随着AI写代码，创业思维现在是Anthropic新人最重要的
Anthropic的工程经理费利克斯·里塞贝格(Felix Rieseberg)取消了经典的编码采访。 他辩称AI使得测试变得毫无意义，野心勃勃的年轻开发者可以利用它来超越更有经验的同事。
里塞贝格曾领导Claude.ai和Claude Code桌面在Anthropic，称的开发，现在已经没有任何理由给候选人小的编程任务。 Claude现在比候选人处理得更好，他称。 在他看来，给某人一个小的React挑战，今天和在大会上盘问一样没有意义。
在他的团队中，良好的测试分数不再预测谁是顶级开发者。 相反，他正在观看相对新颖的工程师们超过经验丰富的同事。 (原始内容存档于2018-07-09). "Hunger beats periols", he 称。 里塞贝格将他的帖子瞄准"Junior"开发者，虽然他有意重新定义了这个词。 他不一定是指刚开始的人他是指任何没有社会地位，权力，或公司内部网络的工程师。
为什么新来者发现问题退伍军人不再看到
Rieseberg认为，新来者具有结构性优势，因为他们注意到长期雇员在几年前接受的问题。 他提到组织研究员詹姆斯·马奇(James March)，他在1991年写道，组织依赖于不断的天真和不知情的新人流。 Stripe创始人帕特里克·科利森(Patrick Collison)也明确敦促新聘请标出任何对他们似乎陌生的东西。 他指出，退伍军人早已停止注意这些摩擦点。
里塞贝格引用他在Notion的时间为例。 他被要求设计公司内部的人认为不可能的离线建筑。 数据模型过于复杂，据推测CEOIvan Zhao绝不会接受唯一可行的解决方案。 技术障碍是真实的，但创始人的抵抗却不是。 没有人问他，因为每个人都把它当作是常识。
便宜的处决会削弱旧的等级
新来者Rieseberg 称的核心问题是公司里没有人听他们的。 因为执行过去很昂贵，所以建立组织是为了让少数决策者能够批准或屏蔽想法。 在管理层信任你之前你不得不努力建立严格的等级制度 Rieseberg写道
他的第一篇建议借鉴了帕特里克·麦肯齐2011年的一段引文，他写道："工程师被雇佣来创造商业价值，而不是编程事物". 开发者应该明白他们的公司到底在试图实现什么，并附着在赚钱的零件上。 一个有用的起点是了解你的经理的目标。
从那里，他建议选择一个雄心勃勃的想法，以大大推进这些目标，并在周末建立一个工作原型。 这超出了产品规格、批准和游说设计师或产品经理的通常周期。 关键是专注于一个好想法，而不是在20个坏想法中隐藏三个好想法。 任何人构建一个包含新编程语言，后端，移动应用程序的原型，都会同时失败。 原型应清楚地传达这一想法，以便领导人立即看到机会和合适的人能够利用这一机会。
新来的人有同情心
里塞贝格区分了两类工程师。 新来者在用户界面产品方面的工作给外部用户带来了自然的同情。 Slack或Notion的长期同事很难从外部用户的角度看待事情。 另一方面，从事基础设施或内部工具工作的工程师却不知道为什么某些东西被认为不可能实现而受益。
他用自己公司的例子来证明这一点。 当时比较新颖的两位Anthropic工程师安东尼·莫里斯和爱丽丝·赵在周末为Claude Code建造了图形界面。 第一个公开版本大约在8天后发货。
Rieseberg现在在工作申请中寻找一些小的原型，这些原型表现出口味、同情心和原则性思维。 经典的编码采访没有回来，他称。
一方面，里塞贝格的建议被解读为赋予权力的信息。 AI成为了一个平等主义者，让新来者通过守门员，他们无视自己公司的问题。 另一方面，它将进入一级工作的酒吧从勤勉和工艺转向创业思维、产品感知和品味。 大多数大学几乎没有教授这些技能，公司传统上都期望这些技能来自有经验的产品经理或创始人。
他所支持的工程师们在周末也建造了原型，没有被询问。 "饥饿战胜了任期"听起来很刺激，但它假设的是，并不是每一个新的开发者都能带给这个工作的一个拼命的心态。
AI News Without the Hype – 由人类治愈
订阅THE DECODER免费阅读，每周AI通讯，我们独家的"AI雷达"前沿报告每年6次，全面存档访问，访问我们的评论部分。
现在订阅
读取全照。
订阅无韵报道。
- 全面查阅THE DECODER的每篇文章
- 没有广告
- 加入评论和社区讨论
- 每周通过邮件转载AI新闻
- 6x/年："AI雷达"——对AI最重要的话题进行深度潜水。
- 每日AI新闻，总是最新的
- 我们完整的十年档案
- 由一个在AI工作10年以上的团队负责
订阅到 The Decoder
本文由 AI图鉴自动抓取并翻译，版权归原作者所有；如需查看原文请点下方链接。
With AI writing the code, entrepreneurial thinking is now what counts most for newcomers at Anthropic
Felix Rieseberg, an engineering manager at Anthropic, has scrapped the classic coding interview. He argues that AI makes the tests pointless and that ambitious young developers can use it to overtake more experienced colleagues.
Rieseberg, who has led development of Claude.ai and Claude Code Desktop at Anthropic, says there's no longer any reason to give candidates small programming tasks. Claude now handles them better than the candidates do, he says. In his view, handing someone a little React challenge makes as little sense today as quizzing them on Assembly.
On his teams, good test scores no longer predict who turns out to be a top developer. Instead, he's watching relatively new engineers overtake more experienced colleagues. "Hunger beats tenure," he says. Rieseberg aims his post at "junior" developers, though he deliberately redefines the term. He doesn't necessarily mean people just starting out. He means any engineer without social status, power, or a network inside their company.
Why newcomers spot the problems veterans stopped seeing
Rieseberg argues that newcomers have a structural advantage because they notice problems that long-time employees accepted as givens years ago. He points to organizational researcher James March, who wrote in 1991 that organizations depend on a steady stream of naive and uninformed newcomers. Stripe founder Patrick Collison has also explicitly urged new hires to flag anything that seems strange to them. Veterans, he noted, stopped noticing those friction points long ago.
Rieseberg cites his time at Notion as an example. He was asked to design an offline architecture that people inside the company considered impossible. The data model was too complex, and CEO Ivan Zhao would supposedly never accept the only workable solution. The technical hurdle was real, but the founder's resistance wasn't. Nobody had asked him, because everyone treated it as common knowledge.
Cheap execution weakens the old hierarchy
The core problem for newcomers, Rieseberg says, is that nobody in the company listens to them. Because execution used to be expensive, organizations were built so that a few decision-makers could approve or block ideas. You had to work your way up a rigid hierarchy before management trusted you to actually build something. AI changes that equation because execution is no longer expensive, Rieseberg writes.
His first piece of advice builds on a 2011 quote from Patrick McKenzie, who wrote that "engineers are hired to create business value, not to program things." Developers should understand what their company is really trying to achieve and attach themselves to the parts that make money. A useful starting point is knowing your manager's manager's goals.
From there, he suggests picking an ambitious idea that would dramatically advance those goals and building a working prototype over a weekend. That skips the usual cycle of product specs, approvals, and lobbying designers or product managers. The key is to focus on one good idea instead of hiding three good ones among 20 bad ones. Anyone who builds a prototype that includes a new programming language, a backend, and mobile apps all at once will fail. The prototype should get the idea across so clearly that leadership immediately sees both the opportunity and the right person to run with it.
Newcomers have the edge on empathy
Rieseberg distinguishes between two types of engineers. Newcomers working on user-facing products bring a natural empathy for outside users. Long-tenured colleagues at Slack or Notion had a hard time seeing things from an outside user's perspective. Engineers working on infrastructure or internal tools, on the other hand, benefit from not knowing why something is supposed to be impossible.
He backs this up with an example from his own company. Two Anthropic engineers who were relatively new at the time, Anthony Morris and Alice Zhao, built the graphical interface for Claude Code over a weekend. The first public version shipped about eight days later.
Rieseberg now looks for small prototypes in job applications that show taste, empathy, and principled thinking. The classic coding interview isn't coming back, he says.
On one hand, Rieseberg's advice reads as a message of empowerment. AI becomes an equalizer that lets newcomers get past gatekeepers who've gone blind to their own company's problems. On the other hand, it moves the bar for entry-level jobs away from diligence and craft toward entrepreneurial thinking, product sense, and taste. Most universities barely teach those skills, and companies have traditionally expected them from experienced product managers or founders.
The engineers he holds up as models also build prototypes on weekends without being asked. "Hunger beats tenure" sounds motivating, but it assumes a hustle mentality that not every new developer can bring to the job.
AI News Without the Hype – Curated by Humans
Subscribe to THE DECODER for ad-free reading, a weekly AI newsletter, our exclusive "AI Radar" frontier report six times a year, full archive access, and access to our comment section.
Subscribe now
Read on for the full picture.
Subscribe for hype-free coverage.
- Full access to every article on THE DECODER
- No ads
- Join the comments and community discussions
- A weekly AI news recap via mail
- 6x/year: "AI Radar" — deep dives on the AI topics that matter most
- Daily AI news, always up to date
- Our full ten-year archive
- Covered by a team with 10+ years in AI
Subscribe to The Decoder
Auto-fetched and translated by AI Atlas. All rights belong to the original author.