The issue with someone like Google, or Microsoft, is that they have a larger moat in many products, so the amount of degradation that internal actors will be happy to push in search for profit is much deeper.

This is the reverse of my desires. Most pages on the current web contain a few useful nuggets buried in a shit sandwich of overly verbose text and aggressive advertising. AI extraction and summarization distills the mess to concentrate the useful bits.

It probably won't be huge but isn't that kind of the point?

Then, they essentially bankrolled their enemy by giving websites a way to monetize spammy content. By the late 2010s, they started losing the game for many types of organic queries - although I many HNers forget that the most common and "important" search queries are things like "Nike shoes", not "efficient implementation of a Bloom filter". And then, in the 2020s, they pioneered LLMs and made the technology public, and drove a final nail in that coffin by making it possible to produce SEO spam on any topic, practically for free.

Now, they see AI answers as their salvation, steamrolling the non-monetizable use cases that never really mattered to them. And this will probably work only until AI-poisoning SEO is in full swing, at which point Google will find themselves with dissatisfied users and diminished ad revenue.

Up until around 2015 desktop traffic dominated, google pre 2015 tended to favour big comprehensive articles that covered not just the question being asked by the searcher but also everything surrounding it. Reading such content would inform you not only of the answer to the question you were asking but also would give you enough context to understand if you were even asking the right question to begin with.

To me this was an absolute golden age driven by the ability for website users to create good comprehensive content and monetise it via ads.

Post 2015 google shifted to prioritising content that directly answered the question being asked and little else. I assume because mobile users just wanted their answers which ultimately lead to the "snippets" that directly answered the query in google itself.

Google invented the tech for LLM's but did basically nothing with it until OpenAI forced their hands because they knew that it would kill the ad driven web. Once OpenAI did force their hands we end up with search as it is now.

Pre 2015 there was a point to creating good information on the web, now there isn't, it'll just be stolen by LLM's, there's literally no point beyond your own personal altruism.

Thankfully for google it owns youtube and there is still an incentive to upload useful information there, so that's increasingly where new information will live, but the web as we have known it most of our lives, is done.

Or for your own reference later. Or for the fun of it (some of us are that type of person!). But yeah, little else.

(The requirement for monetization on YT is 3k "watch hours" per month. If you upload 5-minute videos, and 10% viewers make it to the end, you need hundreds of thousands of views.)

But now, it's worse than altruism. The work your doing doesn't directly benefit the masses, it benefits OpenAI & co. You're not doing charitable work, you're a total chump.

If you're someone providing information for free, because information wants to be free or whatever, then you should be happy when a copyright agnostic (ignorant?) AI company scoops it up for their next model. In a way, you're directly shaping the zeitgeist for everyone that uses those models.

You don't have to wait for that one lone person to stumble upon your site. Now everyone gets an answer from an LLM directly shaped in some small way by small changes to a few of a few billion weights.

The entire premise of this thread is that search is fucked in large part because there's no incentive to upload any information any more beyond altruism.

The LLM's can figure their own shit out, I'm not feeding them for free.

Of course YT creators can do whatever the hell they want. They don't have to listen to me.