Home Artists Posts Import Register

Content

身邊不少朋友對特朗普當選的憂慮,源自他和MAGA陣營顯示出一種對數字的不尊重;而在一般受教育人士眼中,數字等同於科學,科學驗證是負責任政府的唯一依據。如果只有構想,而脫離實證,很容易出現「空想理想主義」的悲劇,極端例子就是中國大躍進、烏克蘭大饑荒,都是那樣煉成的。

然而值得研究的是,這樣的大道理人人都懂,美國是全球教育水平最高的國家之一,怎麼可能有一半以上的人,投票給「反科學」?這入面自然有很多情緒,but let's put aside those sentiments,嘗試思考一下那片森林。似乎很多人對「科學霸權」、「專業霸權」、「數字霸權」的抗拒,都是有客觀原因的:

數字和科學研究的公信力,在於能夠驗證,也可以 falsify。很多微觀層面的專業意見,明顯是經過驗證的,例如吸煙危害健康,例如煙民比非煙民容易有肺癌,諸如此類。有些卻是不能輕易驗證的,例如沒有經濟學者可以在金融海嘯之前預測它的出現和方式,只能事後尋找各種各樣的解釋;也沒有科學數字可以推論蘇聯會崩潰,現在的學術結論,甚麼派系都好,同樣都是事後孔明。

問題是,觀感上,很少人會清晰劃分。

當人類社會發展得越來越規範,如何通過數據去說故事,已經變成一門專業:一門很賺錢的專業。只要賺錢的專業,自然不可能所有人都受規範,而且通常是剛好相反,大多數人都會創造一種行規,默許自己的專門技能去尋租,而逐漸偏離初心。例如在學術界,不知道甚麼時候開始,追求高「影響因子」的學術論文是升遷的唯一準則,而有數字、模型的論文容易發表,又成為了人所共知的潛規則。結果自然是很多根本無需要數字的論文,由不擅長數字的教授,找一個 co-author 或 RA out-source 數字的部份,而且是先有結論、才硬塞數字入去交功課。

學術世界很離地,我們可以視為純粹一種遊戲。但現實世界的大型諮詢顧問公司也是同一做法,client 先有結論,然後顧問就會負責找到合用的數據去論證,無論是政治、經濟、文化、教育,甚麼都好,這一套已經行之有效。例如所有藥物、食品被證明有害之前,都一定經過專家確認對人體「無害」。這自然產生了眾多副作用,例如被一小撮精英壟斷,對廣大群眾則失去了專業的公信力。

問題是隨著科技發展,這種「專業」已經失去壟斷性。就算是對數字一竅不通的人,通過人工智能下達指令:「給我支持說明吸煙有益健康的數據」,一樣可以瞬間找出來。然而又與此同時,由於傳統教育的分工越來越 rigid,畢業生對一切數據、專業越來越 rigid 地視為金科玉律,兩者之間就產生龐大落差。

舉一個實例:美國選舉與經濟。之前我們談論過,特朗普和拜登任內,基本上經濟表現差不多,拜登任內的通脹是誰的責任很難簡單研判,特朗普的政策究竟會進一步刺激通脹還是打擊通脹,同樣有完全不同的專家推論。但反對特朗普的一方,往往傾向認為因為不滿通脹而選擇特朗普的選民,就是受「假新聞」影響,認為如果大家認真做研究,就會投民主黨。

這種閱讀現實世界的方式,看似小事,其實就是難以調和的鴻溝。

(待續)

⏺ 【隨緣家書大長篇 🇺🇸】 Elon Musk 的「政府效率部」,會否成為顛覆世界、製造未來的先知?(下)
https://www.patreon.com/posts/115928374/

Files

Comments

carl mak

但數字總究係人做出來,只要人的參與就無可能係100%可信,任何事情都係可信但不能盡信 另一個問題,依家唔理係政府定專家,好多都唔係根據數字去研究結果,係為左結果而去尋找數字。呢種事,本身就唔科學。 拜登任內的通漲,係天時地利人和的產物,誰上場都無可避免,但問題係選民終究係在佢任內感受到影響,把矛頭指向佢,亦無可厚非~

Good Year

商界都有咁既情況, 數字太多manipulation, 由其係會計啊, 財務啊, 唔熟個行業唔知危險位好易中招, 小弟都多年財務出身, 但商業模型日新月異, 連 fund flow 都可以manipulate 有咩唔得?又咁講專業都係一個框架下工作, 但世事日新月異, 個框架都可以包唔到, 當個框架彊化, 美帝就會跳出Trump咁既case 強勢的影響個框架, 跳出個框架, 逼個框架改變, 唔變就取代

Kin-Wan Chan

There are three kinds of lies…

lyk

We trust numbers when they are not in conflict with our interests. “To persuade, appeal to interest, not reason.” The human brain evolved to help us survive different environments, it reasons rationally to the extent that it gives us an evolutionary advantage to solve problems to stay alive. If it doesn’t increase our chance of survival, for example still firmly believing you can win when the outcome is certain you are about to lose a life and death battle with a lion in 30 seconds, we tend to turn a blind eye to rationality. You believe you can win because you have to, not because it is true. Corporate CEOs have an army of people to prepare fancy projections to justify whatever projects the person decides to do. This is the human condition, nothing is new. To this day from what I see 95% of what the financial world does is make up stories to sell to people. In the end it is about trust, and more importantly “Deserved Trust”, meaning those backed up by their deeds and track record over time that showed their actions are consistent with what they are selling. Numbers simply provide a means for people to establish trust. 所以其實 “大部份” 嘅人從來信嘅都唔係數字,信嘅係講嗰啲數字嘅人 (譬如醫生、學者)或者係講嗰啲數字嘅機構 (譬如以前啲人話 “報紙都有賣啦”). Generative AI like ChatGPT is no different, sure they can throw numbers around, but they are only as good as the data you feed them. Elon Musk claims that some of his rival AI are trained to lie and his generative AI will be “maximum truth-seeking”. The cost of living a civilised life will be prohibitively high for any human if they were to crunch all the numbers themselves. People are going to get advice from whoever they trust (or whoever created the chatbots that they trust) to give them numbers that help them make sense of “their” lives. So what is happening today is probably a reflection that trust in institutions and society has eroded in some places, what Francis Fukuyama called “Low Trust Societies”. Having said all that, I think this is far from the demise of “the monopoly of professions”. Given the human condition is what it is, an unregulated environment will be “far” worse. Yes, most financial products left a lot to be desired, but if those people are not licensed, I can’t imagine what that will look like.