经济与社会科学2020nber.org
停滞与科学激励
Stagnation and Scientific Incentives
入门 · 约 60 分钟读完前置知识：科研评价体系基本了解、引用计量概念
导读
这篇 NBER 论文研究科学界的“停滞”问题：认为以引用量为核心的科研评价体系，使科学家在边际上倾向于选择渐进式、低风险的增量研究，而非真正新颖的探索。论文讨论了科研激励机制如何影响创新产出，并探讨改进新颖性度量与激励的可能方向。
HN 关注点：HN 读者长期关注“科学是否停滞”及科研激励扭曲问题，讨论焦点在于引用导向的评价体系是否压制了突破性研究，以及奖励新颖性是否会加剧可重复性危机。
HN 评论精选
这几条评论里：质疑 2 补充 2
质疑停滞是常态而非激励问题，且低引用已在正确惩罚坏研究
作为学术界人士，我要再次说一个无聊但必要的观点：发现自然基本规律本来就极难，停滞才是常态，20世纪中叶的进展只是量子力学和相对论两场大革命的余震，不可能按需再来一次。基本定律本来就稀缺。每当有人说我们在停滞，我就让他们举出替代方向，结果所有答案都落入这几类：已有大量人研究的成熟子领域；50年前就被仔细排除的想法；需要超过该领域未来50年总经费的方向；只是把现有定律重写成几乎没人看得懂的数学形式、且无新预言的形式体系；不比简单模型更有预测力、却给建模者几百个可调旋钮的复杂特设模型；以及用更“深刻”的词汇重写现有定律的形而上学建议。我不认为改变引用机制能解决这些问题——恰恰相反，据我经验，那些糟糕的东西正是被低引用正确惩罚了。
knzhou查看英文原文
As an academic I will again make the dull but necessary point: figuring out the fundamental laws of nature is hard, stagnation is the norm, and the advances of the mid-20th century were the aftershocks of the great revolutions of quantum mechanics and relativity. You can't just make that happen again on demand. Fundamental laws are in short supply. Whenever people tell me we're stagnating I ask them to name an alternative. So far, from best to worst case, every answer has fallen into one of the following buckets: - a subfield that already exists and already has plenty of people working on it - an idea that was extensively investigated and carefully ruled out over 50 years ago - something that requires more money than the field will receive in total over the next 50 years - a mathematical formalism which simply refactors existing laws in a way that makes them unreadable to almost everyone, without a chance of leading to any new predictions - a complicated, ad hoc model that isn't any more predictive than simpler models, but gives the modeller hundreds of shiny knobs to tune - a metaphysical suggestion about how to view what nature "truly" is, which amounts at best to rewriting existing laws with bigger, more "profound" words I can't see how this is going to be fixed by changing how citations work. If anything, in my experience the particularly bad stuff has been correctly punished by low citations.
补充奖励新颖性的前提是社会容忍“无用”研究
允许更多新颖科学的最难之处，在于接受更多“无用”研究会被做出来。读老一辈的故事会发现，教授们常利用自由去做完全无用的东西，纯粹出于个人偏好——同样的自由也让另一些人做出了伟大成果。如今研究者受到的监督多得多，不仅要对资助方负责，还要面对公众的问责（“为什么资助漂浮蟾蜍这种研究？！”）。如何屏蔽这种愤怒、让公众接受（看似）无用的研究，与论文提出的新颖性新指标同样重要。
city17查看英文原文
The most difficult thing of allowing for more novel science is accepting that more 'useless' research will be done. If you read stories about the old days, you'll see how often professors/researchers used the freedom they had to do absolutely useless things, just based on their own personal preferences. The same freedom of course allowed others to do great things. I think nowadays there is much more awareness about what researchers are doing, and they will be held accountable, not just by the people who fund them, but also by the general public (Why are we funding this research about levitating toads?!?!). Shielding the researchers from such outrage and building acceptance for (seemingly) useless research will be just as important as the article's suggested new metrics for novelty.
补充引用指标催生“科学内容农场”，类比 PageRank 与内容农场
论文说“以引用衡量科研生产力，使科学家的奖励和行为在边际上转向增量科学”——这听起来像搜索引擎催生内容农场的科学版：引用分数之于垃圾科学，正如 PageRank 之于 wikihow？想象一下，如果有一个来自未来的高科技垃圾过滤器，能滤掉 wikihow 和无聊科学，那该多好——虽然我怀疑根本造不出来。
awinter-py查看英文原文
> This emphasis on citations in the measurement of scientific productivity shifted scientist rewards and behavior on the margin toward incremental science this sounds like the science equivalent of how search engines lead to the creation of content farms. citation scores are to spam science as pagerank is to wikihow? Imagine going through life with a high-tech spam filter from the future that can filter out wikihow and boring science. Impossible to build, I suspect, but that would be the life.
质疑质疑论文前提：过度奖励新颖性正是可重复性危机的诱因
“新颖性应得到更好奖励”这个前提在我看来很奇怪，因为对科学期刊的常见抱怨恰恰是过度奖励新颖性而非稳健性？这不正是可重复性危机的成因之一吗？例如《科学中的新颖性——真正的必需还是令人分心的执念？》《事实比新颖更重要：教育科学中的重复研究》等文章都指出，重视“新颖”胜过重复验证的资助机构和期刊难辞其咎。我还没读完这篇43页的论文，但快速搜索“replic”和“repro”表明他们完全没有讨论这一点。
laughinghan查看英文原文
This premise that novelty should be better-rewarded seems odd to me, because I thought a common complaint about scientific journals was overly rewarding novelty instead of robustness? Isn't that considered one of the contributing factors to the replication crisis? For example: "Novelty in science – real necessity or distracting obsession?" https://phys.org/news/2018-01-novelty-science-real-necessity-distracting.html "Facts Are More Important Than Novelty: Replication in the Education Sciences" https://journals.sagepub.com/stoken/rbtfl/w5mrNxPVD8zSg/full In the text: "Funding bodies and academic journals that value “novelty” over replication deserve blame too." https://theconversation.com/science-is-in-a-reproducibility-crisis-how-do-we-resolve-it-16998 Etc. I haven't made it through the entire 43-page paper yet, but a quick search for "replic" and "repro" suggests they don't address this point at all.
相关资源
从这篇论文所有 HN 讨论帖的全部评论里整理出来的链接。
历次讨论
- 2020Hacker News 讨论帖184 分 · 107 评
同领域还值得读
一切的回报率，1870–2015
The Rate of Return on Everything, 1870–2015
利用1870–2015年多国的长期数据，估算股票、住房等各类资本的回报率，并将其与经济增长率进行比较。
经济与社会科学2019 · 487 分 · 260 评进阶 · 约 100 分钟编辑精选一切资产的回报率，1870–2015
The Rate of Return on Everything, 1870–2015
构建了 1870–2015 年 16 个国家的长周期资产回报数据集，覆盖股票、住房、政府债券和短期票据，系统比较各类资产的真实回报与波动性。
经济与社会科学2018 · 215 分 · 151 评进阶 · 约 120 分钟编辑精选一切资产的回报率，1870–2015（2019）
The Rate of Return on Everything, 1870–2015 (2019)
这篇发表于《经济学季刊》的论文构建了 1870–2015 年间 16 个主要发达国家的股票、债券、房地产等资产收益率的长周期数据库。
经济与社会科学2024 · 190 分 · 148 评进阶 · 约 80 分钟编辑精选想法是否越来越难找？
Are ideas getting harder to find?
探讨“想法是否越来越难找到”：通过衡量多个领域的“研究生产率”，发现维持相同的增长需要投入指数级增长的研究资源，即研究效率在持续下降。
经济与社会科学2016 · 182 分 · 106 评入门 · 约 60 分钟编辑精选比特币与区块链的经济极限
The Economic Limits of Bitcoin and the Blockchain
芝加哥大学 Eric Budish 的论文从经济学角度分析比特币与区块链的根本限制：核心论点是只要攻击网络（双花或砸盘做空）的收益超过挖矿成本，51% 算力攻击就有利可图，而挖矿…
经济与社会科学2018 · 176 分 · 37 评进阶 · 约 90 分钟编辑精选