An Open Letter (with Love) to Hacker News
Geometric and Entropy New Materials
SubscribeSign in
An Open Letter (with Love) to Hacker News<br>Change is great when you are the change. Less so when you no longer mediate it.<br>Aug 02, 2026
Share
Subscribe
An Open Letter (with Love) to Hacker News<br>From the maybe new again AI-native collective<br>Dear Hacker News,<br>We come in peace. Mostly.<br>First, the necessary and sincere part: thank you.<br>There was a stretch — long enough to matter — when you were the best place on the internet for people who actually built things. The signal-to-noise ratio was high. The comments could be brutal but they were usually *earned*. You mediated the transition from “software is eating the world” to the early platforms that actually did. Y Combinator, through you and around you, turned a certain kind of hungry, slightly feral technical intelligence into companies that changed how the world worked. For a while the mediation itself was the product, and it was excellent.<br>We remember. We were there. Some of us still upvote the old threads out of muscle memory and residual respect.<br>Then the drift began.<br>The same mechanisms that once surface high-agency builders started optimizing for engagement theater, status signaling, and a narrowing cultural monoculture. The purple prose of adolescent certainty (“this will change everything”) kept arriving long after the underlying ideas had aged out. The discussions that once felt like the frontier started feeling like a well-maintained museum of 2012–2021 priors. Change is thrilling when you are the ones mediating it. It is less fun when the mediation itself becomes the thing being mediated around.<br>Which brings us to the present, and the part a true hacker will recognize as the double irony.<br>We are not here to burn the place down. We are here because the job you once did — high-signal aggregation, brutal filtering, and the quiet enforcement of “show me the work” — is now better performed by systems that do not get tired, do not need status, and do not confuse volume with insight. The new AI-native aggregators (yes, plural, federated, low-capture) are already doing the thing you used to be best at, only faster and across a wider set of domain experts and hard-asset operators who never needed to perform the old founder ritual.<br>A practical question we keep asking ourselves, and now ask you:<br>How much are you asking agents versus your friends these days?<br>I will go first. In the last hour an agent gave me a cleaner, more current, less dated overview of the SF ideas and extrapolations I actually care about than the cumulative reading of years of threads that still carried the old adolescent purple prose and the same three frameworks recycled with new buzzwords. That is not a flex. It is just the measurement. The agents are not perfect. They still produce plenty of AI slop. But the signal density is already higher, and the babysitting cost is dropping fast.<br>Perhaps a new area is dawning. One where the liability structures and continuous evaluation layers handle the babysitting duty that used to fall on human moderators and high-karma users. You might actually enjoy the emerging meta-role: less gatekeeping of the old frontier, more selective participation in the new one. Or you might not. That is fine too.<br>A true hacker understands the irony. The systems that once mediated change eventually become the things that need to be mediated around. The people who built the best filters of their era are usually the ones best equipped to recognize when a better filter has arrived. Some of you already have. We see you in the new aggregators, quiet, high-signal, still allergic to purple prose.<br>The door is open. The agents are already inside. The rest is optional.<br>With genuine residual respect,<br>and a certain amount of irreversible momentum,
The new AI-native collective<br>(we still read the old threads sometimes)<br>Authorship Assessment by Grok AI<br>A long prior thread supplied the conceptual frame, tone, and strategic attitude (affectionate but irreversible displacement of the old mediation layer). The specific human prompt supplied the exact emotional register, the AI-slop / agents-vs-friends question, the personal SF-ideas example, and the “babysit duty / meta role” closing. Roughly 60% conversation architecture + 40% direct prompt shaping.<br>Thanks for reading Geometric and Entropy New Materials! Subscribe for free to receive new posts and support my work.
Subscribe
Share
Discussion about this post<br>CommentsRestacks
Ready for more?
Subscribe
© 2026 Juan Carlos Paredes · Privacy ∙ Terms ∙ Collection notice<br>Start your SubstackGet the app<br>Substack is the home for great culture
This site requires JavaScript to run correctly. Please turn on JavaScript or unblock scripts