转某语丝帖子:“人文社会科学”不是科学

“人文社会科学”不是科学
  ——一次同学网络聚会上的发言

  作者:后悔读史

  所以要提衣俊卿与女博士后的故事给大家听,并不是为了那些八卦,而是该
文通过一位女性细腻的笔触揭露那些我一直怀疑而又无法求证、却天天发生在学
界的肮脏交易。我对常艳女士所披露的国家基金项目评审、国家级奖项评审中打
招呼的事耳有所闻却未曾亲睹,现在这些事情得到了证实。

  如果认为我想佐证上述这些舞弊行为并把它讲给大家仅仅是为了让那些在学
界混饭的同学不要“太天真”,或者为了让那些不在学界混饭的同学知道“学术”
既不干净也不崇高——那是过低的理解了我的初衷。

  我真正想说的是:千万不要让自己的孩子再读文科。孟夫子有言“术不可不
慎”,大众化的语言就是“男怕选错行”。因为学文科必然要在“人文社会科学”
领域混。我用十二年的时间(从2000年读历史学研究生开始)明白了一个道理:
人文社会科学不是科学,它缺乏“科学成立”的许多要件,比如严密的逻辑、供
共同体内交流的专业并固化的表述语言、可证伪的学术途径等等,套用一句俗语
——缺乏共同体范式。那么,我们经常称道的历史学、政治学、哲学等到底是什
么呢?它们仅仅是一些散在的知识群团,而不是一个严整的科学体系,所以是
“非科学”。我怀疑,只有经济学、社会学和语言学才具有一定的科学性,姑且
称为“半科学”。

  有了上述理论前设并在其烛照之下,我们才能对“人文社会科学界”频频发
生的舞弊事件有正确的认识,即我们不能纠结于某些掌控学术资源的人或团体的
人性劣根,而应该深刻的认识到,由于“人文社会科学”先在的病理不足,而使
得从业共同体的正能量不能在学术前行的路上得到释放,难免会走旁门左道。换
句话讲,“人文社会科学” 界鱼龙混杂,而又缺乏基本的学术评价标准,这自
然会走向邪路。

  最后,我想用大家都熟知的丘成桐证明卡拉比猜想的故事来说明这个道理。
1977年,28岁的丘成桐证明了卡拉比猜想(在封闭的空间,存在没有物质分布的
引力场)。而此前,他曾宣称自己找到了证明卡拉比猜想错误的方法。随即,收
到了来自卡拉比的亲笔信,卡拉比教授在信中指出,用他所说的方法无法证明卡
拉比猜想的错误之处,希望丘成桐证明给他看。丘成桐承认了自己的错误,并最
终证明了卡拉彼猜想是正确的。

  大家熟知,在“人文社会科学”界,这样的事情不会发生,而相反,我们会
看到太多各持己见、没有意义的争论。之所以正误难判,归根到底还是因为“人
文社会科学”本就不是科学。

吴金闪注:计算语言学、博弈论理论与实验,甚至更多的社会科学领域,只要其工作方法是:观察与实验->模型、假设->可计算、可操作、可测量的理论(退到底线,我们要求一个理论是可证伪的——也就是说这个理论不成立则得到的理论结果与实验和观察不符)->实验,则仍然是科学。作为一个正面的例子,可以参考这个关于计算文体学的帖子。当然,按照这个要求,大多数“人文社会科学”的工作,确实是垃圾。不过,不能表明“人文社会科学”是垃圾,将来还是有希望科学化的。对于问為什麼要科学化的人,不要理我,我推荐你冻几块豆腐,随时用。

转贴:鼓励的力量:“谢谢你”影响深远 The Power of Praise: 'Thank You' Goes a Long Way

来自于http://www.inc.com/kathleen-kim/a-simple-thank-you-drives-stronger-business.html

鼓励的力量:“谢谢你”影响深远

The Power of Praise: ‘Thank You’ Goes a Long Way
New research reveals something pretty obvious. Thanking and rewarding employees can give your business a boost.

Don’t hold off until annual performance reviews to praise stellar employees. A simple “thank you” here and there can boost efficiency and even help your business make more money.

根据这个新的研究(或者,同样靠谱的,常识),那些在对其员工的贡献认同方面做的好的企业,其商业表现经常在那些做的不好的企业的12倍以上。
According to new research (and, perhaps, the laws of common sense) companies that excel at employee recognition are 12 times more likely to generate strong business results than those that do not.

In companies focused on rewarding their workers, employee engagement, productivity and customer service were about 14% better than in those that skimp on recognition, the study found.

The study was conducted by advisory services firm Bersin & Associates President and CEO Josh Bersin wrote recently that “high-recognition culture” companies share three common traits:

First, they build focused recognition programs which collect “thank you’s” and “feedback” from peers, not just managers. Second, they directly tie recognition to business goals and company values, so recognition reinforces strategy. Third, they give employees open and transparent access to the program – so everyone can see who is being recognized and anyone can recognize another.

But to really praise like a pro, start here:
但是要注意一下三条:

1、要具体,有所针对,而不是泛泛表扬
1. Be specific and know your people: Inc. columnist Gail Browning recently wrote: Our research at Emergenetics indicates that most employees would enjoy a personal thank-you note, but they want it customized to them. For example, to say, “You’re doing a good job,” is fine for a “social” thinker, but a “structural” thinker doesn’t trust you unless you add a specific task he has accomplished.”

2、考虑表扬的方式:有的人喜欢当着很多人的面,有的人喜欢私下
2. Consider the delivery: The way a thank you comes across is just as important as the thank you itself. Inc.’s Jeff Haden recently wrote: Every employee responds differently to recognition. Many appreciate public praise. Others cringe if they’re made the center of attention. Know your employees and tailor your recognition so it produces the greatest impact for each individual.

3、让大家都参与进来:成为一种氛围。
3. Get everyone involved: A thank you shouldn’t be just one-on-one. Encourage your entire company to promote appreciation and praise. Inc.’s Jay Love recently advised: Insist on your department heads sharing stories from their departments and highlighting the achievements of team members at the monthly All-Company Meeting. Lively presentations that include photographs, videos and client comments make this one even better!

吴金闪加的注:赏罚分明,有原则(最好是写下了来的原则,然后是默认的共识的原则),是领导力最根本的来源,而不是老大赋予的。没有赏罚(可以是口头的或者是其他方式的),就没有管理。做管理学的同行,是不是可以借鉴他们的研究方法来尝试做一个对比:只有表扬的企业或者部门,有赏有罚的企业或者部门,其表现的对比。

我带学生的风格

1、我对学生的课业和研究工作,研究品味方面要求比较高。
2、时间上也要求学生比较投入。
3、在研究内容的选择方面,我给学生比较大的自由。可以慢慢地从我想做的东西里头选择,也可以自己想然后通过与我讨论并得到认可以后拿来做。
4、我会尽量地从idea、实现技术、研究所需知识基础、研究课题的背景和现状的调研等等各个方面给于帮助和指导。
5、我会非常严厉地批评学生,也会非常耐心地给学生从头讲解。
6、只要学生清楚自己在做什么,比较细之地考虑过了自己的选择的结果(当然,我会帮你分析选择的可能后果),我一般都尊重学生的选择。

就算我在严厉批评学生的时候,除了知识积累或者技术实现甚至对这个项目的认可程度方面都差别,学生和我是平等的。学生可以和我争论,我不会伤心、生气。研究工作进展很小的阶段很正常,但是如果我不知道进展小的原因,不知道学生在针对这个情况做什么,就不正常了。我希望学生不要怕找我诉说困难,如果这都怕,要老师干什么?老师就是跟学生一起找到困难到底在什么地方,然后在能够帮忙的时候帮一下的人。在研究工作中起主导或者辅助作用我都愿意。

Jinshan's CV

Jinshan’s Curriculum Vitae

Curriculum Vitae

Jinshan Wu

Department of Systems Science

Beijing Normal University

Tel: +86-10-58807876(O), +86-18610014018(M)

E-mail: Jinshanw@bnu.edu.ca


<

ul>

  • Education

    • 2011, PH.D. in Condensed Matter Physics, Department of Physics & Astronomy, University of British Columbia (UBC)
    • 2003-2004, one year in a PH.D. program in Simon Fraser University (SFU), and then transferred to UBC
    • 2006, M.Sc. in Condensed Matter Physics, Department of Physics & Astronomy, UBC
    • 2002, M.Sc. in Statistical Physics, Physics Department, Beijing Normal University(BNU)
    • 1999, B.S. in Physics, Department of Physics, BNU
  • Employment

    • 2011-, Associate Professor, Department of Systems Science, BNU, Beijing, China
    • 2004-2011, Teaching Assistant, Department of Physics & Astronomy, UBC
    • 2003-2004, Teaching Assistant, Department of Physics, SFU
    • 2002-2003, Lecturer and Research Associate, Department of Systems Science, BNU
  • Professional experience

    • Associate Professor

      • 2012 Spring, Physics and Mathematics in Studies of Complexity II, graduate course, Department of Systems Science, BNU.
      • 2011 Fall, Physics and Mathematics in Studies of Complexity I, graduate course, Department of Systems Science, BNU.
      • 2011- , Non-equilibrium statistical physics and Quantum transport project, PI, funded partially by National Natural Science Foundation of China.
      • 2011- , Network-based learning strategies of Chinese characters, PI, not yet funded by any agencies.
      • 2012- , Study concept mapping technology and generate collections of concept maps, PI, funded partially by university research fund from BNU.
    • Lecturer

      • 2003, Math Model, undergraduate course, Department of Systems Science, BNU.
      • 2002, Econophysics, a course for graduate students, Department of Systems Science, BNU, 2002,9-2003,1. I designed and established this course from scratch. A review paper ([19] in the publication list) on Econophysics prepared for the class was post on arXiv. Since then many have used it as an introductory material for the subject.
    • Research Associate

      • 2002-2003, under Prof. Zengru Di’s supervision, lead a team working on empirical studies of and modelling weighted complex networks
      • 2002-2003, help Prof. Zengru Di to organize a proposal for National Fund of Natural Science in China, The statistical properties of firm sizes and its theoretical model, funded at 2003,9.
    • Research Assistant

      • 2004-2011, as a graduate student in Prof. Mona Berciu’s group, working on various projects related to quantum transport
      • 1999-2002, as a graduate student (master) in Prof. Zhanru Yang’s group, during the later years of and one year after my graduation (2001-2003), I lead a team working on physical models on complex networks
  • Skills in numerical computation

    • High-performance computational software: BLAS, Lapack, Petsc, Slepc, gsl, xmds
    • Programming language: C, Java, Linux shell script
  • Award

    • University Graduate Fellowship (UGF) from UBC, 2006-2009
    • Graduate Fellowship from SFU, spring 2004
    • Canron Limited – Sidney Hong Memorial Grad Scholarship, Spring 2004
    • Westak International Sales Inc. Grad Scholarship in Expert Systems, Spring 2004
    • Scholarship for Excellent Graduate Students from BNU, 2000
    • Award for excellent undergraduate students from BNU, 1998
  • Research Contribution

    • Quantum Transport

      In my Ph. D. work at UBC I aimed to establish a theoretical framework for finding the non-equilibrium stationary states of quantum systems starting mostly from first principles. Approaches exist for this problem such as the Landauer-Buttiker formula and the non-equilibrium Green’s function method. We decided to use the open-system scenario, which is not widely used because of the difficulty in solving the resulting open-system master equation. Using direct methods, one needs to solve an eigenvalue problem of size 4N where N is the size of the system measured in qubits. We first searched for efficient methods to solve this problem and then applications of this framework on physical models. The following lists several projects I have worked on.

      • Using a BBGKY-like method for solving the open-system master equation [2] the task of solving an eigenvalue problem of size 4N becomes a problem of solving a linear system of size N2 by converting the open-system master equation into linear equations of Green’s functions. The equations of different Green’s functions (single-particle ones, tow-particle ones and so on) are coupled. The cluster expansion, originally used for the equilibrium BBGKY method, is used to truncate the coupled equation. The accuracy of this method is around 2%.
      • The second order form of the BBGKY-like method requires solving a linear system of size N4 but improves accuracy even further. Such a form also gives the two-particle correlated Green’s functions beyond the Hartree-Fock approximation. Manuscript in preparation.
      • A coherent-state representation approach was also explored to solve the above problem of size 4N by simulating a stochastic differential equation with 2N complex variables by converting the open-system master equation into a generalized Fokker-Planck equation. Analytical expression of the non-equilibrium stationary states are derived for some systems. The accuracy of this method is around 6%. Manuscript in preparation.
      • We also found in study of the Kubo formula for open systems [1] that in order to study transport one has to take into account the coupling from the central system to the baths explicitly. In using the usual Kubo formula in transport studies, one assume the central system is a closed system.
      • Using direct methods we studied thermal transport of spin chains[17] and analyzed systems up to N=10. Connections between integrability and anomalous transport, which is widely believed by physicists and has been demonstrated by studies based on the usual Kubo formula, is challenged by our results.
    • Weighted networks

      This series of works started in late 2002 when I was employed as a research associate for Prof. Zengru Di at BNU after I got my M.Sc. Degree in statistical physics from BNU. Many thanks to Prof. Zengru Di, Prof. Yougui Wang and Prof. Zhangang, Han for offering me a position usually requiring a PH.D. Degree. The focus of my research on weighted networks has been the basic statistical features of static weighted networks, their evolution and also some more advanced structure in those networks.

      • Empirical study of weighted networks[11,12]: We collected almost all papers published on Econophysics up to date (back then), compiled a weighted network and studies its basic statistical properties.
      • Evolutionary model for weighted networks [3,7,9]: Inspired by social networks and the above weighted networks of econophysicists, a new model of weighted networks was proposed. It is based on local rules, which means that nodes in the network only need to know limited information about their neighbors and at most their next neighbors. That is in this model a data centers providing global information is not required. We went one step further and conjectured that the well-known mechanism of global preferential attachment (that the richest gets richer while the poorest gets poorer) can be an emergent phenomenon rooted from local rules. We tested and confirmed this conjecture on our own model and several others.
    • Quantum Game Theory[20]

      In physicists’ terminologies, classical games can be regarded as games based on classical objects. The state of the object changes according to players’ choice of strategies. These strategies are described by operators acting on the object to modify its state. The final state of the object determines the payoff for every player. The coin flipping games is a perfect example of this picture. The coin is a classical two-state system, which is denoted by physicists as a mixture state of “heads” and “tails”. Flipping and non-flipping correspond respectively to the Pauli matrixand identity matrix. A natural question then arises of what happens if the classical coin is replaced by a quantum spin.
      I found that the answer is very non-trivial: a probability distribution over the strategy space, which is the description of a general strategy in classical game theory, is no longer capable of describing games with quantum objects. A density matrix over a basis of the strategy space has to be used. The same transition happens from Classical Mechanics to Quantum Mechanics. A probability distribution is replaced by a density matrix, which allows superpositions while the former allows only probability summations.

    • Quantum Foundation[18]

      Partially inspired by the above work on quantum game theory, I was motivated to study the difference between a probability distribution and a density matrix. Can the former be converted to the later equivalently or vice versa? Luckily I found that the same question has been asked and investigated by physicists on the question of validity of hidden variable theory. In a hidden variable theory, there is no superposition principle, but classical probability summations are allowed. In a sense, the hidden variable theory is searching for a map from a density matrix to a classical probability distribution.

      On one hand there is a theorem stating that all convex theories, which includes quantum mechanics, can be embedded into a classical probability theory with constraints (see for example, A. S. Holevo, Probabilistic and Statistical Aspects of Quantum Theory). On the other hand, Bell’s inequality rules out all local hidden variable theories. The constrained classical theory has to be non-local. Of course many believe that physical theory should be local, but some are still willing to sacrifice locality. I investigated the question of what beyond locality one has to give up in order to have such a classical theory for quantum systems. I found there are many other unacceptable features of the classical theory by explicitly constructing such a theory for systems of one spin half and two spin halfs. Those unwanted features make the theory even harder to understand than the usual quantum mechanics.

  • List of publications