不同國家不同領域DARPA模式的觀察與思考——三思派對話《打造創(chuàng)新型國家》作者Rainer Kattel

【書名】：《打造創(chuàng)新型國家》

【作者】：[愛沙尼亞]雷納?卡特爾、沃爾夫岡?德雷克斯勒、埃爾基?卡羅

【譯者】：于文軒

【出版社】：上海人民出版社 格致出版社

【出版時間】：2025年11月

1、關于政府在創(chuàng)新中的角色

三思派：理論界長期爭論政府在創(chuàng)新中應扮演何種角色。如果用最簡潔的方式概括，在您看來，政府在推動創(chuàng)新過程中主要承擔哪幾種核心職能或角色？

Rainer Kattel：簡而言之，我想這樣概括：政府在創(chuàng)新中的核心角色不僅僅是矯正市場失靈，更在于構建并治理那些使創(chuàng)新成為可能、得以持續(xù)且對社會有益的制度環(huán)境。這意味著政府需要為長期研究提供資金、主動塑造市場、采購新的解決方案、引導發(fā)展方向、協(xié)調跨部門行動者，以及最關鍵的一點——建立能夠隨時間推移不斷學習和適應的公共組織。創(chuàng)新并非由企業(yè)家或企業(yè)獨自創(chuàng)造的，而是依賴于具備合法性、專業(yè)能力以及能夠保持長期定力的公共機構。這就是為什么我們在書中提出：創(chuàng)新需要官僚制。

我還想補充一點，當今政府的角色已經(jīng)遠遠超越了以往許多創(chuàng)新政策辯論的范圍。在一個由人工智能、氣候壓力、地緣政治競爭和數(shù)字基礎設施所塑造的世界中，國家不僅要促進創(chuàng)新，還須構建韌性。因此，真正的問題不再僅僅是政府如何促進創(chuàng)新，而是它如何在動蕩的環(huán)境中發(fā)展出相應的實力（capacities）與動態(tài)能力（dynamic capabilities）來實現(xiàn)這一目標。這正是我們在工作論文中主張從“敏捷性穩(wěn)定”（agile stability）邁向“敏捷性韌性”（agile resilience）的動因。

2、關于成功與失敗的案例

三思派：《打造創(chuàng)新型國家》一書深入剖析了多個創(chuàng)新官僚組織的成功實踐。在這些案例中，哪一家機構您認為是最佳的案例？其成功的關鍵因素是什么？此外，在研究過程中，您是否也觀察到一些失敗的案例？這些失敗背后反映出哪些問題？

Rainer Kattel：如果非要挑選一個最能體現(xiàn)本書邏輯的機構，我會選擇德國的帝國物理技術研究院——這是一個在19世紀末幫助奠定了現(xiàn)代科學和工業(yè)能力基礎的德國機構。這一案例在今天鮮有人知，但它卻是官僚制與創(chuàng)新相互交織的典型體現(xiàn)。它的故事展示了創(chuàng)新官僚組織的完整演變歷程：首先是“魅力型網(wǎng)絡”（charismatic network）的構建與制度創(chuàng)新；然后是將這種能量逐漸“慣例化”（routinization），轉變?yōu)槌志玫摹皩＜倚徒M織”（expert organization）。換言之，它體現(xiàn)了我們所說的“敏捷性穩(wěn)定”：即從非常規(guī)的創(chuàng)舉向穩(wěn)定的能力建設的轉變。在當代，DARPA（美國國防高級研究計劃局）當然是一個極其顯著的例子，即便在那里，關鍵經(jīng)驗也不在于這個機構如何孤立運行并取得輝煌。其真正的優(yōu)勢源于一個更廣泛的系統(tǒng)：采購權力、穩(wěn)定的資金、大學，以及一個能夠接納不確定性的國家安全架構。

至于失敗案例，我們的研究表明，失敗通常不會表現(xiàn)為某個機構的突然崩潰，而是一個碎片化、空心化的過程，或是缺乏制度厚度的盲目模仿。在美國，我們描述了后期的“DARPA化”現(xiàn)象：到處散布著類似DARPA的創(chuàng)新“孤島”，卻未能重建更廣泛的官僚實力，而這是使命導向型政策實現(xiàn)系統(tǒng)化所必需的能力。歐洲的模式則通常是有使命卻無具備相應執(zhí)行能力的官僚機構：口號響亮、項目眾多，但用于規(guī)模化、協(xié)調和學習的中層機構卻極為缺乏。這屬于制度架構的失敗。

3、關于制度移植的挑戰(zhàn)

三思派：許多國家試圖借鑒他國成功的創(chuàng)新官僚組織模式（例如設立本國版的 DARPA），但在實踐中，這類制度移植往往成效有限。您在研究中是否發(fā)現(xiàn)典型的“復制失敗”案例？要使此類制度有效落地，哪些本土條件或治理要素是不可或缺的？

Rainer Kattel：是的，這是我們發(fā)現(xiàn)的最清晰的規(guī)律之一。許多政府試圖復制某種組織形式——比如一個DARPA、一個創(chuàng)新實驗室、一個使命導向的小組——卻未能復制使原機構行之有效的更廣泛的生態(tài)條件。但成功的創(chuàng)新官僚機構從來不僅僅是“一個聰明的機構”。它們植根于政治支持、資金安排、采購系統(tǒng)、專業(yè)規(guī)范、技術人才以及更廣泛的國家實力之中。當這些基礎缺失時，被移植的機構往往淪為一種象征性的存在，而非真正的變革引擎。

因此，本土條件不可或缺，絕非做做表面文章就行的。它們包括政治合法性、長期承諾、免受短期績效壓力的保護、精英隊伍、與大學和企業(yè)的聯(lián)系，以及足以支撐跨組織協(xié)調的權威。這就是為什么我會對創(chuàng)新政策領域的“最佳實踐移植”持謹慎態(tài)度。制度無法像藍圖那樣被簡單照搬，充其量只能作為一種啟發(fā)式經(jīng)驗被借鑒。本書更深層次的啟示在于，每個國家都必須立足于自身的行政傳統(tǒng)、政治安排和戰(zhàn)略優(yōu)先事項，構建獨特的“敏捷性穩(wěn)定”版本。

4、關于使命導向型組織的當代意義

三思派：阿波羅登月計劃催生了以使命驅動的創(chuàng)新組織范式。在當今應對氣候變化、公共衛(wèi)生、數(shù)字轉型等復雜挑戰(zhàn)的背景下，這種“使命導向”模式是否依然有效？您認為當前全球有哪些組織較好地繼承并發(fā)展了這一模式？

Rainer Kattel：“使命導向型”模式不僅依然可行，而且比以往任何時候都更具必要性。但我們需要以一種不同于阿波羅時代的眼光來理解它。阿波羅計劃是一個邊界相對清晰、技術明確可解的使命。而今天所面臨的挑戰(zhàn)——氣候變化、人口老齡化、公共衛(wèi)生、人工智能治理、數(shù)字化轉型——則更加開放和充滿政治爭議，且更具系統(tǒng)意義。因此，當代的使命不能簡單地作為具有特定終點的英雄主義項目來運作，而是需要以組合、生態(tài)系統(tǒng)以及長期的制度承諾的形式來推進。

最能推動這一進程的組織，是那些能夠將戰(zhàn)略方向與學習能力相結合的組織。在我們的研究中，中國之所以重要，正是因為它表明了強大的中央統(tǒng)籌可以與去中心化的分散試錯共存。更廣泛地說，當代最有趣的組織并不將使命視為口號，而是將其視為一種組織公共能力的方式，包括資金、監(jiān)管、采購、試驗和交付等各個環(huán)節(jié)。這也是我們?yōu)槭裁凑J為那些最成功的機構正日益呈現(xiàn)出“新韋伯式”特征：它們將長期的合法性與專業(yè)性，同開放性、試驗性以及網(wǎng)絡構建結合在了一起。

5、關于敏捷的公共投資

三思派：當前科技創(chuàng)新快速發(fā)展，技術路徑高度不確定，投資風險顯著上升。從治理角度出發(fā)，政府應如何設計其創(chuàng)新投入機制，使其既能支持長期、高風險的探索，又能靈活適應技術創(chuàng)新的快速迭代？

Rainer Kattel：政府需要設計一種具有耐心視野、運作敏捷的資助機制。在實踐中，這意味著一種“投資組合”策略：支持探索性、高風險和長期投資的組合，而不是要求每個項目都用短期回報預先證明其合理性。創(chuàng)新領域的公共投資應該視不確定性為常態(tài)。任務不在于消除風險，而在于智慧地把控風險。這需要分階段注資、定期的學習反饋閉環(huán)、國家內部的技術評估能力，以及隨著技術和戰(zhàn)略格局變化重新配置資源的能力。

同樣重要的是，資金機制不應脫離更廣泛的公共能力。單靠金錢是無法催生創(chuàng)新的。政府還需要采購權力、監(jiān)管的敏捷性、內部的專業(yè)知識，以及能夠跨越部門壁壘進行協(xié)調的機構。本書的核心論點之一是，動態(tài)能力（capabilities）根植于長期積淀的實力（capacities）之中。因此，敏捷的公共投資仰賴于持久性的制度：即那些能使國家從失敗中汲取教訓，而不致陷入短期主義或政治恐慌的技能、合法性和制度深度。

6、關于新時代創(chuàng)新官僚制的新興形式

三思派：當前，人工智能正深刻重塑科研范式，而地緣政治緊張又加劇了技術競爭與供應鏈重構。這些新趨勢將如何影響傳統(tǒng)創(chuàng)新官僚組織的運作邏輯？您是否預見到未來可能出現(xiàn)新型的公共創(chuàng)新組織形態(tài)？如果有，它們可能會具備哪些關鍵特征？

Rainer Kattel：人工智能和地緣政治的發(fā)展正在改變創(chuàng)新官僚機構的運作邏輯。過去，許多創(chuàng)新機構可以預設一個相對穩(wěn)定的全球環(huán)境，并主要關注增長、競爭力或產(chǎn)業(yè)升級。如今，這些已經(jīng)遠遠不夠了。今天的創(chuàng)新官僚機構還必須應對戰(zhàn)略依賴、數(shù)字基礎設施、算力和數(shù)據(jù)瓶頸、網(wǎng)絡安全、供應鏈風險以及技術主權政治等問題。換言之，它們不僅僅作為創(chuàng)新組織，還必須扮演韌性機構的角色。這正是我們的工作論文主張從“敏捷性穩(wěn)定”向“敏捷性韌性”轉變的原因。

我的確認為新型公共創(chuàng)新組織將會涌現(xiàn)。然而，其關鍵特征將不在于單純追求新穎性，而在于融合。最有效的組織可能會更具“新韋伯式”特征：它們具備專業(yè)基礎和合法性，但同時保持開放、網(wǎng)絡化、具有試錯精神，并且能夠進行跨部委、跨部門和跨政府層級的協(xié)同工作。它們需要更強的內部技術能力，需要創(chuàng)新政策與安全政策之間更緊密的聯(lián)系，以及為應對危機建立儲備和緩沖機制的能力。未來需要的不是更少的官僚制，而是更有能力的官僚制：即在一個更加嚴酷的世界中，能夠將方向引領、學習能力和制度耐心充分結合的公共組織。

1. On the Role of Government in Innovation.

Science Pie: The role of government in innovation has long been debated in academic circles. In the most concise terms, how would you characterize the core functions or roles that government should fulfill in fostering innovation?

Rainer Kattel: In the most concise terms, I would say this: the core role of government in innovation is not merely to correct market failures, but to build and govern the institutional conditions under which innovation becomes possible, durable, and socially useful. That means funding long-term research, shaping markets, procuring new solutions, regulating direction, coordinating actors across sectors, and, crucially, building public organizations that can learn and adapt over time. Innovation is not produced by entrepreneurs or firms alone; it depends on public institutions with legitimacy, competence, and the ability to hold a long-term course. That is why our book argued that innovation needs bureaucracy.

I would add that government’s role today is broader than in many older innovation-policy debates. In a world shaped by AI, climate stress, geopolitical rivalry, and digital infrastructures, the state must not only foster innovation, but also build resilience. So the real question is not simply how government promotes innovation, but how it develops the capacities and dynamic capabilities to do so under turbulent conditions. That is the move we make in the working paper from agile stability to agile resilience.

2. On Successful and Failed Cases

Science Pie: How to Make an Entrepreneurial State offers in-depth analyses of numerous successful innovative bureaucratic organizations. Among these cases, which institution do you consider the best exemplar of an effective entrepreneurial state? What were the key factors behind its success?

Additionally, during your research, did you also encounter cases where similar organizations failed to deliver results? What institutional or structural problems underlay those failures?

Rainer Kattel: If I had to pick one institution that best captures the logic of the book, I would choose the Physikalisch-Technische Reichsanstalt, the late nineteenth-century German institute that helped build the foundations of modern scientific and industrial capacity. Today, nobody really knows this example but it is archetypical for the way bureaucracy and innovation are interwoven. It’s story shows the whole arc of innovation bureaucracy: first, charismatic coalition-building and institutional entrepreneurship; then the gradual routinization of that energy into a durable expert organization. In other words, it embodies what we call agile stability: the movement from exceptional initiative to stable capability. In the contemporary era, DARPA is of course a highly visible example, but even there the key lesson is not the brilliance of one agency in isolation. The real strength came from a wider system: procurement power, stable funding, universities, and a national-security architecture able to absorb uncertainty.

As for failure, our research suggests that failure often does not come in the dramatic form of one institution collapsing. More often it appears as fragmentation, hollowing-out, or imitation without institutional depth. In the United States, we describe a later phase of “DARPA-fication”: the spread of DARPA-like islands without rebuilding the broader bureaucratic capacities needed to make mission-oriented policy systemic. In Europe, the pattern was often missions without sufficiently mission-capable bureaucracies: strong rhetoric, many programmes, but too little organizational middle layer for scaling, coordination, and learning. These are not failures of ambition; they are failures of institutional architecture.

3. On the Challenges of Institutional Transplantation

Science Pie: Many countries have attempted to replicate successful models of innovative bureaucratic organizations from other nations—such as establishing their own version of DARPA. Yet in practice, such institutional transplants often yield limited results. Have you identified any notable “failed replication” cases in your research? What local conditions or governance elements are indispensable for such institutions to take root effectively?

Rainer Kattel: Yes, this was one of the clearest patterns we found. Many governments try to copy an organizational form—a DARPA, an innovation lab, a mission unit—without reproducing the wider ecology that made the original effective. But successful innovation bureaucracies are never just “one clever agency.” They are embedded in political support, funding arrangements, procurement systems, professional norms, technical talent, and broader state capacities. When those foundations are missing, the transplanted institution often becomes a symbolic layer rather than a real engine of transformation.

So the indispensable local conditions are not cosmetic. They include political legitimacy, long-term commitment, protection from short-term performance pressures, capable staff, links to universities and firms, and enough authority to coordinate across organizational boundaries. This is why I would be cautious about “best practice transfer” in innovation policy. Institutions do not travel as blueprints. They travel, at best, as heuristics. The deeper lesson of our book is that each country has to build its own version of agile stability out of its own administrative traditions, political settlements, and strategic priorities.

4. On the Contemporary Relevance of Mission-Oriented Organizations

Science Pie: The Apollo moon landing gave rise to a mission-driven paradigm for organizing innovation. Today, as we confront complex challenges like climate change, public health crises, and digital transformation, is this “mission-oriented” model still viable? Which organizations around the world do you believe have best inherited and advanced this approach in the current era?

Rainer Kattel: The mission-oriented model is not only still viable; it is more necessary than ever. But it needs to be understood differently from the Apollo era. Apollo was a relatively bounded, technically legible mission. Today’s challenges—climate change, ageing, public health, AI governance, digital transition—are more open-ended, more politically contested, and more system-wide. So contemporary missions cannot be run simply as heroic projects with a single end point. They need to be organized as portfolios, ecosystems, and long-term institutional commitments.

The organizations that best carry this forward are those that combine direction with learning. In our work, China is important precisely because it shows that strong central direction can coexist with decentralized experimentation. More broadly, the most interesting contemporary organizations are those that do not treat missions as slogans, but as ways of organizing public capabilities across funding, regulation, procurement, experimentation, and delivery. That is also why we argue that the most successful agencies are increasingly neo-Weberian: they combine long-term legitimacy and professionalization with openness, experimentation, and coalition-building.

5. On Agile Public Investment

Science Pie: Technological innovation is accelerating rapidly, with high uncertainty in technological trajectories and significantly elevated investment risks. From a governance perspective, how should governments design their innovation funding mechanisms to simultaneously support long-term, high-risk exploration while remaining agile enough to adapt to the fast-paced evolution of technology?

Rainer Kattel: Governments need to design funding mechanisms that are patient in their horizon but agile in their operation. In practice, that means a portfolio approach: supporting a mix of exploratory, high-risk, long-term investments rather than demanding that every project justify itself in advance by near-term returns. Public investment in innovation should accept uncertainty as normal. The task is not to eliminate risk, but to govern it intelligently. That requires staged financing, regular learning loops, technical evaluation capacity inside the state, and the ability to reallocate resources as the technological and strategic landscape changes.

Just as importantly, funding mechanisms should not be separated from broader public capability. Money alone does not create innovation. Governments also need procurement power, regulatory agility, in-house expertise, and institutions able to coordinate across silos. One of the book’s core arguments is that dynamic capabilities are nested in longer-term capacities. So agile public investment depends on durable institutions: the skills, legitimacy, and organizational depth that allow states to learn from failure without collapsing into short-termism or political panic.

6. On Emerging Forms of Innovation Bureaucracies in a New Era

Science Pie: Artificial intelligence is profoundly reshaping scientific research paradigms, while geopolitical tensions are intensifying technological competition and supply chain reconfiguration. How do you expect these trends to transform the operating logic of traditional innovation bureaucracies? Do you foresee the emergence of new forms of public innovation organizations? If so, what key characteristics might they possess?

Rainer Kattel: AI and geopolitics are already changing the operating logic of innovation bureaucracies. In the past, many innovation agencies could assume a relatively stable global environment and focus mainly on growth, competitiveness, or sectoral upgrading. That is no longer enough. Today innovation bureaucracies must also deal with strategic dependence, digital infrastructures, compute and data bottlenecks, cyber-security, supply-chain risk, and the politics of technological sovereignty. In other words, they must operate not only as innovation organizations, but also as institutions of resilience. That is precisely why the working paper argues for moving from agile stability to agile resilience.

I do think new forms of public innovation organizations will emerge. The key features, however, will not be sheer novelty. They will be hybridization. The most effective organizations will likely be more neo-Weberian: professionally grounded and legitimate, but also porous, networked, experimental, and able to work across ministries, sectors, and levels of government. They will need stronger in-house technological competence, closer links between innovation and security policy, and a greater ability to build reserves and buffers for crisis response. The future is not less bureaucracy. It is more capable bureaucracy: public organizations able to combine direction, learning, and endurance in a much harsher world.

Rainer Kattel（雷納·卡特爾），倫敦大學學院（UCL）創(chuàng)新與公共目標研究所（IIPP）副所長，教授，《打造創(chuàng)新型國家》作者。本文由上海市科學學研究所李輝研究員采訪。文章觀點不代表主辦機構立場。

◆ ◆ ◆

編輯郵箱：sciencepie@126.com