何毓琦的个人博客分享 http://blog.sciencenet.cn/u/何毓琦 哈佛(1961-2001) 清华(2001-date)

博文

On Research #9 - three things to remember 精选

已有 10325 次阅读 2007-10-19 18:14 |系统分类:海外观察

(For new reader and those who request 好友请求, please read my 公告栏 first)

On “How to do research” - #9



“Sanity checks in Research” ,  “欲速则不达”, and “Two wrongs

sometimes make one right”


下面有中文翻譯


There are many aspects of doing research successfully. Today, I

want to illustrate three aspects of this general problem using a real

incident which may be helpful to young researchers starting out.



First, when you “discover” a new results or formula, particularly if

it is after a series of long arguments or calculation, it is always a

good practice to do a “Sanity Check” on the result in order to

avoid making a silly mistake in the process of discovery or

derivation.  By sanity check I mean that you apply the result or

formula you derived to a known situation or an extreme situation

for which you already know the answer from independent

considerations.  If the result of application does not agree with the

independently known answer, then something is wrong. You need

to go back to check your derivation step-by-step.



Second, the pressure to get results in research often makes us

careless. The pressure to “publish”  also tends to rush us. But

research is basically built on painstakingly careful experiments

and analysis.  Sure we often have inspirations. But that is only the

beginning. The famous inventor Thomas Edison said it well

“success is 10% inspiration and 90% perspiration”.  The well

known Chinese saying “欲速则不达” cannot be more applicable here.



Lastly, “serendipity” does play a part in research. Lucky or

unlucky accidents happen. Sometimes two mistakes cancel each

other out, and you accidentally get the right answer. But pure luck

should not make you forget the first two dictates above. In this

case, it is even more important to carefully check the process of

getting to your results.



These three lessons of research combined to produce a comedy of

errors in my blog article of “Beijing sights and thoughts III –

energy cost of drying your clothes” dated Oct.19, 2007”



One of my Beijing experience during my recent trip to China was

the accidental observation of the Chinese and the US approach to

drying your laundry. http://sciencenet.cn/blog/user_content.aspx?id=9302 . I thought

this was an interesting observation on energy consumption and

environmental protection that have not been addressed before in

the public media. Thus, in the rush to get my blog article out, I

committed the second cardinal error above - “欲速则不达”. I did

not check my calculations carefully enough. The mistake was

detected quickly enough by a reader with the penname “confused”

(see comment associated with above blog article). So I quickly

admitted the error. But since the error did not materially change

the conclusion I was making, I did not bother to carefully check

the source data as the calculation mistake was plain enough.  Had

I done a “sanity check”,  I’d have discovered that my source data

was copied wrong. It takes 5 KWH of electricity to dry one load

of cloth not the 50KWH in the article. How should I have known

this? Here is my sanity check if I had done it. Yes, on the average

it takes one hour to dry one load of cloth (if you have lived in the

US, this is common knowledge). A simple hairdryer consumes

power at the rate 1KW. The heat produce by an electric dryer is

approximately that of five hair dryers NOT fifty hair dryers (again

this is common experience). Thus, I should have realized that the

figure 50KHW/load of cloth does not pass the sanity check. Since

I so readily agree with the reader “confused” about the calculation

mistake, I again should have realized that the figure  $750

billion/year of cloth  drying cost based  on the correct calculation is

unreasonable. The GNP of the US is only on the order of a few

trillions  dollars.  Finally this bring up my third lesson,  namely, the

conclusion of $75 billion dollars per year is not an  unreasonable

figure .  In my original article, I simply made two mistakes on

calculation and on source data that canceled each other out. The

final result does pass the sanity check.



In any case, here is my “mea culpa (我的錯)”. Let this be a lesson

to young researchers. One can never be too careful. Double

check and don’t rush!



教育与研究随笔(九)

科学研究中需要牢记的三件事:“可靠性检查”、“欲速则不达”和“负负得正”

科学研究要想取得成功,需要多个因素。今天,我想通过一个实例来说明其中的三个方面,或许能帮助刚上路的年轻科学家。

首先,当你“发现”了一个新的结果或者公式,特别是如果你是在经过了一系列冗长的论证或者计算之后得到它们的,那么你最好对结果进行“可靠性检查”,这能帮助你避免在推导或发现过程中犯下愚蠢的错误。所谓可靠性检查,指的是将你得到的新结果或者新公式应用于一个已知的问题或者一个极端问题在那裡你已经通过其它方法得到了这个问题的正确答案。因此,如果你得到的结果和已知的答案不一致,那么肯定什么地方出了问题。你需要回过头去一步步检查推导过程。

其次,研究中需要得到结果的压力常常使我们变得粗心,而“发表文章”的压力同样催促着我们快一点、再快一点。但是从本质上说,科学研究是建立在辛辛苦苦精心设计的实验和分析的基础之上的。有时候我们当然会灵机一动,想出个好主意,然而那只是一切的开始。著名发明家爱迪生有一句名言“成功是10%的灵感加上90%的汗水”,而中国谚语“欲速则不达”则能最好地说明这一点。

最后,“意外的惊喜”确实是研究的一部分。幸运或者不幸的事情总是在发生着。有时两个错误之间会相互抵消,负负得正,结果你得到的却是正确答案。但是你不能因为运气好,就忘记上面提到的两点。在这种情况下,仔细检查整个研究过程就变得尤其重要。

以上三个关于研究的教训结合起来,就产生了2007年10月19日我所写的《北京所见所想3——衣物烘干中的能量消耗》一文中所发生的喜剧性错误,。

我在最近一次的中国之行中,在北京意外发现中国人把衣服弄干的方法和美国人不一样。(博文请见:http://sciencenet.cn/blog/user_content.aspx?id=9302)。我认为这是一个关于能量消耗和环境保护方面很有意思的事情,而且此前没有媒体报道过。因此,为了尽快发表这篇博文,我犯了上文中提到过的第二个错误——“欲速则不达”。我没有仔细检查我的计算过程,结果一位名为“困惑(confused)”的网友很快发现了我犯的错误(你可以在该博文的评论中找到)。于是我马上承认了这个错误,但是由于这一错误并未在实质上改变我的结论,而且那个计算错误非常明显,所以我懒得去仔细核查数据来源。而如果我进行了“可靠性检查”,就会发现源数据在复制的时候出了问题。烘干一缸衣物其实只需要5度电,而不是文章中写的50度。而我怎么才能够发现这个原始错误呢?如果我当时进行了可靠性检查,那么过程应该是这样的:一般烘干一缸衣服需要一小时(凡是在美国住过的人都知道)。一个简单的吹风机的功率约为1度。而烘干机产生的热量大约相当于5个吹风机而不是50个!(这也是常识。)因此,我本应该想到烘干一缸衣物要费50度电是不可能的。在我欣然接受网友指出的计算错误的同时,也应该发现即使计算正确,仅仅烘干衣服也不可能每年花掉美国人7500亿美元,因为美国的国民生产总值也不过在数万亿美元的水平上。最后,这印证了我提到的第三点,即文中得出的计算结果——每年烘干衣服花掉750亿美元——并不是不合理的,因为在原文中,我犯了计算和数据源方面的两个简单错误,它们之间相互抵消,最后的结果反倒通过了可靠性检查。

不管怎样,这就是我自己犯过的一个错误。让它成为年轻科学家们的教训吧,要记住,在科研中,再细心也是不够的,一定要检查检查再检查,千万别着急!(何宏辉译 何姣校)


原文链接:http://www.stimes.cn/blog/user_content.aspx?id=9328



http://blog.sciencenet.cn/blog-1565-9328.html

上一篇:Beijing sights and thoughts- III
下一篇:A Plea for Help
收藏 分享 举报

1 李天成

该博文允许注册用户评论 请点击登录 评论 (0 个评论)

数据加载中...

Archiver|手机版|科学网 ( 京ICP备14006957 )

GMT+8, 2017-9-21 00:36

Powered by ScienceNet.cn

Copyright © 2007-2017 中国科学报社

返回顶部