Siri有点像天气。人人嘴上都爱抱怨它，但只有投资银行Piper Jaffray的吉恩•蒙斯特真正采取了行动。

    今年六月，蒙斯特测试了iOS 5版本的苹果（Apple）声控个人助理Siri。他向Siri问了1,600个问题，其中有800个问题是在美国明尼阿波里斯市的街头问的，而另外800个则是在一间安静的房间中问的。

    结果显示Siri尚有很大的改进空间。在最理想的情况下（即没有街道噪音），Siri正确理解了89％的问题，正确回答了68％的问题。（一个很典型的例子是，Siri听见问：下次哈雷彗星出现是什么时候？它的回答是：“您没有安排与哈利会面。”

    上周四，明斯特报道了在安静的房间（没有街道噪音）里对“改进后的”iOS 6版Siri再度进行测试的结果。这一次，Siri理解了91％的问题，并正确回答了77％的问题。

    按照学校的打分标准，Siri的得分从D提高到了C。

    （明斯特没有解释他为何将Siri在六月份的正确率从68％调高到了76％，从而使得iOS 6版Siri的进步看起来小了很多。或许他现在是按正态曲线评分了。）

    Siri is a little like the weather. Everybody complains about it, but only Piper Jaffray's Gene Munster seems to do anything.

    In June he put the iOS 5 version of Apple's (AAPL) voice-activated personal assistant to the test -- asking her 1,600 questions,800 in the streets of Minneapolis and 800 in a quiet room.

    The results left much to be desired. In the best of circumstances (i.e. no street noise), Siri correctly interpreted 89% of the questions and correctly answered 68%. (Typical exchange, as Siri heard it: When is the next Haley's comment? "You have no meetings matching Haley's".)

    On Thursday, Munster reported the results of a quiet-room (no street noise) re-test of the "improved" iOS 6 version. This time Siri understood 91% of the questions and correctly answered 77%.

    In school-grade terms, she's gone from a D to a C.

    (For reasons Munster doesn't explain, he's raised Siri's June score from 68% to 76%, making the improvement look more modest than it was. Perhaps he's now grading on a curve.)