But that changes when you help a friend overcome a…
I said, “Just think about graph theory. ” And she asked, “What do you mean graph theory?” Like, really confused. Obviously graph theory! I assume she knows what I mean. From there, I realized that not everyone knows everything I know, and maybe I have an advantage and can really help people. I was looking for a book to write, and now that I’ve written it, that’s great.
You write, “Where people are vague, often contradictory, and confusing, science is reliable and clear. It doesn’t deceive you…” But science is a human endeavor. Doesn’t that make it as wobbly as the people who made it?
Yes, that is indeed the case! At first, I was quite nervous, like “Oh my god, the infrastructure is so stable [to me] was just as confused as the rest of us. ” But the whole point of the scientific process is not just the storage machines; it’s our ability to use our instincts to know when to take the leap and ask questions, and also to have the patience to troubleshoot. Through working with it, asking questions, and recognizing inconsistencies in the process, I realized [science] beautiful in its uncertainty. It’s not just logic, it’s also a very instinctive endeavor.
Regarding the point of uncertainty, you explain that we cannot optimize our lives unless we “study and understand noises, errors, and deviations from the mean.” Are you telling err is science?
Completely. Many people think that to be the perfect scientist you need to not do everything wrong, but “wrong” is a somewhat elusive concept. False in one context is actually true in another. Like, when it comes to evolution, what is the proper way to fertilize an egg that develops into an embryo? Evolution is not perfect. It is an adaptive process where many different ways of doing things can develop. Science is not perfectionist.
So let’s take a concrete example of how science has informed and changed your behavior. You argue machine learning can help us analyze information and make better decisions. How so?
Growing up with neurosis, you cling to fixed categories of hell or high water because you need that sense of security. But I realized that this is a very unsustainable and inflexible way of doing things – what I call “box-based”.
You explain what in the context of machine learning would be called supervised learning, where, “you have a specific result and you program the algorithm to achieve that result”.
Yes. Over time, I learned that I needed to be more flexible in order to enjoy my life. So machine learning [offers] another discrete process called unsupervised learning, which specifically looks at the data around you and then re-clusters it accordingly. You are not trying and clustering the data according to your preset conditions. You look around, sort through what you have, and from there, ask “what’s the best option?”
