作者與出版商就Anthropic人工智能15億美元和解協議展開爭奪（2/2）

 Recently The New York Times reported the following:

Authors Wrangle with Publishers Over $1.5 Billion Anthropic A.I. Settlement (2/2)

The tech giant must pay $3,000 per pirated book that it used to train its chatbot. Many authors fear they could lose funds to others in the book business.

The NYT - By Neil Vigdor

Sept. 5, 2026

(continue)

Another author, Amy Lupold Bair, who has written guidebooks about blogging and how families can navigate the digital world, vented on social media over her potential diminishing share of the settlement.

“My publisher is fighting the default 50/50 payout split from the Anthropic class action settlement,” she wrote on Threads. “They only want me, the author whose entire work was stolen, to get 10%. So that’s a fun email to wake up to.”

Her post, which did not name the publisher, received more than 1,000 likes and started a lengthy discussion about the settlement process.

She declined to comment.

John Wiley & Sons Inc., which was listed as the publisher for two of her books on her website, “Raising Digital Families for Dummies” and “Blogging for Dummies,” did not address whether it was seeking a 90 percent cut.

In a statement on Friday, a representative for the company said that it was following the settlement terms when it filed claims for “all Wiley-published” works covered by the Anthropic agreement.

Allocations for educational titles were set by individual contracts, said the representative, which did not immediately say whether Wiley put its Dummies guidebook series in that category.

The Association of American Publishers, a trade group that represents most of the major publishing houses in the United States and has hundreds of members, did not immediately provide a comment on Friday.

In 2024, a group of authors filed a class-action lawsuit against Anthropic after learning that the company, seeking huge amounts of digital text needed to train A.I. technologies, had used their work without paying for it and stockpiled millions of pirated books.

Anthropic chose to settle the case after a judge paved the way for the case to go to trial, finding that the company’s stealing of books had given the plaintiffs cause.

At the same time, the judge said that the company’s training of its chatbot with books that it had legally acquired was a “fair use.”

Anthropic referred to a statement in May by Aparna Sridhar, the company’s deputy general counsel, as the settlement was going through final approval.

“We are pleased that more than 91 percent of authors and publishers covered by the settlement have claimed their share of the payment, and we’re looking forward to bringing this matter to a close,” the statement said.

JND Legal Administration, the settlement administrator, did not immediately respond to requests for comment on Thursday. Neither did a lawyer for the plaintiffs in the case.

Kristelia García, a professor at Georgetown Law who specializes in copyright law, said the dynamics of divvying up the Anthropic settlement funds were reminiscent of a 2007 federal lawsuit over the distribution of digital royalties for the music of the rapper Eminem.

Producers who discovered Eminem sued a subsidiary of the Universal Music Group and argued that they should receive a higher percentage of royalties from downloads at iTunes and other digital stores.

The two sides reached a settlement in 2012.

Not all book contracts have language about who is entitled to settlement payments in copyright infringement cases or revenues from emerging technologies, including A.I., García said.

“So I think this is just kind of the publishing industry having that sort of streaming moment where their contracts didn’t contemplate this,” she said.

Muddying the settlement process further, according to the Authors Guild, are more recent book ownership changes.

Rasenberger said an author contacted the guild this week with a dilemma: Rights to the books had reverted to the author this year.

“It depends on who owned the rights when the books were downloaded when the infringement happened, which is, you know, 2021 and ’22,” Rasenberger said.

In instances when ownership disputes remain unresolved, Rasenberger said, a court-appointed arbitrator will have the final say, but she added the guild “will fight tooth and nail for them if they are, in fact, being screwed over.”

Translation

作者與出版商就Anthropic人工智能15億美元和解協議展開爭奪（2/2）

這家科技巨頭必須為每本用於訓練其聊天機器人的盜版書籍支付3,000美元。許多作者擔心他們的資金可能會被其他出版業人士所瓜分

（繼續）

另一位作者 Amy Lupold Bair 撰寫了關於部落格寫作和家庭如何駕馭數位世界的指南，她在社交媒體上表達了對和解協議中可能減少的份額的不滿。

她在Threads上寫道:「我的出版商正在反對Anthropic集體訴訟和解協議中預設的50/50賠償分配方案」; 「他們只想讓我 - 一個作品全部被盜用的作者 - 拿到10%。所以，一早醒來就收到這樣的電郵，真係幾開心喎」。

她的貼文並未指明出版商，但獲得了超過1,000個贊，並引發了關於和解流程的長時間討論。

她拒絕置評。

約翰威立父子出版公司（John Wiley & Sons Inc.）是她網站上列出的兩本書的出版商，兩書分別是《養育數位家庭入門指南》（Raising Digital Families for Dummies）和《部落格入門指南》（Blogging for Dummies）。該公司並未回應是否要求90%的版稅。

該公司一位代表在週五的聲明中表示，他們已按照和解條款，就所有受Anthropic協議約束的 “威立(Wiley)出版” 作品提出索賠。

這位代表表示，教育類書籍的版稅分享比例由單獨的合約決定，但並未立即說明威立是否將其「入門指南」系列歸入此類別。

美國出版商協會（Association of American Publishers）是一個代表美國大多數主要出版社的行業組織，擁有數百名會員。該協會週五未立即對此事發表評論。

2024年，一群作家對 Anthropic 公司提起集體訴訟，因為他們發現該公司為了訓練人工智能技術，未經許可使用了他們的作品，並囤積了數百萬冊盜版書籍。

在法官裁定該公司盜版書籍的行為構成原告的訴訟理由，允許案件進入審判程序後，Anthropic公司選擇與原告達成和解。

同時，法官表示，該公司使用合法取得的書籍訓練聊天機器人屬於「合理使用」。

在和解協議最終獲批之際，Anthropic 公司援引了其副總法律顧問 Aparna Sridhar 於5月發表的聲明。

聲明中寫道：“我們很高興超過91%的受和解協議約束的作家和出版商已領取了他們應得的賠償金，我們期待著此事​​能夠圓滿結束。”

和解管理人 JND Legal Administration 週四未立即回應置評請求。本案原告律師也未回應。

喬治城大學法學院專攻版權法的 Kristelia García 教授表示，Anthropic 和解金的分配方式讓人想起2007年一起關於說唱歌手 Eminem 的電子音樂版稅分配的聯邦訴訟。

發掘 Eminem 的製作人起訴了環球音樂集團的子公司，並主張他們應該從iTunes和其他電子商店的下載中獲得更高的版稅比例。

雙方於2012年達成和解。

García 指出，並非所有圖書合同都包含有關在版權侵權案件中指出誰有權獲得和解金，或源自新興技術（包括人工智能）的收入的條款。

她說:「所以我認為這只是出版業在經歷串流媒體時代之後，合約中沒有考慮到這一點的一種體現」。

據美國作家協會稱，近期圖書版權所有權的變更進一步加劇了和解過程的複雜性。

Rasenberger 表示，本週有一位作家聯繫了協會带來了一個難題：今年圖書的版權又返回到了作者手中。

Rasenberger 說: 「這取決於侵權發生時 - 也就是2021年和2022年 - 圖書被下載時的版權歸屬」。

Rasenberger 表示，如果版權糾紛無法解決，法院指定的仲裁員將擁有最終裁決權，但她補充說，如果作家協會「確實受到了不公平對待，會將竭盡全力為他們爭取權益」。

              So, in 2024, a group of authors filed a class-action lawsuit against Anthropic after learning that the company, seeking huge amounts of digital text needed to train A.I. technologies, had used their work without paying for it and stockpiled millions of pirated books. Now, thousands of authors are waiting for their share of the largest copyright settlement in U.S. history from Anthropic. In instances when copyright ownership disputes remain unresolved, a court-appointed arbitrator will have the final say. Apparently, training the chatbot is a costly endeavor.

作者與出版商就Anthropic人工智能15億美元和解協議展開爭奪（1/2）

Recently The New York Times reported the following:

Authors Wrangle with Publishers Over $1.5 Billion Anthropic A.I. Settlement (1/2)

The tech giant must pay $3,000 per pirated book that it used to train its chatbot. Many authors fear they could lose funds to others in the book business.

The NYT - By Neil Vigdor

Sept. 5, 2026

The sum is gargantuan: $1.5 billion.

But as thousands of authors wait for their cut of the largest copyright settlement in U.S. history from Anthropic, the tech giant that used pirated books to train its artificial intelligence chatbot, some are accusing the publishers of their books of trying to squeeze them out of a percentage of the payout.

The haggling is coming into plain view as a settlement administrator attempts to clear up who owns the rights to some of the more than 482,000 books covered by the agreement in the class-action case, which was approved in July by a judge in the U.S. District Court in the Northern District of California.

For many of those titles, ownership is undisputed, but on repeated occasions, authors and publishers have staked competing claims with the settlement administrator, which began recently informing them of those discrepancies, said Mary Rasenberger, chief executive of the Authors Guild, a professional association that says it has more than 18,000 members.

In an interview on Thursday, Rasenberger said the settlement process was ripe for such misunderstandings.

“My big fear when this was all set up is that not all publishers keep great records of what books they’ve reverted rights to,” Rasenberger said. “And so you know, they should be taking it off their catalog.”

Rasenberger said she did not believe that publishers were acting in bad faith.

“I don’t see this as a grab by the publishers,” she said, adding, “I don’t think they’re specifically trying to screw any author over.”

Authors could receive up to $3,000 for each one of their books that a judge ruled last year were illegally downloaded and stored by Anthropic in the development of its online chatbot Claude.

But they must split the proceeds with publishers they granted book rights to and with co-authors, according to the Authors Guild, a trade group.

Rasenberger said the percentages were determined by a class-action council that sought input from the guild and publishers.

Textbook authors will receive far less — in some cases, 10 to 15 percent of the total — as determined by their contracts with their publishers, she said.

“It’s the textbooks where there are a lot of unhappy authors right now,” Rasenberger said.

April Henry, a New York Times best-selling author of more than 30 mysteries and thrillers for teens and adults, said she was baffled when she logged into the settlement claims portal to see that HarperCollins, her former publisher, had claimed partial ownership of her first book, “Circles of Confusion.”

Henry, 67, said in an interview on Thursday that the rights to the book, published in 1999, had reverted to her in 2007.

“I was like, well, that’s not right,” said Henry, echoing a complaint she made on social media on Wednesday.

Her agent provided her with a letter confirming that she owned the book’s rights that Henry said she uploaded to the portal. When she checked back in the portal, she said, she learned that she would receive the full $3,000 for that book.

“I don’t think Harper was deliberately trying to cheat,” she said.

HarperCollins declined to comment.

Henry said she has 22 titles on the list of books that Anthropic had illegally obtained, and that she expected to receive in the mid-$20,000 range after her publishers and co-authors collect their cut.

Having to share part of the settlement with a publisher was a source of frustration for Henry, who said she was first told that the payments could start flowing in August, only to hear that it might not happen until later this year.

(to be continued)

Translation

作者與出版商就Anthropic人工智能15億美元和解協議展開爭奪（1/2）

這家科技巨頭必須為每本用於訓練其聊天機器人的盜版書籍支付3,000美元。許多作者擔心他們的資金可能會被圖書業的其他人所瓜分

金額高達15億美元。

數千名作者正等待著從Anthropic（這家科技巨頭使用盜版書籍訓練其人工智能聊天機器人）那裡分得美國歷史上最大的版權和解金，但一些作者指責他們的出版商試圖從中榨取一部分賠償金。

隨著和解管理人試圖釐清這件集體訴訟案中涉及的48.2萬餘本圖書的版權歸屬問題，各方討價還價的局面也逐漸浮出水面。該集體訴訟案已於7月獲得加州北區聯邦地區法院法官的批准。

美國作家協會（Authors Guild）執行長Mary Rasenberger表示，許多圖書的版權歸屬並無爭議，但作者和出版商卻屢次向和解管理人提出相互衝突的版權主張。該協會自稱擁有超過1.8萬名會員。

Rasenberger 在週四接受採訪時表示，和解過程本身就容易出現此類誤解。

Rasenberger 說: “我當初最擔心的是，並非所有出版商都會妥善保存已歸還版權書籍的記錄” ， “所以你知道喇，他們應該把這些書從目錄中移除。”

Rasenberger 表示，她不認為出版商有惡意行為。

她說道:“ 我不認為這是出版商的掠奪行為”，又說:“我不認為他們是故意想害任何作者。”

去年，一位法官裁定，Anthropic公司在開發其線上聊天機器人Claude的過程中非法下載並儲存了作者的書籍。根據該裁決，作者每本書最高可獲得3,000美元的賠償。

但據美國作家協會（一個行業組織）稱，作者必須與獲得書籍版權的出版商，以及共同作者平分這筆賠償金。

Rasenberger表示，這些分攤比例是由一個集體訴訟委員會决定的，該委員會徵求了作家協會和出版商的意見。

她也表示，教科書作者獲得的賠償金要少得多 - 在某些情況下，只有總額的10%到15% - 而具體數額取決於他們與出版商簽訂的合約。

Rasenberger 說: 「現在最讓作者們不滿的就是教科書了」。

《紐約時報》暢銷書作家April Henry著有30多部青少年和成人的懸疑驚悚小說。她說，當她登入版權索賠入口網站時，發現她之前的出版商哈珀柯林斯（HarperCollins）聲稱擁有她第一本書《混亂之環》（Circles of Confusion）的部分版權，這讓她感到困惑不已。

67歲的 Henry 在周四的採訪中表示，這本書於1999年出版，版權已於2007年歸還給她。

Henry 說道: 「我當時就想，這不公平」，這與她週三在社交媒體上的抱怨如出一轍。

她的代理人給她提供了一封信函，確認她擁有該書的版權。Henry說，她已將這封信上傳到了索賠入口網站。她說，當她再次登入門戶網站查看時，得知她將收到該書的全部3,000美元。

她說:「我不認為哈珀是故意作弊」。

哈珀柯林斯出版社拒絕置評。

Henry 表示，她有22部作品被列入Anthropic非法取得版權的名單中，她預計在扣除出版商和合作作者的分成後，她將獲得2萬美元左右的賠償。

Henry 感到沮喪的是，她不得不與出版商分享部分賠償金。她說，她最初被告知款項可能在8月開始支付，但後來又聽說可能要到今年晚些時候才能支付。

（待續）

科學家發現1,000多個與人格相關的基因變異（2/2）

Recently The New York Times reported the following:

Scientists Find More Than 1,000 Genetic Variants Linked to Personality (2/2)

More than one million genomes helped identify DNA variants linked to extroversion, agreeableness, neuroticism and more. Researchers measured how those traits influenced how people lived.

The NYT - By Emily Baumgaertner Nunn (Emily Baumgaertner Nunn is a national health reporter for The Times, focusing on public health issues that primarily affect vulnerable communities.)

Sept. 3, 2026

(continue)

“It was kind of a conspicuous absence,” said Dr. Tucker-Drob, whose academic work focuses on “individual differences,” a realm of psychology that examines the ways people diverge in their thoughts, feelings and behaviors. To him, personality is not just an abstract concept: The way people engage with themselves and the world can help determine how much money they make, the types of medical care they seek, and how strong their friendships are.

Personality is “meaningfully related to all sorts of outcomes that I think everyone in society values,” he said.

The research team scanned roughly 10 million genetic variants across the genome in order to map the “Big Five.” By aggregating disparate studies on various traits, researchers were able to uncover the ways personality predispositions rippled out into health, habits and lifestyle choices.

Researchers found 258 different genetic variants linked to extroversion, for example, a 15-fold increase in the number of variants previously known. A genetic profile with high extroversion seemed to reduce a person’s risk of anxiety, depression and post-traumatic stress disorder, all of which involve internalizing distress. But the trait was also linked to neurodevelopmental conditions such as A.D.H.D.

People who were predisposed to conscientiousness — or a capacity for productivity and organization — appeared more likely to exercise, choose low-calorie foods and sleep well. Genetic scores for conscientiousness helped predict who would move out of urban centers and into affluent suburbs by age 50.

Neuroticism, which is defined as a tendency toward emotional instability, was predictably correlated with 10 different psychiatric disorders, plus a higher body mass index and increased rates of smoking. Study participants with genetic profiles high in neuroticism seemed to second-guess themselves in real time, frequently selecting options such as “I don’t know” on research survey questions.

Since neuroticism can be a strong predictor of mental illness, some researchers are particularly interested in how the new findings illuminate possible triggers and avenues of disease. Dalton Conley, a sociologist at Princeton University who was not involved in the research, was particularly intrigued by the way the consortium’s results challenged a long-held theory among psychologists that neuroticism is primarily driven by differences in serotonin levels. Depression and anxiety are often successfully treated with medications that boost the chemical.

The new analysis showed that neuroticism variants were indeed especially active in brain cells — but revealed nothing pronounced about the serotonin system in particular. “I wouldn’t say it’s slaying that theory forever,” Dr. Conley said, “but it’s definitely casting doubt.”

Another unexpected finding came in the context of relationships. Behavioral geneticists know that people tend to select life partners with similar DNA profiles in areas such as educational attainment. But in the new study, personality traits did not show much so-called “assortative mating” at all: In fact, personality variants among partners were matched mostly at random.

In contrast, spouses’ actual personalities — specifically their openness to experience — were sometimes correlated, perhaps because sharing daily life makes people more similar over time. Dr. Conley, who studies the interplay between genes and the environment, said that such hidden phenomena, scattered throughout the data set, are helpful for researchers like himself who are interested in understanding cause and effect.

“The next study that has double the sample size will find 3,000 variants, no surprise,” he said. “But the fact that we can now look into the dynamics between personality and marital stability, or personality and the labor market, or personality and the education system — that’s what makes this a really useful tool.”

Translation

科學家發現1,000多個與人格相關的基因變異（2/2）

超過一百萬個基因組幫助科學家辨識出與外向性、親和力、神經質等人格特質相關的DNA變異。研究人員測量了這些特質如何影響人們的生活方式

（繼續）

Tucker-Drob博士，一個在心理學上專注去研究人在思想、情感和行為上的差異， 叫 “個體差異” 的學術研究領域的人說: 「這確實是一個明顯的空白」。對他而言，人格不僅僅是一個抽象的概念：人們與自身和世界互動的方式會影響他們的收入、他們尋求的醫療保健類型，以及他們交友的倾向度。

他說：“個人性格與我認為社會上每一個人都會重視的各種實況都有密切相關。”

研究團隊掃描了基因組中約1,000萬個基因變異，以繪製「五大人格特質」圖譜。透過整合各種特質的不同研究，研究人員得以揭示人格傾向如何影響健康、習慣和生活方式的選擇。

例如，研究人員發現了258種與外向性相關的基因變異，比先前已知的變異數量增加了15倍。具有高度外向性的基因特徵似乎可以降低個體焦慮症、憂鬱症和創傷後壓力症候群的風險，這些疾病都涉及內化痛苦。但這種特徵也與過動症（ADHD）等神經發育障礙有關。

那些具有責任感（或稱為高效能和組織能力）的人似乎更有可能鍛鍊身體、選擇低熱量食物並擁有良好的睡眠。責任感的基因分數有助預測哪些人會在50歲時從市中心遷往富裕的郊區。

神經質被定義為一種情緒不穩定的傾向，它與10種不同的精神疾病、更高的體重指數，以及更高的吸煙率均有顯著相關性。研究參與者中，神經質基因評分較高的人似乎會即時地自我懷疑，經常在問卷中選擇「我不知道」之類的選項。

由於神經質可以作為精神疾病的強烈預測指標，一些研究人員對這些新發現如何揭示疾病的潛在誘因和途徑特別感興趣。普林斯頓大學的社會學家Dalton Conley並未有參與這項研究，但他對該研究結果挑戰心理學家長期以來所持有的一個理論 - 即神經質主要由血清素水平的差異所驅動 - 尤為感興趣。憂鬱症和焦慮症通常可以透過提高血清素水平的藥物來有效治療。

新的分析表明，神經質變異體確實在腦細胞中特別活躍 - 但並未揭示血清素系統有任何顯著差異。Conley博士說: “我不會說它會永遠徹底推翻那個理論”，“但它無疑提出了質疑。”

另一項出乎意料的發現來自人際關係領域。行為遺傳學家知道，人們傾向於選擇在教育程度等方面擁有相似DNA特徵的伴侶。但在這項新研究中，人格特質幾乎沒有表現出所謂的「同型交配」：事實上，伴侶之間的人格變異體大多是隨機匹配的。

相較之下，配偶的實際人格 - 特別是他們對新體驗的開放情度 - 有時卻存在相關性，這或許是因為共同的日常生活會隨著時間的推移使人們變得更加相似。研究基因與環境交互作用的Conley博士表示，這些散佈在資料集之中的各種隱藏現象，對於像他般致力於去理解因果關係的研究人員來說，是非常有幫助。

他說:「下一項樣本量翻倍的研究將會發現3,000個變異，這並不令人意外」 ，「但我們現在可以研究人格與婚姻穩定性、人格與勞動力市場、人格與教育體系之間的動態關係 - 這才是真正令它成為有用的工具」。

So, scientists have long known that personality is influenced by a number of factors — including DNA. The challenge has been deciphering the specific pathways through which a genetic blueprint influences the direction of a person’s life. Recently, a group of researchers find a symphony of 1,260 genetic markers that have ties to personality, nearly two-thirds of them are identified for the first time. Apparently, we can now look deeper into the dynamics between personality and marital stability, or personality and the labor market, or personality and the education system.

科學家發現1,000多個與人格相關的基因變異（1/2）

Recently The New York Times reported the following:

Scientists Find More Than 1,000 Genetic Variants Linked to Personality (1/2)

More than one million genomes helped identify DNA variants linked to extroversion, agreeableness, neuroticism and more. Researchers measured how those traits influenced how people lived.

The NYT - By Emily Baumgaertner Nunn (Emily Baumgaertner Nunn is a national health reporter for The Times, focusing on public health issues that primarily affect vulnerable communities.)

Sept. 3, 2026

Are you naturally anxious? Highly adventurous? More argumentative than you’d like to admit?

Those tendencies are often measured through what psychologists call the “Big Five” dimensions of personality: extroversion, agreeableness, conscientiousness, neuroticism and openness to experience. Each trait exists on a continuum and tends to remain fairly stable throughout adult life.

But what determines where you land on those spectra? Scientists have long known that personality is influenced by a host of factors — including DNA. The challenge has been deciphering the specific pathways through which a genetic blueprint nudges the direction of a person’s life.

To map this sprawling architecture, a group of experts formed a research consortium and analyzed more than one million genomes, searching for variants associated with personality. The researchers also compared the genes of thousands of siblings and parent-child pairs to rule out other factors that could be at play. With data from 46 different study cohorts, they found a symphony of 1,260 genetic markers tied to personality, nearly two-thirds of them identified for the first time.

The findings, published on Wednesday in the journal Nature, reaffirmed scientists’ understanding that temperament and disposition are shaped not by a few particular genes, but by thousands of tiny variants that can be expressed in different circumstances. The sheer scale of the project also brought new statistical power to the study of personality genetics, giving researchers a tool kit to investigate the biological mechanisms behind how we think, feel and behave.

“I view the 1,260 variants as kind of a testament to the fact that we’ve now got the power to answer all of these other questions that we couldn’t answer before,” said Elliot Tucker-Drob, a professor of psychology at the University of Texas at Austin and a leader of the study.

Common genetic differences identified in this study still account for only about 5 to 10 percent of personality variation among people. But experts caution against thinking about personality drivers as a pie chart, divided into neat percentages of genes versus environment. In reality, a person’s DNA and context are constantly interacting; experiences are the very tool that genetic predispositions use to mold who a person becomes.

In studies of genetic influence on educational attainment, much of what appears driven by DNA often turns out to be parental privilege or home environment in disguise. With personality, however, family comparison studies that looked for such confounding variables revealed that about 96 percent of the genetic influence indeed came directly from the DNA itself. That makes intuitive sense: Growing up in the same household might give siblings similar launching pads, but it almost never gives them similar temperaments.

Wednesday’s publication is, in some ways, a revitalization. In the late 20th century — the era of classical twin studies — personality was a central focus of behavioral genetics. But over the last decade, personality genetics has not been studied with nearly the intensity that other traits have, such as the risk of developing schizophrenia, in part because most personality traits were not considered medically relevant.

(to be continued)

科學家發現1,000多個與人格相關的基因變異（1/2）

超過一百萬個基因組幫助科學家辨識出與外向性、宜人性、神經質等人格特質相關的DNA變異。研究人員測量了這些特質如何影響人們的生活方式

你天生焦慮嗎？你充滿冒險精神嗎？你比自己願意承認的更愛爭辯嗎？

這些傾向通常透過心理學家所說的「五大人格特質」來衡量：外向性、宜人性、盡責性、神經質和開放性。每一種特質都存在於一個連續光譜上，並且在成年生活中往往保持相對穩定。

但是，是什麼決定了你在這些特質譜上的位置呢？科學家早已知道，人格受多種因素影響，包括DNA。真正的挑戰在於，如何解讀基因藍圖如何影響人的一生。

為了繪製出這套龐大的基因圖譜，一群專家組成了一個研究聯盟，分析了超過一百萬個基因組，尋找與人格相關的基因變異。研究人員還比較了數千對兄弟姐妹和親子對的基因，以排除其他可能的影響因素。他們利用來自46個不同研究隊列的數據，發現了1,260個與人格相關的基因標記，其中近三分之二是首次被發現。

這項研究成果於週三發表在《自然》雜誌上，再次印證了科學家們的認知：人的氣質和性格並非由少數特定基因決定，而是由數千個微小的基因變異共同帶來作用的結果，這些變異會在不同的情況下表達出來。該計劃的龐大規模也為性格遺傳學研究帶來了新的統計力，為研究人員提供了一套工具，用於探索我們思考、感受和行為背後的生物學機制。

德克薩斯大學奧斯汀分校心理學教授、該研究的負責人之一Elliot Tucker-Drob說道:「我認為這1,260個變異體證明了我們現在有能力回答以前無法回答的所有問題」。

這項研究中發現的常見基因差異仍只佔人格差異的5%到10%。但專家提醒，不要把性格驅動因素看成一個圓餅圖，簡單地劃分成基因和環境的百分比。事實上，一個人的DNA和環境因素一直在相互作用；個人經歷正是基因傾向塑造一個人性格的工具。

在關於基因對教育成就影響的研究中，許多看似由DNA驅動的因素，最終往往被證明是父母的優勢或家庭環境所變相驅動的。然而，在人格方面，針對此類混雜變數的家庭比較研究表明，約96%的遺傳影響確實直接來自DNA本身。這符合直覺：在同一個家庭中成長，或許能為兄弟姊妹提供相似的起點，但幾乎不可能讓他們擁有相似的個性。

週三發表的論文在某種程度上是對人格研究的復興。在20世紀後期 - 經典雙胞胎研究的時代 - 人格是行為遺傳學的核心研究方向。但在過去十年中，人格遺傳學的研究強度遠不及其他性格特徵，例如罹患精神分裂症的風險，部分原因是大多數的個人格特質被認為是無醫學上的關連。

（待續）

每年48萬美元的藥丸反映癌症藥物價格的新常態（2/2）

Recently The New York Times reported the following:

A $480,000-a-Year Pill Reflects a New Normal for Cancer Drugs (2/2)

Rasonque, the new pancreatic cancer treatment from Revolution Medicines, illustrates the huge inflation in cancer drug prices over the past two decades.

The NYT - By Rebecca Robbins (Rebecca Robbins is a Times reporter covering the pharmaceutical industry. She has been reporting on health and medicine since 2015.)

Aug. 31, 2026, 5:03 a.m. ET

(continue)

A drug’s sticker price is not the real price paid by patients or their insurers — it’s just the starting point for negotiations that lower the final price for the employer or the government. But while that dynamic sharply reduces the true prices of diabetes and autoimmune drugs, the final price for cancer drugs is typically only slightly lower than the sticker price.

Cancer drugmakers and their defenders say the prices reflect a boom in innovation that has transformed deadly diseases into chronic ones and even produced cures.

And particularly in the case of smaller companies, they say, there’s a need to recoup considerable investments in clinical trials.

Revolution Medicines, based near San Francisco, has registered $4 billion in losses since its founding in 2014. That is largely because of the $3.3 billion the company has reported spending on research and development this decade. Buoyed by high investor hopes that Rasonque, the company’s first approved product, will become a blockbuster, Revolution has surged to a market valuation of $45 billion.

The company set up an expanded access program that, over the past few months, granted more than 2,000 patients free, early access to the drug. Now, with its drug approved, the company is shifting participants to getting the drug covered through insurance.

Pharmacy benefit managers play a key role in determining whether new pills like Rasonque are broadly covered, by recommending to their employer clients whether to pay for them.

Two of the largest benefit managers, CVS Health’s Caremark and UnitedHealth Group’s Optum Rx, said they had begun a review process for Rasonque.

David Whitrap, a spokesman for CVS, said Rasonque appeared to be a breakthrough. “At the same time,” he said, “it is critical that drugmakers realize that break-the-bank prices can severely limit the number of patients who are able to access these breakthroughs.”

Experts said they expected that most insurance plans would cover Rasonque, even at such a high price, because of the strong evidence that it benefits patients and the lack of alternative options.

Still, the experts said, plans could aggressively seek to limit coverage to people who meet the criteria tested in the clinical trial and approved by regulators: those with advanced pancreatic cancer who have already tried or can’t try chemotherapy. That could shut out many patients seeking it and cause battles over whether a patient is really eligible.

To make sure that ineligible patients aren’t getting the drug, experts predict, plans will use levers like prior authorization, an often cumbersome review that has been criticized for dangerously delaying care.

Revolution Medicines said it would give patients experiencing insurance delays free access to a few weeks’ supply of pills to get them started on treatment while any coverage issues were being sorted out.

Still, commercially insured patients with high deductibles could face huge sticker shock this January, when their deductible resets and they fill their first Rasonque prescription of the year. (The company said such patients might be eligible for assistance with their out-of-pocket costs.)

Commercially insured people typically have an upper limit on their annual out-of-pocket drug and medical costs. But that maximum can be high — in some cases, more than $6,000.

“It will be really a tragedy if this price makes it hard for patients to get it,” Dr. Kesselheim said.

Pancreatic cancer disproportionately affects older people, so many of the patients who will take it are covered under Medicare. Federal law should require Medicare drug plans to cover Rasonque because cancer is a protected class.

The drug offers a vivid example of how Medicare patients benefit from an annual cap on out-of-pocket drug costs created by the Inflation Reduction Act, which was passed in 2022 by Democrats without any Republican votes. That cap, which was $2,100 this year, will rise to $2,400 next year. It covers the pills and injections that patients get from a pharmacy — including Rasonque.

That means that a patient on Medicare who takes several other pricey prescription drugs may pay only a few hundred dollars a year, or even nothing at all, by starting Rasonque.

Had the drug arrived before the law went into effect, many patients would have been on the hook for over $25,000 a year for Rasonque, Dr. Dusetzina calculated.

Translation

 每年48萬美元的藥丸反映癌症藥物價格的新常態（2/2）

Revolution Medicines 公司推出的新型胰臟癌治療藥物 Rasonque，凸顯了過去二十年來癌症藥物價格的巨額上漲

 （繼續）

藥品的標價並非患者或其保險公司實際支付的價格 - 它只是談判的起點，最終由雇主或政府支付的價格會降低。雖然這種機制大幅降低了糖尿病和自體免疫疾病藥物的實際價格，但癌症藥物的最終價格通常只比標價略低而已。

癌症藥物生產商及其支持者表示，高昂的價格反映了創新蓬勃發展，這種創新已將致命疾病轉變為慢性疾病，甚至催生了治癒方法。

他們也指出，尤其對於規模較小的公司而言，是極需收回在臨床試驗方面的大量投資。

總部位於舊金山附近的Revolution Medicines公司自2014年成立以來已累積虧損40億美元。這主要是由於該公司在這十年在研發方面投入了33億美元。由於投資者對其首款獲批產品Rasonque寄予厚望，認為它將成為傑作級藥物，Revolution的市值已飆升至450億美元。

該公司設立了一項擴大用藥計劃，在過去幾個月中，該計劃已為2,000多名患者提供了免費、提前使用該藥物的機會。如今，隨著該藥物獲得批准，該公司正引導參與者透過醫療保險獲得藥物報銷。

藥品福利管理機構在決定像Rasonque這樣的新藥是否能被廣泛醫保覆蓋方面發揮關鍵作用，他們會向雇主客戶建議是否應該支付這些藥物的費用。

兩家最大的藥品福利管理機構 - CVS Health旗下的Caremark和聯合健康集團旗下的Optum Rx - 表示，他們已經啟動了對Rasonque的審查流程。

CVS發言人David Whitrap表示，Rasonque似乎是一項突破。他說道: “與此同時” ， “製藥商必須意識到，高昂的價格會嚴重限制能夠獲得這些突破性療法的患者人數。”

專家表示，他們預計大多數保險計劃都會承保Rasonque，即使價格如此之高，因為有強有力的證據表明該藥對患者有益，而且目前缺乏其他替代方案療法。

不過，專家也指出，一些保險計劃可能會積極限制承保範圍，只允許符合臨床試驗中測試並經監管機構批准的條件的患者使用：即那些患有晚期胰腺癌且已經嘗試過或無法嘗試化療的患者。這可能會將許多尋求該藥物的患者拒之門外，並引發關於患者是否真正符合用藥資格的爭論。

專家預測，為了確保不符合資格的患者無法獲得該藥物，醫保計劃將使用諸如事先授權之類的手段。事先授權通常是一種繁瑣的審查程序，並因其可能危險地延誤治療而備受批評。

Revolution Medicines公司表示，將為遇到保險延誤的患者免費提供幾週的藥量，以便他們在解決任何保險問題期間能夠開始治療。

然而，對於擁有高額自付額的商業保險患者來說，今年1月他們的自付額重置後，首次購買Rasonque時，可能會面臨巨額費用。 （該公司表示，此類患者可能有資格獲得自付費用方面的援助。）

商業保險通常對年度自付藥品和醫療費用設有上限。但這個上限可能很高 - 在某些情況下，甚至超過6000美元。

Kesselheim醫生說: 「如果這個價格讓患者難以負擔，那將是一場真正的悲劇」。

胰臟癌主要影響老年人，因此許多需要服用藥物的患者都享有聯邦醫療保險（Medicare）。聯邦法律應該要求聯邦醫療保險的藥物計劃涵蓋Rasonque，因為癌症屬於受保障的疾病類別。

該藥生動地展現了聯邦醫療保險患者如何受益於《通貨膨脹削減法案》所設立的年度自付藥品費用上限。該法案於2022年由民主黨人通過，當時並未獲得任何共和黨人的支持。今年的上限為2,100美元，明年將提高到2,400美元。此上限涵蓋患者從藥房獲得的藥片和注射劑，包括Rasonque。

這意味著，一位服用多種其他昂貴處方藥的聯邦醫療保險患者，如果開始服用Rasonque，每年可能只需支付幾百美元，甚至無需支付任何費用。

Dusetzina醫生計算，假如這種藥物在法律生效之前就已上市，許多病患每年將需支付超過 25,000 美元的 Rasonque 費用。

             So, as typical with other cancer drugs, the vast majority of the cost of Rasonque is expected to be covered through insurance. This drug offers a vivid example of how Medicare patients benefit from an annual cap on out-of-pocket drug costs created by the Inflation Reduction Act in the US. Buoyed by high investor hopes that Rasonque, the company’s first approved product, will become a blockbuster, the market valuation of Revolution has surged to about $45 billion. Apparently, high drug prices reflect a boom in innovation that has transformed deadly diseases into chronic ones and even may produce new cures for patients.

  

每年48萬美元的藥丸反映癌症藥物價格的新常態（1/2）

Recently The New York Times reported the following:

A $480,000-a-Year Pill Reflects a New Normal for Cancer Drugs (1/2)

Rasonque, the new pancreatic cancer treatment from Revolution Medicines, illustrates the huge inflation in cancer drug prices over the past two decades.

The NYT - By Rebecca Robbins (Rebecca Robbins is a Times reporter covering the pharmaceutical industry. She has been reporting on health and medicine since 2015.)

Aug. 31, 2026, 5:03 a.m. ET

Revolution Medicines set a sticker price of about $480,000 a year, or $663 a tablet, for its new pancreatic cancer drug — a cost that would have once been unthinkably high.

Today, such prices are routine for cancer drugs.

One cutting-edge CAR-T therapy now goes for $600,000 for a one-time infusion. A competitor, given the same way, is listed at $550,000. A daily pill for gastrointestinal cancer has a sticker price of $520,000 a year. A twice-daily pill for leukemia? It’s $430,000 a year.

In 2024 alone, nine cancer drugs were approved and then introduced with sticker prices above $400,000 annually, according to a tally by the Institute for Clinical and Economic Review, which evaluates the value of medicines.

That these prices have become commonplace would have been hard to imagine two decades ago, when even the prospect of a $100,000-a-year cancer drug evoked fierce criticism.

But since then, cancer drug prices have steadily increased, with few constraints beyond what a company thinks the market will bear. The expense is borne by government programs like Medicare, employers, patients and Americans who pay taxes and health insurance premiums.

A spokesman for Revolution Medicines, Brian Crawford, said the price of its new drug, Rasonque, “reflects its differentiation and the meaningful benefit it provides for patients.”

As is typical with other cancer drugs, the vast majority of the cost of Rasonque is expected to be covered through insurance, with varying out-of-pocket costs for patients. Mr. Crawford said uninsured and underinsured patients who met certain financial and medical criteria would receive the drug for free. The drug, also known as daraxonrasib, is taken as two pills a day.

In the medical world, much of the anger over cancer drug prices has been directed at pricey medicines that either failed or did not undergo clinical trials assessing whether they extend lives. Those drugs won approval because they could shrink tumors or delay their growth.

Revolution Medicines’ drug is not a cure, but it has shown that it can prolong life expectancy. In a key clinical trial, patients with advanced pancreatic cancer who got the pill lived for a median of 13.2 months, compared with 6.7 months for those who got chemotherapy.

Those results have electrified pancreatic cancer specialists and patients because they occurred in people who had no good treatment options and little time left to live.

In that context, it’s hard to say whether Revolution Medicines’ price is fair, said Stacie Dusetzina, a health policy professor at Vanderbilt University who studies drug pricing. “This drug does have substantial clinical benefits over existing treatment, which isn’t always the case for high-priced cancer drugs,” she said.

Cancer patients often stay on a drug for only a matter of months, stopping because it has become too toxic or their cancer has progressed. For example, in the clinical trial that won approval for Rasonque, patients were on the drug for a median of six months. But even over several months, the costs can pile up, and some patients who respond well can stay on a cancer drug for years.

Over the years, the public and lawmakers have expressed outrage at the high costs of insulin and drugs for H.I.V., hepatitis C and, most recently, obesity. One-time therapies for rare genetic diseases now routinely carry price tags of a few million dollars.

But cancer holds a unique place in the system, because care and drugs are so expensive and affect so many people.

For the past five years, employers have said cancer is the top disease category driving their medical expenses, according to a survey published in late August by the Business Group on Health, which represents large employers that provide health insurance to their workers.

Compounding the rising costs is that in some cancers, like multiple myeloma, patients are now routinely given three or four therapies at once.

The high prices aren’t limited to drugs that treat small numbers of patients. Oncology’s biggest blockbuster is Keytruda, which has a sticker price of $213,000 a year. It is given as an infusion or an injection every few weeks and has been used to treat several million patients.

For many older drugs, generic competition and government negotiations have sharply driven down costs. But there are hardly any limits on manufacturers’ ability to set high prices for newly approved medications.

“We allow pharmaceutical companies to charge whatever they want, and without any clear rationale, other than how much they think they can get,” said Dr. Aaron Kesselheim, a professor of medicine at Harvard and Brigham and Women’s Hospital who studies drug pricing.

(to be continued)

Translation

每年48萬美元的藥丸反映癌症藥物價格的新常態（1/2）

Revolution Medicines公司推出的新型胰臟癌治療藥物Rasonque，凸顯了過去二十年來癌症藥物價格的巨額上漲

Revolution Medicines公司為其新型胰臟癌藥物設定的標價約為每年48萬美元，即每片663美元 - 這在過去是難以想像的高昂價格。

如今，這樣的價格已成為癌症藥物的常態。

有一種尖端的CAR-T療法，一次性注入的費用就高達60萬美元。同樣的藥物，競爭對手的定價為 55 萬美元。一種治療胃腸道癌症的每日服藥，每年的價格高達 52 萬美元。一種治療白血病的每日兩次服藥呢？每年的標價為 43 萬美元。

根據評估藥物價值的臨床與經濟評估研究所 (ICER) 統計，光是 2024 年，就有九種癌症藥物獲準上市，其標價均超過每年 40 萬美元。

二十年前是難以想像的高昂價格如今已司空見慣。那時，即使是每年 10 萬美元的癌症藥物都會引發激烈的批評。

但自那時開始，癌症藥物的價格一直在穩步上漲，除了製藥公司認為市場能夠負擔的價格之外，幾乎沒有其他限制。最終，這筆費用由政府計劃如Medicare (聯邦醫療保險）、雇主、病患以及繳稅和醫療保險費的美國民眾承擔。

Revolution Medicines公司的發言人Brian Crawford表示，其新藥Rasonque的定價「體現了該藥的獨特性及其為患者帶來的顯著益處」。

與其他癌症藥物一樣，Rasonque的大部分費用預計將由保險覆蓋，患者只需自付一部分費用。Crawford先生表示，符合特定經濟和醫療條件的無保險或保險不足的患者將免費獲得該藥。該藥也稱為daraxonrasib，每日服用兩片。

在醫學界，人們對癌症藥物價格的憤怒大多集中在那些價格昂貴但療效不佳，或未經過臨床試驗評估是否能延長患者壽命的藥物。這些藥物之所以獲批，是因為它們能夠縮小腫瘤或延緩腫瘤生長。

Revolution Medicines公司的這款藥物並非治癒癌症的藥物，但已證明它可以延長患者的預期壽命。在一項關鍵的臨床試驗中，接受口服藥物治療的晚期胰臟癌患者的中位存活期為13.2個月，而接受化療的患者中位存活期僅為6.7個月。

這些結果令胰臟癌專家和患者都感到振奮，因為這些患者先前幾乎沒有有效的治療選擇，生命所剩無幾。

Vanderbilt大學研究藥物定價的衛生政策教授Stacie Dusetzina表示，在這種情況下，很難判斷Revolution Medicines公司的定價是否合理。 她說:「這種藥物確實比現有療法具有顯著的臨床效益，而高價抗癌藥物並非總是有明顯效益的」。

癌症患者通常只能服用一種藥物幾個月，就會因為藥物毒性過大或癌症擴展而停藥。例如，在為Rasonque贏得批准的臨床試驗中，患者的中位用藥時間為六個月。但即使只有幾個月，費用也會累積起來，一些有療效顯著的患者可以服用抗癌藥物數年之久。

多年來，公眾和立法者一直對胰島素以及治療愛滋病、丙型肝炎和最近肥胖症的藥物的高昂費用表示憤慨。如今，罕見遺傳疾病的一次性治療費用通常高達數百萬美元。

但癌症在醫療體系中佔有特殊的地位，因為癌症的治療和藥物費用極為昂貴，而且影響著許多人。

根據代表為員工提供醫療保險的大型雇主的「商業健康集團」（Business Group on Health）8月下旬發佈的一項調查，過去五年，雇主表示癌症是推高其醫療支出的首要疾病類別。

雪上加霜的是，對於某些癌症，例如多發性骨髓瘤，患者現在通常需要同時接受三到四種療法。

高昂的價格並非僅限於治療患者人數較少的藥物。腫瘤科最暢銷的藥物是Keytruda，年定價高達21. 3萬美元。該藥每隔幾週透過輸液或註射給藥，已用於治療數百萬名患者。

對於許多舊藥而言對於許多，，仿製藥的競爭和政府的談判已大幅降低了價格。但對於新核准的藥物，藥廠幾乎不受任何限制地設定高價。

哈佛大學和布萊根婦女醫院研究藥品定價的醫學教授Aaron Kesselheim博士說:「我們允許製藥公司隨意定價是沒有任何明確的理由，除了他們認為這藥能夠賺到多少」。

（待續）

有些科學家在實驗室裡擁有「魔術手」。人工智能正在學習其中的奧秘。 (2/2)

Recently The New York Times reported the following:

Some Scientists Have ‘Magic Hands’ in the Lab. This A.I. Is Learning Why. (2/2)

Even the most adept researchers may not know exactly what they do to get successful results. An A.I. model is trying to figure it out by watching every move.

The NYT - By Carl Zimmer

Aug. 27, 2026

(continue)

Renee Wegrzyn, a co-founder of Transfyr, became keenly aware of this gap while doing research as a postdoctoral researcher. She jotted down records of her experiments on proteins in a lab notebook, but it captured only a fraction of her efforts.

When Dr. Wegrzyn published her results, she had to distill her records even further. “I did a postdoc that was three years long, and then it got summed up in a three-page paper,” she said.

Missing from those notes and reports were all the little decisions Dr. Wegrzyn had made while performing her research.

“All of those details could matter, but we just don’t know if they matter,” said Brian Nosek, an expert on replication at the University of Virginia who is not involved in Transfyr.

When a replicated study fails to produce the original results, that may mean the original study was wrong. Or maybe it was right, and the replicating scientists made an unwitting blunder. It can be hard for teams of scientists to agree on what happened.

In 1993, for example, the psychologist Frances Rauscher and her colleagues reported that people who listen to Mozart get a temporary boost on reasoning tests. The buzz around the so-called Mozart effect led other researchers to run experiments of their own.

Many of them couldn’t find any significant benefit, but Dr. Rauscher brushed off these failures. The other scientists didn’t find the Mozart effect because they weren’t good enough at running experiments, she said — they were not good scientific chefs.

The search for the Mozart effect went on for years. In the end, other scientists found little evidence that it exists.

Conflicts like these led some scientists to search for new ways to communicate tacit knowledge. In 2006, the biologist Moshe Pritsker created a journal called JoVE, which published videos of scientists carrying out complex experiments.

In 2014, the biologist Lenny Teytelman created a place where scientists could share their protocols and talk about them. Researchers use the website, called Protocols.io, to help train new members of their labs when the seasoned scientists with magic hands have moved on.

With Transfyr, Dr. Wegrzyn and Ms. Wagner are building a system that can slurp up enormous amounts of data about what happens in a lab, in the form of video, audio and sensor logs from lab equipment. Transfyr even tracks the source of supplies, down to the lot numbers of glove boxes.

(Why? “Maybe there’s a fume coming off the glove that is changing your experiment,” Ms. Wagner said.)

Dr. Nosek said that this new approach might reveal some secrets about why experiments succeed or fail. “What I really like about them is they’re going gangbusters into trying to unpack every detail,” he said. “It makes a lot of sense to go all in, and then figure out what actually matters.”

The company feeds its digital record of experiments to computers that have been trained to recognize the equipment used in biochemistry experiments. Artificial intelligence systems track the equipment, along with the hands of scientists, and decipher each action.

This analysis has revealed that lab workers are performing the same experiment in many different ways. “The variation we see even among well-trained scientists is pretty jaw-dropping,” Ms. Wagner said.

Sometimes those variations matter.

Recently, Transfyr hosted technicians from Dr. Livny’s lab, to watch them extract RNA from cells. Charlie Low, famous for her magic hands, was unwittingly doing a key step wrong, letting chemicals react for 120 seconds instead of 90 seconds.

“I didn’t realize that was even happening,” Ms. Low said. The mistake happened because she turned her timer on after starting the chemical reaction, not before. But it turns out that the extra time from that mistake improved the experiment.

In another trial, Transfyr has found that a protocol that takes six hours for one researcher may take eight hours for another. “It’s not a hard thing to see,” Ms. Wagner said. “But there’s no way to capture it at scale manually.”

Transfyr has signed up a diagnostics company and other clients who want the software to track their research. Ms. Wagner said that ultimately, it might be possible for the system not just to uncover mistakes but to document unexpected breakthroughs.

Rather than sifting notebooks for clues, the system can look back at every second of the work while querying A.I. about unusual things the workers did without realizing it.

Dr. Nosek said that Transfyr would have to present detailed results to prove that this system really does make science better. “Take 50 labs, track half the experiments with this stuff and don’t track the other half — then look for differences in progress,” he suggested.

Harry Collins, a sociologist at Cardiff University who has studied tacit knowledge for over 50 years, said he was optimistic about the effort because it was focused on molecular biology — a field that has matured to the point that many of its fundamental mysteries have been worked out.

But he was skeptical that Transfyr’s approach would reveal much in  fields where scientists agree on far less, even about which experiments will reveal something important.

“You might be lucky now and again, but it’s not a silver bullet for scientific problems,” Dr. Collins said.

Translation

有些科學家在實驗室裡擁有「魔術手」。人工智能正在學習其中的奧秘。 (2/2)

即使是最熟練的研究人員，也可能不完全清楚自己是如何取得成功的。一款人工智能模型正試圖透過觀察他們的每一個動作來找出答案

（繼續）

Transfyr 的聯合創始人Renee Wegrzyn在擔任博士後研究員時，敏銳地意識到了這一資訊缺口。她將自己對蛋白質的實驗記錄在實驗筆記本中，但這只記錄了她工作的一小部分。

當Wegrzyn博士發表她的研究成果時，她不得不進一步精簡記錄。她說: 「我做了三年的博士後研究，最後卻被總結成了一篇三頁的文件」。

這些筆記和報告中缺失了Wegrzyn博士在研究過程中所做的所有細微決定。

並未參與Transfyr計劃的維吉尼亞大學的重複性研究專家Brian Nosek說道:「所有這些細節都有機會很重要，但我們並不能確定它們是否真的這樣重要」。

當一項再次重複研究未能得出與原始研究相同的結果時，這可能意味著原始研究是錯誤的。或者，原始研究本身是正確的，只是重複研究的科學家們無意間犯了一個錯誤。科學家團隊很難就究竟發生了什麼事達成一致意見。

例如，1993年，心理學家Frances Rauscher及其同事報告稱，聽莫札特音樂的人在推理測驗中的表現會暫時提升。圍繞所謂「莫札特效應」的熱議促使其他研究人員進行了自己的實驗。

他們中的許多人未能發現任何顯著的益處，但Rauscher博士對這些失敗不以為意。她說，其他科學家之所以沒有發現“莫札特效應”，是因為他們不夠擅長進行實驗 - 他們不是優秀的科學廚師。

對「莫札特效應」的探索持續了數年。最終，其他科學家幾乎沒有發現任何證據表明它存在。

諸如此類的衝突促使一些科學家尋找新的方式來交流融會知識。 2006年，生物學家Moshe Pritsker創辦了名為JoVE的期刊，該期刊發表了科學家們進行複雜實驗的影片。

2014年，生物學家Lenny Teytelman創建了一個平台，讓科學家分享和討論他們的實驗方案。研究人員利用這個名為Protocols.io的網站，在有魔術手的經驗豐富科學家離開後，幫助訓練實驗室的新成員。

連同Transfyr，Wegrzyn博士和Wagner女士正在開發一個系統，該系統能夠收集實驗室中發生的海量數據，包括視頻、音頻以及來自實驗室設備的傳感器日誌。 Transfyr甚至可以追蹤耗材的來源，精確到手套的装箱批號。

（為什麼？Wagner女士說:「也許是手套散發出的氣味影響了你的實驗」。）

Nosek博士表示，這種新方法或許能揭示實驗成敗的一些秘密。 他說:「我真正欣賞的是，他們正如火如荼地探究每一個細節」;「全力以赴是很有意義，之後找出什麼是真正重要的」。

該公司將實驗的電子記錄輸入電腦，這些電腦經過訓練，能夠識別生物化學實驗中使用的設備。人工智能系統會追蹤這些設備以及科學家的操作，並解讀每一個動作。

這項分析表明，實驗室工作人員會以許多不同的方式進行相同的實驗。 Wagner女士說: 「即使是訓練有素的科學家之間，我們也看到了令人震驚的差異」。

有時，這些差異至關重要。

最近，Transfyr公司接待了Livny博士實驗室的技術人員，讓他們觀摩從細胞中提取RNA的過程。以巧手聞名的Charlie Low無意中犯了一個關鍵錯誤，她讓化學反應持續了120秒，而不是90秒。

Low女士說: 「我根本沒意識到這一點」。這個錯誤是因為她在化學反應開始後才打開計時器，而不是之前。但結果證明，這個錯誤帶來的額外時間反而改善了實驗結果。

在另一項試驗中，Transfyr發現，一個研究人員需要6個小時才能完成的實驗方案，在另一個研究人員可能需要8個小時。 Wagner女士說: “這是很容易理解” ， “但人手地大規模捕捉這種差異是不可能的。”

Transfyr已經與一家診斷公司和其他客戶簽約，希望利用該軟件追蹤他們的研究。Wagner女士表示，最終，該系統不僅能夠發現錯誤，還能記錄意想不到的突破。

系統無需翻閱筆記本尋找線索，而是可以回顧實驗的每一秒，並可向人工智能系統查詢關於做實驗的人在不知不覺中做出的不尋常的事情。

Nosek博士表示，Transfyr公司必須提供詳細的實驗結果，以證明這套系統確實能促進科學發展。他建議道: “選取50個實驗室，用這套系統追踪一半的實驗，另一半不追踪 - 然後觀察兩組實驗進展的差異。”

Cardiff大學的社會學家Harry Collins研究融會知識已有50餘年，他表示對這項研究持樂觀態度，因為它專注於分子生物學 - 一個發展成熟的領域，許多基本奧秘都已被解開。

但他懷疑Transfyr的方法能否在其他領域取得突破性進展，因為科學家們對許多問題的共識都遠不如分子生物學，甚至連哪些實驗能夠揭示重要資訊都難以達成共識。

Collins博士說: 「你或許偶爾會走運，但這並非解決科學問題的靈丹妙藥」。

So, failure to repeat an experiment can take many forms. Chances are that a biotech company creates a promising new drug; when it starts with large-scale production, suddenly it just doesn’t work. When scientists carry out experiments, they keep careful records, but when other researchers use that information to repeat the experiment, scientists may discover that essential knowledge has not been written down. Scientists have generally come to agree that tacit knowledge is an essential part of research. Now some scientists are building a system that can generate enormous amounts of data about what happens in a lab when an experiment is done. It is hoped that ultimately it may be possible for the system not just to uncover mistakes but to document unexpected breakthroughs.