tante argues this is not a transitional UX experiment but the productization of a model where Google captures the full value of web content while returning approximately zero traffic. The two-decade implicit contract — you publish, Google crawls, Google sends traffic, you monetize — is being formally terminated on the supply side while content ingestion continues unchanged.
By submitting tante's piece and driving it to 484 points, the submitter amplified the framing that AI Mode represents a deliberate war on the web rather than a neutral product evolution. The submission resonated with a community already experiencing the described traffic collapse firsthand.
The editorial reframes the debate: usefulness of AI summaries is not the interesting question — the question is who pays to keep producing the underlying material once the eyeballs that funded it disappear. Publishers, indie devs, and SEO professionals in the HN thread corroborate this with analytics showing impressions flat or up but clicks down 30–60% year over year, with informational queries hit hardest.
tante counters Google's official framing by noting that 'helps users get answers faster' is technically true but economically devastating — faster for the user means fewer destinations for everyone else. The user-benefit narrative is used to launder what is functionally a unilateral renegotiation of the value exchange between Google and content producers.
In a post titled *On Google Declaring War on the Web*, German technologist tante (Jürgen Geuter) argues that Google's accelerating rollout of AI Mode — the chatbot-style search experience that synthesizes answers instead of returning links — represents the formal end of an implicit two-decade contract between Google and the open web. The deal was simple and never written down: you publish, Google crawls, Google sends you traffic, you monetize the traffic. Everyone wins except, occasionally, the user wading through SEO sludge.
AI Mode breaks that contract on the supply side while preserving it on the demand side: Google still ingests your content, but the answer is now generated in the SERP, and the click never lands on your domain. tante's framing is that this isn't a glitch or a transitional UX experiment — it's the deliberate productization of a model where Google captures the full value of the web's content and returns approximately zero of it. The piece landed on Hacker News with 484 points and a long comment thread of publishers, indie devs, and SEO professionals describing the same thing in their own analytics: impressions flat or up, clicks down 30–60% year over year, with the steepest drops on informational queries.
Google's own framing — that AI Mode "helps users get answers faster" — is technically true and economically devastating. Faster for the user means fewer destinations for everyone else.
The open web was never a charity. It worked because a search engine sent eyeballs to pages, and pages either sold ads, sold products, or built reputation that converted somewhere else downstream. Strip the eyeballs, and the economic substrate collapses. The interesting question isn't whether AI summaries are useful — they obviously are — it's who pays to keep producing the source material once the referral traffic that funded it dries up.
Three dynamics make this worse than the last decade's slow squeeze (featured snippets, knowledge panels, zero-click results). First, AI Mode isn't a snippet — it's a full answer, often hundreds of words, synthesized from multiple sources with citations buried behind an expand toggle that almost nobody clicks. CTR data from publishers who've shared it (Tom's Hardware, Stack Overflow before they pulled out of the conversation entirely, several independent tech blogs in the HN thread) suggests the click-through rate on AI Mode citations is roughly 1/10th of a traditional blue-link result for the same query.
Second, the training-data argument has run out of road. Even publishers who block GPTBot, Google-Extended, and CCBot can't easily block Googlebot itself without disappearing from search entirely. The same crawler that historically delivered traffic is now the one extracting content for a product that withholds it. tante's strongest line — and the one the HN thread quoted most — is that this isn't a negotiation, it's an enclosure.
Third, the alternatives don't exist at scale. Bing's market share is rounding-error in most countries. Kagi is a paid niche product (~50k subscribers, last public number). DuckDuckGo proxies Bing. The structural problem is that one company controls roughly 90% of search worldwide and has now decided the destination is itself.
The community reaction in the HN thread split along predictable lines. SEO professionals are pricing in a 40–70% traffic loss over 18 months. Publishers running ad-supported sites are quietly modeling shutdown scenarios. Indie devs and open-source maintainers are noting that documentation pages — the things developers actually need to find — are getting summarized and decontextualized, with AI Mode confidently producing wrong answers that look right. The Stack Overflow precedent (traffic down ~50% since ChatGPT) is now everyone's precedent.
If you ship a developer-facing product, an open-source library, or any content business that depends on search referrals, three concrete shifts are worth making this quarter.
Stop optimizing for Google's funnel and start owning the relationship: email list, RSS, Discord, a real account system. Anything where the user shows up because they chose to, not because an algorithm allowed them to. The publishers surviving this transition all share one trait: a direct channel that doesn't route through a referral graph they don't control. If your docs only exist as HTML behind Google, build a versioned CLI `--help`, a downloadable PDF, a `llms.txt`, and an MCP server. Make your content addressable through channels Google doesn't intermediate.
For open-source projects specifically: the README is now training data, not documentation. Assume every code example will be paraphrased into someone's AI Mode answer with the attribution stripped and the version number wrong. Defensive moves: pin authoritative answers to your own domain with strong canonical signals, publish a structured `/llms.txt`, and — uncomfortably — consider whether public docs should require even a soft sign-in for the deep technical pages where wrong answers cause real damage.
For anyone running an ad-supported content site: the math is now hostile. CPMs are flat, traffic is declining, and the synthesizer eating your lunch is also your largest referrer. Either move upmarket (paid newsletters, paid community, paid tools) or accept that you're producing free training data for a competitor.
tante's piece is angrier than most takes on AI search, and the anger is load-bearing — the polite framing ("publishers need to adapt," "the web is evolving") obscures that one trillion-dollar company is unilaterally rewriting the economics of every other site on the internet. Regulators will eventually arrive; the EU's DSA and the UK CMA's market investigation into Google search are already circling. But regulation moves in years and AI Mode rolls out in weeks, so the operational question for anyone reading this is simpler: what does your business look like when Google referral traffic is 30% of what it is today, and you have 18 months to get there? If the answer is "the same," you're not paying attention. If the answer is "I don't know," that's the most important problem on your roadmap.
To stop this, I a month or two ago put most of my Amazon Redshift research web-site behind a basic auth username/password wall.It all remains free, but you need to email me for a username and password.If I put in time and effort to make content and OpenAI et al copy it and sell it through their
I don't understand the endgame here. Websites let Google crawl their content in exchange of traffic. If Google cuts that out completely, what incentive do websites have to not block the Google crawlers?I understand that Google is feeling an existential threat from other AI products that provide
We abrogated getting traffic to our websites to Google long ago. Mostly because Google was so good at it that the alternatives became significantly less useful.Now that Google is focusing on becoming 'self contained', so to speak, we should find a better way to drive traffic to websites. I
As a website owner I have seen major upticks in viewership myself but really it hits hard when you see an Ai summary that is wrong and your sites there. The whole Ai for everything push unfortunatly will downskill the world I fear and nothing can be done about it.
Top 10 dev stories every morning at 8am UTC. AI-curated. Retro terminal HTML email.
I feel like AI has gotten to the point where the message is: If you want to make something (art/code/music/writing) you can do it for your own enjoyment, but you aren't allowed to make money from it anymore; only the large corporations can make money from content. If you do relea