What are Singularitarians and the Singularity?
There are several definitions out there on the web but they are not really general enough or concise.
A Singularitarian is a person that believes that the increasing rate of technological progress is an exponential curve. That at some point as time progresses humanity will change substantially. It is also a horribly awkward term. Let’s just call them Sings – as if to say when humanity Sings.
There are lots of different types of Sings. On this blog you see what could be called the parents of Sings.
And while there are lots of people that speak about the Singularity very few talk about the mechanics of how the Singularity might come in to existence. Ray Kurzweil has even put a date on when the Singularity will occur – 2045.
As a project manager I believe in data based decision making. Based on the information that I can come up with I think 2045 is a bit early and I have ideas as to one of the mechanisms for the Singularity to occur.
First, let’s have a look at a simple graph and a small dataset:
Linear_Vs_Curve_Growth (excel file with dataset)
The dataset contains two observations or columns. One increases linearly 1,2,3….23, and the other multiplies by 2 starting at two.
The curved line closely represents Moore’s Law – which is really an observation not a natural law like gravity, which basically states that computing power (or the number of transistors in a CPU) doubles every 2 years. Some people quote 18 months as the time to double processing power in CPUs.
The years on the graph and dataset are arbitrary just to illustrate a timeline and computing power increases.
Later in the article it discusses the declining costs of electronics and computing on an annual basis. That decline in cost was 16% per year for many decades. Then there were fluctuations. For simplicity, I will use 16% per year as the declining costs of computing systems. This could introduce errors in the hypothesis later; however, in the future I will work to refine the calculations and use statistical methods to try to predict the price decreases per year.
People seem to buy PCs on three pricing scales for their home use. Low cost – $500, Medium, $1000, and expensive $2500. There are gamers out there that spend as much as $5000 or more for a PC, but for now we will only consider the mainstream groups.
In addition, there are households such as mine that have 6 computers – from different pricing ranges. Computing prices in the future may not be based on individual PCs, but on computing power.
If we consider the amount of money a family is willing to spend on computing power to be in those three ranges (typically) we can use it as a kind of three-point estimate like as in Project Management Cost estimate technique so the low cost PC would be the Optimistic value, the medium cost PC would be the Most likely Value and the expensive home PCs would be the Pessimistic Value.
Watson is an amazing piece of computing hardware and software. If you haven’t seen the Jeopardy episode where Watson beats two Jeopardy champions you should definitely watch it. It is kind of a horse vs. car competition. The car (Watson) beat the pants off the horse (two human jeopardy champions).
We could even go on and use a PERT style estimate and weight one value over another, but we won’t get in to that here. Just keep it in mind.
Watson costed about $3 million dollars to create (hardware). If the costs of Watson decline at 16% per year, there should be a point in time where the inflation increased costs of home computers is intersected by Watson’s declining costs. Imagine, a computer that can answer every question correctly in every middle-class living room (assuming there still is a middle-class). The rich will of course get the technology sooner than the middle-class. Eventually, the technology would become ubiquitous and everyone would have it and think nothing of it – much the way smart phones are today.
If we take Ray Kurzeil’s estimate that we will have a computer that can simulate the human brain accurately by 2030 – we can take the cost of that hardware and also reduce its cost by 16% per year and find when an artificial brain computer could be in every middle-class home.
End of Part 2