不用尋思大概也能知道，過去10年是前所未有的科技增長期之一。它帶來的不光是智能手機，還有大量app，可以滿足所有能想到的愿望、需求或者愛好。在這10年中，優步變成了名詞、動詞，也成了一種比喻手段，比如“某某行業的優步”。也是在這10年里，虛擬現實頭盔和無人機成了眾多日常玩具中的一員，社交支付app Venmo挽救了80、90后的友誼，比特幣成了投資者的最愛，盡管這種無國界數字貨幣的大多數粉絲還沒完全弄懂它的原理。得益于名為Crispr的工具，基因編輯變得格外“平易近人”。人工智能則學會了下圍棋和“自己”開汽車。如今，以我們口袋里的通訊設備為基礎，移動互聯撐起了一個完整的經濟體系。在2007年，這些創新中的大多數都還沒有誕生。

考慮到所有這些因素，斷言創新沒有以前那么有力，或者說，我們想出的所有這些新點子不知怎的都沒有以前那么“物美價廉”，似乎并不合理。但這正是一些經濟學家的結論。它值得探討，原因很多，而且這條結論會對企業、投資者乃至整個社會產生巨大影響。

10年來，盡管涌現出大量創新，但總生產率以及經濟產出的增長速度卻放慢了。大衰退以來，這兩項指標一直偏低，近來的上揚也未能改善情況，它們的年增長率仍介于1-2%之間，而以往的GDP平均增長率接近5%，生產率增速則為3%。此外，收入差距逐年擴大。企業創意帶來的經濟效益似乎并未得到那么廣泛的分享。原本打算改變人們生活方式的新點子并沒有結出那種覆蓋范圍廣闊的經濟成果，從而真正改善大多數人的生活。

原本打算改變人們生活方式的新點子并沒有結出真正改善生活的經濟成果。

換句話說，創新的雄心壯志與其影響范圍之間存在差異。了解出現這種差異的原因則有助于我們消除它，進而將企業的更多重大創意轉換成那種我們一直苦苦求索的共同繁榮。

進入工業化時代后，我們的增長速度一直很迅猛，接近指數級別。經濟學家們認為，從某種程度上說，這樣的增長源于人數大體不變的研究者和科學家堅持不懈地完成了他們作為研究者和科學家的工作。經濟學家們的看法是，這樣的持續性應該可以穩定產生一系列新創意，進而推動經濟發展。只要這些研究者和科學家的人數基本不變，我們基本上就可以保持飛速發展的潛力。

但人們最近發現這項經濟理論有不足之處。今年夏天，麻省理工和斯坦福大學的一批經濟學教授和研究生更仔細地考察了研究開支和經濟增長的關系。美國國家經濟研究局發表了他們題為：《發現創意的難度正在增大嗎？》的論文，題目剛好切中要害。讀了這篇論文后就會很快明白，沒錯，創意的難度是變大了，而且成本也大幅上升。

老理論認為持續的研究可帶來指數級增長，其最佳例證是摩爾定律，預測說計算機芯片的處理能力差不多每兩年就會提高一倍。在40多年的時間里，情況基本如此，我們的手機恰好就能證明這一點，如今手機的計算能力要遠高于許多擺滿了70年代計算機的建筑物。

但問題在于，帶來這種巨變的研究根本沒有持續性。實際上，上述論文的作者們發現，和尼克松當政時相比，如今同等研究成果的成本在剔除通脹因素后是那時的78倍。計算機可能比那個時候小得多了，但現在推動計算機發展的人則會站滿許多許多建筑物。

這種情況不僅限于科技行業。開展此項研究的教授和研究生們還考察了農業，特別是玉米、大豆、棉花和小麥的產出水平。他們比較了年產量和用于提高農作物產出而投入的研究資金。開支增多了，產量也上升了，但方式可能和大家想的不一樣。1960-2015年間農作物平均產量提高了一倍，而用于提高產量的年度研究投資在剔除通脹因素后至少上升了三倍。在某些情況下，或者說某些年份的某些農作物，研究投資增加了25倍。農業在研發方面的投資似乎越來越多，取得的成果則越來越小。

其他領域的情況也是這樣。斯坦福大學和麻省理工的經濟學家們還研究了醫療科研的生產率，特別是在癌癥領域。他們發現后者全面下降。盡管發表的論文變多了，上馬的臨床試驗也增加了，但在每10萬人中，得救患者的增長速度不斷放慢。雖然取得進展的時刻讓人感到驚嘆和舒心（在一定程度上就像我們看到癌癥免疫療法那樣），但總的來說，付出的更多，效果卻更小。如果還想以幾十年前那樣的速度來救人，我們就得發表更多的論文，并為臨床試驗投入更多的資金。

當然，研究開支的增長往往會在短期內讓個別公司獲得回報?！敦敻弧冯s志對標普500指數的研究表明，2007-2017年披露研發支出增長的公司有155家，其中逾三分之二，具體來說是108家的股票回報率超過了標普500指數。但跑贏大盤和對整個經濟產生全面而積極的影響并非一回事。上述論文的作者們發現，對所有上市公司來說，要讓經濟增速達到30年前的水平，他們的研發支出就得提高到原來的15倍。結論就是，重大創意，也就是真的能促進經濟增長或改善生活水平的創意變得更難發現了，原因是其成本達到了前所未有的高度。

為什么會這樣？一個原因是越發難以進行此類創新是因為隨著知識進步，基礎知識的規模不斷擴大。如今，精通多個科學或工業領域所需的教育或培訓投資遠高于上一代人需要的水平。

另一個原因是純研究成本。沒錯，它也在上升。設備價格普遍上漲，而且更加專業化?？梢越佑|到設備的人變少了。同時，就像我們從一臺一臺的計算機轉向超級計算機一樣，我們的文化也從個人取得研究突破變為受過高等教育并且要拿酬勞的大型專家團隊來設法解決遠比以前復雜的問題。我們早已越過了那些低垂的果實，目前正在嘗試通過制作工具和建立系統來達到樹的頂端。

上述論文以及其他類似著述像是在說，面對創新的枯竭我們只能干瞪眼。別那么絕望。我們也許，甚至很有可能只是在某些領域走到了繁榮期的末尾，就像某些頂端果實已被基本采摘完畢的樹木。從發明計算機到IT和互聯網的崛起，摩爾定律一直暗流涌動。如今，這股暗流的速度正在放慢，其他因素則有可能在某一天為經濟增長貢獻自己的力量。

上述報告的作者之一、斯坦福大學經濟學博士生邁克爾·韋伯說：“因為遇上了困難，所以創意就見頂了，情況并不是這樣。也許，更好的看法類似于石油勘探。在既定的油田中，人們會把大多數石油開采出來，而要開采剩下的那些石油，成本就會非常高。我們在IT這塊油田里已經開采了很長時間，但還有一大塊我們尚未發現的新油田。”

韋伯指出，其中一塊新油田是基因組學，得益于被稱為Crispr的基因編輯工具（請參見《正在待命的美國大創意》），這個領域正在涌現諸多新的應用。我們仍處于非常早期的探索階段，而這些探索有可能發現醫學和經濟上的大油田，其中那些成本低、數量足的創意將幫助我們降低醫療成本并延長有質量的壽命。虛擬現實也是如此，它出現的時間更長，而且有可能成為未來的首選媒介，進而實現爆炸性增長。

這就引出了同樣重要的第二點，那就是現在有眾多看起來成本效率不高的技術，但只要它們真的具備了改變的能力，按照以往經驗，到那時它們就會變得較為劃算。在90年代以前，互聯網一直都是學術怪咖的愛好，而今它已經成為經濟支柱。

只有我們重視培養和扶持人才，人才群體才會擴大（從而降低創意成本）。

在我接觸過的經濟學家中，幾乎所有人都認為人工智能特別有可能成為創意激發增長和生產率實現極大提升的渠道。在說明新創意會有多貴方面，人工智能也是一個絕佳的例子——它需要巨大的計算能力，眾多專業知識，從業者的工資也很高（所以對企業來說很貴）。

美國西北大學經濟學家本杰明·瓊斯研究的是創業問題，而且即將就人工智能發表論文。他說人口增長可能有助于降低獲得新創意的成本，這也許和常識相悖。瓊斯解釋說：“我們完全可以讓更多的人來解決某個問題。”雇用成本應該下降，至少理論上是這樣，因為人口遠超其他國家的中國和印度仍在融入推動全球經濟的研究引擎之中。

但只有在更多的人受過高等教育之后，增加解決復雜問題的人數才會對我們有幫助。換句話說，只有我們重視培養和扶持人才，人才群體才會擴大（從而降低創意成本）。就像進化論生物學家史蒂芬·杰伊·古爾德曾說過的那樣：“不知道為什么，我對愛因斯坦大腦的重量和上面的溝回不那么感興趣，我更關注的是幾乎一定有具備同樣才能的人在棉花地和血汗工廠里勞作和死去?！?

麻省理工經濟學家、同樣參與上述論文撰寫的約翰·范瑞恩曾研究美國的人才培養問題，方法是嘗試判斷某人做出發明的可能性。他發現，收入排在前1%的家庭生下的小孩成為發明者的幾率是收入排在后50%的家庭所生小孩的10倍。他還發現，男性成為發明者的可能性更高，少數族裔進行發明創造的可能性則較低。

范瑞恩的研究完美地展示了機遇問題。如果我們相信來自任何地方的聰明孩子都有可能成為發明者，那么我們就需要支持讓這種信念變成現實的政策，因為這種信念對經濟以及我們告訴自己的有關美國夢的故事來說都不可或缺。范瑞恩說：“許多孩子甚至都沒有機會想象一下成為發明者是什么樣，那是什么樣的工作以及它看起來像什么。這在很大程度上跟教育、眼界以及允許別人為自己夢想更好的可能性有關?！?

范瑞恩意識到，和建起有玻璃幕墻的建筑并讓出色的研究者入駐其中相比，他的解決方案似乎不那么漂亮。但基于政策的解決辦法，比如為欠缺財力的公立學校提供更多資金這樣簡單的做法，有可能比其他旨在填補創新缺口的復雜辦法便宜得多。范瑞恩說，在一個越發以知識為動力的經濟中，“這就是那些低垂的果實”。

It’s possible, without squinting, to gauge the past decade as one of unprecedented technological growth. It ushered in not only the smartphone, but with it a cornucopia of apps for every imaginable want, need, or obsession. It was a decade in which Uber became a noun, a verb, and an analogy too—as in, the “Uber of X.” In these same 10 years, virtual reality headsets and drones joined the litany of everyday playthings; millennial friendships were saved by Venmo; and a completely stateless digital currency became an investor darling, even if most of Bitcoin’s fans still don’t quite grasp how it works. Genetic editing became extraordinarily accessible thanks to a tool named Crispr, and A.I. went from mastering Go to becoming the “self” in self-driving cars. There’s now an entire economy based on the mobile connectedness that’s built on the devices in our pockets. In 2007, most of these innovations didn’t exist.

So given all that, it would seem absurd to suggest that innovation is somehow less potent than it once was—that, somehow, all those new ideas we’re generating are offering less bang for the buck. But that is precisely what one group of economists have concluded. And it’s worth assessing their argument for a number of reasons, because it has enormous implications for businesses, investors, and society as a whole.

Over the past decade, even as innovation has soared, growth in overall productivity as well as economic output has slowed. Both remain at anemic levels even after recent upticks, hovering at annual rates between 1% and 2% since the Great Recession, whereas historical averages are closer to 5% GDP growth and 3% productivity rates. Plus, year by year, income inequality has widened. The economic benefits of the business world’s new ideas do not seem to be so widely shared. The new ideas that were meant to transform the way we live aren’t unleashing the kinds of broad economic benefits that actually make life better for most people.

Ideas that were meant to transform the way we live aren’t unleashing the economic benefits that actually make life better.

Put another way: There’s a gap between the ambitions of our innovations and the scale of their impact. Understanding the causes of that gap could help us close it—and, in turn, convert more of business’s big ideas into the kind of broadly shared prosperity we’ve been pining for.

We have been in a period of remarkable, near-?exponential growth since the beginning of the industrial age. Economists have explained this growth as the result, in part, of a more or less constant number of researchers and scientists diligently doing what researchers and scientists do. That constant, economists held, was supposed to be capable of producing a steady stream of new ideas that would, in turn, generate economic expansion. So long as the number of researchers and scientists is held in a relative constant, so is the potential for very strong growth.

But lately, cracks have begun appearing in this economic theory. This summer, a group of economics professors and graduate students at MIT and Stanford took a closer look at the link between research spending and economic growth. The title of their working paper, published by the National Bureau of Economic Research, got right to the point: “Are Ideas Getting Harder to Find?” Reading the paper, one quickly learns that, yes, ideas are getting harder to find, and a lot more expensive too.

The best example of the old theory—that constant research leads to exponential growth—is Moore’s law, which holds that processing power on a computer chip doubles approximately every two years. This has pretty much been the case for more than 40 years, as happily evidenced by our mobile phones, which hold more sheer computing power than many buildings full of 1970s computers.

The problem, however, is that the research that has driven such a tremendous change hasn’t been a constant at all. In fact, as the “Ideas” paper’s authors found, the cost of such research efforts requires 78 times the funding today in inflation-adjusted dollars as it did when Nixon was President. The computers may be much smaller, but now it’s the people driving their advances that fill many, many buildings.

This trend isn’t limited to the tech sector. The study’s authors also looked at agriculture, specifically the yields of corn, soybean, cotton, and wheat. They compared the annual yields with funding for research into improving crop productivity. Spending increased, and so did yields, but not in the way you might expect. While the average yield doubled between 1960 and 2015, the annual inflation-adjusted investment in research to improve those yields increased at least threefold. In some cases—for certain crops, over certain years—the research investment increased by a factor of 25. Agricultural businesses seem to be spending more and more on R&D, all while getting less and less out of it.

It’s the same story in other fields too. The Stanford and MIT economists also looked at productivity in medical research—specifically in cancer. They found declines in productivity across the board. Even as more papers were published and more clinical trials mounted, the growth in the rate of lives saved per 100,000 people has continued to slow. While individual moments of progress bring astonishment and relief (as we’ve seen to some extent in cancer immunotherapy), on the whole more effort is yielding less impact. We need to publish even more papers, spend even more on clinical trials, if we hope to keep up the pace of lifesaving treatments we were developing decades ago.

Sure, increases in research spending tend to pay off in the short term for individual companies. A Fortune study of the S&P 500 showed that of the 155 that reported increased levels of R&D spending between 2007 and 2017, more than two-thirds—108 companies in all—had stock returns that beat the index’s. But generating stock market outperformance isn’t the same thing as having a broad, positive impact on the economy as a whole. Across all publicly traded companies, the “Ideas” paper’s authors found that to produce the same rate of growth our economy experienced 30 years ago, companies would need to spend 15 times as much on R&D. The bottom line: Big ideas—the ones that truly drive economic growth or a change in the standard of living—are harder than ever to find because they’re more costly than ever before.

Why is that so? One reason that such innovations are increasingly difficult to come by is because, as knowledge advances, the base of fundamental knowledge grows. Just to be proficient in many fields of science or industry today requires an investment in education or training that is significantly higher than it was a generation ago.

Another factor is the cost of pure research. Yes, that’s growing too. Equipment has generally become more expensive and, relatedly, exclusive. Fewer people have access to it. And just as we’ve moved from single computers to supercomputers, we’ve moved from a culture of individual researchers making breakthroughs to one of massive teams of highly educated and compensated experts attempting to solve far more complicated problems. We’re way past the low-hanging fruit, trying to build the tools and systems that will get us to the tippy-top of the tree.

The “Ideas” paper, and others like it, might make it sound as though we’re staring down the barrel at the end of innovation. Do not despair. It is entirely possible, even likely, that we’re simply approaching the end of boom times in particular fields, the tops of certain already well-trimmed trees. From the discovery of computers to the rise of IT and the Internet, we have the undercurrent of Moore’s law, churning away. Now, that current is slowing down—but others could someday lend their velocity to a growing economy.

“It’s not true that because we’ve hit a hard place, we’ve reached peak ideas,” says Michael Webb, a Ph.D. candidate in economics at Stanford and one of the study’s authors. “A good way to think of it, perhaps, is like prospecting for oil. Within a given oilfield, you get most of the oil out, and it becomes really expensive to get the rest. We’ve been pumping out the oil from IT for a very long time now, but there’s a whole new oilfield out there we haven’t found yet.”

One such field, Webb suggests, is genomics, which is experiencing a flourishing of new applications thanks to the gene-editing tool known as Crispr (see “America’s Big Ideas in Waiting”). We’re still in the very early, exploratory days of what may become the medical and economic equivalent of a rich oilfield of cheap and abundant ideas that pay off in reduced medical costs and longer productive life spans. The same goes for virtual reality, which has been around for longer, and may end up being the media of choice in the future—driving explosive growth.

Which brings up a secondary but no less important point: There are plenty of current technologies that won’t look cost-effective unless and until they really do become transformative—at which point, in hindsight, they will appear to be relative bargains. The Internet was a geeky academics’ hobby until the mid-1990s. Today it’s an economic pillar.

The talent base only grows (making ideas cheaper) if we make a point of cultivating and supporting talent.

Nearly every economist I spoke to pointed to artificial intelligence as an especially likely avenue of ideas that could catalyze enormous leaps in growth and productivity. A.I. is also a great example of how expensive new ideas can be: The computing power required is enormous, the expertise tremendous, and the jobs incredibly well-paying (and thus costly for employers).

Benjamin Jones, an economist at Northwestern who studies entrepreneurship and has a forthcoming paper on A.I., says that, in perhaps a counterintuitive way, population growth can help reduce the cost of generating new ideas. “We’re fully able to throw more people at the problem,” he explains. Costs of employment should level off, in theory at least, as China and India—by far the two most populous nations—continue to integrate into the research engines feeding the global economy.

But having more people to throw at complex problems helps us only if more people are highly educated. In other words: The talent base grows (making ideas cheaper) only if we make a point of cultivating and supporting talent. As the evolutionary biologist Stephen Jay Gould once put it, “I am, somehow, less interested in the weight and convolutions of Einstein’s brain than in the near certainty that people of equal talent have lived and died in cotton fields and sweatshops.”

John Van Reenen, an economist at MIT and another of the “Ideas” paper’s authors, has performed studies that look at talent cultivation in the U.S. by trying to determine the likelihood of someone becoming an inventor. He found that those born into the top 1% income level were 10 times as likely to become inventors as those born into the bottom 50%. He also found that men were more likely to be inventors, and minorities less likely.

Van Reenen’s research nicely illustrates a problem of opportunity. If we believe that a smart kid from anywhere could become an inventor—a belief that is integral to both the economy and the story we tell ourselves about the American dream—then we need to support the policies that make that possible. “Many kids don’t even get the opportunity to imagine what it would be like to become an inventor, what that job is, what it looks like,” says Van Reenen. “A huge part of this is education, exposure, allowing someone to dream of better possibilities for themselves.”

Van Reenen realizes that his solution might not seem as sexy as erecting a steel and glass-walled building filled with brilliant researchers. But policy-based solutions as simple as improved funding to underfunded public schools could be a much cheaper solution to closing the innovation gap than other, more complex fixes. In an increasingly knowledge-fueled economy, Van Reenen says, “This is the low-hanging fruit.”??