.C. § 102(a) under the AIA? What about a mere term paper? Actually, the distinction has nothing to do with the sophistication of  the  publication  or  the context in which it was created, but instead has everything to do with whether the document was “publicly accessible”.

A reference is “publicly accessible” whenever:

“… persons interested and ordinarily skilled in the subject matter . . . exercising reasonable diligence, can locate it and recognize and comprehend there from the essentials of the claimed invention without need of further research or experimentation”.

Several cases have required that for a publication to provide prior art effect, the publication must allow those interested in the art a sufficient amount of time to “capture, process and retain the information conveyed by the reference”, or to locate the material “in a meaningful way”.

It turns out those tests are fairly easy to satisfy because they merely address technical accessibility. In Bruckelmyer v. Ground Heaters, Inc., for example, the Federal Circuit found that canceled drawing figures in a Canadian patent application were publicly accessible even though neither the figures nor even the application had been specifically indexed, and there were no copies known to have been made or sent outside the Canadian patent office during the prior art period. In that instance the figures were considered to be prior art merely because the Canadian prosecution file had technically been open to the public more than a year before the inventor filed his U.S. application. The bottom line is that, at least during the searching process, one should assume that any relevant documents that the searcher can find will be considered publicly accessible.

Another factor to consider is that inventors often become unnecessarily discouraged by wild-eyed teachings or suggestions in the prior art. Even under that AIA, disclosures that fail to teach how to make and use what they are describing are “not enabled”, and cannot validly be used by the Patent Office against a patent application.

      1. (1)Choose An Appropriate Search Strategy

With all this in mind, the next steps in searching the prior art are to figure out how much effort to put into the search, and how to go about doing the search. The answer depends entirely upon the specific circumstances.

Figure 11 – What Level Of Search Is Best?

  • Is the search readily susceptible to keyword searching? Keyword searching works best where the invention can be readily described with a few distinctive words. Thus, a clothing perspiration shield that sticks to the skin would be readily searchable, because the searcher could simply search for a combination of perspir*, shield or cover, and adhere* or stick*, where the asterisks indicate a symbol for a wild card. Searching for suntan lotion packaged in single use packets the way mustard and ketchup are sold is also readily searchable. The searcher would simply search for a combination of the terms suntan, packet, and possible single-use. But searching for an internet invention that uses a database to store information is very difficult to search. Terms such as server, database, and user interface can pull up millions of hits.
  • Is the field crowded? Patentability searching often works poorly in crowded fields, even if the invention is readily described by distinctive keywords. For example, a searcher should probably invest minimal time searching a baby toy in which shaped pieces are inserted into correspondingly shaped holes. There are so many competitors that anything more than a quick search is probably a waste of money.
  • Would a thorough search help in developing the product? Sometimes it is useful to conduct a very thorough search even though the subject matter is difficult to search and the field is crowded. The benefit lies less in determining if the idea is patentable than in guiding the R&D effort by identifying what is known in the field. The development of technology is like building an arch. No matter how many blocks are used to build the foundation of the arch, it won’t stand up by itself until someone inserts the keystone at the top. In technology, there may be hundreds or even thousands of people working on a project, each adding to the foundation. But the person who adds that last missing piece gets the patent and all the credit, even though that keystone piece is almost always quite small compared to the whole. A good search can go a long way towards finding that keystone.

Figure 12 – The One Who Adds The Keystone Gets All The Credit

  • Is the technology likely or unlikely to be published? The various keyword databases are only useful in searching inventions that are published. For example, using keywords to search for the use of a particular sewing stitch on the hem of a swimsuit would be almost impossible. Swimsuits have been manufactured for well over a hundred years, with different manufactures regularly trying out new stitches without ever publishing information on what they are doing. Use of one stitch rather than another would likely never find its way into the local newspaper, in a patent, or in any publication for that matter.
  • Does the invention require a lot of R&D? In some instances, a company is planning to spend hundreds of thousands or even millions of dollars on R&D for a product. Knowing early on that the product is unpatentable can save huge sums of money, or at least re-direct that money to something more worthwhile. In those instances, the cost-effective approach is to invest in a thorough search as soon as possible, and update the search as the research develops.
  • Can the marketplace provide the search? If the invention is fairly simple, and there is little or no additional conceptualization to be done, then the most cost-effective strategy is probably to file an inexpensive provisional application to secure patent pending status, and then go out in the marketplace and try to find a licensee or customer. If the product is old, those people will surely say so, and the application can be abandoned. If everyone loves the invention, then it is probably cost-effective to run a more thorough search, and file a formal patent application (utility or PCT).

Of course, the filing costs and prosecution speed should be taken into account. The out of pocket filing costs for filing a U.S. utility application runs about $1,700 for a large entity (in general, at least 500 employees) and takes about 27 months for the applicant to receive the first substantive office action. Small entities pay half that, and under the America Invents Act, micro entities pay only a quarter of the large entity fees for filing and certain other charges. See additional details in Chapter V, section G(1).

Pendencies have improved somewhat.  In 2012 the pendency from filing to first office action was reported to be about 21 months.  At the end of 2015 the pendency was down to about 16 months.  Pendencies and other interesting information can be viewed in graphical form at the USPTO Data Visualization Center.  Here’s an example of their pendency dashboard.

Figure 13 – Portion of USPTO Dashboard From November 2015

Filing costs for a PCT application through the U.S. receiving office runs about $3,500, and a PCT filing through the International Bureau (where searching is done by the European Patent office) runs about $4,000. But while the out of pocket costs are greater, the searches are faster. Where the PCT is filed with no prior priority claim, the applicant is supposed to receive a search report within 9 months of the filing date, and where the PCT claims priority to an earlier application, the applicant is supposed to receive a search report within 16 months of the filing date of that earlier application.

In deciding on an appropriate search strategy, one should focus on both patent and non-patent materials. In some fields, such as computer science, this is especially critical because a great deal of the prior art never makes it into patent publications. As a result, time and money are wasted on patent applications that have no chance whatsoever of being issued as a valid patent with broad claims.

And yet, such applications are filed all the time. Below is a claim from US6874084 assigned to IBM™ with a priority claim to May 2000.

The patent lists only three U.S. patents references, and no references outside the patent literature. If the patent drafter or the examiner had just spent even a short time looking for references outside the patent literature, the patent would never have been issued. The patent was still in force at the end of in 2015, and one can only hope that IBM won’t ever waste anyone’s time trying to enforce it.

The Internet is such a treasure trove of information, many people wonder whether they can access historical “snapshots” of web pages to use in prior art searching. It turns out that one can access old Internet pages through the WayBack Machine.

      1. (3)Tailor The Searching Effort To The Purpose

The first rule is to tailor the searching effort to the needs of the application. Below is a diagram of a broad patentability search, in which the arrows represent searching efforts going off in many different directions, and the squares represent the patents or other documents being sought. The search here is wasteful because it goes off in way too many directions (lacking focus), and several of the searches go deeper into the field (are much longer) than is necessary.

Figure 14 – Wasteful Search

If the goal is to find out whether an idea is worth pursuing, the better approach is to narrow the search. The reason is that the searcher need only find one piece of invalidating prior art to support a conclusion that the invention is old. Drowning in ten feet of water is the same as drowning in 100 feet of water – the person is still dead.

 Figure 15 – Sometimes A Simple Search Will Suffice

      1. (4)Use Iterative Searching

Probably the best strategy is to run searches iteratively, i.e. in stages. Start with very specific keywords in a proximity-based system to see if there is an easy match. When searching for a screw with interleaved threads, for example, one should search for a patent where the terms “screw”, “inter*”, and “thread” are all in the same sentence, or within 10 words of each other. If you find just what you are looking for, then the search is over. If you don’t, then you need to keep broadening the scope (usually by removing keywords from the search) until he finds at least a few patents or other prior references that match. This works on paid services such as Lexis™ Total Patent™, Delphion™, and Westlaw™, and on some free search engines including Google Patents WebCrawler, and FreePatentsOnline.

Unfortunately, with Google™ and most of the other free search engines, the only proximity searching available is exact matches, e.g., “cancer treatment” or “brake pads”.

The second step in iterative searching is to examine the documents that were found in the first step, and then use them to refine the keywords. People use a great many terms and phrases to describe the same thing. A sawhorse may be described in one reference as a sawhorse, but in another reference as a cutting or sawing table. Similarly, the Internet may be alternatively described as a global network, or as a package switched network. One should definitely look in each new reference for keywords that the searcher hadn’t thought of in the first place. In addition to identifying new words, the searcher may even realize that he needs to search for plural terms as well as singular terms. By refining the keywords the searches are both broader (in terms of using more alternatives) and more focused (because the searcher can combine more search terms into a given search).

I filed a patent application several years ago on a searching system that simplifies iterative searching, by providing a user with a concordance of nearby words and phrases. So far, that has not been commercialized.

Figure 16 – Iterative Keyword Searching Focuses The Effort

An alternative to iterative keyword searching is iterative parent/child searching. Once a searcher finds patents or applications in step one (the parents), he then looks at all the references cited in those patents and applications (the children). That second-generation search will likely widen the catch to at least 40 or 50 references. The next step is to examine all those references to find the closest ones, and then look at children and parents of that second generation of references to find a third generation. Repeat that process until critical references are found, or until the subsequent searches find only references that were already identified.

Figure 17 – Iterative Parent/Child Searching Is Also Extremely Effective

      1. (5)Use The Most Cost-Effective Sources For Searching

No matter what level of search is undertaken, and what strategies are decided upon, it is extremely important to choose the right information sources. It therefore behooves a prior art searcher to choose the very best search engine for the job, and indeed to use multiple search engines. This section addresses strengths and weaknesses of specific search engines. Additional information is available on the Internet in a well-researched article put out by Cambia.

Figure 18 – Use The Right Tools For The Search

There are two main criteria for evaluating searches, recall and precision. Recall is the proportion of retrieved documents to relevant documents. A search that missed a large number of relevant documents might have a recall of only 80%. Precision is the proportion of relevant documents to the total number of documents retrieved. A search that includes a large number of “garbage” hits would have a low precision.

Obviously, a searcher wants a high level of both recall and precision, but increasing one of the metrics tends to decrease the other. It does little good for a search to capture every relevant item if the results set has tens of thousands of entries. The key is to use a search engine that balances recall and precision in an appropriate manner, which means extensive use of Boolean logic, stemming, and user-definable proximity searching. Very sophisticated systems also rank the results using various vector space models, such as latent semantic analysis and Cluster Analysis, and inference models. There are numerous choices, in both free and subscription services, that provide these features.

      1. (6)Searching For EPO Applications

The EPO (and several other countries and regions) focus allowability decisions on whether the claimed invention recites a novel and non-obvious “technical effect”. The discussion is relevant here because whoever is searching patent claims that might be filed abroad needs to focus on searching technical effect.

What does that mean? Basically, the EPO uses a problem/solution approach to patentability. The examiner will try to identify the key problem(s) that the inventor is trying to overcome (which are hopefully set forth in the Specification), and will then search for the single document that is the most promising starting point for an obviousness determination leading to the claimed invention. If the claimed invention is a novel composition of steel, the examiner is most likely to focus on the alleged benefit(s) of the new composition, (e.g., greater rust resistance or perhaps greater tensile strength) and search for the closest prior art reference that addresses those characteristics. If the technical effect (benefit) is known in the same or neighboring field for the same component, then that reference is likely to be used in the obviousness analysis.

The bottom line for a searcher is that mere keyword searches for the components or steps of the claims is likely to miss the very references deemed most relevant by the EPO examiner.

    1. C)Free Patent Databases
  • Proximity Searching

All of the widely used Internet search engines, Google ™, Bing™, and Yahoo!™  support inherent proximity search using quotes around two or more search terms, (e.g., “black cat”). Unfortunately, while that is an effective good way of narrowing excessively broad searches, it also tends to render the search under-inclusive.

There are Web search engines that support explicit proximity searches.  Walhello allows users to define proximity by the number of characters between the keywords, while Exalead and Yandex define proximity by the number of words between keywords.

Figure 19 – So Many Choices For Searching

Perhaps the easiest explicit proximity searches use the NEAR and AROUND operators. Yahoo! supports an undocumented NEAR operator (keyword1 NEAR keyword2), and Bing goes one better by  permitting user definition of what is meant by NEAR.  The syntax is keyword1 near:n keyword2 where n is the number of maximum separating words. Google supports a similar operator, AROUND(#). The World International Patent Office (WIPO) search engine supports both explicit word distances, but uses a different syntax.  “electric car”~10 searches for documents where “electric” is within 10 words of “car”.

Google and Yahoo! both offer ordered searching using the asterisk character “*” as a wildcard. In Google the asterisk matches one or more words, while in Yahoo! each asterisk matches exactly one word.  WIPO supports both single character wildcard search use the “?” symbol, and multiple character wildcard search use the “*” symbol.

  • CAMBIA PatentLens

Probably the best free search service is that provided by CAMBIA™ PatentLens™. The system can search U.S. patents and published applications, EPO (European Patent Office) publications, the life sciences portion of WIPO publications, EPO patents, and Australian patents (beginning 2007), all in single search. The system provides ranking, automatic stemming, segment searching, (title, abstract, inventor, applicant, agent, references, description, and claims), and proximity searching using the “near/” connector. Searchers can retrieve documents in full text or as .pdf files. One of the very best features of PatentLens is that it highlights the search terms within the text files. This is extremely useful for quickly ruling in or ruling out the relevance of a particular document.

One of the drawbacks of PatentLens is that it can choke on complex searches, reporting “Wild Card Limit Exceeded”. For such searches, one probably has to use Delphion or one of the other paid subscription services.

  • Google Patents

Google™ Patents provides free text searching for all U.S. patents and published applications, as well as those for Canada, China, European Patent Office (EPO), Germany, Japan, and WIPO.

For the most part the service is good, although there are numerous text errors since the text apparently comes from machine optical character recognition of the USPTO images. One big advantage over many other free web sites is that Google Patents alls narrowing of searches using the field codes: priority date, assignee, inventor, patent office, language, filing status, citing patent, and Cooperative Patent Classification (CPC) codes.

There are several sites that provide tips and tricks for searching Google. One interesting (but so far not terribly useful) feature is Google’s Find Prior Art button, which is accessible from its Google Patents interface.

  • USPTO Website

The USPTO website is a fairly good place to start. The search engine is easy to use, with both the quick-search and the advanced-search interfaces. Unfortunately, the downloading of images on the USPTO website is awkward and unnecessarily time consuming, and patents from 1790 through 1975 are searchable on the USPTO site only by Patent Number and Current U.S. Classification). The website of the U.S. patent office is reasonably convenient for quick searches, and is very convenient for copy full text into a word processing document. One can view and download prosecution history documents through the Public PAIR site.

One of the best ways to use the USPTO database is to combine keyword searching with classification searching. Wikipedia has a decent summary of the Cooperative Patent Classification (CPC) that has been used since January 1, 2013. The patent office provides instructions, but the system is clumsy to use.

          • Espace (European Patent Office)

Another good free source for patent searching is the European patent office. The website includes access to US, European patents and applications, as well as PCT (WIPO) applications, but does not support proximity searching. The site also includes numerous links to Japanese and other government patent sources. One can view and download patents and patent applications through the main search site, and publications through the European Publication Server. To download patent prosecution history documents you need to use the Register.  Once you find the right application or patent, click on the “All Documents” tab in the left menu to see links to available documents, click the boxes designating the items you want to see, and then click on the “▼Selected documents” to download.

          • WIPO (World Intellectual Property Office)

The WIPO site has a very good search engine that supports proximity searching, wildcards, and so forth. The site is available in many different languages, and includes documents from all of the large patent offices. Also on the plus side, WIPO provides a natural language interface, and natural language expansion.

  • PatentInformatics (Sequence Searching)

PatentInformatics™ provides database sequence searching for nucleic and amino acids. The company also provides a service for producing genealogy trees of patents, and its own general patent searching database derived from USPTO, EPO and other public databases.

  • The Dark Web

Although most people don’t know it, the portion of the web you can search using Google and other common search engines is only a small fraction of the Internet. Google, for example, is said to have indexed more than 30 trillion web pages, but that is estimated to be much less than 1% of the total web pages available. To search the invisible web you need to use specialty search engines such as the following: Yippy, Intute, Infomine, Librarians’ Internet Index, or Direct Search. Historical web pages are available through the WayBack Machine.

    1. D)Subscription Services 
  • Lexis™ TotalPatent™

TotalPatent™ is probably the best pay service currently available, and the one we use the most at Fish & Tsang. TotalPatent can search the full text databases of more than 25 countries or regions, including machine translations of non-English filings, all in the same search. TotalPatent supports complicated Boolean logic, as well as semantic logic (where, for example, a search for “elephant” could pick up references containing “pachyderm” but not “elephant”). One huge advantage of Total Patent is that the system can show windows of text around the search terms, thus obviating the need to open individual documents to see the context. Another huge benefit is that one can download specific segments (patent number, title, date, priority date, claim language, etc) for a large number of patents. The downloaded information can then be searched and manipulated off-line using a word processing program. Still another benefit is that PDF files downloaded from TotalPatent are already optical character recognized. TotalPatent has a no-cost-added alert service.

  • Delphion (formerly the IBM Intellectual Property Network)

Delphion™  does all the basics, but is not so sophisticated as TotalPatent. Delphion can concurrently search U.S. Patents, European Patents, Patent abstracts of Japan, PCT Patents, and U.K. patents , can access World Patents Index (WPI) titles and other information, can secure patent family listings. Delphion can also graphically summarize large sets of data. Delphion does allow proximity searching, which can be done using the “near” connector. For example, one can find instances of “lipofuscin” within five words of “deanol” by typing “lipofuscin <near/5> deanol”. Delphion’s ranking program (results scoring) is excellent, and greatly facilitates searching.

  • Derwent WPI (General, Chemical & Sequence Searching)

Derwent™ World Patent Index™ is a collection of value-added databases. They have a team of 350 editors that assess, classify and index patents and applications from 41 patent-issuing authorities around the world. Not only is the database huge, (over 31 million patent documents in their databases), but every document is classified according to the Derwent classification system, and the subject matter has been re-abstracted into a standard format that rationalizes the often obscure language used by patent applicants. That method goes a long way towards minimizing differences in terminology employed by different patent drafters.

Other significant benefits are that WPI provides patent family listings, on-going search capability, and even statistics on patent issuances by subject matter, inventor, assignee, and so forth. The biggest drawback is high expense. The service charges a per-record charge for accessed data. There are several strategies for minimizing those charges, and the customer service people are extremely helpful in assisting users to achieve cost-effective searching. But the cost can still be quite high.

Derwent’s Merged Markush Service™ (also called Markush DARC) allows chemical structure searches from 1987 onwards. Note that STN and CAS (see below) provide chemical structure searching from the early 1900s. STN/Derwent’s GENESEQ provides sequence searching for nucleic and amino acids.

Derwent databases are available through Delphion, Dialog™, Orbit™ , STN™, and Westlaw™, and possibly other sources, but almost always at an added cost. Delphion does provide free the Derwent titles as part of its standard subscription service.

  • STN (General, Chemical & Sequence Searching)

STN™ relies upon databases of the Chemical Abstracts Society (CAS) to provide what is probably the most comprehensive chemical searching available. Use of the system, however, does require considerable training, and searches can run many thousands of dollars. Those who want the benefits of STN should run their searches through Science IP, which is run by CAS.

Another CAS option is STNeasy™ ( STNeasy focuses on chemistry and related fields, but accommodates keyword searches. The system charges on a transactional basis (about $2 per search), but viewing the titles of references is free, and viewing the abstracts is only about $4 – $7 each. Full-text articles cost as much as they would elsewhere, about $25 – $40 each.

Still another CAS option is SciFinder™ ( SciFinder is an extremely powerful tool that provides access to possibly the world’s largest collection of biochemical, chemical, chemical engineering, medical, and other related information. The search engine is first rate, there are no extra charges for viewing abstracts, and SciFinder includes marvelous graphical tools for summarizing large sets of references. The only real drawback is that one cannot “pay as you go”. Subscribers must purchase a block of “tasks” for several thousand dollars, and then use up those tasks within one year. Fortunately, one can access the SciFinder resources through SicenceIP. That service uses professional, full time researchers to run searches against all available databases.

As noted above, STN and Derwent collectively provide sequence searching for nucleic and amino acids through their GENESEQ databases.

  • SureChem™ (Chemical Searching)

SureChem™ allows users to find chemical structures in patents, journal articles and other text documents using chemical names (CAS, IUPAC, common, SMILES) to chemical structures. The names are automatically translated into chemical structures using thesauri, and then the chemical structures are used to search structure databases.

SureChem currently gives users access to USPTO full-text patent documents and MEDLINE journal abstracts. The website advises, that additional patent data, including European and Japanese patents, WO applications and New Zealand and Australian patents, will be added in coming months. There is also a version of SureChem (SureChem Enterprise™) that allows users to search their in-house databases.

  • NERAC (Patent Databases And Journals)

Searching full text journal articles, newspapers and other publications can be frightfully expensive. One of the most cost-effective ways to do so is NERAC.  For a fixed cost of (as low as) a few thousand dollars per year, NERAC provides unlimited access to tens of thousands of publications, as well as worldwide patents and published applications. Searches against the databases cannot be performed directly by NERAC customers, but instead are performed by professionals with particular experience in different technical fields. Thus, a pharmaceutical search may well be performed by someone with an advanced degree in microbiology or biochemistry. Moreover, the person doing the search is available by telephone or email to discuss search strategy, and to fine-tune subsequent searches. Turnaround is usually a day or so, although NERAC is very good at handling emergency searches. NERAC is also wonderful for on-going searches through their on-line TechTrak™ facility.

    1. E)Patent Alerts

The various pay services, (TotalPatent, Delphion, Westlaw, etc) all offer provisions for storing searches, and running them periodically. In most cases, Alerts are included in the monthly subscription fee, but do be careful. Some services charge per Alert.

There don’t seem to be any robust, truly free patent alert services, although PatentAlert™ does provide alerts in several specific fields of technology.

Another good choice is Automated PAIR Alerts from Cardinal IP. There is no long term subscription requirement, and the fee is only $10 per month per Alert.

    1. F)Professional Searching Services

Our office almost never relies on the results of outside searching services for run-of-the-mill prior art searches. For one thing, about 20% of the searches we run from Fish & Tsang result in a critical “hit” within about 10 minutes. In those instances it seems silly to charge the client $750 or more for using an outside service. In other instances, the search justifies the cost, but the very act of searching brings considerable insight to the searcher. That insight cannot be adequately transferred to the attorney for use in drafting the patent application. Indeed, much of the benefit of searching lies in gaining ideas about how the claims of a new application can be written to take advantage of loopholes in the prior art. Use of an outside searching service eliminates that benefit entirely.

We do use outside services to supplement our own searches, including for example, GreyB Services in India and Cardinal IP in the U.S..

There are several contingency fee searching services, including Research Wire. The typical plan is charge several thousand dollars if anticipatory / invalidating prior art is found for independent claims, with the charge reduced to a much smaller base fee if such art is not found.

For an inventor who is drafting and filing his own patent application, perhaps the best thing to run his own preliminary searches, and then if the invention still seems patentable, take the invention to a patent attorney or agent for further searching. Just remember, that if the attorney saves you money by finding critical prior art, you still need to pay for the searching.

    1. G)Don’t Allow Searching Costs To Get Out Of Hand

Many patent firms and independent searching services charge a standard, fixed fee (usually about $750 – $1500) for patentability searching. That makes sense from a marketing standpoint, but a one-size-fits-all standard fee doesn’t work well across completely different technologies. Some inventions are simple to search and others are notoriously expensive to search. A $750 fee, for example, overcharges for the simple searches and provides too little funding to adequately run complicated searches. In some instances Fish & Tsang doesn’t charge anything for searching, because we find invalidating prior art while we’re on the phone with a potential new client.

At the high end, patentability searching can run many thousands of dollars because of the databases used. For example, chemical inventions often require searching of chemical structures (as opposed to merely chemical names). There are only a few such databases available, and they can charge a small fortune. Searches performed according to the USPTO patent search templates require applicants to search non-patent literature, which is often available only through Dialog.

It is counterintuitive, but patentability searches can also be expensive for simple inventions directed to common items. Novel baby bottle designs, improvements to doorknob latches, new pens and pencils, and other very common consumer items can be notoriously difficult to search. Such inventions usually have an enormous amount of prior art, and it is extremely time consuming to sift through that much material. Simple inventions are also usually described with common language, so that a word search on a typical database can yield tens of thousands of hits.

Still further, simple inventions can be difficult to search because the relevant products are never described in patents or applications. A good search requires searching of non-patent literature, including for example newspaper and journal articles, and advertising materials.

One important caveat is that a searcher should use multiple sources for any given search. Limiting one’s search to issued patents is especially dangerous. Published patent applications, whether in the US, EPO, Japan, or elsewhere, can all be used as prior art to preclude patentability. Journals and other non-patent publications can also be used against a patent application. For simple consumer products, we often search catalogs on the Internet. Often an “invention” is already in the Hammacher Schlemmer™ catalog!

    1. H)U.S. And International Classification Systems 

Searching by technology area can be a double-edged sword. On the one hand, such searches can be extremely helpful in gaining a general background on the technology in a field, and in limiting on-going alert searches. On the other hand, the fact that any given technology usually spans many classes and subclasses can be extremely misleading.

Admittedly, the USPTO has made a valiant effort to simplify classification, and render it accessible to ordinary users. The best place to start is the USPTO classifications web page, which provides an overview of the system as well as detailed instructions. If one wants to mimic what the patent examiner is likely to search, one should follow the patent office templates.

The classification system is getting better all the time. On January 2, 2013 the patent office announced the formal launch of the Cooperative Patent Classification (CPC) system, a global classification system for patent documents. According to the press release, “CPC is the product of a joint partnership between the USPTO and the EPO to develop a common, internationally compatible classification system for technical documents used in the patent granting process that incorporates the best classification practices from both offices. It will be used by the USPTO and more than 45 patent offices – a user community totaling more than 20,000 patent examiners – all sharing the same classifications helping to establish the CPC as an international standard.”