Sunday 28 February 2010

sake

went to blue fin yesterday for dinner and tasted sake for the first time. there is a sampler special that you can choose 3 kinds of sake. always thought sake is very strong, like the chinese rice wine (baijiu), turns out it's usually 18%-20% alcohol. comparing to the baijiu which is 40-60% alcohol, sake is like nothing. turns out i'm confused between sake and shōchū. shōchū is also a japanese alcohol drink which is 25% alcohol. shōchū may be similar to the korean soju.

can't remember the names of the 3 types of sake we ordered. one was nigori which appears cloudy because it's unfiltered. contrary to its cloudy appearance, it tastes clear and refreshing. no strong taste. not very strong in alcohol either. can taste the the sweetness of the fermented rice as well. The second one is a dry one. don't quite like it. only tasted alcohol. the third one is koshu (古酒), "aged sake." it has an orange coloring and an interesting and distinctive taste.

Friday 5 February 2010

一年

 Those of us who fell for the professor cast ourselves as Cinderella intellectuals, waiting for the phrase - rather than the slipper - that fit us perfectly. We waited, at fourteen, at nineteen, at twenty-five or even thirty-five, for the figure who would see what was hidden and special and glorious in us, who would love us for our smart selves alone and not our yellow hair - or so we thought - who would play out every fantasy of sibling rivalry we ever had and choose us from among all our peers for attention. -- regina barreca "contraband appetites"
一年前,当我走进6.824的教室,一切没有什么不寻常。这么多年的求学生涯,上过的课无数,而这门课也只是其中之一。虽然很多人都说这是系统课里最难、作业量最大的课,我还是很坚定的选择了它,因为我觉得分布式系统是现在和今后很重要的领域,而且我也很喜欢写程序,特别是用c++。

教室很小,而第一堂课来听的学生却不少,都坐到门外了。走进来一位高高瘦瘦的教授,做了自我介绍,然后就开始讲分布式系统的一些大的概念和课程安排。一切还是那么寻常,波澜不惊。

课业确实很重,每堂课前都要读一篇paper,还要回答一个问题,并且要交的。每周都一个lab要完成,要看要写不少程序。学生也渐渐少了, 三十来个吧。我不算是个刻苦的学生,况且还有研究项目要做。 我喜欢做lab, 每次新的lab一出,我就花上5,6个小时做完,bonus的部分也做。但是读paper就没什么耐心,不是整篇读完,只是读到能回答问题就行了。我自以为这样很有效率,反正看了重点就行了,等以后考试前再仔细看一边,那时更清楚什么是重点。可是这样在上课时,教授提的关于paper上的问题,我并不都能回答上来。现在想来有点后悔,如果我更认真一点,上课更积极发言,也许能给他留下更深刻的印象。可那时我还没有要留下好印象的冲动。就像以前任何一门课一样我只是安静的听课,有时即使知道答案,也懒得回答。不过我还是回答了几次问题的,有些一反常态,可能是因为人比较少,所以我也有胆量了。回答问题的感觉真的很好,我似乎能感到教授投来赞许的目光。可是有时我就是提不起劲来回答,或是怕回答错了。但后来发现其实我知道答案后,就会非常懊悔。特别是临近学期末的时候,那时我真的很想给教授留下好印象。可是越是这样想,越是心有顾忌,越是不敢发言。每次这样,下课后我就情绪低落,甚至厌恶自己的怯懦和内向。

学期过半时,有一天课后,我问助教一个关于lab的问题, - 我一般总是先问助教, 因为对教授有种敬畏感, - 我说觉得一个squence number没必要。教授碰巧也在旁边,听后又在黑板上画了可能出现的communication sequence and failure mode, 觉得不用sequence number是可以的,"if you are being  smart and careful in other places",(但其实sequence number对于以后的部 分还是有用的, 如server and client failure),但他还是对于我的提问给予了肯定。至今我还能在脑海中回放他笑着对我说"You stunned us." 那一瞬间,我觉得有些惊愕,似乎"stunned us"有些夸大了,可还是满心欢喜,看着那满眼的笑意,他那炯炯的眼神直透我心底。老师,你太强大了,一个不经意的赞许把我彻底击败。

期中考试考得还不错,lab也都是满分,我觉得我对系统的兴趣确实很大,就像以前在ubc上的computer architecture, 我也花了很多精力去做branch prediction competition, 最后得了第一,那时的兴奋也是难以言表,老师也对我特别欣赏,那也是个很帅的教授。正如regina barreca在contraband appetites中提到的:
the confusion between loving the profession and loving the professor
我有时真的不知道我是因为喜欢系统这个领域才喜欢教授呢,还是反之。或许两者都有吧。

快到学期末时,我发觉上课的时候我越来越不敢看教授了,只是盯着黑板或看paper或写笔记。因为我害怕,害怕与那智慧的眼神交汇,害怕电流从视神经直击心脏,使之飞速跳动,害怕我会脸红,因为我无法掩饰我的爱慕。

最后一堂课是project presentation, 由于电脑与投影仪不兼容,准备好的demo也不能演示。虽然其他和我用一样mac book pro的同学都不能演示,我还是很郁闷,辛辛苦苦准备的演示,就白忙活了,而且也没有机会impress教授了。整整一天心情都很沮丧,直到教授发来email问大家能不能把report放在网上,在回复的email中,我说我对没有做第二手准备感到抱歉,没想到教授马上回复说是他没有测试好投影仪,还说我口头的解释也很好。看到这,我心花怒放,低落的心情烟消云散。

教授还给每个同学发email, 感谢我们上课的参与。我想他是给每个同学都发的,虽然email里写着我的名字,当然自动的生成不一样的名字也不是什么难事,我还是感到很窝心。教授真是细心啊。

Wednesday 3 February 2010

ai and safety

there are more and more electronics and intelligent devices installed in the car. however as the safety issue is really crucial for the car, we need really robust artificial intelligent systems in the car if we'll ever have one. i'm very pro-technology. i think the world will become more and more digitalized in very fabric of our life. intelligent car should be the trend, however there is still a long way to go from the current state of ai development. as a first step, we can have intelligent system as an aid that augments human's ability, but it cannot replace human as the main controller and there should be easy ways to over ride the automatic systems.
http://tinyurl.com/ye6qvsn