這是一個極其荒謬的現實：當煤炭或算力的效率提升時，我們會瘋狂地消耗更多；但當「人力」的效率提升時，公司卻急著把人踢出大門。這難道不違反「傑文斯悖論」嗎？
其實不然。傑文斯悖論之所以在人力市場失效，是因為勞動力與資源在權力結構上有本質的區別。資源（如電力或石油）是被動的消耗品，成本降低會誘發新的用途；但人是主動的成本中心。在資本家的邏輯裡，提升效率的目的不是為了「僱用更多人來做更多事」，而是為了「用更少的人完成一樣的事」，從而省下那筆最昂貴的開支：薪水。
機器不會要求勞健保，AI 不會抗議加班。當技術讓一個員工能做三個人的工作時，老闆絕不會再請兩個員工來陪你，他會直接裁掉那兩個人，然後把省下的錢變成報表上的淨利。這就是人性的陰暗面：我們對物質的欲望是無限的（所以資源消耗激增），但對「分享利潤」的意願卻是極其有限的。在技術的軍備競賽中，人不再是需要被「更高效利用」的資源，而是被視為一種「待解決的瑕疵」。當我們把人變成了工具，而工具又變得太好用時，人就成了多餘的零件。
It is a delicious irony of our age. When coal gets efficient, we use more coal. When data gets efficient, we use more data. But when human labor gets efficient, we use fewer humans. Why does the Jevons Paradox suddenly stop working when the "resource" being optimized is a person in a cubicle?
The answer lies in the cold, hard logic of ownership and substitution. You see, Jevons Paradox triggers because the costof the resource drops, stimulating massive new demand. If electricity gets cheaper, I want more of it because it improves my life. But if a worker gets "more efficient"—thanks to AI or automation—they aren't becoming a cheaper, more desirable resource for the market to consume more of. They are becoming redundant. Unlike coal, a human being is a "multi-purpose resource" that comes with annoying overheads: health insurance, lunch breaks, and the inconvenient tendency to ask for a raise.
In the eyes of a corporation, a human is not a resource to be "saved" and reallocated; they are a cost center to be eliminated. When technology improves, we don't use the "saved" human time to let people write poetry or work more deeply. We simply replace the human component with a digital one. In the capitalist business model, the "efficiency dividend" of human labor doesn't go back into hiring more humans—it goes straight into the pockets of the shareholders. We’ve managed to create a world where everything gets consumed more voraciously as it gets cheaper, except for the one thing that actually needs a paycheck to survive.
1865 年，英國經濟學家傑文斯（William Stanley Jevons）提出了一個讓所有環保主義者心碎的觀察：當技術進步讓資源使用效率提高時，總消耗量反而會增加。這就是著名的「傑文斯悖論」。道理很簡單：當你讓一件事情變得更便宜、更有效率，人類那種永無止境的貪婪就會被釋放，最終消耗掉更多的資源。
看看我們的生活。LED 燈泡比傳統燈泡省電 90%，結果我們現在連公園的樹木和建築物的外牆都要裝滿裝飾燈，甚至開上一整晚，總用電量不降反升。汽車變省油了，我們就開得更遠、買更大的車；網路速度變快了，我們就瘋狂看 4K 直播。到了 2026 年，最典型的例子莫過於 AI 與算力。即便運算模型變得更省電、更高效，Google 與 Microsoft 等巨頭的能耗依然激增，因為 AI 變得太好用了，企業部署的量直接蓋過了省下的能源。
傑文斯悖論給了現代社會一個冷酷的教訓：「效率」不等於「節省」。如果我們只是一味追求技術改良，而不從總量管制或人類行為去自我約束，那麼每一項「綠色技術」都只會變成加速地球資源耗盡的催化劑。這是一場永遠跑不贏的馬拉松：當你以為省下了一杯水，人類的欲望就會立刻蓋出一座游泳池。
William Stanley Jevons must be laughing in his grave. In 1865, he noticed that as steam engines became more efficient at burning coal, England didn't use less coal—it used vastly more. This became known as the Jevons Paradox, and it remains the ultimate middle finger to our modern dreams of "green growth." The logic is simple and brutal: when you make a resource cheaper to use through efficiency, you don't save it; you just find more ways to burn it.
We see this everywhere. We invented LED bulbs that use 90% less energy, so we decided to light up our trees, our building facades, and our driveways all night long. We made car engines more fuel-efficient, so we built massive SUVs and moved to the suburbs to drive longer commutes. Even in the digital realm, 5G and high-speed fiber were supposed to make data "leaner," but instead, we just started streaming 4K cat videos in the shower. Now, in 2026, AI is the ultimate Jevons monster. Every time we optimize a Large Language Model to run on less power, a thousand new startups sprout up to use that "saved" energy for even more mindless automation. We aren't solving the energy crisis; we are just making the fire more efficient at spreading.
公用事業的「可加可減機制」聽起來像是恩賜，實則是一場精準的數字遊戲。這條方程式的本質並非與民共進退，而是為了確保壟斷企業在任何經濟氣候下都能旱澇保收。當我們將收費標準掛鉤於「入息中位數」時，就已經陷入了一個邏輯陷阱。
問題的核心在於薪酬的滯後性與黏性。在經濟衰退時，薪水很少會直接「調減」，取而代之的是「裁員」。那些失去收入的底層勞工，一旦進入失業狀態，就不再被計入薪酬中位數的數據中。換句話說，這條公式只看「還活著的人」領多少錢，而無視那些被時代馬車拋下的人。再加上外賣平台、Uber 等「零工經濟」的收入難以被精準統計，導致數據庫本身就帶有強烈的倖存者偏差。
更荒謬的是，公務員調薪參考私營市場，而公用事業加價又參考薪酬漲幅，這形成了一種自我實現的循環加價邏輯。從 2015 到 2025 年，薪資漲幅竟高出通膨一倍，這意味著如果機制完全掛鉤人工，收費將徹底脫離購買力的現實。這條方程式從來就不是為了「減價」設計的，它只是給壟斷者遞上一把合法的屠刀，讓他們在割韭菜時，還能優雅地宣稱：這是科學。
The "Plus-or-Minus" price adjustment mechanism is a masterpiece of bureaucratic gaslighting. In theory, it’s a fair formula designed to keep public service fees—from transport to utilities—in sync with the economy. In reality, it acts like a ratchet: it clicks forward easily but is physically incapable of turning back. The culprit isn't just corporate greed; it’s the mathematical DNA of the formula itself, which is hardwired to favor the "plus" and ignore the "minus."
The fatal flaw lies in tying prices to the Median Monthly Household Income. On paper, this sounds populist—linking costs to what people earn. But "wages" are notoriously "sticky." In a downturn, companies don't usually lower salaries; they just fire people. Those who lose their jobs—the most vulnerable—are conveniently scrubbed from the median income data. Furthermore, the burgeoning "gig economy" of Uber drivers and delivery riders, whose incomes are volatile and often shrinking, is rarely captured accurately in these formal statistics. When the formula only looks at the "survivors" of the labor market who haven't had a pay cut, the data stays artificially high, providing a "scientific" justification to hike fees even while the streets are struggling.
美國現在每 30 個孩子 就有一個被貼上自閉症標籤。這場「自閉症大流行」背後，除了篩檢普及，更多的是診斷標準的無限擴張。在現行制度下，診斷不再是為了瞭解孩子，而是為了「換錢」。沒有那張醫生證明，家長就拿不到保險給付，學校就不會提供補助。於是，診斷成了一張進入社會福利系統的「入場券」。
這塊肥肉引來了最貪婪的獵食者：私募基金。當應用行為分析（ABA）治療變成按小時計費的生意時，過度治療就成了常態。有些孩子一週要接受 40 小時 的訓練，比大人上班還累。諷刺的是，這數十億美元的市場裡，第一線的治療師往往薪水微薄、流動率極高。我們投入了天文數字的資源，卻只是在肥了醫療財團，而孩子們則成了流水線上的實驗品。
英國的情況同樣慘烈，SEND（特殊教育需求） 的學生人數激增，學校預算被徹底壓垮。我們必須問一個殘酷的問題：我們是不是正在把「正常的人性差異」給「病理化」？當一個孩子只要跟不上標準化的進度就被視為有病，我們毀掉的不只是孩子的自信，還有家庭的韌性。我們正在創造一個「被診斷的世代」，讓孩子從小就學會躲在標籤背後，而不是學習如何面對世界的粗糙。這不是進步，這是一場披著慈悲外衣的集體平庸化實驗。
發表文章
所有留言
