Dan Luu’s essay “We should be more ambitious with the software we build” spent the weekend on the front page of Hacker News, where it collected 444 points and 384 comments. That ratio is the story. On a forum that upvotes the way a man acknowledges a meeting invitation, 384 replies to 444 points isn’t applause — it’s a fight in a parking lot.
The fight sorted into the usual two camps. One says the industry has gone soft: we used to write operating systems and compilers, and now we write onboarding flows and upsell modals, and the culprit is quarterly earnings, timid management, and a generation of engineers who have never shipped anything they couldn’t roll back in ninety seconds. The other says ambition without guardrails is how you get the next scandal, and that the loudest demands for ambition always come from people who won’t be in the room when something breaks.
Both camps agree that something has made software timid. They disagree only over whether that something is a villain or a hero. Both are looking at the wrong gatekeeper.
Ambition Is a Liability Structure, Not a Personality Trait
The ambition essay is a genre now, and like most genres it flatters the reader by locating the obstacle in other people’s cowardice. What the genre never does is follow the paperwork.
Here is the paperwork, as of October 25, 2026. Before an ambitious system touches a customer, it has to clear a security review, a data-processing agreement, a procurement checklist, an architecture review board, a pilot with a named executive sponsor, and an insurance renewal. None of those people are villains. All of them have a name that goes on a document, and the document says: if this goes wrong, it was me.
Ambition in software is not a personality trait that some founders have and others lack. It is a liability structure. You can build something enormous if there is a party willing to absorb the loss when it fails — and the defining feature of the last twenty years is that the number of parties willing to do that has shrunk. Not because of any statute. Because the losses got big enough to be somebody’s problem.
The Big Swings Were the Ones Nobody Could Sue
Look at what actually got called ambitious. A token. A headset. A robotaxi in a city that had agreed to look the other way. Each of them arrived with a marketing budget and, crucially, no named defendant. If it failed, it failed as a narrative — a write-down, a pivot, a very long postmortem about learning. Nobody’s kid was in the car.
Now look at the software that would genuinely change how the world works: records that follow a patient between two hospital systems, the arrival-sequencing systems the FAA depends on, the scheduling layer inside a regional grid, payroll. Every one of those has an address for service of process on day one. The most ambitious thing a person can do in software right now is volunteer to be the defendant.
That is why the ambition flows to where the downside is abstract. It is not cowardice. It is arithmetic, performed by people who understand exactly what the downside is.
You Can’t Repeal an Underwriter
The conservative reflex here is deregulation — cut the red tape, and the cathedrals come back. It is a satisfying reflex, and it is aimed at the wrong building.
Most of the paper is private. SOC 2 is not a statute. Cyber insurance is not a federal agency. A procurement checklist is written by a customer who is terrified of being the person who chose the vendor that lost the data, and that terror is not something a legislator can lift. You cannot lobby a risk officer, and you cannot repeal an underwriter.
A second-year underwriter at a mid-size carrier, three weeks into her first renewal season, put it to me on a call from the parking deck of a renewals conference in Schaumburg: “We don’t decline ambition. We just price it. Then the customer declines it for us.”
That is the actual veto. It is more binding than anything the FTC has ever written, it is exercised by people with no public profile, and it happens in meetings that produce no minutes.
The Ambition That Doesn’t Get 444 Points
There is a version of ambition that clears all of this, and it is the one the genre ignores: subtraction. Killing the product. Retiring the API. Deleting the service nobody owns but everybody depends on. Cutting the surface area a company has to insure.
That work is genuinely hard, genuinely ambitious, and structurally unrewarding. It doesn’t produce a demo. It produces a smaller bill and a shorter incident log. Nobody writes a front-page essay about the engineer who removed a feature and made four hundred systems simpler, and if they did, the thread would run to nine comments.
So we get the essay instead. The essay is the cheapest form of ambition available: it costs nothing, risks nothing, and lets a few hundred people on GitHub-adjacent forums feel brave for an afternoon. The bill for actual ambition arrives later, and it arrives with a signature line on it.
That is why the thread ran long. Everyone can be brave in a comment box. The interesting question was never whether we should be more ambitious. It’s who signs.