jameshzd的个人博客分享 http://blog.sciencenet.cn/u/jameshzd

博文

A forever pursuit for better science

已有 1719 次阅读 2017-1-24 15:59 |个人分类:自我感悟|系统分类:科研笔记

 What does a PhD student do in a daily basis? Research! But do they know how to do good quality research? Maybe. But it is always important to give an exact definition about what really is "good science". In this paper I will talk a little bit of my ideas on that matter. Hope we could have some discussions. I also think this should be taught in grad school for young graduate students, because this knowledge can make huge impact on one's scientific career.


 What is a good science? In my opinion, science is all about "finding things out". Well, it is a rather general statement, let's break it down. Finding means we have to think something new, something that hasn't been thought before, whether it is a new way of doing experiment; a new technology or facility; a new experimental discovery; a new theory; a new understanding/explanation for some old problems. If you want your science to be "good quality", novelty is the first issue.


 We all read papers everyday, there are papers well-written, well-thought. Reading them is the most rewarding experience. It can give you insight, help you forsee the future and embark some new ideas in your brain. But some papers will drain your energy even for just understanding what the authors are talking about. Once we have finished that paper, we cannot grasp the main point. And the worst case is, by the time you've finished it, you suddenly realized the paper's foundation is wrong, it's like a house of cards. Then you will regret why you spent that much time in such unrewarding paper.


 How to become a good scientist in both research and writing? It is an ongoing task for scientists. Many books have been published, for example Beveridge <The art of scientific investigation>, which is widely-recognized book on that subject. In my opinion, to think clearly in both your scientific problem and your scientific concepts are both equally important. We will talk about this in further detais because it is such an important topic.


 Everything we have learned in college or even in graduate school, is the concepts developed by many brilliant men and women in the past several hundred years. For example, mathmatics, physics and chemistry. What's interesting about these subjects is that they are all built on concepts, which comes from experimental results. All the concepts we can learn in the books today have been tested thoroughly and used for explanation the world around us and experimental finding done in the laboratory. Even though, we should keep in mind that they are "temporary" correct, with new finding and new way of looking at things, these concepts or "building stone" of our science will change or simply disappear. It is very important to know how the concepts have been constructed, and whether they have a solid basis or not. Some quantities, like mass and electric charge, they are very basic quantities in our science, they are like the alphabet in our scientific languages. We cannot develop more complex concepts without them. I think when we learn a new concept, it is very important to think about how it has been constructed, is the foundation solid and what circumstance can it be applied. By keep doing this and recheck our scientific knowledge, we will have more powerful and more clear concepts for us to address new phenomena and find new things.


 In previous we talked about how to make your scientific concepts clear, now we talk about how to think about your scientific research clearly. Everyone studies a specific question by using a set of specific methods in order to get specific answers from those findings. In science, the most important skill is to ask important questions. Important questions aren't big questions, they have clear scientific importance, and by answering them, it can benefit for a wide range of researchers. The first thing we think about the problem is in what scale I'm going to study it. If we look at a glass of water, in macroscopic level, we will talk about its density, its temperature, its viscosity and other properties; in mesoscopic level, we will talk about the "substructure" of the water, we will find the density and temperature are not isotropic through the whole water range, they can fluctuate; in microscopic level, we will talk about the water cluster and hydrogen bonds between water molecules, and the formation of network when the liquid water becomes ice. By using the same glass of water, by thinking in different level we will ask different questions, and then by using different methods to study those questions in order to formulate the answer. What is really important in here is the results in a certain level cannot be used to understand/predict the other level's results. We muct perfectly match the theory and experiments in the same spatial and temporal dimension, only in this manner we can create clear information of the system. It doesn't say we can't "guess" what is going on in microscopic world by looking at "macroscopic" results, sometimes it can also give us some hint and potential resesarch questions. But it is better to study the fish if we can look at the fish.


 We can use 2 axis to represent 4 kinds of quantities: (measurable, unmeasurable) * (computable, uncomputable). If a quantity is both measurable and computable, then we should put that into our "basis quantity list". If a quantity is computable but not measurable, then we can use it as an intermediary quantity for our calculation (at last our calculation should contain only the "basis quantities"). If a quantity is measurable but not computable, the reasons could be: (1) There is no clear definition of the quantity. In this case we should throw it away immediately. (2) The calculation of the quantity is extremely hard. This can drive us to come up with new ways for calculating the quantity, could be a new angle and a new publish. If a quantity is either unmeasurable or uncomputable, we should say "si yo na la" to that.


 The other thing I want to address is how we can "prove" or "ensure" that a theory/explanation is reasonable and could be used in a predictable way. The answer is very simple: Assume it's right, see what we could predict from that. If the predictions agree with experimental results (in here we use only computable and measurable quantities for direct comparison, and when we say "agree", we mean the trend and key characteristics, but not the exact match, that'll be very improbable because of the error we included in both experiments and calculations), then we can see at this moment the new theory/explanation holds. But it will endure more and more tests and maybe eventually it will fail. But it will never be "reality", because the reality is hidden in every detail. We can only find the partial truth, but not the whole truth.


 In my opinion, we can use the enginnering point of view to look at research: Formulate a problem --> Determine what you would like to know --> Clarify the key concepts will be used in discussions --> Choose the methods --> execute those methods and gain results --> analyze the results to the same level you want to know --> If we can make a logic story --> Publish it ! (If not, go back and do more work!!!). Even in some details you don't know, that is OK. There is no study that can contain all the information of the system. If we look the papers published, even in very good journals, the authors often "guess" what is going on and they call it the "discussion section", in a personal level, I don't like it. It doesn't make any sense until you bring me the solid evidence or data to prove your point. Before that, it's bullshit!! When we write a paper, every work we say should have backup, whether it is some else's work or our own experimental data. We cannot say "it could be ....." without providing any information.


 We must clearly know what we are talking about. It contains 2 parts: know what you do know and know what you don't know and even better to know how you can gain insights in what you don't know.


 Those are my opinions of good science, it is all about clarity. Without clarity, we cannot build our scientific world. This is my pursuit for better science. And I hope this article can help you with your research and the way you think about scientific discoveries in general.


 The beauty of science is everywhere, and it is really a nice way of spending one's life in finding and figuring out interseting things.


 Happy new year my dear fellow scientists, let's do more good science in 2017!


 PS: I'm thinking about an efficient way to read extensive amount of papers. And this way looks very promising: Whenever you are reading a paper, figure out these things in order:

(1) What question will the author solve?

(2) What is the novelty of this paper?

(3) Depend on the question, what things do you want to know? What evidence will you collect?

(4) Read the data the authors collected, by only looking at those results, are you convinced that these results can tell you something about the problem raised in the paper?

(5) If it can, what other variations can you do? If it can't, what data should be included in order to make clear of the question?

After these 5 steps, you can put the paper in your folder. Remember this: You will read each paper several times if it is really important to your own research. Also the literature review in each article is really helpful for us to generate a time line for a particular small area in science. But this part will be read until we finish the whole paper first.



http://blog.sciencenet.cn/blog-376446-1029660.html

上一篇:Short report for paper reading
下一篇:George Whitesides --- The man who reinvent chemmistry

1 张海权

该博文允许注册用户评论 请点击登录 评论 (1 个评论)

数据加载中...

Archiver|手机版|科学网 ( 京ICP备14006957 )

GMT+8, 2020-1-26 13:46

Powered by ScienceNet.cn

Copyright © 2007- 中国科学报社

返回顶部