One of the core concepts behind the Technological Singularity is that technology matures at an exponential rate . (The famous example is Moore's Law ) The basic assertion is that as we go forward through time we observe that all sorts of technological measurements (e.g. computing speeds, bytes transferred on the internet, number of words written by humans world-wide) are exponentially increasing. The x-axis is time. The y-axis is some measurement of technology.
It follows, that the amount of time required to for an innovation to go from novelty to commodity compresses as you move into the future. That is the thesis of this article.
Accordingly, research is being performed to verify this claim. Assuming the research results in a validation of the exponential doubling effect, then the thesis in this posting should follow as valid.
The conventional way to think about innovation goes something like this: It takes a long time. It can be very expensive. If it took 10 years to develop a product by a competent inventor, barring anything extraordinary, it would probably takes a similarly competent inventor 10 years as well. Under the singularity concept, a great idea kept perfectly secret might take 10 years now. As time moves on, a similar innovation might only take 5 years. As time moves on, a similar innovation again might only take 1 year. Then only months. Then only weeks and so on. At some point certain conventions (like patents stop making sense)
Get into any sufficiently long discussion on patents and one of these two conventional examples might come up
We need the patent system to protect the ideas made by a small operators (Dr Bob working in his garage for 10 years). If we don't protect Dr. Bob's idea, then not as many Dr. Bob's will innovate. That's bad.
We need the patent system to protect big pharmacology companies because drug approvals take forever.
These two examples represent the conventional wisdom on the subject of "Why do we need patents?"
In the case of software, an idea, once implemented in software and distributed could be freely copied or used by anyone. Software can often-times be reverse engineered (especially when using scripting languages) the great ideas understood and then reimplemented in another language or with a different technology stack.
The corollary to the Singularity is that as we approach it, the time to create and disseminate new ideas, compresses -- asymptotically.
In the example of the small operator (Dr. Bob), the argument is about protecting someone spending 10 years in the garage. If the corollary is true, then as we approach the Singularity, less and less time is needed to innovate.
Anecdotally, I have experienced this myself; the time-distance between inception and development is very short. If something is possible, then it's been implemented by someone, somewhere, already. Most things (in software) appear to be incremental steps.
The operative word here is "investment" -- not "idea", not "invention", not "process". The patent concept makes sense when we are protecting investments, especially time investments. In the example of a drug, it demonstrably takes a long time. In the case of an incremental improvement to software concepts, it takes very little time (anecdotally). What are software patents protecting? Ideas maybe?
The patent concept works - that is to say it supports invention - when it protects major time and resource investments. But, do we actually need 10 years or more to innovate in software? Do we really need to invest vast sums of cash to innovate?
Think about that. A closed-door software development team, working for years, in secret, and then releases something new. I'm having trouble thinking of examples of that. Can you?
The Google Brain maybe? Not a good example. They don't distribute it. It's used internally only. It's protected not by patents, but by non-disclosure agreements.
In the software world, people invent things constantly. Usually, we think about software innovations enabling something that couldn't be done before, such as full immersion virtual reality. However, some innovations decrease the complexity of a process. Thoses processes shorten someone else's attempt at innovation, which in turn enables more innvoation. As time progresses, process innovations have an exponential effect on all technology.
The singularity concept implies that as we approach it, the time to create and disseminate innovation, compresses asymptotically.
What is the use of a patent in such a reality? What is the use of any notion of intellectual property if the ideas themselves are constantly getting easier to invent?
In areas where innovation takes years or mega-millions to create, patents make sense. In software, things don't take that long. If the Singularity is true, then we may not be protecting the right things with our software patent system. If the Singularity is true, what are software patents protecting?