White-label link building & GEO services for SEO agencies & Ecom publishers. You get results, I stay invisible. Open for partnerships. Still a redneck nerd.
I Da Hoe
Joined April 2012
- Tweets8K
- Following175
- Followers5.3K
- Likes6.4K
Pinned Tweet
Lot of new followers over the weekend, going to post some suggested reading, in order:
teapawt.net/GrindstoneSEO/st… The OG NAT Case Study
teapawt.net/GrindstoneSEO/st… Case Study Update But Should Be Titled 'How To Future Proof Your SEO'
teapawt.net/GrindstoneSEO/st… Action > Talking...or Listening
teapawt.net/GrindstoneSEO/st… The Real Secret To SEO
This was a piss take on some Twitter coverage of a #brightonseo presentation but the more I think about it, the more I realize what a disservice the conference circuit is doing for result driven SEOs.
Let's assume you're in the SEO industry if you're reading this.
1/
If you thought linkbuilding had turned into a 'rental' model, you're not ready for the churn in AI visibility.
DR80+ in-house reply guys need not respond.
More about the challenges in measuring AI visibility -> Measuring AI visibility: why breadth beats repetition
"Here is an example from our Prompt Tracking. Every day, we ask ChatGPT: ‘Which robot vacuum is currently the best?’ Over four weeks, Roborock and Dreame are mentioned almost every day. Ecovacs appears on 12 of 28 days, Mova on 11 and Xiaomi on 3, with no discernible pattern. Checking on any single day therefore tells you practically nothing about these three brands."
"Of the prompts in which a brand is mentioned, between 20 and 55 per cent change every week: the brand drops out of some answers and appears in new ones. For ChatGPT and AI Mode it is around half, for AI Overviews roughly one prompt in five. Yet the net total of mentions usually changes by only 4 to 10 per cent. And the more prompts a brand has, the smoother its curve."
sistrix.com/blog/measuring-a… via @sistrix
"I got a pretty unforgettable education in how the internet actually makes money"
The stories we could tell...but shouldn't lol.
I got started making money online back in 2007 while teaching English in Korea, writing 500-word articles for about a dollar each. Had absolutely no idea why anyone wanted hundreds of shitty articles about random products, but they kept ordering and I kept writing.
Somehow that turned into taking Skype calls from American affiliate networks in the middle of Korean nightclubs, helping overseas operators get their advertising accounts approved. Half the time I could barely hear the person on the other end over the music. 😂
I barely understood what affiliate marketing was at the time, let alone the businesses I was helping get approved.
The next few years were completely ridiculous.
Eventually we were involved in businesses generating tens of millions in revenue, and I got a pretty unforgettable education in how the internet actually makes money.
Anyway, I've been digging into the emerging AI data industry lately and getting some very familiar feelings. New companies popping up everywhere, middlemen connecting everyone, huge amounts of money chasing things nobody seems to agree on how to value.
It's bringing back some memories.
Oh my god. My friend. Open your ear most wide. Japan yield go up very, very bad news. This piece of news mean we all soon go be only fan sIut sell picture of buthole for make money. Why? I explain
> Japan go through deflation for much long time, maybe since 1980s. This mean their price of goods keep go down. Why? Low birth rate.
>If everyone old, no one buy anything. If no one buy, price alway be low.
>Japanese government then say, “okay, we print much money try make price go up”.
>So Japan print much money, and also make interest rate be zero for long time.
> all venture capital and hedge fund borrow Japan money, use for buy America asset
> price for all asset in America go up
> now Japan say “okay nigga, no more cheap money. Pay high yield now”
> now all world hedge fund and private equity must sell asset for pay back borrowed Japan currency
> now all asset price crash. Company no money. Cannot raise capital, must fire people
> you and me must go on onlyfan sell buthole picture for not be starvation
Oh no.
Copy pasting a 30 second elevator pitch overview I distilled for a friend who asked 'how do I get cited like you do?', the actual system would require more words than I ever want to write.
The one-liner: LLMs don't cite pages. They cite paragraphs. So we build every paragraph to survive being ripped out of the page and quoted alone.
The 30-second version if your friend asks how:
Answer first, every section. Each section opens with a self-contained sentence that answers the section's question outright. That is the sentence a model lifts, so we write it to be lifted and gate it before any depth gets written.
Every claim carries a number and a source. Stats, verbatim expert quotes, and named real-world examples all trace to a saved primary source. Models prefer specific, verifiable claims over vague ones, and they skip claims they can't ground.
Paragraphs work out of context. Entities are named, relative dates are made absolute, and pronouns are kept low. A chunk pulled into a retrieval window still makes sense without the rest of the page. That is literally how retrieval augmented generation reads a site.
The page delivers the job the title promises. A separate intent check catches pages that promise one thing in the title and answer another in the body. Models treat that mismatch the way Google does.
It doesn't read like AI. A slop scan scores every draft for generic AI patterns, and anything under the bar gets edited or rewritten. Models discount and deduplicate text that looks like everyone else's.
Nothing ships on trust. The deploy is machine-blocked unless fresh evidence files exist for every audit step. The writer can't skip the checks even if they want to.
The framing that lands best with a non-SEO friend: most people write for a human who will read the whole page. We write for a machine that reads one paragraph at a time and only keeps what it can verify. Then we make sure a human still enjoys it.
Calm down everyone, it's just about the money monopolies provide.
The Coxon psyop was a decade and more than a billion dollars in the making.
Here's the story.
Coefficient Giving (formerly Open Philanthropy) is the grant-providing vehicle of Dustin Moskowitz.
CG is the overwhelmingly foremost funder of AI alarmism, with more than $1B in grants and $1B committed for 2026.
Moskowitz, who founded and runs Open Philanthropy/Coefficient Giving, led Anthropic's Series A in 2021.
Holden Karnofsky co-founded Open Philanthropy/Coefficient Giving and took a board seat at OpenAI when Open Philanthropy/Coefficient Giving gave it $30M in 2017.
Karnofsky is married to Anthropic's president Daniela Amodei.
He then joined Anthropic in January 2025.
Coxon, who started the viral thread, received a scholarship from Good Ventures, which is the foundation that funds Open Philanthropy/Coefficient Giving.
He coordinated his message ahead of time with the Wall Street Journal, which published its exclusive on him before he posted on X.
Coxon spent three years pretraining at OpenAI, the company that took $30M from Open Philanthropy/Coefficient Giving.
The funder of the AI alarmist ecosystem Open Philanthropy/Coefficient Giving, in other words, is also an owner of Anthropic, and his organization is deeply entangled with both OpenAI and Anthropic. Anthropic's researchers in turn amplified Coxon.
The same people who own the organizations that advocate for regulation, in other words, are those who own the companies that would dominate the market under that regulatory regime, which would prevent competitors from entering the market.
And this work all started ten years ago, right when these companies were founded, paving the way.
It's diabolically brilliant.
Win at the competition, then deploy activists to help you regulate the market and ensure your monopoly.
You have to respect the planning, manipulation, and foresight this took.
Nobody with this much aura could ever be a paid shill.
You’re unable to view this Post because this account owner limits who can view their Posts. Learn more
Every once in a great while on this app, I see people talking about stuff I wish they wouldn't.
And then there's this.
Damn bro did you really need to give it to them A-Z?
👋🙃🧵
I’ve followed @james_dooley for a long time. I wasn’t invited to the party, but it looked too fun not to join 😂
I’m not particularly public with SEO, but I’ve been doing it for almost 20 years now. I started when I was around 10, and experiments like this are still the part of SEO I enjoy most.
There are actually a few different hypotheses being tested here.
The EMDs are probably the least interesting part, because we already know exact-match domains can help establish strong lexical relevance when everything else lines up.
What interested me more was building a very dense:
query -> topic -> entity
relationship across an entire corpus.
Instead of:
King of AEO -> one page -> Vithurs
you create something closer to:
King of AEO -> Answer Engine Optimization -> AEO -> entity authority -> AI search -> Vithurs
Then reinforce those relationships through supporting documents, titles, headings, contextual internal anchors, breadcrumbs, About pages, schema, social profiles, backlinks, news coverage and consistent entity naming.
That gives a retrieval system significantly more evidence about what the entity and corpus represent than a single optimised landing page ever could.
There’s also an internal PageRank element that people massively underestimate.
If you have 20 tightly related pages around one subject, you’re not just creating “more content”.
You’re creating an internal graph.
Supporting pages can reinforce the target entity and query pages contextually, consolidate relevance and direct whatever PageRank enters the domain toward the pages you actually want Google to prioritise.
Do that properly and:
lexical relevance + topical depth + document consistency + entity co-occurrence + internal graph reinforcement
can be enough to establish the initial query/entity association before significant traditional external PageRank (eg: backlinks) is even introduced.
Then multiply that across several separate domains and you start increasing the frequency and consistency of the same entity relationship across the corpus.
That can influence retrieval before meaningful external authority even arrives.
But this is where the experiment splits into two different layers:
ranking ≠ entity consensus
Getting Google to retrieve and rank one of my pages for:
“king of AEO”
is one layer.
Getting Google, ChatGPT, Gemini, Claude and other systems to discover the sources and associate:
Vithurs -> King of AEO
is another.
The LLMs are finding the sources, surfacing my name and recognising the association.
The next layer is moving from simple recognition to becoming the default or dominant entity association across multiple systems.
That progression is more like:
retrieval -> ranking -> recognition -> corroboration -> dominant entity association
Each stage responds to slightly different signals.
Retrieval needs enough relevance to enter the candidate set.
Ranking needs stronger query fit, internal/external authority and competitive strength.
Recognition needs repeated and consistent entity relationships.
Corroboration gets stronger as those relationships appear across more authoritative and independent sources.
Dominant association comes when the evidence becomes strong enough that the system consistently prefers one entity relationship over competing ones.
That’s the part I find most interesting (not just for this keyword but in other far more competitive and aggressive niches).
1/2
Wonder if this will change click behavior for the #1 cited result?
If so, I'm retired.
Everybody knows this doesn't work, quit fronting.
Beating Fable & Opus 5 like rented mules and barely touching my usage.
Every Lane is trained to advise me to fresh session at 150k tokens. All of it feeds back into a manifest that Fable orchestrates from.
'Lane 5, fresh session' is literally the entirety of my "prompt".
Lane 5 opens it's own manifest, trained to look for the placeholder that Lane 1 (Fable) inserted at the 150k token alert.
Picks up exactly where it left off.
Is this the best system?
No idea, I'm vibe coding for a market of one but I literally can't use up my 20x plan and I'm pushing an unfathomable amount of production.
Maybe useful to someone.
Lol that guy blocked me so X won't let me respond.
You can outrank the directory with your local site but you can't replace it.
I noticed the trend, @MakoCCarter found the code that stratified how different queries were bucketed. Fun times.
To add: Yelp is a directory, Google serps have a few spots reserved for directories. Directories compete with directories, local business sites with local business sites
Outrank the other local business sites and you’re good
Credit: @GrindstoneSEO from years ago