马斯克的民调跟曾经的纽约时报一样自嗨

文摘   2024-08-25 00:00   北京  

《纽约时报》曾在2016年美国大选时伪造民调,说希拉里有九成以上的机会胜选,而且在那届大选中,多数民调,因门外汉特朗普的加入,在大数据上都预测错了最终结果,舆论哗然。于是,之后有许多人说民调不靠谱。

但今年马斯克的民调跟2016年时的《纽约时报》一样不靠谱。有这么几个原因:

一是在马斯克声称要买下推特之后,许多左翼人士和粉丝都不再信任这个社交媒体会是他们满意的言论之地,纷纷关闭了这个社交媒体的账号,即便那些依然留下的账号,上社交媒体平台发推的频率也比之前低了许多。更不要说还是在民主党大会召开的时候,多数左翼人士都聚焦于如何给哈里斯加油,顾不上刷推。

二是关注了马斯克的,多半都是特朗普铁粉,因为马斯克自买下推特之后,就一直为特朗普说话,今年更甚,他还公开发推说支持特朗普,还给特朗普阵营捐款,还用自己的平台打广告给特朗普作访谈。他这样只能吸引更多的特朗普粉丝,因而他的民调,顶多只能是他的粉丝对特朗普的支持度。

三是最重要的也是容易被忽视的一点。有人说,民主党和左翼主流媒体在操纵民调,因而不可信。但其实特朗普一样有操纵民调的历史。在给特朗普当了十余年私人律师的迈克尔·科恩的有忏悔性质的书《不忠》里,第10章一整章都是披露特朗普如何操纵民调的细节故事。摘译如下:

...at the time, CNBC was conducting an online poll to determine the twenty-five most influential business people alive to celebrate the twenty fifth anniversary of the network. Trump was one of the two hundred businessmen listed as contenders. I learned about the poll when I received an email from the Boss’s assistant, Rhona Graff, telling the staff of the Trump Organization that the vote was being conducted and requesting everyone to click on the hyperlink and vote for Mr. Trump.
......当时,CNBC 正在进行一项在线投票,以选出 25 位在世的最具影响力的商界人士,以庆祝该电视网成立25周年。特朗普是被列为竞争者的200名商人之一。我是在收到老板助理Rhona Graff的电子邮件时得知这次投票的。这封邮件告诉特朗普集团的员工正在进行投票,并要求大家点击邮件链接,为特朗普先生投票。

...

It was irresistible to Trump: he would kill to win that competition, I knew the second I read about the poll. I immediately voted on my desktop, and then I voted again on my tablet, followed by votes registered from both of my cell phones, using four different IP addresses to disguise the multiple clicks. I figured if everyone in the company did the same thing, Trump would at least make the top ten.
这对特朗普来说是不可抗拒的:我一看到投票信息就知道,他一定会拼尽全力赢得竞赛。我立即在我的台式机上投了票,然后又在平板电脑上投了票,接着又用我的两部手机注册了投票,使用了四个不同的 IP 地址来伪装多次点击。我想,如果公司里每个人都这么做,特朗普至少能进入前十名。

But when I looked at the results, with the running totals available online, I discovered Trump was near the bottom of the list – around 187 out of 200. Word around the office was that Trump wasn’t happy – “pissed” was the exact term – which was confirmed when one of his assistants brought a note from Trump to me.
但当我在网上查看总结果时,发现特朗普的排名几乎垫底——在 200 人中的排位仅位于187 名左右。办公室里弥漫着特朗普不高兴的消息——准确地说应该是生气——他的一位助手把特朗普的一张纸条拿给我时证实了这一点。

It was early in the morning, before most others had turned up for work, but Trump often woke before dawn, particularly if he was mad about something, as he frequently was. I was always an early bird myself, turning up before eight most mornings, our shared habit of waking early likely a function of our bond as teetotalers.
那是一个清晨,大多数人还没有上班,但特朗普经常在黎明前就起床了,尤其是在他为某事生气的时候,他经常这样。我自己也是个早起的人,大多数情况下早上八点前就起床了,我们共同的早起习惯很可能是我们作为滴酒不沾者的纽带。

The note consisted of a printout of the poll rankings, with the humiliating place Trump occupied highlighted by a black Sharpie circle, and in the margin in his distinctive, manic, all uppercase handwriting, “SEE ME ASAP. “
纸条上是一张民调排名的打印件,上面用黑色夏普笔圈出了特朗普所处的屈辱位置,并在空白处用他独特、狂躁、全大写的笔迹写道:“尽快见我。

I entered Trump’s office carrying a notepad and pen.
“You want to see me, Boss?” I asked.
“Yeah,” he said. “What can we do about this poll? I’m at the bottom the f*cking list. Check into this immediately and let me know.”
“Of course, Boss, I’m on it,” I said.
我拿着记事本和笔走进特朗普的办公室。
“你想见我吗,老板?” 我问道。
“是的,"他说。我们能为这次民调做些什么?我在这X蛋的名单上排倒数第一。马上整一下,然后告诉我。
“当然,老板,我正在弄。"我说。

...

Back in my office, I called my friend John Gauger, the chief information officer at Liberty University. I’d met John in 2012 when Trump had been invited by Jerry Falwell Jr. to address the school, and I’d accompanied him to Virginia. John also had a side business called RedFinch Solutions LLC, which provided services for search engine optimization and Internet reputational management.
回到办公室,我给朋友John Gauger打了电话,他是自由大学的首席信息官。我是在 2012 年认识John的,当时特朗普应Jerry Falwell Jr. 的邀请在学校发表演讲,我陪同他去了弗吉尼亚州。John还有一家名为RedFinch Solutions LLC的副业公司,提供搜索引擎优化和互联网声誉管理服务。

...

“I can do this very easily, “ John said. “The algorithmic code they’re using is very basic. My team and I have already cracked it. We can manipulate the voting by inserting IP addresses casting votes for Mr. Trump based on the overall number of total votes, so the votes aren’t visible. That way, we will be totally undetected while we move Mr. Trump higher in the rankings. But we need to buy IP addresses since we don’t own enough to make a dent in the rankings. At the same time, we’ll perfect the algorithm to ensure a seamless strategy. “
"我可以很轻松地解决它,“ John说,“他们使用的算法代码非常基础。我和我的团队已经破解了它。我们可以在总票数的基础上插入为特朗普先生投票的 IP 地址,从而操纵投票,这样投票就不可见了。这样,我们就可以在完全不被发现的情况下,将特朗普先生的排名提高。但我们需要购买 IP 地址,因为我们拥有的 IP 地址还不足以影响排名。同时,我们将完善算法,确保策略天衣无缝。

...

“Wow,“ Trump said. “Go do it. I want to be number one. “
“Boss, you don’t want to be number one,” I said. “That will potentially attract unwanted attention. Let’s go for, say number nine. Then you’re in the top ten. “
“Good,” Trump said, looking very pleased that he was going to be able to manipulate the poll.
“哇,"特朗普说。去干吧。我想排名第一。
“老板,你不能想着成为第一的," 我说。这会引起不必要的关注。我们还是争取第九吧。这样你就能进入前十了。
“很好,"特朗普说,他看起来非常高兴,因为他将能够操纵民调。

...

“How are we doing?” Trump asked, without having to specify what he meant.
“We’re at number twenty-nine and climbing,” I said. “I have a status call with my friend at three. I will update you then. “
“Man, your friend is great,” Trump said.
“我们做得怎么样?” 特朗普问道,无需具体解释他的意思。
“我们现在是第29 位,而且还在上升。"我说。在三点我要跟我的朋友通个电话。到时我会向你通报最新情况。
“老兄,你的朋友真不错," 特朗普说。

如果两人票数差一点,还有一定的可信度。差这么多,则完全不可信,甚至可笑。如果说上述那个故事中是特朗普通过律师的朋友间接操纵民调,那么这次就更简单了,只需要推特老板马斯克写几行代码,特朗普自然就可以赢得绝对多数。而现如今的马斯克,就是狂热的支持着特朗普。即便没有足够的证据表明他们操纵了民调,也有足够的理由相信有着操纵民调历史的特朗普有一定的概率会这么做。

而且马斯克似乎对美国选举制度完全不懂,也不屑于懂。毕竟,如果按照马斯克的网络普选民调的逻辑,在2016年大选中,希拉里的民众普选票数更多(比特朗普多近300万张选票)应该胜选,但依据美国大选规则中的选举人投票制,以各州的选举人团之和来决定选举结果,希拉里反而输掉了大选。美国大选并不是许多人认为的直选,而是一种为了确保相对公平的间接选举。

还有,网络民调本身就存在着巨大问题。即便是在21世纪的美国,也有相当数量的穷人和草根没法上网。这一点,不论是在《你当像鸟飞往你的山》还是在《乡下人的悲歌》中都有类似的笔触,所以网络民调并不能代表实际民调,更何况,本届大选,两方都吸引了大量的“贫下中农”的草根人士。

我平时看的美国民调,只关注两个人旗下的。一个是538民调网站创始人Nate Silver的,他自己本来是一位资深的扑克牌玩家,知道怎么从扑克牌做起,搞专业的数据统计、概率论、事件和心理情绪分析:

另一个是弗吉尼亚大学政治中心主任Larry Sabato的,他自己本身是研究美国历届大选民调的,而弗大政治中心有着美国最全的总统数据库与各时期人口迁移状况的数据库,那里对美国各段历史都有着第一手的信息源:

这两人旗下机构的预测,每届大选(包括中期选举)都比其他机构的要准很多。

说到Larry Sabato,这里顺便提一句,他是美国最资深的肯尼迪总统研究专家,写过《肯尼迪半世纪》,剖析了肯尼迪对之后美国各总统的直接影响力。因而当某些媒体评论说,刚刚退选的小罗伯特·肯尼迪退选转而支持特朗普让特朗普稳赢,不如去看Sabato教授对这种微不足道小事的评论。Sabato在电视上说,“小肯尼迪就是个没头脑且没人气的小丑。几个月前,他还发了长文,表示要坚决反对特朗普呢。” 接着他转发了这样一条推特,曝光了这个小肯尼迪当时大肆批判特朗普的话语:

何况,小肯尼迪退选之后,虽然他支持特朗普,但他们家族有另几位肯尼迪们也都发表声明支持哈里斯。连小肯尼迪这么没有影响力的人物都能上头条,不是更表明特朗普那边着急了么。两个小丑一起跳梁,也就刷个一两天的微博热门话题,谁还真把他们当回事了似的。

另外劝特朗普粉们好好读读德桑蒂斯自传。德桑蒂斯不止一次对民调表示了根本不屑一顾的态度。他有一处这样说:

It is one thing to examine poll results, but those results depend on how the question is asked and often do not even ask the most relevant questions. Plus, a poll cannot tell you how voters will respond to issues once they are properly articulated and framed. Of course, while in an ideal world the media would report on issues of vital concern to people, the filter of corporate media narratives does far more to obscure legitimate issues than to enlighten about them. A candidate blindly accepting such narratives, much less fashioning an agenda in response to them, is a candidate who will fail to garner enthusiastic support.
研究民调的结果是一回事,但这些结果取决于提问的方式,甚至往往没有问到最相关的问题。此外,民调并不能告诉你,一旦问题得到恰当的阐述和界定,选民会如何反应。当然,在理想的世界里,媒体会报道与人们息息相关的问题,但在商业媒体叙事的过滤下,对合理问题的遮蔽远远大于对其的启发。如果候选人盲目地接受这些叙事,更不用说针对这些叙事制定议程,那么他将无法获得热烈的支持。

选前民调重要吗?重要,但的确不应过于为了民调数据的好看而在民调本身的数字上动脑筋,否则这种民调有可能不科学,甚至造假、被操纵。

那么,真正的投票日当天,选票会真实反映选民的投票结果吗?同样未必。哈里斯在其回忆录中记录了一段美国相关部门是如何检测投票机安全漏洞的故事,摘译如下:

To help members of Congress and their staffs understand the nature of the risk, I invited a computer science and engineering professor from the University of Michigan to visit the Capitol and demonstrate the ease with which a hacker could change an election’s outcome. We gathered in a room in the Capitol Visitor Center, where the professor had set up a paperless voting machine used in numerous states, including swing states like Florida, Pennsylvania, and Virginia.
为了帮助国会议员及其幕僚了解风险的本质,我邀请密歇根大学的一位计算机科学与工程教授访问国会大厦,演示黑客可以轻易改变选举结果。我们聚集在国会游客中心的一个房间里,这位教授在这里架设了一台在许多州使用的无纸投票机,其中包括佛罗里达州、宾夕法尼亚州和弗吉尼亚州等摇摆州。

Four senators participated – Senators Lankford, Richard Burr, Claire McCaskill, and me – and the room was filled with staffers who had come to better understand the process.
参议员LankfordRichard BurrClaire McCaskill和我等四位参议员参加了测试,会议室里坐满了前来想要更好地了解过程的工作人员。

The professor simulated a vote for president, where we were given a choice between George Washington and the infamous Revolutionary War traitor Benedict Arnold. As you might imagine, all four of us voted for George Washington. But when the result came back, Benedict Arnold had prevailed. The professor had used malicious code to hack the software of the voting machine in a way that assured Arnold’s victory, no matter how the four of us had voted.
教授模拟了一次总统投票,让我们在George Washington(美国首位总统)和臭名昭著的革命战争叛徒Benedict Arnold之间做出选择。如你所想,我们四个人都投给了George Washington但结果出来后,Benedict Arnold占了上风。教授使用恶意代码入侵了投票机软件,确保Arnold获胜,不论我们四人投了什么票。

He told us that the machine was very easily hacked, enough so that, in a demonstration elsewhere, he turned one into a video game console and played Pac-Man on it. Can you imagine?
他告诉我们,这台机器很容易被黑客入侵,以至于在其他地方的一次演示中,他把这台机器变成了一台视频游戏机,并在上面玩起了吃豆人游戏。你能想象吗?

我们可以大胆想一下,如果按哈里斯他们所试验的,投票机上可以出现电脑游戏吃豆豆,那么你们说那上面会不会出现一只时下热门的悟空呢?

因而从这个角度上来说,共和党人积极倡导的完全使用纸质票投票的做法不无道理。其实民主党人也意识到了这一点,但他们在疫情期间的那次大选上依然鼓励选民用另一种方式投票,甚至有许多未经注册的邮寄选票。觉察到这里面会有舞弊情况的德桑蒂斯,当时在佛州就用强制手段让大家都用纸质选票投票,因而即便在2020年大选中特朗普输给了拜登,但德桑蒂斯确保佛州变成了一个共和党绝对占优的深红色州。

美国大选的民调可以被造假,正如大选投票结果也可以被操纵。两党都有可能在各环节上破坏公正、大搞舞弊。没有什么不可能。这就是为何两党都有一批政客表态要改革选举体系。但他们的改革究竟是为了更公正,还是更有利于其中一个党派更好操纵,那就请诸位擦亮眼睛,继续吃瓜看戏。

熊法通则
高楼高,高楼美,我在楼下往上瞧。哎呀呀,掉了我的小花帽。
 最新文章