Human based evolutionary computation and genetic algorithms #humancomputation #crowdmining #globalbrain #microwork #crypto
Evolving algorithms typify soft computing. In soft computing it is recognized that there is no precise or approximate solution which can be arrived at without a trial and error approach. This is similar to how natural systems work and it is why many soft computing techniques are based on biomimicry such as genetic algorithms, or particle swarm optimizations, or swarm intelligence. These techniques can be found in nature and why is this?
First we have to define nature for purposes of this discussion. Nature in this discussion are biological-based computers, molecular computers, wet wear (brains) and similar mechanisms which the universe has evolved to allow for computation. At the same time nature in this discussion also includes everything a biological entity has created, so this includes all of our computers in whatever form. So in essence anything man made for purposes of this discussion is natural.
Nature inspires algorithms and solutions can be found by looking at how biology has solved the problem. Biology is imprecise in it’s solutions and uses trial and error approaches such as evolutionary computation but these approaches work because over a period of time the algorithm approximates to an answer sufficient for the query. It is in this way that an anternet can be similar to an Internet or that a bee hive can produce swarm intelligence with great predictive qualities. Markets also act in a similar way and are also imprecise but trend toward efficiency.
Human based evolutionary computation is a technique to solve problems using the mechanics of human nature. For example Steemit uses human beings as the human computers in the system who curate and that is an example of a collaborative filtering. In a human based evolutionary computation the human being is the innovator acting as initializer, mutator, and recombinator. In other words the human acts as the selector (curator), and innovator (content producer). In Steemit the most fit content rises to the top which we could state is an evolutionary process which uses a human based evolutionary computation to encourage and reward value in terms of a fitness function.
Human computation is a commodity just as any other form of computation
When people think of human computers they might think back to the times before there were electronic computers. During the time when only a human being could compute and before the inventions of Alan Turing the human computer was the only way to solve problems. The human computers were eventually augmented and supplemented by the mainframe style computers and it was in this era that precision improved. The formalist way is about precision and it is this precision which put a man on the moon. At the same time when problems are so difficult that don’t know a good algorithm which can lead to a precise answer then you may have to try the brute force trial and error method.
Human computation is coming back in style with the advent of crowdsourcings. Crowdminig and human based computation games are going to allow human beings to contribute to a virtual computer made up of both human computers and AI. This convergence will allow for the most difficult problems to be solved where all of the problems a human cannot solve will be delegated to a non-human artificial intelligence and all of the problems a non-human intelligence cannot solve will be delegated to a human intelligence.
Non-human intelligences include artificial intelligence but are not restricted to it
Non-human intelligence can include molecular computers, biological computers such as DNA or any form of computation which can be found. It may be that it’s artificial intelligence or it may be a different genetically engineered species. The point here is that human beings are not going to be the only thinkers going into the future which means the market will be able to gather resources in the form of computation from all available sources. Problems will be presented, broken into sufficiently small pieces, incentives will be offered.
Human computation intro
Crowdmining
Crowdlang
Conclusion
A risk of over reliance on hard computing is over engineering. Engineers may believe they can use pure logic and calculate with precision and for certain problems this may be the case but it fails for other cases(NP complete problems). On the other hand a risk of soft computing is over reliance on evolutionary strategies in cases where precision is available and can lead to greater efficiency.
In a market you sacrifice some precision and some efficiency when it comes to solving problems. You may gain in innovation but lose an ability to build large infrastructure or do very precise kinds of efforts because it might not be profitable in the short term. The space race was not a result of the market competition but of military competition where all expenses for scientists were paid and basic research could be funded which expanded the knowledge base.
In the future due to blockchain technology we can now combine the benefits of soft and hard computing. Proofchains, correct by construction, precise formal specifications for smart contracts, all offer the engineered benefits of hard computing such as efficiency, predictability, security. At the same time we can get the benefits of soft computing through platforms like Steemit, market mechanisms, human computation, and more.
References