Are We Harold Bloom?

我们都是哈罗德·布鲁姆吗？

A human wrote this essay. Proof of work. Harold Bloom was a professor at Yale University and a famous author and literary critic: think Roger Ebert but for books. Many considered him a prick and colleagues bristled at him. This grumpy professor’s in my noggin lately, because I happen to participate in an online community, Lobsters, which I find is bursting at the seams with Harold Blooms. (Except we’re software engineers, not generally book authors.) 这篇文章是由人类撰写的，这就是工作证明。哈罗德·布鲁姆（Harold Bloom）曾是耶鲁大学的教授，也是一位著名的作家和文学评论家：你可以把他想象成图书界的罗杰·伊伯特（Roger Ebert）。许多人认为他是个混蛋，同事们也对他敬而远之。最近，这位脾气暴躁的教授总是在我脑海中挥之不去，因为我恰好参与了一个名为 Lobsters 的在线社区，我发现那里到处都是“哈罗德·布鲁姆”。（只不过我们是软件工程师，通常不是书的作者。）

This isn’t criticism per se because I’ve discovered that, oh my God, I’m Harold Bloom too! What do I mean by this? I bet my life savings you’ll understand If I shared a minute of him on TV. I invite you to watch the professor’s vituperation of Harry Potter while that mass phenomenon was in its cradle. (Potter Heads are well-advised to skip this one as it’s a little brutal.) 这本身并不是批评，因为我发现，天哪，我自己也是哈罗德·布鲁姆！我这是什么意思？我敢拿我的全部积蓄打赌，如果我分享一段他在电视上的视频，你就会明白。我邀请你观看这位教授在《哈利·波特》这一大众现象尚处于萌芽阶段时对它的抨击。（哈利·波特迷们最好跳过这一段，因为它有点残酷。）

At first I was amused by the unabashed elitism, yet as the full interview unfolded I began to see myself in Harold Bloom. I was startled to hear “slop” lobbied a quarter-century ago against the work of another author. The resemblance grew more unnerving as his peers reacted to the berating. 起初，我被那种毫不掩饰的精英主义逗乐了，但随着采访的深入，我开始在哈罗德·布鲁姆身上看到了自己的影子。令我震惊的是，早在四分之一个世纪前，就有人用“垃圾”（slop）这个词来攻击另一位作者的作品。当他的同行们对这种斥责做出反应时，这种相似性变得更加令人不安。

Indeed when I trash the AI hype profiteers my friends will mirror the above opinion, labeling me reactionary. A contrarian. I could also tell they were reaching for “snob” or “asshole” and stopped just short of it. It’s important to make fun of ourselves. I do appreciate Lobsters: I adopted their invite tree system for my own community. However, I can’t unsee the Bloom filter. 事实上，当我抨击那些靠 AI 炒作牟利的人时，我的朋友们也会表达上述观点，给我贴上“反动派”或“唱反调者”的标签。我能感觉到他们其实想说“势利眼”或“混蛋”，只是话到嘴边又咽了回去。自嘲是很重要的。我很欣赏 Lobsters：我甚至为我自己的社区采用了他们的邀请树系统。然而，我无法摆脱这种“布鲁姆滤镜”。

I read Loris Cro’s lovely North Star and agreed with it. I also couldn’t help myself picturing a smug smirk, his extended pinky poking me in the eye as he picked up his tea. Do we want to pull folks back to more sanity? For now I’d stash our monologues on craft and reach for everyone’s wallets. 我读了 Loris Cro 那篇精彩的《北极星》（North Star），并表示赞同。但我也忍不住想象他那自鸣得意的假笑，以及他端起茶杯时那根戳到我眼睛里的翘起的小指。我们想把人们拉回更理智的状态吗？目前，我宁愿先收起我们关于技艺的独白，转而谈谈大家的钱包。

You might’ve heard AI usage is priced more realistically this year. Data centers are stalled or blocked. Investors are getting nervous. No one knows how to map AI output to a productive workforce: Seriously, somebody please show me a company spending millions of dollars on AI tokens that can also express a clear, indisputable return on investment. Show me the actual returns. Show me the processes automated and what those processes being automated do to offset these remarkable costs. 你可能听说过，今年 AI 的使用定价变得更加现实了。数据中心建设停滞或受阻。投资者开始感到紧张。没人知道如何将 AI 的产出转化为生产力：说真的，请给我展示一家在 AI Token 上花费数百万美元，同时又能明确、无可争议地体现投资回报的公司。给我看看实际的回报。给我看看哪些流程被自动化了，以及这些被自动化的流程如何抵消这些高昂的成本。

All of this fucking bloviating about how AI is inevitable and real and so powerful never seems to result in a profit. This is a powerful argument to me. Because the value isn’t palpable so people are panicking at our new cost of compute: Uber’s COO admits it’s “harder to justify AI costs within the company.” (They blew past their annual budget.) 所有这些关于 AI 是不可避免的、真实的、强大的废话，似乎从来没有带来利润。对我来说，这是一个强有力的论点。因为价值并不明显，所以人们对我们新的计算成本感到恐慌：Uber 的首席运营官承认，“在公司内部证明 AI 成本的合理性变得越来越难。”（他们已经超出了年度预算。）

Now that it’s gone, Copilot users are astonished how much their token subsidy was worth. Axios revealed a company unknowingly spent half a billion dollars (!) in a month after failing to place usage limits. Orgs are desperately moving away from so-called tokenmaxxing. “Throwing AI licenses at the wall and seeing what sticks isn’t leading to tangible returns.” Here’s a funny one that made me blink rapidly: One CTO told Axios that employees were using AI models to check the weather. That gets expensive fast. 现在补贴没了，Copilot 的用户们才惊讶地发现他们之前的 Token 补贴有多值钱。Axios 披露，一家公司因为没能设置使用限制，在一个月内不知不觉地花费了五亿美元（！）。各机构正拼命摆脱所谓的“Token 最大化”。“把 AI 许可证到处乱扔，看看哪种有效，并不能带来实际的回报。”这里有一个让我目瞪口呆的趣事：一位首席技术官告诉 Axios，员工们竟然在使用 AI 模型来查询天气。这烧钱的速度可太快了。

I’m the rare low-level programmer who loves bars, clubs and meeting absolute strangers. When I tell the Seattle tech(-adjacent) scene that GitHub is conclusive proof our industry fell into disrepair I get blank stares. However, if I pause my Harold Bloom impression for a moment and talk about executives scrambling to measure productivity? Suddenly their eyes widen and they’re nodding with ease. They can feel software hasn’t gotten any better to justify the investment. 我是一个罕见的、喜欢酒吧、俱乐部和结识陌生人的底层程序员。当我告诉西雅图科技圈（及其周边）的人，GitHub 是我们行业陷入衰退的铁证时，他们只会一脸茫然地看着我。然而，如果我暂时放下“哈罗德·布鲁姆”的架子，谈谈那些忙着衡量生产力的高管们？他们会突然瞪大眼睛，轻松地点头。他们能感觉到，软件并没有变得更好，不足以证明这些投资是合理的。

The focus of Lobsters is narrowed to computing. Business rants are off-topic. But what is my blog post? If engineers are increasingly charged for the “privilege” of compute it becomes impractical to disentangle the two. I found it’s very persuasive to cease lamenting our craft and instead point to a defective business enterprise. I’m seeing it fold the more neutral devs into our camp. Lobsters 的焦点仅限于计算领域。商业吐槽被视为离题。但我的博客文章又算什么呢？如果工程师们越来越多地为计算的“特权”付费，那么将两者分开讨论就变得不切实际了。我发现，停止哀叹我们的技艺，转而指出商业企业的缺陷，是非常有说服力的。我看到这种观点正在将更多中立的开发者拉入我们的阵营。

-Abner