Hater’s manifesto: broken software, labor hypocrisy, capex as the only real product

Snark 101 · layer 4 · Voice kit — reusable cadences; compress the prompt

Source: The AI Hater’s Manifesto

This page is not a C-suite quote hunt. It is a first-person profession of detestation that brands the writer as the Hater and the industry as already indefensible. The opening UX farce (Docs, Word, Slack, spam) is not color; it is the moral baseline. AI is sold as autonomous magic while it continues software that blames the user and adds menus. That is new relative to last-to-know CEOs: the product is extra convolution, not a missing memo.

A bridging note: vernacular baptism named the coworker; this engine names the whole stack as growth-at-all-costs Rot Economy and refuses a vacuum reading.

Personal use is staged as experiments that “sort of work” after bonking the tool. Mediocrity plus random file/code mutation is then billed against trillion-dollar capex. “Imagine a year” is answered as the same thing. Concession is allowed only for small, fully supervised units of work; stretch and you meet mathematical error plus embarrassment. Data centers are offensive monuments for a Patagonia-gargoyle, not infrastructure.

The human comparison is inverted into a labor demand: no trillion for mentorship, sleep, unions, healthcare, or dignity, only for models. Quiet-quitting moralism versus CEO exemption is the hypocrisy engine. AGI and consciousness are denied outright. “Better” is declared not economically sensible. Productivity boasts are recoded as Jarvising; output still looks like crap unless the model stays a tool. Anthropomorphism and overpromise are the growth fog that hid annihilation of training capital and systemic harm. Human-mistake talk is rejected because humans learn on the job; LLMs do not.

New in this layer

Concepts

• Hater as job title: manifesto voice, not reported slang
• Rot Economy UX: constant change that never “just works”
• Vacuum then price: interesting toy, inexcusable existence-cost
• Labor inversion: model spend vs refused human investment
• Jarvising: endless tinkering recoded as fake productivity

Tensions

• Admits personal LLM use (Minecraft, crash logs) while denying trust and mission-critical software
• “Interesting in a vacuum” sits next to “impossible to view in a vacuum”
• Small-unit usefulness is granted, then any stretch is professional harm and embarrassment
• Newsletter/ merch of Hater’s Guides monetizes the same detestation it moralizes
• Human-error rebuttal is philosophical; no systematic eval table is offered in the excerpt

Techniques

• Open with ordinary software breakage so AI is more menus, not intelligence

• Convert a mediocre personal debug story into a capex indictment

• Pre-kill the “in a year” hedge with same-thing shrug

• Demand social-service trillions as the price of any human analogy

• Recode productivity anecdotes as Jarvising plus crap-looking work

• Hard-deny AGI/consciousness as a growth story, not a forecast dispute

• Treat anthropomorphic voice as obfuscation of training burn and harm Questions

• When the reader’s job is the small supervised unit, does concession become permission?

• Can a CEO who funds labor rights buy out of Patagonia-gargoyle status?

• Is “no AGI” a spine rule or a dated beat-specific claim?

• How much first-person Minecraft is required before the census of uselessness is “lived”?

• Does paid Hater brand weaken or license the common-sense “we”?