Ever since I first became aware of how LLMs were trained and was being pulled into using them in my day-to-day work, a dichotomy has been pretty self-evident.
These tools were trained on the wealth of knowledge sharing that happened via internet communities such as Stack Overflow, but the nature of the tools and how they have been designed is that the learnings and solutions that were being defined within the LLM environments (like Cursor, Claude Code, Codex etc.) were not being re-shared.
To my reckoning, this means that the very means by which the LLM providers had generated their value had been removed from the equation. I could—and still can—very easily see a time when LLMs are not as effective, because the diversity of source data (questions asked and solutions shared) will disappear.
The other day I saw a breakdown of Koto’s rebranding of Stack Overflow where this clearly and plainly highlighted:
Then came 2022 and the boom of mainstream AI. LLMs scraped data from the platform, and AI search changed user behavior overnight. By inadvertently powering the very models eating into its own traffic, Stack found its utility and cultural relevance at a critical crossroads. …
Is there a place for Stack Overflow in an AI-first world? This wasn’t just a brand strategy task; it was an existential reckoning. Our response? Change the narrative. Because AI models are only as good as the data they consume, Stack’s archive of expert discussion, problem-solving and human-validated knowledge became a goldmine. We repositioned the platform as the ultimate arbiter and hub of truth for the entire industry.
Of course, this isn’t just about code. Recently, we saw Google announce that AI search will be the default, and we have seen radical reductions in traffic going to the sites that they were trained on. Many “content creators” are unable to continue operating and developing that content because the key revenue stream they relied on to fund those activities is gone.
I presume the thinking within Google is that people will pay for placement within the LLM output. They’ve seemingly forgotten that the trust their value proposition rests upon is dependent on the credibility of the content their search results returned, not on the ads served beside them.
At the time I characterised it as big corporations and their offerings “eating themselves.” It’s short-termism at its most blatant. Of course, we have another well-worn parable that warns against this—that of “The Goose That Laid the Golden Eggs.”
Of course, if I search for this parable in my browser, I get an LLM response. Oh, the irony. But here’s what it says:
The parable you’re thinking of is “The Goose That Laid the Golden Eggs.” Here’s the story and its meaning in a clear, short form.
The Parable A farmer owned a goose that laid a single golden egg every day. Each egg was solid gold, and over time the farmer became quite wealthy.
But he grew impatient and greedy. Instead of waiting for one egg each day, he wanted all the gold at once. Convinced the goose must be full of gold inside, he decided to kill it and cut it open.
When he did, he found it was just like any other goose inside—no hidden stash of gold. Now he had no goose and no more golden eggs. By trying to get everything at once, he lost the steady source of his wealth.
Meaning The central lesson is that greed and impatience can destroy a good, sustainable source of benefit. In modern terms, it warns against:
- Sacrificing a long‑term, reliable gain for a short‑term grab.
- Over‑exploiting people, resources, or systems until they break.
- Ruining something valuable by pushing it beyond its natural pace.
People often use this parable about jobs, investments, health, or relationships—any time someone is tempted to push too hard for “more now” and risks losing everything.
They are striking at the very thing that provides them value. And yet these profit-making enterprises seem oblivious to the ramifications on their own utility, and thus business model. Do they think they’ll just be able to magic up a new content and revenue stream?
I feel forced to participate in this downward spiral because of commercial and professional pressures. But I really don’t like it, and can clearly see the folly…
It presents to us a very dangerous and problematic future, if you ask me. The knowledge and innovation that comes from the human component of this equation—the valuable synthesis and problem-solving that is occurring—is in the most locked down and hidden of “walled gardens.” After usurping all of this value, LLM companies are further locking us into only “sharing” that information with them.
For all its failings, the Internet has been monumental in advancing the sharing of knowledge. And its value has been completely usurped and is being monetised and privatised—not just in the capitalistic sense, but also in the knowledge economy sense. The complete antithesis of what the Internet aimed to achieve.
That’s one of the reasons I’ve been posting a few “solutions” that I’ve found to this blog is to help ensure that, in my own small way, I’m contributing to the commons of knowledge. I have the privilege of having some “spare time” to write. Writing is a passion of mine, and I have the skills to be able to build a site etc. to share this. I’m doing it because I hope it helps others. But a lot of folks don’t have this opportunity. Or choose not to exercise it.
I personally find it difficult to envision Stack Overflow finding a model that works again. But I do hope that a band of passionate, insightful innovators find something that can help open up the knowledge commons again. Or that we somehow, collectively, come to our senses and course correct in our own behaviour, reconnecting to the ethos and spirit that has been stolen from us.