送交者: mangolasi 于 2006-1-26, 10:29:23:
回答: I am confused on why do you think I am arrogant. 由 mangolasi 于 2006-1-26, 09:41:33:
to put it mildly.
"如果我们无法判断一种神秘事件的真假，但又必须做出选择，科学的方法是
考察相关事件的先验概率和后验概率，并使用最大似然准则进行推断。例如以前
发生相关事件多起，其中绝大多数被证明是假的，那么我认定这个新的神秘事件
是假的，这样的推断是合理可靠的。
如果有人反问我“投掷硬币时，连续6次是正面，按你的说法第7次是正面的
可能性大喽，可正确答案是正面的可能性还是50％”。说这种话的人是了解一点
概率论的初步知识，但还不懂什么是概率分布，也没有学过后续的《参数估计》
等课程，让我用几句话讲明白是不现实的，呵呵"
From here, I guess he said "prior and posterior" (先验概率和后验概率), and I suppose his argument is more or less a Bayesian framework (though the maximum likelihood is not that essential in Bayesian : likelihood yes, but maximize--doubt).
Now, Mr./Ms. Bayesian can not answer the guy's question (which can be interpret in different ways, in which you can say right/wrong/senseless accordingly) in one line: "连续6次是正面" means the coin is probably biased (in a Bayesian's point of view), and of course "第7次是正面的
可能性大" rather than 50-50. Instead, this Mr./Ms. Bayesian had to tease the other guy for not having taken the advanced courses (had you taken every advanced courses in Economics, biology, Space Engineering, Nuclear Physics, History etc etc SIMULTANEOUSLY, Mr./Ms. Bayesian?) Not having taken an advanced course is not a shame, and the REAL expert with aim to extend his/her own expertise to the layman need to explain problem in a neat and understandable way, rather than prescribing specialized reading list. Not to mention I had the impression that this guy just throwing out some terminology rather than really knowing something.
So patronizing.
More comment (non-technical) to be continued...
I am not a Bayesian, but I think it's a good approach for me to have action on the new arrival: my last few sampling of it had the impression of very low quality, it is time to believe it is really of low quality.