|||
经典句子:
now if you think this is just a problem for psychology, neuroscience, or medicine, consider the pentaquark, an exotic particle made up of five quarks,as opposed to the regular three for protons or neutrons. particle physics employs particularly stringent requirements for statistical significance referred to as five sigma or one chance in 3.5 million of getting a false positive. but in 2002 a japanese experiment found evidence for the theta plus pentaquark and in the two years that followed, eleven other independent experiments then looked for and found evidence of that same pentaquark with very high levels of statistical significance. from July 2003 to may 2004, a theoretical paper on pentaquarks was published on average every other day, but alas it was a false discovery. further experimental attempts to confirm that theta plus pentaquark using greater statistical power failed to find any trace of its existence
如果您认为这只是心理学、神经科学或医学的问题,请考虑五夸克是由五个夸克组成的奇异粒子,而不是质子的常规三个夸克或中子。粒子物理学对统计显着性采用了特别严格的要求,即 5 西格玛或 350 万分之一的概率出现假阳性,但在 2002 年,日本的一项实验发现了 theta+ 五夸克的证据,并在随后的两年里进行了其他 11 个独立的实验。然后通过实验寻找并发现了具有非常高统计显着性的同样的五夸克的证据,从 2003 年 7 月到 2004 年 5 月,平均每隔一天就会发表一篇关于五夸克的理论论文,但可惜的是,这是一个错误的发现,进一步的使用更严格的统计,实验证实theta+ 五夸克,未能找到其存在的任何痕迹,
the problem was those first scientists weren't blind to the data. they knew how the numbers were generated and what answer they expected to get, and the way the data was cut and analyzed or P hacked produced the false finding. now most scientists aren't P hacking maliciously. there are legitimate decisions to be made about how to collect analyze and report data, and these decisions impact on the statistical significance of results
那些第一批科学家并不是对数据视而不见,他们太知道数字是如何生成的,他们期望得到什么答案,以及数据被切割和分析的方式 或 要得到所期望的P值而产生了错误的发现,大多数科学家都没有恶意操纵P值,有关于如何收集分析和报告数据才能做出合法的判断,这些判据影响统计显着性的结果,
scientists have huge incentives to publish papers in fact their careers depend on it
科学家有巨大的动力去发表论文,事实上,他们的职业生涯取决于发表论文
the thing I find most striking about the reproducibility crisis in science is not the prevalence of incorrect information in published scientific journals. after all, getting to the truth we know is hard and mathematically not everything that is published can be correct. what gets me is the thought that even trying our best to figure out what's true using our most sophisticated and rigorous mathematical tools, peer review, and the standards of practice, we still get it wrong so often. so how frequently do we delude ourselves when we're not using the scientific method. as flawed as our science may be, it is far and away more reliable than any other way of knowing that we have
这是我发现最引人注目的可重复性 科学危机并不是已发表的科学期刊中错误信息的普遍存在,毕竟要了解我们知道的真相是很困难的,而且从数学上讲,并不是所有已发表的内容都是正确的,让我感到困惑的是,即使我们尽了最大努力去弄清楚什么是真实的 使用我们最复杂和最严格的数学工具同行评审和实践标准,我们仍然经常出错,所以当我们不使用科学方法时,我们经常欺骗自己,我们更能肯定的是科学报导可能有缺陷.
视频文件:
Is Most Published Research Wrong1.mp4
Is Most Published Research Wrong.docx
出处:
https://www.youtube.com/watch?v=42QuXLucH3Q
Is Most Published Research Wrong?
More information on this topic: http://wke.lt/w/s/z0wmO
https://wakelet.com/wake/35a80e1c-0033-4db1-a5c1-28337f57d75b
The Preregistration Challenge: https://cos.io/prereg/
Resources used in the making of this video: Why Most Published Research Findings Are False: http://journals.plos.org/plosmedicine...
https://journals.plos.org/plosmedicine/article?id=10.1371%2Fjournal.pmed.0020124
Why Most Published Research Findings Are False
参考文献:
https://www.sohu.com/a/423577113_788170
https://zhuanlan.zhihu.com/p/389134254
诺奖得主本庶佑:CNS这些顶刊观点有九成不正确,不要盲从迷信,搞科研做到六个C更重要
https://www.163.com/dy/article/FEE1RTDF05419EOY.html
历史学家李伯重:无论何种“学术垃圾”,都是有害的,而且都是公害
扩展听力资料:
https://www.youtube.com/watch?v=bVG2OQp6jEQ
This is How Easy It Is to Lie With Statistics
https://www.youtube.com/watch?v=IwkPEAY4D1c
Your Favorite Research Is (Probably) Wrong: The Replicability Crisis
https://www.youtube.com/watch?v=VcgO2v3JjCU
Is Science Reliable?
https://www.youtube.com/watch?v=5xgbRxA_7DI
ACADEMIA IS BROKEN! The publishing scandal happening right now
https://www.youtube.com/watch?v=rPkBGe8XTAk
6 Publication Bias: Why Most Published Research Findings are False: Part I (FR)
https://www.youtube.com/watch?v=wM0vXVclQZg
"Why Most Published Research Findings are False" Part I
https://www.youtube.com/watch?v=NWjN67vqXOo
"Why Most Published Research Findings are False" Part II
https://www.youtube.com/watch?v=_GAeTwLB24c
"Why Most Published Research Findings are False" Part III
https://www.youtube.com/watch?v=dvk2PQNcg8w
Post-Truth: Why Facts Don't Matter Anymore
https://www.youtube.com/watch?v=7JwElC2pkkM
Fraud & Deception in Science | Elisabeth Bik, PhD
https://www.youtube.com/watch?v=QGWeVbYduOI
The scandal that shook psychology to its core
英汉对照(机器翻译)
in 2011 an article was published in the
2011年,
reputable Journal of Personality and
著名的《人格与
Social Psychology it was called feeling
社会心理学杂志》上发表了一篇文章,题为“感受
the future experimental evidence for
未来对
anomalous retroactive influences on
cognition and effect or in other words
认知和效果的异常追溯影响的实验证据”,换句话说,
proof that people can see into the
证明人们可以预见
future
未来,
the paper reported on nine experiments
该论文报告了九个实验
in one participants were shown two
一名参与者
curtains on a computer screen and asked
在电脑屏幕上看到两块窗帘,并被
to predict which one had an image behind
要求预测哪一块背后有图像,
it the other just covered a blank wall
另一名
once the participant made their
参与者在参与者做出选择后,将
selection the computer randomly
positioned an image behind one of the
图像随机放置在其中一个
curtains then the selected curtain was
窗帘后面,然后将所选窗帘覆盖在空白墙上。
pulled back to show either the image or
拉回以显示图像或
at the blank wall the images were
在空白墙上,
randomly selected from one of three
从三个类别之一中随机选择图像,
categories neutral negative or erotic if
中性负面或色情如果
participants selected the curtain
参与者选择
covering the image this was considered a
覆盖图像的窗帘,那么
hit now with there being two curtains
现在有两个窗帘
and the images positioned randomly
并且图像随机放置,这被认为是热门
behind one of them you would expect the
其中一张图片的
hit rate to be about 50% and that is
命中率约为 50%,这正是
exactly what the researchers found at
研究人员发现的,
least for negative and neutral images
至少对于负面和中性图像而言,
however for erotic images the hit rate
然而对于色情图像,命中
was 53% does that mean that we can see
率为 53%,这是否意味着我们可以预见
into the future is that a slight
未来 是一个轻微的
deviation significant well to assess
偏差,可以很好地评估显
significance scientists usually turn to
着性,科学家通常会转向
p-values a statistic that tells you how
p 值,这是一种统计数据,可以告诉您
likely a result at least this extreme is
if the null hypothesis is true in this
如果零假设为真,则至少出现这种极端结果的可能性有多大,在这种
case the null hypothesis would just be
情况下,零假设只是
that people couldn't actually see into
人们不能' 没有真正预见到
the future and the 53% result was due to
未来,53% 的结果是由于
Lucky guesses for this study the p-value
本研究的幸运猜测,p 值为
was 0.01 meaning there was just a 1%
0.01,这意味着只有 1% 的
chance of getting a hit rate of 53
机会
percent or higher from simple luck
从简单的运气中获得 53% 或更高的命中率
p-value is less than 0.05 are generally
p 值为 小于 0.05 通常被
considered significant and worthy of
认为是重要且值得
publication but you might want to use a
发表的,但在
higher bar before you accept that humans
您接受人类
can accurately perceive the future and
可以准确感知未来并
say invite the study's author on your
邀请该研究的作者参加您的
news program but hey it's your choice
新闻节目之前,您可能需要使用更高的标准,但嘿,毕竟这是您的选择
after all the point O 5 threshold was
5 阈值是
arbitrarily selected by Ronald Fisher in
Ronald Fisher 在
a book he published in 1925 but this
1925 年出版的一本书中任意选择的,但这
raises the question how much of the
提出了一个问题:
published research literature is
已发表的研究文献中有多少
actually false the intuitive answer
实际上是错误的,直观的答案
seems to be 5% I mean if everyone's
似乎是 5% 我的意思是,如果每个人都
using P less than 0.05 as our cutoff for
使用小于 0.05 的 P 作为我们的截止值 出于
statistical significance you would
统计显着性,您会
expect five of every hundred results to
期望每 100 个结果中有 5 个
be false positives but that
是误报,但
unfortunately grossly under
不幸的是,严重
estimates the problem and here's why
低估了问题,这就是为什么
imagine you're a researcher in a field
想象您是某个领域的研究人员,
where there are a thousand hypotheses
currently being investigated let's
目前正在研究 1000 个假设,让我们
assume that 10% of them reflect true
假设 10% 它们反映了真实的
relationships and the rest are false but
关系,其余的都是错误的,但
no one of course knows which are which
当然没有人知道哪个是哪个,这就是
that's the whole point of doing the
research now assuming the experiments
现在进行研究的全部意义,假设实验
are pretty well designed they should
设计得很好,他们应该
correctly identify around say 80 of the
正确地识别出
hundred true relationships this is known
一百种真实关系中的大约 80 种,这被称为
as a statistical power of 80% so twenty
统计功效为 80%,因此 20 个
results are false negatives perhaps the
结果为假阴性,可能是
sample size was too small or the
样本量太小,或者
measurements were not sensitive enough
测量结果不够敏感。
now consider that from those 900 false
现在考虑一下,从
hypotheses using a p-value of 0.05 45
使用 p 值为 0.05 的 900 个错误假设中,45 个
false hypotheses will be incorrectly
错误假设将被错误地
considered true as for the rest they
视为正确,如下所示 其余的
will be correctly identified as false
将被正确识别为虚假结果,
but most journals rarely published no
但大多数期刊很少发表任何
results they make up just 10 to 30
结果,它们仅占论文的 10% 到 30
percent of papers depending on the field
%,具体取决于领域,
which means that the papers that
这意味着
eventually get published will include 80
最终发表的论文将包括 80 个
true positive results 45 false positive
真阳性结果 45 个假阳性
results and maybe 20 true negative
结果 也许 20 个真实的阴性
results nearly a third of published
结果,即使系统正常工作,近三分之一的已发表
results will be wrong even with the
结果也会是错误的,
system working normally things get even
worse if studies are underpowered and
如果研究动力不足,并且
analysis shows they typically are if
分析显示,如果
there is a higher ratio of false to true
hypotheses being tested or if the
正在测试的假说与真假的比例较高,或者如果研究的动力不足,情况会变得更糟,情况通常会更糟
researchers are biased all of this was
研究人员
pointed out in a 2005 paper entitled why
在 2005 年发表的一篇题为《为什么
most published research is false so
大多数发表的研究都是错误的》的论文中指出了所有这一切,因此
recently researchers in a number of
最近许多领域的研究人员
fields have attempted to quantify the
试图通过复制过去
problem by replicating some prominent
一些突出的
past results the reproducibility project
结果来量化这个问题。可重复性项目
repeated 100 psychology studies but
重复了 100 项心理学研究,但
found only 36% had a statistically
发现 第二次只有 36% 的结果具有统计
significant result the second time
显着性,
around and the strength of measured
测量到的
relationships were on average half those
关系强度平均为
of the original studies an attempted
原始研究的一半,尝试
verification of 53 studies considered
验证 53 项被认为具有
the landmarks and the basic science of
里程碑意义的研究,即使密切合作,癌症基础科学也只能
cancer only managed to reproduce six
重现 6 项研究
even working closely with the original
对于原始
study's authors these results are even
研究的作者来说,这些结果
worse than I just calculated the reason
比我刚刚计算的结果还要糟糕,
for this is nicely illustrated by a 2015
2015 年的一项研究很好地说明了其原因,该
study showing that eating a bar of
研究表明每天吃一块
chocolate every day can help you lose
巧克力可以帮助您
weight faster in this case the
更快地减肥,在这种情况下,
participants were randomly allocated to
参与者被随机分配到
one of three treatment groups one went
一个 三个治疗组中,一组
on a low-carb diet
采用低碳水化合物饮食,
another went on the same low-carb
另一组采用相同的低碳水化合物饮食,
plus a one point five ounce bar of
chocolate per day and the third group
每天加一点五盎司的巧克力,第三组
was the control instructed just to
是对照组,指示在治疗
maintain their regular eating habits at
the end of three weeks the control group
结束时保持正常的饮食习惯。 三周后,对照组的
had neither lost nor gained weight
体重既没有减轻也没有增加,
but both low-carb groups have lost an
但两个低碳水化合物组
average of five pounds per person the
每人平均减轻了五磅,但
group that ate chocolate however lost
吃巧克力的组比
weight ten percent faster than the non
不吃巧克力的组减轻了百分之十的速度,
chocolate eaters the finding was
这一发现具有
statistically significant with a p-value
统计学意义,p -值
less than 0.05 as you might expect this
小于 0.05,正如您所预料的那样,这条
news spread like wildfire to the front
消息像野火一样传到了欧洲
page of build the most widely circulated
发行量最广的
daily newspaper in Europe and then to
日报的头版,然后传到了《
The Daily Star the Irish Examiner -
每日星报》、《爱尔兰观察家报》、《赫芬
Huffington Post and even shape magazine
顿邮报》,甚至是《形状》杂志,
unfortunately the whole thing had been
不幸的是,整个事情都被报道了
faked
kind of I mean researchers did perform
我的意思是研究人员
the experiment exactly as they described
确实按照他们描述的那样进行了实验,
but they intentionally designed it to
但他们故意设计它以
increase the likelihood of false
增加误报的可能性,
positives the sample size was incredibly
样本量非常
small just five people per treatment
小,每个治疗组只有 5 个人,
group and for each person 18 different
每个人跟踪了 18 种不同的测量
measurements were tracked including
结果,包括
weight cholesterol sodium blood protein
体重胆固醇 钠、血液蛋白质
levels sleep quality well-being and so
水平、睡眠质量、幸福感等等,
on so if weight loss didn't show a
所以如果体重减轻没有显示出显
significant difference there were plenty
着差异,那么可能还有
of other factors that might have so the
很多其他因素,所以
headline could have been chocolate
标题可能是巧克力
lowers cholesterol or increases sleep
降低胆固醇或提高睡眠
quality or something the the point is a
质量或其他重点
p-value is only really valid for a
p 值仅对
single measure once you're comparing a
单个测量值真正有效,一旦您比较了
whole slew of variables the probability
一大堆变量,
that at least one of them gives you a
其中至少一个变量给出
false positive goes way up and this is
误报的概率就会大幅上升,这就是
known as P hacking researchers can
所谓的“P黑客”研究人员可以
make a lot of decisions about their
做一个 有关分析的许多决策
analysis that can decrease the p-value
可能会降低 p 值,
for example let's say you analyze your
例如,假设您分析
data and you find it nearly reaches
数据,发现它几乎达到
statistical significance so you decide
统计显着性,因此您决定
to collect just a few more data points
收集更多数据点,
to be sure
以确定
then if the p-value drops below 0.05 you
p 值是否下降 低于 0.05 时,您
stop collecting data confident that
会停止收集数据,并确信
these additional data points could only
have made the result more significant if
如果确实存在真正的关系,这些额外的数据点只会使结果更加显着,
there were really a true relationship
there but numerical simulations show
但数值模拟表明,
that relationships can cross the
significance threshold by adding more
data points even though a much larger
即使样本更大,通过添加更多数据点,关系也可以跨越显着性阈值
sample would show that there really is
会表明确实
no relationship in fact there are a
没有关系,事实上有
great number of ways to increase the
很多方法可以增加
likelihood of significant results like
显着结果的可能性,例如
having two dependent variables adding
让两个因变量添加
more observations controlling for gender
更多观察结果来控制性别
or dropping one of three conditions
或放弃三个条件之一,将
combining all three of these strategies
所有这三种策略结合
together increase
在一起增加显着结果的可能性
the likelihood of a false positive to
假阳性的可能性
over 60% and that is using P less than
超过 60%,即现在使用小于
0.05 now if you think this is just a
0.05 的 P,如果您认为这只是
problem for psychology neuroscience or
心理学、神经科学或
medicine consider the pentaquark an
医学的问题,请考虑
exotic particle made up of five quarks
五夸克是由五个夸克组成的奇异粒子,
as opposed to the regular three for
而不是质子的常规三个夸克
protons or neutrons particle physics
或中子粒子物理学对
employs particularly stringent
requirements for statistical
统计显着
significance referred to as five sigma
性采用了特别严格的要求,即 5 西格玛
or one chance in 3.5 million of getting
或 350 万分之一的概率出现
a false positive but in 2002 a japanese
假阳性,但在 2002 年,日本的一项
experiment found evidence for the theta
实验发现了 theta
plus pentaquark and in the two years
加五夸克的证据,并在随后的两年里进行了
that followed eleven other independent
其他 11 个独立的实验。
experiments then looked for and found
然后实验寻找并发现了具有
evidence of that same pentaquark with
very high levels of statistical
非常高统计
significance from July 2003 to may 2004
显着性的同一个五夸克的证据,从 2003 年 7 月到 2004 年 5 月,
a theoretical paper on pentaquarks was
published on average every other day but
平均每隔一天就会发表一篇关于五夸克的理论论文,但
alas it was a false discovery further
可惜的是,这是一个错误的发现,进一步的
experimental attempts to confirm that
实验试图证实
theta plus pentaquark using greater
theta 再加上五夸克使用更大的
statistical power failed to find any
统计能力未能找到
trace of its existence the problem was
其存在的任何痕迹,问题是
those first scientists weren't blind to
那些第一批科学家并不是对
the data they knew how the numbers were
数据视而不见,他们知道数字是如何
generated and what answer they expected
生成的,他们期望得到什么答案,
to get and the way the data was cut and
以及数据被切割和
analyzed or P hacked produced the false
分析的方式 或 P 黑客攻击产生了错误的
finding now most scientists aren't P
发现,现在大多数科学家都没有
hacking maliciously there are legitimate
恶意黑客攻击,
decisions to be made about how to
关于如何
collect analyze and report data and
收集分析和报告数据需要做出合法的决定,
these decisions impact on the
这些决定会影响
statistical significance of results for
结果的统计显着性,
example 29 different research groups
例如 29 个不同的研究小组
were given the same data and asked to
被给予 使用相同的数据,并被要求
determine if dark-skinned soccer players
确定深色皮肤的足球运动员
are more likely to be given red cards
是否更有可能被红牌罚下,
using identical data some groups found
一些小组发现
there was no significant effect while
没有显着影响,而
others concluded dark-skinned players
另一些小组则得出结论,深色皮肤的足球运动员
were three times as likely to receive a
收到
red card the point is that data doesn't
红牌的可能性是该点的三倍 数据
speak for itself it must be interpreted
本身并不能说明问题,必须根据
looking at those results it seems that
这些结果进行解释,似乎
dark-skinned players are more likely to
深色皮肤的球员更有可能
get red carded but certainly not three
被红牌罚下,但
times as likely consensus helps in this
在这种情况下,共识的帮助肯定不是三倍,
case but for most results only one
但对于大多数结果来说,只有一项
research group provides the analysis and
研究 小组提供了分析,
therein lies the problem of incentives
其中存在激励问题,
scientists have huge incentives to
科学家有巨大的动力去
publish papers in fact their careers
发表论文,事实上,他们的职业生涯
depend on it as one scientist Brian no
依赖于此,正如美国证券交易委员会(SEC)一位科学家布赖恩(Brian no
SEC puts it there is no cost to getting
SEC)所说,犯错是没有成本的,
things wrong the cost is not getting
代价是不让
them published journals are far more
他们发表期刊。 更有
likely to publish results that reach
可能发表达到
statistical significance so if a method
统计显着性的结果,因此,如果
of data analysis results in a p-value
数据分析方法的 p 值
less than 0.05 then you're likely to go
小于 0.05,那么您很可能会
with that method publications also more
采用该方法
likely if the result is novel and
如果结果新颖且
unexpected this encourages researchers
出乎意料,那么您也更有可能发表论文,这会鼓励研究人员
to investigate more and more unlikely
去研究越来越多不太可能的
hypotheses which further decreases the
假设,这些假设进一步降低了现在测试的
ratio of true to spurious relationships
真实与虚假关系的比率,
that are tested
now what about replication isn't science
复制是不是科学
meant to self-correct by having other
意味着通过让其他
scientists replicate the findings of an
科学家复制
initial discovery in theory yes but in
最初发现的结果来进行自我纠正,理论上是的,但在
practice it's more complicated like take
实践中 更复杂的是,就像
the precognition study from the start of
从本视频开始进行的预知研究一样,
this video three researchers attempted
三名研究人员试图
to replicate one of those experiments
复制其中一个实验,
and what did they find
他们发现了什么,令人惊讶
well surprise surprise the hit rate they
obtained was not significantly different
的是,他们获得的命中率与
from chance when they tried to publish
他们试图
their findings in the same journal as
在 与
the original paper they were rejected
原始论文相同的期刊,他们被拒绝的
the reason the journal refuses to
原因是该期刊拒绝
publish replication studies so if you're
发表复制研究,所以如果你是
a scientist the successful strategy is
一名科学家,成功的策略很
clear and don't even attempt replication
明确,甚至不要尝试复制
studies because few journals will
研究,因为很少有期刊会
publish them and there is a very good
发表它们,而且有一个非常
chance that your results won't be
无论如何,你的结果很可能不会具有
statistically significant anyway in
统计显着性,在
which case instead of being able to
这种情况下,
convince colleagues of the lack of
reproducibility of an effect you will be
你不会被指责为没有正确执行,而不是能够让同事相信效果缺乏可重复性,
accused of just not doing it right so a
因此
far better approach is to test novel and
更好的方法是测试新颖的和
unexpected hypotheses and then P hack
your way to a statistically significant
result now I don't want to be too
我不想
cynical about this because over the past
对此过于愤世嫉俗,因为在过去的
10 years things have started changing
10 年里,事情已经开始变得
for the better
更好,
many scientists acknowledge the problems
许多科学家承认了
I've outlined and are starting to take
我概述的问题,并正在开始 为了采取
steps to correct them there are more
措施纠正这些问题,
large-scale replication studies
undertaken in the last 10 years plus
在过去 10 年里进行了更大规模的复制研究,此外
there's a site retraction watch
还有一个网站撤回观察,
dedicated to publicizing papers that
专门用于公布已
have been withdrawn there are online
撤回的论文,有在线
repositories for unpublished negative
存储库来存储未发表的负面
results and there is a move towards
结果,并且有提交假设的趋势
submitting hypotheses and methods for
peer review before conducting
在进行实验之前进行同行评审的方法,
experiments with the guarantee that
保证只要遵循
research will be published regardless of
results so long as the procedure is
程序,无论结果如何,研究都会被发表,
followed this eliminates publication
这消除了发表
bias promotes higher powered studies and
偏见,促进了更高水平的研究,并
lessens the incentive for pee hacking
减少了小便黑客的动机,
the thing I find most striking about the
这是我发现最引人注目的可
reproducibility crisis in science is not
重复性 科学危机并不是
the prevalence of incorrect information
in published scientific journals after
已发表的科学期刊中错误信息的普遍存在,毕竟要
all getting to the truth we know is hard
了解我们知道的真相是很困难的,而且从
and mathematically not everything that
数学上讲,并不是所有已
is published can be correct what gets me
发表的内容都是正确的,让我感到困惑的是,即使我们
is the thought that even trying our best
尽了最大努力去
to figure out what's true using our most
弄清楚什么是真实的 使用我们最
sophisticated and rigorous mathematical
复杂和最严格的数学
tools peer review and the standards of
工具同行评审和
practice we still get it wrong so often
实践标准,我们仍然经常出错,
so how frequently do we delude ourselves
所以
when we're not using the scientific
当我们不使用科学
method as flawed as our science may be
方法时,我们经常欺骗自己,因为我们的科学可能有缺陷,这
it is far and away more reliable than
是遥远的事情 比
any other way of knowing that we have
任何其他方式都更可靠,我们知道
this episode of veritasium was supported
这一集的 veritasium
in part by these fine people on patreon
部分得到了 patreon 上这些优秀人士
and by audible.com the leading provider
和 audible.com 的支持,audible.com 是领先的
of audiobooks online with hundreds of
在线有声读物提供商,拥有包括小说、非小说
thousands of titles in all areas of
在内的所有文学领域的数十万种图书
literature including fiction nonfiction
and periodicals audible offers a free 30
和期刊 audible
day trial to anyone who watches this
为任何观看此频道的人提供 30 天免费试用,
channel just go to audible.com slash
只需访问 audible.com 斜线
potasium so they know I sent you a book
钾,这样他们就知道我给你寄了一本书,
I'd recommend it is called the invention
我推荐这本书叫《
of nature by Andrea Wolfe which is a
自然的发明》,作者是 Andrea Wolfe,这是一本
biography of Alexander von Humboldt an
传记 亚历山大·冯·洪堡 (Alexander von Humboldt) 是一位
adventurer and naturalist who actually
冒险家和博物学家,实际上
inspired Darwin to board the Beagle you
启发了达尔文 (Darwin) 登上小猎犬号 (Beagle)。您
can download that book or any other of
可以在 audible.com 下载该书或
your choosing for a one-month free trial
您选择的任何其他书籍,免费试用一个月。
at audible.com slash veritasium so as
因此,
always I want to thank audible for
一如既往,我要感谢 audible 对
supporting me and I really want to thank
我的支持 我真的很感谢
you for watching
你的观看
Archiver|手机版|科学网 ( 京ICP备07017567号-12 )
GMT+8, 2024-11-22 00:14
Powered by ScienceNet.cn
Copyright © 2007- 中国科学报社