The Unexamined Invention is Not Worth Inventing

A Socratic Approach to Getting a Patent

Socrates is the founder of Western philosophy.  His method of questioning and argumentative dialogue laid the foundation for critical thinking, forming the basis for the Scientific Revolution, the Age of Enlightenment, and inevitably, the Industrial Revolution.  We have Socrates to thank for the advances in how modern humans think and perceive the world and for the logical and technological tools that make our lives easier, more efficient, comprehensible, and solvable.

 Socrates questioned those in society that were assumed to be wise, such as the esteemed poets, politicians, and skilled workers of Athens.  For Socrates, his questioning was not only a way of testing our understanding, but to find a common truth and understanding in a collaborative fashion.  He did this by breaking down commonly held beliefs into logical elements, called presuppositions.  By examining the logical links within a conclusion individually and holistically, a belief could be verified, modified, or completely rejected.

If you ask the average patent attorney about the Socratic method, you might hear a loud groan followed by an anecdote of a traumatic law school experience, likely a first year cold calling episode when the attorney was repetitively questioned in the style of our beloved Greek philosopher.  But in practice, that same attorney likely uses the Socratic Method every day to help clients, whether its formulating questions to better understand an invention or arguing the invention’s patentability.

When an inventor approaches a patent attorney with, “I have an invention,” the typical response is, “what is it?”  The question, “what is it?” was Socrates’ favorite question.  Socrates understood definitions to be fundamental to the pursuit of knowledge and understanding.  Two people cannot examine the validity of a statement until they both agree on the meaning of each word in the statement.  If you’ve ever debated someone on an incredibly nuanced topic, then you’ve likely danced in circles to the point of someone shouting, “that’s just semantics!”  In essence, the two of you have reached an impasse regarding the very meaning of the words being used, and thus, cannot reach a shared truth.

Patent attorneys also care a lot about definitions.  Like the validity of a statement, the validity of a patent often turns on the meaning of an individual word.  Understanding an invention is all about defining the invention, and furthermore, defining the words used to describe the invention.  Let’s use an example. An inventor invents a fraud detection system and calls up his patent attorney:

            Inventor: My invention helps businesses identify fraud.  We sell our payment processing software to stores so they can securely accept digital payments.  When a store processes a transaction, the software identifies other transactions that relate to it and links them together to form a chain.  We then compare the chain to a fraud list to determine if the current transaction is fraudulent.

Now, the attorney will think about what was just explained and maybe the first instinct might be to ask, “how?”  But, according to the Socratic method, before you can move onto any other question, you must ask, “what?”   Namely, the Socratic method is based on deduction, which is a top-down approach starting from the general to the specific.  Being an invention about fraud detection, here the first question might be, “what is fraud in this context?”  It's likely that if the attorney has worked with this client before, he likely already knows the answer to this question.  However, let’s assume this is the first meeting with this client and the dialogue goes like this:

            Attorney:  This invention sounds interesting. So, what exactly is fraud in this context? Is it when a person uses a credit card that’s stolen?

            Inventor: It can be.  In our industry, fraud detection can relate to detecting any type of behavior that is suspicious or illegally conducted for monetary gain, whether it’s by using fake or stolen payment information to buy things or by using falsified identities to hide the fraudster’s tracks.

            Attorney: I see.  So then this “fraud list” would basically be a list of data that is associated with negative behavior.  We might call it a “negative list” in that case?

            Inventor: Yes, that’s right.  And that data associated with negative behavior are what we call “identities.”  Things like an IP address, credit card number, name, billing address, email.  Anything that can be used to identify someone.  The suspicious or “negative” behavior, as you say, could be related to how many times a credit card was used over a short period.  If someone uses one card to conduct dozens of transactions in one day, we’ll likely put all the identifying information for those transactions on a negative list. 

            Attorney: So, is it as simple as reading off the negative list when a new transaction comes in?  What is the “chain” that you mentioned before?

Inventor: Fraudsters typically aren’t that dumb and won’t use just one card several times in a short period.  They might be smart enough to ship merchandise to different addresses and find a way to hide their IP address.  They’ll probably even use different emails for each site they buy from too.  That’s where the chaining comes in.  Even if they’re mixing and matching identifying information in each transaction, we can chain individual transactions together when they have common data and make an inference that it is likely the same person, even if the link from one piece of data to another is many hops away.

Attorney: In that case, can we define “chaining” to mean continuously linking together all transactions that have common identities?

Inventor: Yes, it’s continuous and its dynamic.  The chain morphs in real-time as more and more transactions come in. And these chains are mapping transactions throughout the payment network from several different merchants using our system.   This is probably on the order of thousands of transactions per second.  That’s what makes this so much better than anything else out there.

Attorney: And then, which of those thousands of transactions are considered fraudulent?

Inventor: Any transaction linked to another transaction in the chain that matches the negative list.  That link provides an inference that it’s the same fraudster behind both transactions, and we use that inference to alert the merchants where that fraudster has transacted.

Here, we can see the goal and aims of the Socratic method taking place.  By asking seemingly dumb questions like “what is the fraud” and “what is a chain,” both the questioner and the person questioned uncover a shared truth and both form a better understanding of what the invention really is and what it’s not.  Here, we might define the invention as follows:

            The invention is a computer program that collects transaction data from a plurality of merchant computers and dynamically links several of the transactions into a chain based on shared identities within the collected transaction data.  When a new transaction is received, the computer program updates the chain in real-time and compares the chain to a negative list.  If any transaction contains an identity marked on the negative list, all transactions linked to that transaction in the chain are flagged as fraudulent and the merchant computers where the transactions were made are alerted.

This definition of the invention is our “patent claim,” which will be examined by the patent office.  Up until now, all we’ve talked about are the meaning of words and the formulation of the words into an acceptable definition.  To determine patentability, we must test if our patent claim meets two main requirements.

The invention must be:

1) novel; and 2) nonobvious.

An invention is novel if it was not previously known to the public.  Simply put, your invention is novel if it brings something new to the table.  In practice, this means that any public information that describes that new thing should not predate your patent application.  If a patent examiner concludes an invention lacks novelty, it can be rejected by the patent office.  Using the Socratic method, we can break down this conclusion into the following presuppositions:

1)    The inventor defines the invention as X in his patent application.

2)    Item 1 describes X.

3)    Item 1 is older than the patent application.

            =>Therefore, the invention lacks novelty.

As you can see, there are three presuppositions to examine here, meaning three opportunities to disprove that the invention lacks novelty.  We can disprove presupposition 1 by showing that the invention is not defined as X but instead something else.  For example, we can clarify our definition of the claim through argument and/or adjusting our patent claim.   We can disprove presupposition 2 by uncovering that item 1 does not actually describe X.  Maybe the patent examiner misunderstands the documents describing item 1, for example.  Or, we can disprove presupposition 3 by showing that item 1 came after the invention, essentially indicating a clerical error.

Regarding obviousness, whether an invention is obvious depends on the existing knowledge and skills of those working in the field of the invention.  This is examined through the eyes of a hypothetical person having ordinary skill in that field.  In practice, this means no combination of public information predating the invention should render the invention obvious in the eyes of that hypothetical person.  Here are what the presuppositions for an obviousness conclusion may be:

1)    The inventor defines the invention as X in his patent application.

2)    The combination of Y and Z makes X.

3)    Publication 1 describes Y.

4)    Publication 2 describes Z.

5)    Both publication 1 and publication 2 existed before the invention.

6)    Prior to the invention, a person having ordinary skill in the field of X, considering publication 1 and publication 2, would understand to combine Y and Z to make X.

=>Therefore, the invention is obvious.

Again, we can examine presupposition 1 by clarifying whether the invention is truly defined as X.  Additionally, we can also examine whether the combination of Y and Z truly makes X.  We can examine whether publication 1 truly describes Y.  We can examine whether publication 2 truly describes Z.  Or we can examine whether both publication 1 and 2 existed before the invention.

Lastly, presupposition 6 is complicated, because it requires you to stand in the shoes of a hypothetical third person.  The ordinary level of knowledge and skill of scientists and engineers at the time of the invention might be hard to uncover through deduction alone.  One way to examine presupposition 6 is to uncover whether publication 1 or 2 cautions against combining Y and Z.  This shows that our hypothetical person, having obtained the knowledge from publication 1 and 2, would not understand to combine Y and Z to make X because the person would have been led away from doing so. 

Let’s walk through novelty and obviousness using a simulated dialogue.  At the patent office, a patent examiner determines an invention is unpatentable.  The inventor calls up the examiner and starts off this dialogue:

            Attorney: Hi, examiner.  I’m here to talk about this fraud detection invention.  Just as a refresher, the invention is a software program that alerts merchants when a transaction is linked to fraudulent identities.  The program dynamically links several transactions into a chain based on shared identities within the transaction data collected from several merchant computers and then compares the chain in real-time to a negative list with each update of the chain.  When a transaction contains an identity on the negative list, any transactions linked to the transaction in the chain are alerted as fraudulent.   I’m wondering where you see that being described elsewhere.

            Examiner: Thanks for that explanation.  Yes, this other patent application that was filed three years before your client’s patent application describes alerting a store anytime someone tries to use credit card information that’s been red flagged on a list of stolen credit cards.  The list of stolen credit card information is continuously updated based on data collected from several merchants, and each transaction is compared to the list as the transaction occurs.

Now, at this point, the average person at first glance will hear the description of the other invention and think, “yeah, I can see how that’s really similar.”  Usually this is due to mental shortcuts that we use in our everyday lives, namely inductive reasoning.  Inductive reasoning, in contrast with the Socratic method which uses deductive reasoning, means logically flowing from the specific to the general rather than from general to specific.  We do this every day when we stereotype, generalize, or establish a pattern from only a few observations of something. 

“That movie was bad, so all movies by that director must be bad.  I’ll skip the next one.” “I had a bad time last year, so I’m not going this year.” “That French person was rude, so I guess all French people are rude.”  Any of these statements are all forms of inductive reasoning, and we can see clearly why they might be based on faulty logic. 

When we look at the explanation of the other patent application in the dialogue above, we might think to ourselves, “our invention is about catching fraud through comparisons of new transaction data to negative data, like stolen credit card information, so all inventions that compare new transactions to stolen credit card information must be the same as our invention.”  But remember our patent claim defining the invention.  And, remember the meaning of the words contained within our patent claim, which we established in our dialogue with the inventor.  We might respond to the patent examiner as follows:

Attorney: I see some similarities between the inventions, but those similarities seem a bit superficial.  Can you help me understand where there is chaining occurring in the older invention?

Examiner: Well, we have this list of stolen credit information that’s continuously updated as new transactions come in.  Is that not chaining?

Attorney: I wouldn’t say so.  My client’s invention does incorporate a list of negative data, including stolen credit card information like credit card numbers, names, shipping addresses, IP addresses, basically any sort of identifying information associated with a transaction.  In other words, those are “identities.”  But that’s not the chaining part.  What did you understand chaining to mean?

Examiner: Okay, I think a chain might just be a collection of transaction data then.

Attorney: You’re correct in that the chain includes that’s updated using collected transaction data, but what’s being updated are the links between identities in the chain.  In the chain, identities are linked when they are contained in the same transaction data.  For example, if two transactions involve the same credit card information but with two separate IP addresses, we can link the two IP addresses in a chain.  You see that this chaining is more than just matching credit card numbers, right?

Examiner: Maybe.  Credit card information is just a type of identity though.

Attorney: Yes, let’s stick with the credit card example.  Does the previous invention link different credit cards together?  We do this in our invention to catch fraudsters that are clever enough not to use the same information every time.

Examiner: I guess not.  They only compare credit card information in a new transaction to the list of stolen credit cards to see if there’s an exact match.  I still think the invention might be obvious though.

Attorney: Why’s that?

Examiner: Well, even though that older invention doesn’t describe chaining, I found this other old reference that describes connecting transactions together in a “web.”

And now, you might be thinking, “well, we had them in the first half, but they got us now.”  A web sure does sound like a chain. But we’re not falling back on mental shortcuts now.  Remember our presuppositions for obviousness?  We’re bound to expose faulty logic in one of them.  Continuing our dialogue:

            Attorney: Can you explain this “web” a little bit more?  How do they describe it in the reference that you found?

            Examiner: Well, there’s a computer program that allows data analysts to view an entire transaction network by generating a visualization of a web.  Each dot in the web indicates a credit card number and each of those large circles in the web indicate a store location. Each connection between the dots and the circles indicates a transaction.  The data analyst can mark suspicious behavior when they see a pattern in the web that is irregular, like a dot connecting to several different locations that are far away from each other.

            Attorney:  So, are you establishing that chaining is obvious when you consider the combination of a list of stolen credit cards with a web of interconnected transactions?  Where are the “shared identities” in this web?

            Examiner: In the web, the credit cards are identities and so are the store locations.  So here, like in your client’s invention, we have identities connecting to each other in a chain.

            Attorney: In my client’s invention, a transaction not marked fraudulent is then marked as fraudulent if linked to a fraudulent transaction.  Did I understand you correctly when you said the links in the "web” are the transactions themselves?  I’m having a tough time understanding how transactions are linked in this additional reference that you’ve found.

            Examiner: Yes, the connections in the web are the transactions linking credit card information to locations.  It’s not a complete 1:1 comparison, but I think considering the other patent application in tandem, this makes your client’s invention look obvious.

Here, the examiner is conceding a little bit, but still has enough confidence to hold his ground.  You’ve exposed some flaws in his logic, but he jumped straight to his conclusion anyways.  We’ve already established that the first reference he found doesn’t teach X and instead teaches Y.  He then showed you a second reference that teaches Z.  He wants to show that X is obvious because Y and Z make X.  Question him again so that he revisits his thinking. 

            Attorney: In the first reference that you showed me, the invention involved detecting fraud by comparing a credit card numbers with a stolen credit card list for a match.  Now, this second reference you showed me involves catching fraud by looking at suspicious behavior in a visualized map of credit card information being used at different locations.  What about catching a fraudster that uses different credit card information at the same location?

            Examiner: I guess they don’t provide a solution for that.

            Attorney: My client’s invention does.  By chaining identities together it essentially identifies a fraudster masking his identity when using different credit card information.  We can link multiple transactions to the same fraudster by linking all his different identities together.  I don’t think the credit card list and the web from these references allude to anything like that.  I don’t think someone reading them would come away with that idea either.  It appears to be highly unconventional; don’t you agree?

            Examiner: It does seem to go about it in a very different way.  Thanks for clarifying that, all these fraud detection cases look the same at first until you really dig into them.

So, there you have it.  You’ve successfully deduced your way to patent.  You can thank Socrates.  There are additionally two other patentability requirements, which the Socratic method can also be used for.  Each of those topics – sufficient written description and eligible subject matter – we will save for another day.  They could each warrant their own separate dialogues.  At the very least, you now know enough about the Socratic method and about patentability to cause some damage at the patent office.  Good luck!