T********i
发帖数: 2416 | 1 他只不过说出了一个真理而已。搞电脑和数学之类的，90%以上都是吃屎的。没有了他
们，生产率，
生产周期，以及质量都成倍增长。也就是说，绝大多数team，干掉至少50%，绝对没有
任何不良影响。 |
|
|
P***t
发帖数: 1006 | 3 Who the hell is this guy? Is his stuff still used in Grok? |
|
M*****a
发帖数: 2054 | 4 他只是看不惯忽悠而已
绝大多数公司，忽悠实在太多。。。水平不够就很容易被忽悠
英文
我可
耻辱 |
|
h*i
发帖数: 3446 | 5 我看了他在github的code, 他的编程水平比这儿绝大多数人高出一个数量级是保守的说
法。 |
|
h*i
发帖数: 3446 | 6 计算机系的各个领域，我唯一佩服的就是做程序语言研究的。这是计算机专业最核心的
东西，是计算机系唯一的别的专业的人搞不了的领域。别的领域，都是外专业的人可以
作得很好的。
认识不到上述观点的的计算机科班出身的人，都算还没入计算机科学的门。呵呵。 |
|
h*i
发帖数: 3446 | 7 这样的人自己创不了业，太注重技术了。创业成功和技术无关。 |
|
z***t
发帖数: 2374 | 8 Do you mean the programming language research?
Do you really understand what is this research area?
Many math people can do such topics very well
E.g. the lambda calculus was introduced 1930 before the computer appears |
|
h*i
发帖数: 3446 | 9 Not really. Programming language research may USE lambda calculus, but
lambda calculus is NOT a programming language.
Math people cannot do programming language research, but they can do pretty
well in complexity theory, machine learning, and any other "theory" fields
of computer science. Computer science theory people here I work all consider
themselves mathematicians, but NONE of them can do programming language
research. |
|
h****r
发帖数: 2056 | 10 Do you know the reason why they can not do programming language research?
pretty
consider |
|
h*i
发帖数: 3446 | 11 They don't have the training, which unfortunately is only offered in a small
subset of computer science departments.
I am glad Mr. Wang finally found the place where he can get good training in
programming language research. |
|
h****r
发帖数: 2056 | 12 Since when a training is so essential for a soft science (not a rock science
certainly) field? And who trained the trainer then?
small
in |
|
h*i
发帖数: 3446 | 13 This Steve guy happens to be one of the gods in the programming world today
:-).
With such a recommendation, Mr. Wang can easily get a programming job in any
IT company. |
|
h*i
发帖数: 3446 | 14 By training, I mean the freedom, the time and the effort spent actually
implementing all kinds of novel ideas of programming languages.
For that, you don't need a trainer, but you do need such a luxury, which I
simply call "the training".
science |
|
z***t
发帖数: 2374 | 15 I said "E.g.". You extend it to equivalent by yourself.
PL is Logic research, which was a branch in math and merged to CS.
Please read papers from POPL to understand PL.
Some researchers work on both PL and system, and publish papers in applied
conference PLDI.
Computer science theory people are mathematicians, including PL theory
people.
http://www.cs.princeton.edu/~appel/papers/science.pdf
pretty
consider |
|
h*i
发帖数: 3446 | 16 I don't have very high opinions on people who do pure theory in computer
science. These people tend to be 2nd or 3rd rate mathematicians who cannot
survive in pure mathematics. In computer science, they tend to attack small
problems and do incremental work. I don't have too much respect to these
people. Sorry.
Computer *science*, by definition, is empirical. |
|
h****r
发帖数: 2056 | 17 Most people would probably respect someone dares to say "I don't know" more. |
|
z***t
发帖数: 2374 | 18 PL is a pretty theoretical research area.
small |
|
s******g
发帖数: 3841 | 19 如果计算机科学家做的叫incremental的话，那么最近几十年数学家做的就是灰尘的
我怀疑你是否真的知道数学家在做什么，理论计算机科学家在做什么，很大可能你两者
都不了解
small |
|
h*i
发帖数: 3446 | 20 What do you mean? If you think you know something I don't know, just point
it out. Enlighten us.
BTW, you need to learn the habit of making yourself understood instead of
resorting to some vague suggestions. That doesn't work so well in the
English speaking world (since you obviously want to talk in English).
more. |
|
|
h*i
发帖数: 3446 | 22 数学家在做什么我不太了解，不认识什么数学家。
但理论计算机科学家在做什么我很清楚，认识的人很多，包括大牛，基本上符合我的描
述，没有冤枉他们。 |
|
h*i
发帖数: 3446 | 23 I've seen this cute article before. What a lame excuse!
It's not like we don't have an agreed-upon definition of science. For most
people, the distinction between science and mathematics are pretty clear,
there is no need to come up with such an indirect measure.
That article only serves to show the kind of smallness and cynicism popular
among those "theoretical computer scientist who think they are mathematician
". I honestly don't like them.
PL researcher who don't implement PLs are just fakes... 阅读全帖 |
|
h****r
发帖数: 2056 | 24 honesty. Most important, be honest to ourselves.
Can it be more clear? |
|
h*i
发帖数: 3446 | 25 Yes. It can be.
No, you are not clear at all.
★ 发自iPhone App: ChineseWeb 7.3.1 |
|
h*i
发帖数: 3446 | 26 No, it's not clear at all. "honesty. Most important, be honest to ourselves.
" What does that mean?
Are you trying to say that I am not honest? If so, you have the burden to
point out where I am not honest. Apparently you are not able.
In an English speaking context, yours doesn't work at all, because the only
thing we are clear so far is that you are not happy about what I am saying,
but we don't know why.
This might work in Chinese society, where I might feel obligated to figure
it out on my o... 阅读全帖 |
|
G*****7
发帖数: 1759 | 27 elaborate please.
what's his key strength apart from frequent use of design patterns? |
|
D***h
发帖数: 78 | 28 You are digging a hole, dude.
ourselves.
only
,
forum
like |
|
w***g
发帖数: 5958 | 29 我觉得把公司内部的事情这么具体地写在blog上是非常不professional的做法。 |
|
C*******s
发帖数: 231 | 30 就事论事，他说的那些google的事还是挺靠谱的。google像个巨大的黑洞，骗了一堆
super star进去，然后从此无影无踪。 |
|
A***g
发帖数: 1816 | 31 好象这个王垠很多年前写过一篇骂中国计算机教育的文章，然后到美国来念书。 |
|
z***e
发帖数: 5393 | 32 what?你是没上过compiler这个课还是咋回事？ |
|
n*y
发帖数: 26 | 33 看了他的sina博客。有一篇讲语言的，很靠谱。不论对错，他思考的问题是大多数码工
一辈子也不会想的。 |
|
D***h
发帖数: 78 | 34 神经病思考的问题, 别说是码工, 大多数人一辈子也不会想的. |
|
d***q
发帖数: 1119 | 35
我想不是什么码工都需要想这种问题。如果都需要，这才是有问题。 |
|
M*****a
发帖数: 2054 | 36 大把
这东西只要会c懂kernel就行了
心的
可以 |
|
a*****e
发帖数: 1700 | 37 我挺你。PL 理论不能简单归结为数学分支，尽管它和 logic, category theory
等联系紧密。
王垠本人，我今年 POPL 的时候见过，外表挺谦虚的小伙儿，虽然骨子里肯定是骄
傲的。他的其它博文有些看法我不能认同，但是这篇讲 google 的写得不错，和我
的经验类似。 |
|
|
n*********2
发帖数: 357 | 39 You have a very interesting perspective.
Now, to explain your point, could you please start with a _concrete_ book or
technical paper? A concrete example will be much more convincing.
small |
|
|
n*********2
发帖数: 357 | 41 Most people are NOT Chomsky. There is only one Chomsky in the world and
anybody else who claims Chomsky is simply false.
HCI indeed has a point. Most CS people are doing minor/incremental
improvements. Doing minor/incremental improvements in practice-oriented
systems & programming languages is very likely more meaningful than doing
minor/incremental improvement in theory.
The real questions is how to do meaningful minor/incremental improvements in
system and programming language. That is the rea... 阅读全帖 |
|
z***t
发帖数: 2374 | 42 If the fundamental design is wrong, the practical implementations have no
meaning.
Both theory and system are important. These points have been recognized by
CS people for long long time.
in
example |
|
n*********2
发帖数: 357 | 43 I guess hci's point is that not all CS problems have the same level of
impact. A minor/incremental improvement in those non-PL/systems areas
probably has no impact on our daily life. In contrast, a minor/incremental
improvement in programming languages (PL) and systems is more likely to have
a relatively bigger impact.
So, how to do good things in programming language and system? |
|
|
y**e
发帖数: 49 | 45 记得以前碰到过一个软件学院的学生问：课程为什么没有VC
today
any |
|
z***e
发帖数: 5393 | 46 算法和AI都是数学的子领域，看不出来你这句话和你的结论有任何联系。
计算机软件和PL本身都是一种针对性很强的实用工具，PL实现本身大量考虑的是数据结
构和具体的一些细节。至于 research topic，坦白说全是些唬人的东西。 |
|
a*****e
发帖数: 1700 | 47 你原帖的意思是学了 compiler 原理就了解 PL 领域了
然后我指出这是以偏概全
你现在又说 PL 领域（除开 compiler）其它的 research topic 都是唬人的
这就可以直接看出来你没有正规学习过，所以对此有误解
比如在 PL 领域占很大比重的 type theory，就非常实用
PL research 的精髓是 abstraction，这在 CS 领域是独树一帜的。
不理解这个，其它都无从谈起。 |
|
z***e
发帖数: 5393 | 48 我意思是PL的概念都不难理解，我不是搞这个领域的，也就是上课看了几本书而已，当
然不敢说我对此很了解。但是就我各人经验，programming language的各种概念很多，
但是不难理解，难处在于具体有效的实现上。一个大家都懂的garbage collection，各
种实现都有，难处还是在具体的performance和针对场合，而不是该概念本身多fancy.
还有就是各种理论的实用性都是正反两面的，反正在我看的几本书里面对比一些语言的
设计和背后思维的时候，基本都是各有各的目的和道理。
你指的abstraction我不是很清楚具体是什么，是指programming language research中
比较抽象化的那些概念么？那么我认为这还是落入“凡是越抽象越用数学表现的，就越
高级/深奥/etc."这样的无聊论点，那大家别学cs了，全部去做数学，数学还必须是理
论数学才高级。。。 |
|
a*****e
发帖数: 1700 | 49 概念抽象并非是说它们都是所谓“高级”可以拿来炫耀的东西。
相反，很多时候它们是很初级很直白的，很容易被忽视掉的。
但正是因为抽象，所以无法用举例的方式来具象描述，必须使
用数学工具来精确表达。形式语义 (formal semantics) 类型
系统 (type theory) 等往往看起来充斥着各种奇怪的符号难以
理解，但通过认真学习则不难发现它们所代表的涵义。多数行
外人只要看到没有接触过的数学符号和表达式就产生畏难甚至
抵触情绪，才是最要不得的。
前些时 Philip Walder 在 QCon 给了一个演讲：Faith, Evolution,
and Programming Languages.
http://www.infoq.com/presentations/Faith-Evolution-Programming-
里面反复提到，这个领域很多伟大的发现，实际上都是 50 年
后的再发现，而第一次发现通常都是数学家或逻辑学家完成的。
现在看来，这些又都是多么直白浅显的概念。
Philip 是一个很棒的老师和演讲家，你有时间可以看一下，他
所谈的那些东西，就是我提到的 abs... 阅读全帖 |
|
c*****e
发帖数: 3226 | 50 大牛。 关于测试绝对是真的，其实谷歌不需要英雄，更需要的是螺丝钉。 |
|
