1. About Probability of Collision

Some thoughts in this context occurred to me for the first time when I formed an argument for “how not to be scooped”. I know there must exist profound strategies. But my current shallow strategy is that I strive to do the kind of science which is more like composing a song. No two songs are exactly same. It's very unlikely that two composers, sharing the knowledge of already-released songs but without communication, end up writing identical songs. Even when we allow for similar tastes and communication about main melody, chords or rhythms, one can still see how plenty a room there is for two songs to be different. So, if I can develop a mindset that bridges song composing to scientific paper writing, I may lead myself to the wonderland where it's very unlikely for someone else to happen to write a paper the same as what I want to write next. I'm very much driven towards this wonderland.

Last year, I heard someone say that they had some idea, but, given the fast pace of publication in AI, they said "I have to start this immediately and get this done within the next two months. Otherwise, the same idea will be published by someone else." Such a stressful situation is really a real thing in my field in the current days. I am trying everything to stay away from that sea of anxiety. I want to get to a place (train my mind to a certain level) so that I am better immune from such worries. When you liken doing science to composing songs, you would see how ridiculous those worries about “my ideas will be published by someone else” are. Imagine a composer saying "I have to finish this song by next month, otherwise, the same song will be released by someone else”. Wow! How could that be the case? We are not living in a world so densely enriched by thoughts that people's ideas easily collide onto others', yet!

I know how the above arguments may sound absolutely wild to many others working in my field. So I will elaborate why I find this perspective so attractive and why I feel being urged to abide by it. Perhaps it is because, by heart, I see science as being more similar to art than to engineering, or mechanics. It is more about the act of creation, creating something out of nothing, creating something that otherwise does not exist. In art, you create beauty, or thought-provoking representations of reality. In science, you create knowledge, yet another form of thought-provoking representations of reality.

One might raise a counterargument that scientific progress is somewhat “inevitable”, and this degree of determinism undermines its resemblance to art. From the stance of this counterargument, if calculus had not been invented by Newton and Leibniz, it would have been invented by someone else younger. But each art piece is unique. Yes, I agree with this argument. But the fallacy here is that the scale of progress underpinning something like invention of calculus is different from the scale of progress underpinning a singular act of creation. It is the progress on the large scale that bears some degree of determinism. The invention of calculus is one such progress on the large scale. The counterparts of large-scale progresses in art are perhaps those “isms”. You'd say if Monet had not initiated impressionism, someone else younger would have done it. Fairly likely. But no one would have painted impressionist drawings identical to those of Monet's.

2. About Unexplainable Skills

I bought a book at the bookstore of Yale entitled “101 things that I learned in Architecture school”. I resonate with it a lot. Science and architect have lots of best practices in common. Assuming that it is easier for a reader to see why architecture is art (or why architecture is closer to art than science appears to be), my second argument for why science is art will hinge on why science bears resemblance to architecture.

(a) Both start with an “idea”, with the entire creation process revolving around the idea. Through the training process you learn the unexplainable skill of differentiating between when to revise an idea vs. when to revise the way you approach the idea.

(b) Both is more of a holistic process rather than a sequential process. In a holistic process you try to let multiple things fall into place at a time rather than letting them one by one. Through the training process you learn the unexplainable skill of differentiating between when to stick to previous decisions vs. when to make a new ones.

(c) Both require you to peacefully live with unresolved problems, or put it in a nicer way, to turn unresolved problems into opportunities. The book tells me that an architecture is an eloquent restatement of the problem. A scientific product is by and large the same! An eloquent restatement is the other face of the same coin embodying the problem you start with, or the problem you meet on your way. Through the training process you learn the unexplainable skill of flipping the coin.

(d) Both have a fallacy the juniors in the field may fall into, in which one waits for things to become clear before starting to do something. This is a tempting but upside down view of the right process. Doing something is the way you make things clearer. And of course, as one gains experience, this process tends to be more organized, principled, and accelerated. The foremost barrier one must get over is taking the first step when there isn't any compelling external reason to do so. The way to concur this barrier is by reinforcing compelling internal reasons (which can be numerous). Letting internal reasons to emerge and learn to perceive, appreciate, listen to and manage them are the prerequisites of developing those unexplainable skills for doing science and doing art.

A parenthetical comment on "the prerequisites". The way I phrase it makes it sound like an introduction-level accomplishment, a mere knock on the door to a vast space of the “real” things to learn. But my impression is, unfortunately, that many PhDs never make that knock. Thus, they never open the door they are meant to open and never explore the space they are meant to explore. This view challenges many people's assumption of how high the entry barrier is. But in my opinion, it is indeed that high. It can take years to make the single move of "entering".

3. About Discipline Serving for Communication

So far, I have argued at length about why science, in its most beautiful form, is extremely creative, blurring its border with art. A critic would say: “No, science requires much more discipline than art does, especially quantitative science”. I agree with this statement. But one shall not overlook the fact that art also requires discipline —— the ability to communicate with the history of art. I quote the historical definitions of art, and express special thanks to Wenjie Li for making me aware of this illuminating fact.

… All of them (the definitions) are, or resemble, inductive definitions: they claim that certain entities belong unconditionally to the class of artworks, while others do so because they stand in the appropriate relations thereto.

Switching back to science, being artful and exhibiting discipline do not raise a contradiction. The artistic value should not be dismissed as some exotic flavor and repelled to the periphery of “properties of good science”. Also, the appeared higher degree of discipline in science does not do much to threaten the view that science is art. There can be various forms of self-control or order, depending on the scientific field. And all of them, in the limit, serve for the communication with the history of science. Creating something that has not been existing before is a fundamental feature of science, and the “thingness” of the created thing will be warranted by its ability to communicate with the history of science.

I have finished presenting my three arguments. They must be naive in many senses, which I expect the future me to spot and correct. Naivety aside, I believe it is never a bad thing that I see tremendous amount of artistic value hosted in science.