Exploration v.s. Exploitation
In one of my last blog posts, I wrote about how easily accessible information, learning sources, and tools are in the modern world and how lucky we are to feed our curiosity with interesting information. I read a post about curiosity conflict that made me think more. Basic idea behind the curiosity conficlt says:
High curiosity is what makes us interested in following ideas and generate new creative ideas of our own. At the same time, it makes difficult to focus on one idea when encountered with multiple new ideas.
I could relate strongly with this conflict. During my student days, it was difficult for me to choose a single project topic for course projects. I searched and generated a lot of potential interesting topics but could not decide which one I should choose as my final topic. When I discover a new interesting blog, I spend most of my time exploring the blog rather than reading through the initial post that brought me to the blog in the first place.
The accessibility of information on the internet makes exploration quite easy and, to some extent, quite fun. I get quite excited whenever I find a new website or book that somehow relates to me and my experiences. But the important point is that exploration alone does not bring much benefit. All the new interesting courses, websites, and articles end up in a "to read later" list, and the list keeps growing and growing. The real benefit comes from focusing on the information and engaging with those sources.
I had a course at university about optimization algorithms. Two basic concepts in stochastic optimization algorithms are exploration and exploitation. To find a good solution for complex problems, it is very important to balance between the two. Exploration is about searching in new areas of the solution space to find potentially better solution areas. Exploitation is about refining the current known good solutions to converge to the best possible result. Too much exploitation and you risk getting stuck in premature solutions (or a local optimum), and too much exploration risks wasting too much time/resources and slowing down convergence to the final solution.
I think this is a good analogy for the curiosity conflict in the real life as well.