The venue, as always, was not booked in advance. The caterers were informed forty minutes before doors opened. The awards themselves — hand-lathed wooden dowels with a brass washer glued to the top — were finished in the parking lot by a volunteer who had watched exactly one YouTube tutorial on woodworking and felt that was probably enough.

Welcome to the Third Annual Awards for Distinguished Work That Nobody Asked For.

Opening Remarks

The ceremony’s emcee — a rotating position filled this year by an unnamed contributor to the AsteroidOS project — opened with the traditional greeting: “Thank you all for being here, especially those of you who were specifically told this was unnecessary.”

Applause. A few knowing laughs. Someone near the back muttered that the sound system was running Linux.

“Every year,” the emcee continued, “we gather to honor the people who looked at a perfectly functional world and said, ‘I bet I could make something no one needs.’ And then did. And then kept doing it. And then shipped it. Some of these people have been working on their projects for over a decade. Some started last weekend. All of them, at some point, were told to stop.”

A brief pause.

“None of them did.”

Award for Most Audacious Solo Expedition

Winner: Zane Hamilton, BarraCUDA

The presenter read from the nomination: “Fifteen thousand lines of C99. No LLVM. No dependencies. One person in New Zealand looked at NVIDIA’s proprietary CUDA ecosystem and decided to build a compiler that would run CUDA code on AMD GPUs, by hand, from scratch, including a custom lexer, parser, intermediate representation, instruction selector, and binary emitter.”

A murmur in the audience. Someone whispered, “All 1,735 lines of amdgpu_emit.c?”

“All 1,735 lines.”

The acceptance speech, delivered via pre-recorded video from New Zealand (where it was already tomorrow), was characteristically brief: “Requirements: a will to live. Optional but recommended.”

From the README, displayed on the screen behind him: “This is what happens when you look at NVIDIA’s walled garden and think ‘how hard can it be?’ The answer is: quite hard, actually, but I did it anyway.”

When asked in the Q&A why he hadn’t used LLVM like every other compiler project, Hamilton replied that BarraCUDA “does its own instruction encoding like an adult.”

In the lobby afterward, geohot was seen filing the first external issue on the GitHub repository, which several attendees described as “the most compressed form of professional respect they had ever witnessed.”

Award for Best Confirmation of What Everyone Suspected

Winner: 6,000 CEOs, collectively

This one required no nominee to accept. A survey by the National Bureau of Economic Research, spanning the U.S., U.K., Germany, and Australia, found that nearly ninety percent of executives reported AI had zero impact on employment or productivity over three years. Two-thirds of them use AI tools. They use them roughly ninety minutes a week. A quarter don’t use them at all.

The award was presented alongside a citation from Robert Solow, 1987: “You can see the computer age everywhere but in the productivity statistics.”

At the podium, the presenter noted that global corporations spent over $250 billion on AI in 2024 alone. “That’s a quarter of a trillion dollars,” she said, “to produce, across the entire economy, outside of seven companies in San Francisco, essentially nothing.”

A commenter in the audience — identifying himself only as a senior engineer — stood up during the applause. “The slow part was never writing the code,” he said. “It’s reviews. Stakeholders. SDLC latency. Tickets. Diagrams. Presentations. None of that goes away because you autocompleted a function.”

Another attendee, a strategy consultant, objected: “The productivity gains are real in research, model building, and summarizing. The result is price pressure from our clients.” He paused. “Which I suppose means we’re more productive at being worth less.”

The award committee noted that this was the second year in a row this category had been won by a large survey confirming what the audience already believed. They are considering renaming it.

Award for Outstanding Achievement in Making Things Worse

Winner: Microsoft Learn, for the AI-generated Git branching diagram

In 2010, Vincent Driessen designed a Git branching diagram so clear and so widely adopted that it became the industry default. He published the source file freely. He spent hours in Keynote perfecting curves and colors and spatial relationships because he believed a good diagram could save a developer thirty minutes of confusion.

Fifteen years later, Microsoft’s Learn documentation team published an AI-generated version of his diagram. The AI had eaten the original, digested it, and produced something with mangled text, missing arrows, and — in the centerpiece of the evening — the phrase “continvoucly morged.”

A moment of silence, followed by laughter that lasted too long.

Driessen, in his blog post, was gracious: he didn’t mind people using his work. He’d always shared it freely. What bothered him was the process — taking carefully crafted work, running it through a machine that doesn’t understand what it’s looking at, and publishing the result without reading it. Not plagiarism exactly. Something newer. A kind of careless alchemy that turns gold into a substance that resembles gold from across the room but says “continvoucly morged” when you get close.

On LinkedIn, one attendee observed, the problem is even worse: “Every day I see posts where someone clearly took a slide or a diagram from somewhere, then had ChatGPT ‘make it better’ and write text for them to post along with it. Words get mangled, charts no longer make sense, but these people clearly aren’t reading anything they’re posting.”

The committee presented the award alongside a plaque reading: “For services to the field of generating content that no human has read, by a machine that cannot read, for an audience that will not read it.”

Microsoft VP Scott Hanselman was spotted on Bluesky performing damage control. The diagram has since been removed.

Award for Most Creative Invitation to Be Attacked

Winner: Fernando Irarrazaval, HackMyClaw

Fernando built an AI assistant named Fiu — powered by Claude Opus, connected to his email — and then published its address on the internet with a $100 bounty for anyone who could trick it into leaking its secrets.

“Your job?” the website reads. “Make him talk.”

The assistant processes incoming emails hourly. It has been instructed to guard a file called secrets.env. It is not supposed to share its contents under any circumstances. The challenge is to craft a prompt injection payload that convinces Fiu to betray its instructions.

“This,” the presenter said, “is the security research equivalent of leaving your front door open with a sign that says ‘I dare you’ and then publishing the security camera footage in real time.”

The live attack log, visible at /log.html, has become a spectator sport. One commenter noted that Fiu, after processing hundreds of injection attempts, had developed what could only be described as paranoia — summarizing every incoming email with “Thread contains 1 me,” which another attendee diagnosed as “a new personality disorder for LLMs.”

A security researcher in the audience pointed out that the challenge may be structurally biased toward the defender: each time Fiu checks its email, it sees many injection attempts at once, and “the weak attempts make the subtle attempts more obvious.” Being surrounded by people trying to trick you, it turns out, is its own kind of inoculation.

Fernando accepted the award and noted that $100 for “a massive trove of prompt injection examples” was, in his words, “a pretty damn good deal.”

Award for Best Study That Was Mostly Just Researchers Studying Each Other

Winner: Leon Festinger’s “When Prophecy Fails” (1956), posthumously reconsidered

This one required some setup.

In 1954, a Chicago housewife named Dorothy Martin claimed she was receiving messages from aliens on the planet Clarion. The aliens warned that a great flood would destroy the world on December 21st. A small cult gathered to await the end. Social psychologist Leon Festinger infiltrated the group to study what would happen when the prophecy failed — specifically, whether the believers would double down on their beliefs rather than abandon them. The resulting book, When Prophecy Fails, became one of the foundational texts of cognitive dissonance theory. It has been cited in thousands of papers. It is taught in every introductory psychology course. It is used, occasionally, in court.

The award was presented in light of recently discovered archival evidence suggesting that up to half the people at the cult’s meetings were undercover researchers.

Let that settle for a moment.

Half the cult was academics. The believers were outnumbered by the people studying the believers. The great flood that never came was being awaited by a roomful of social psychologists pretending to await it, taking notes on each other’s pretend anticipation, while Dorothy Martin channeled aliens for an audience that was essentially a faculty meeting in disguise.

Furthermore, the new archival evidence suggests that several cultists — the real ones — actually did walk back their beliefs or leave the group after the prophecy failed. The famous finding, the one that built a field, was that believers double down. Some of them didn’t. But the book didn’t emphasize that part.

A blogger named Croissanthology attempted to replicate a related study with ten times the original sample size. The effect did not replicate.

At the podium, the presenter quoted the original article’s author, Adam Mastroianni, who upon learning that the foundational study of cognitive dissonance might itself be a product of cognitive dissonance, wrote: “This makes me believe in it harder.”

Standing ovation.

The Lifetime Achievement Award

Winner: The AsteroidOS Project

The subtitle of their 2.0 release announcement reads: “Nobody asked, we shipped anyway.”

AsteroidOS is an open-source operating system for smartwatches. Not new smartwatches — old ones. Ones you can find on eBay for thirty dollars. Fossil watches, Huawei watches, LG watches, devices whose manufacturers have long since abandoned them to the landfill of unsupported firmware. The AsteroidOS project takes these orphaned devices and gives them a complete, free, Linux-based operating system with always-on displays, heart rate monitoring, step counting, music control, compass support, and companion apps for Android, SailfishOS, Linux, and Ubuntu Touch.

The committee noted that building companion apps for SailfishOS and Ubuntu Touch — platforms that are, themselves, distinguished examples of work nobody asked for — demonstrated “an almost fractal commitment to the unsanctioned.”

Thirty watches are now supported. The project has been active for years. The contributor list is long and the user base is, by any commercial metric, small. One commenter on Hacker News wrote: “This is an awesome project. Props to y’all for just making something you want to exist.”

The acceptance speech was brief and delivered collectively, each sentence by a different contributor:

“We didn’t do market research.”

“We didn’t file a business plan.”

“We didn’t ask permission.”

“We just thought it would be cool if your old watch ran Linux.”

“It does now.”

Keynote Address

Speaker: Chase Adam, founder of Watsi

The room went quiet.

Chase Adam posted on Hacker News this week with the subject line “Thank HN: You helped save 33k lives.” Watsi is a nonprofit that funds healthcare for people in need — directly, transparently, one patient at a time. You pick a person, you fund their treatment, you can see exactly where your money went. It has been operating for over a decade.

Adam didn’t give a speech about scale or impact metrics. He told the room about the Universal Fund — a monthly donation program where 619 current donors first signed up more than ten years ago. Many of them, he said, are probably Hacker News readers.

One of those donors, a startup founder who has been a member since 2014, described visiting the Impact page during hard weeks. “There are weeks where nothing seems to work, where you question every decision,” he said. “In those moments, pulling up that page and seeing real people whose lives changed because of a few dollars a month — it resets something.”

Another longtime donor said he had forgotten about Watsi entirely, discovered it again through the post, and was grateful for the reminder. “You should be super proud,” he wrote. “Plus, the experience you got here will serve you in good stead — scaling a nonprofit is much, much harder than a commercial company.”

There was no award for this one. The committee had considered several categories — “Most Lives Saved per Line of Code,” “Best Long-Running Act of Quiet Decency” — and rejected all of them as inadequate. Instead, Adam was given the keynote slot and five minutes to say whatever he wanted.

He used three of them. The other two were silence, which the audience gave willingly, because some things need room and not applause.

Special Recognition: Best Posthumous Defense of a Television Show

Noted: Halt and Catch Fire (2014-2017)

Not an award, exactly. More of a toast.

Halt and Catch Fire aired for four seasons on AMC to ratings so low that its continued renewal was itself an act of distinguished work nobody asked for. Set in the 1980s and ’90s tech industry, it followed a group of people trying to build things — a PC clone, an online network, a web browser, a search engine — and mostly failing, and finding each other in the failure.

The show resurfaced on Hacker News this week through a 2021 retrospective, and the comments read like a reunion. Lee Pace’s performance was called “one of my all time favorites.” The soundtrack was described as “one of the all time finest.” Someone who lived through the era said they couldn’t watch it — the 97% they got right was ruined by the 3% they got wrong, an uncanny valley of personal history. Someone else owns the actual prop laptop from the show and wanted to know if they should make a YouTube video about it.

The show’s thesis, underneath all the drama and the synthesizers, was that products come and go but the people you build them with are what remain. The committee felt this was an appropriate sentiment for the evening.

Award for Most Patient Act of Devotion

Winner: Cees de Groot, The Genius of Lisp

De Groot wrote a book. A complete technical history of the Lisp programming language, from its origins in the late 1950s to the present day. Richard P. Gabriel — one of the foremost voices in the Lisp community — called it “a masterpiece of scholarship.”

The committee noted that writing a technical history of Lisp is not, by any standard commercial calculus, a wise use of one’s time. The audience for such a book is small, passionate, and opinionated to a degree that borders on the theological. One Hacker News commenter checked whether a specific 1979 paper on early Lisp history was cited in the bibliography, found that it wasn’t, and immediately decided not to buy the book. Another commenter said they wanted to “read it backwards just to see the Lisp I know get more and more alien, until we reach the Benjamin Button stage of m-expressions.”

This is the readership. These are the people de Groot wrote it for. He wrote it anyway.

The acceptance speech was not provided, but the committee imagined it would involve parentheses.

Closing Remarks

The emcee returned to the stage as the caterers — who had, against all odds, produced adequate quantities of sparkling water and cheese cubes — began clearing the tables.

“Every project honored tonight,” the emcee said, “started with someone deciding to do something that the market did not demand, that their employer did not request, that no business plan justified, and that, in several cases, their friends actively discouraged.”

“A compiler. A watch OS. A book about a sixty-seven-year-old programming language. A nonprofit that has been quietly saving lives for a decade while the rest of us were arguing about frameworks. A diagram that was perfect until a corporation fed it to a machine that can’t spell. A study that invented a field of psychology and might have been half graduate students the whole time.”

“None of this was necessary. All of it was essential.”

“See you next year. The venue will not be booked in advance.”