Imagine you're sitting in your research lab when you come across a really sweet art generating AI tool. To use the tool, you simply have to upload a bunch of images that look vaguely like something you would want to make, and press a button. The tool reads all the images and generates a new piece of art. It's absolutely breathtaking. It's so beautiful, that you look into creating a patent for the created art so that no one replicates it and profits from what you created.
However, you run into an issue at the patent office. Who really owns that piece of art? Can you really claim ownership after just uploading a bunch of images you didn't create and pressing the button? Can the author of the tool really claim ownership for simply just making the tool but not any subsequent piece of art? Or even more bizarrely, can the AI itself claim ownership over the piece of art?
This story is not a hypothetical. Technology, in particular machine learning, is evolving at a pace that will severely stress test the patent laws we have in place. These laws are predicated on the idea of intellectual property - that if you came up with an idea you get to claim ownership over the idea. But technology is beginning to expose the question: can you really claim ownership over an idea? If the answer is no, what would a world that got rid of patents start to look like?
The Oxymoron of Intellectual Property
Intellectual Property is legally defined as any product of the human intellect that the law protects from unauthorized use by others. This traditionally includes creative work like articles, music, logos, movie characters, etc. In the last half century, software has entered this definition, leading to all sorts of ridiculous patents like Amazon's one click checkout. Breaking down exactly what makes this patent so ridiculous is the realization that "Intellectual Property" as an idea is logically inconsistent.
Property itself when it was first defined in the Adam Smith days of early capitalism all shared one important characteristic: if I own it then you don't. This could apply to food, household items, and even the servers that software runs on. In all these cases, one person using the property physically prevented the use of someone else from using it during the time of use.
To consider software intellectual "property" is to artificially impose the same physical barrier onto ideas. It operates under the premise that if you get a hold of my idea, then I somehow no longer have that idea. But, if I teach you a new song, your newfound knowledge of the song does not prevent me from singing it. In the same exact vein, another retailer implementing one click checkout does not directly impede Amazon's implementation of the same software.. This characteristic of property does not apply to ideas.
To also claim sole ownership over the idea is also to say that no one else contributed to the creation of the invention. But every idea in practice is a remix of everything the creator has learned to that point. We are heavily influenced by what we learn from each other, to the point where it's almost impossible to trace back the origin of where the idea truly came from. Amazon did not invent clicking, checkout, or ease of use, they were just one of the first to remix them together.
So if intellectual property is an idea that does not make intuitive sense, why does it exist?
History of Patents
The origin of patents comes from incentives.
Think about what it takes to come up with a new drug. Not only is the medical field really complex in its own right, but testing new drugs is an extremely expensive process. One needs to conduct months of research, satisfy several regulation requirements, acquire approval from the FDA, and then run several clinical trials with the new drug formulas that could span several months, even years. This creates an extremely high cost of creation. Without patents, the cost of replication is really low. As soon as another competitor could see what the formula is for your new drug, they could bypass all of those costs of creation that you experienced and start producing, possibly out-competing you. Why would you as an inventor want to invest the resources to create the drug in the first place?
To incentivize creators to invest time in new ideas, whether they are researchers in the medical field or engineers in software, patents were created to artificially introduce a high cost of replication. With a cost of replication that's higher than the cost of creation, individual actors are now incentivized to innovate and create new ideas as opposed to stealing those of their competitors. This is why it still makes some sense to keep patents in fields where the cost of creation is still high, such as in medicine.
But in software? It is absurdly antiquated.
Cost of creation in software has gotten extremely low. We could use the drug creation process for comparison to illustrate this. Several asynchronous communication services and user analytics have been developed to increase the amount of user research available. Satisfying regulation requirements has become as easy as downloading existing open source libraries. Acquiring approval from a higher authority has been removed in favor of peer to peer code reviews which have a faster developer cycle. Finally, running trials testing your software happens on the order of milliseconds with several frameworks like jest and junit making the process easier. Because the cost of creation is already as low if not lower than the cost of replication, patents in software should not apply in the same way it originated in other fields.
What Could We Replace Patents With?
Can we take this a step further to argue for the removal of patents altogether? Under the framework that they are an artificial cost to replication to incentivize creation, the question could be rephrased. Can we incentivize creation without adding an artificial cost to replication?
Part of why it's hard to imagine how this is possible in the current world of patents is because we reward creators for their work after the work is done. This leaves them vulnerable to replication, since if you wait until after the work is fully complete to reap the benefits someone else could just copy the code and reap it from you. This means to incentivize creation in a world with low cost of replication, we need to create ways to reward creators during the process of work.
You can see the world already trending in this direction. Sponsorship models through Github and Patreon allow creators to build an audience of sponsors to fund them to continue creating work, not transacting on the work itself.
"Building in public" is also an idea that is gaining popularity, where sharing what you are learning is providing value to others, regardless of whether you achieved the desired end result or not. This is typically done through people sharing and selling their personal notes, newsletters, articles, and courses. This is inherently valuable since we could learn from each other's mistakes and progress regardless if the software works or if the drug relieves the targeted symptom.
Attribution is another pattern that's growing, where for any given piece of work there is often a link or git history referencing who was the original author of the work. This way even if there is replication, you as a consumer will have the context of where it originally came from and allocate a proportion of your value back to the relevant creator.
All of these patterns have a consistent theme. The more visibility we gain into other people’s work in progress, the more value we are able to benefit from it. Systems are being built right now as a result to capture some of that value for the creator. There’s a concern that this could result in a market of existing ideas gaining momentum at the expense of the market of new ideas. But again, all new ideas are simply a remix of existing ideas, making both intellectual property and patents in general unfit models for incentivizing creation.
To achieve a world without patents, we need to build towards a world where we reward the process of creating rather than the end result of the creation.