How Technology Evolves
Source: http://difficultday.com/
In my previous article I discuss the fact that technology does evolve. However, that discussion was limited to the general heuristics of evolution and technology. I did not include any human agency into the article. Effectively, I was treating these events as solely within the realm of technology where humans did not guide the process at all. In this article I will discuss how people actually interact with the technology to create selection environments.
Adding agency into this discussion changes things in two significant ways. First we must look at the methods in which technologies are modified and we must look at how technologies are adopted. Each of these represents a type of selection method. I will argue that the first selection environment is closer to artificial selection and the second selection environment is closer to natural selection.
What is the first selection environment? Well it’s the lab, the garage or wherever the technology is being improved or invented. Let’s assume it’s an R&D lab at some company in this case. Generally, when researchers are looking to improve a technology they have an idea of what sort of results they will have when they change a specific component or material. However, they do not know how these will interact with other pieces of the technology. If you are working on a battery there are many different chemicals that react with each other and react with different intensities. If you change the chemical composition of the battery you will have different reaction rates and thus different battery life.
The researcher knows that there are some general requirements for the technology they are developing. Let’s say size, weight and battery life. Let’s also say that in this battery there are only three components that they can change, and it’s either in or out (1 or 0). In this case, since we’re changing an existing battery, these are the only three factors that will impact size, weight and battery life. From these requirements we could create a number that implies the general fitness of these changes to a general optimal.
So, if we change one part at a time we’ll have a different fitness value for the new battery. After each change we can see if the part is better or worse than it was before and then we can reject or keep the change. There could be a few different ways to actually find the optimal fitness for the battery, lowest weight smallest and longest battery life, that could be by testing every combination possible. For 3 different components ( 2^3 = 8 ) testing every combination might not be too expensive. However, if you have 5 (32) or 10 (1024) the costs quickly go out of control. So, researchers typically make their choices based off of changing one component or several at a time using an experimental design. However, once they find a local optimal compared to the starting point, they will use this new design as their starting point and look for a new local optimal (Further reading). This creates a type of path dependency where the random start changes will have a long lasting impact on the design of the technology.
Source: Frenken, 2006
This type of path dependency can eventually lead to costly re-engineering as well. One historic example of this is the adoption of retractable landing gear in planes. One company simply covered them with “pants” others created retractable landing gear. It actually took ten years to determine which was the better solution as planes got faster and the difference in drag finally forced the company to shift to retractable landing gear.
Using this type of search heuristic it’s difficult to guarantee discovery of the global optimal, but it will quickly find an optimal that is good enough for the product to go to market (See picture of fitness cube number in parenthesis is the “fitness value” arrows indicate search direction starting at 000). This is what Edison did with his invention of the telephone. He was able to afford a huge range of search costs through his lab to test a wide range of materials. Eventually, he had to make a move to get a patent and went with something that was extremely good, but may have not been the global optimal.
The second selection environment is the market. Once the technology is out of the lab and into the wild of the market things behave differently. For instance the value of the telephone wasn’t obvious immediately because there were other network effects involved. If you were one of the early adopters of the telephone you may only be able to talk to one or two people. However, if you bought a telephone two or three years after they were released on the market, it would be more valuable to you than the early adopter because a network of users already existed. Because of the additional effects of other technologies, in this case the telephone connections to people you know, the value of a technology could be more or less depending on when you purchased it and who you know that has also purchased it.
In this way, technology operates within something of a biosphere of technology. Another great example of this is the adoption of the jet engine from the propeller. Airports had to be completely redesigned around the jet engine. The run ways had to be longer, made of stronger materials (heavier planes), the hangers had to be larger (bigger planes) and airtraffic control needed better radars to plot better flight paths because of the higher speeds. Because of this, even though the product was superior to propeller airplanes they jet planes weren’t adopted right away because of airport network requirements and other technologies that needed to advance to keep up with the jet.
The selection of many technologies is much more complicated than simply walking into a hardware store and picking up a hammer. If you’re just going to be driving big nails into thick boards, sure get a claw hammer. However if you’re going to be doing sculpting or finishing work you’ll want a different hammer. You’ll end up doing research and based on that you may buy a specific type of hammer or brand based on recommendations. According to engineer and economist Brian Arthur the final selection of one product over the other can’t be predicted ahead of time. Eventually one technology will win and the users will be locked into that technology.
A great example of the end result of this lock in and selection process is the difference in the usage of communication technology around the world. The riots in the UK were coordinated through Blackberry messenger (BBM), the revolt in Egypt caused by Facebook / Twitter and the Occupy Wall Street was started on several different websites including 4Chan, Reddit, Facebook, Twitter and other messaging services. BBM isn’t popular in the US at all, so a similar type of coordinating method would fail there. There were specific reasons why BBM is popular in the UK while not in the US. Most likely because of Apple and Android, which dominate the smart phone market there.
For technological evolution there are two types of selection environments at play. The first models more closely to artificial selection because some one is designing the technology to meet specific specifications. The second has more random chance and network effects that drive the adoption of technology. Each of these selection environment impacts the next generation of technology. The selection at the market level changes the requirements for the engineers and the improvements by the engineers and the previous market generation impact expectations for the product coming to market.
Edison invented the electric light not the telephone that was Bell.
yes, and Edison also received a patent on the telephone. There was a huge lawsuit over the patent. Bell claimed that Edison was infringing on Bell's patent. In the book I mentioned the chapter about Edison's drawings are based on drawings that were used in the court case. Edison was one of the first research labs in the US. He invented thousands of things including a type of phone, the light bulb, DC power and the phonograph.
[...] We, our government and our corporations need to understand that there is an incredible sci-tech research and development resource just waiting to be unfettered and financially supported. We must [...]
[...] my last two articles [Does Technology Evolve], [How does Tech Evolve?] I’ve painted a very rosy picture of technological change. I describe a picture where [...]
[...] will learn is becoming ever more apparent, with the birth of new technologies, ideas and innovation. Whether it is my school, a state school near London, or groundbreaking, modern schools around the [...]
[...] Technology is the root of this change. In a situation where salaries are completely elastic, meaning they will change as soon as there is a need to, it maybe possible to keep many more jobs. However, in cases where they are not elastic, mainly the real world, these jobs will simply disappear as workers will not do the same work over time for less and less money. Despite the lack of elasticity worker wages have declined over the past 30 years in the US. Automation is driving this downward trend and as we have seen in the previous chapter is expected to continue increasing. Eventually, the salary will fall to the point where a worker can’t pay to survive. When this is the case there is no point in working any more. [...]
[...] and telecom systems do not have this restriction. Powered by rapidly evolving technology that seems to know no limits, the electro-digital matrix is growing at a pace unmatched [...]