@RyanJones
Account based inUnited States!
About this account
- Account based in
- United States
- Connected via
- United States App Store
X says this location may be affected by a proxy or VPN.
Account-level information from X, not a live location or the device used for a specific post.
SVP, SEO & Stable Genius @ Razorfish. Founder https://nitter.cf/t.co/hiC5n0dzWE Works from early in the morning until late in the evening. Makes many calls and has many meetings.
Trenton, MI
Joined March 2007
- Tweets81.8K
- Following1.3K
- Followers13.1K
- Likes14.8K
Pinned Tweet
Made a Chrome extension to bring semantic SEO to everybody for FREE!
Excited to launch our FREE Chrome extension that uses Chrome's built in gemini model to do semantic semantic analysis of websites. Extract keywords, score content, find the most relevant "chunk" and more! serprecon.com/chrome
Ryan Jones retweeted
Looks like it dates back to June 2026 (dated late May 2026), before the August spam update and now September spam update. It's Google’s second system identified in 2026 designed to catch AI slop -> Google Has Deployed A New AI Spam Detector Called SAFE
"Google has published a research paper about detecting spam that mimics a human manual review that catches content that violates the “spirit” of policy violations and platform guidelines. The system is called Scaled Abuse Forensics Examiner (SAFE) and it is expressly designed to identify AI-generated content."
"The paper says SAFE identifies “spirit of policy” violations primarily with a few-shot-trained LLM. The goal of SAFE is to catch content that may not match an existing rule or known violation pattern but still violates the intent of the policy or platform guideline and can go undetected by traditional classifiers and fine-tuned violation-detection models."
searchenginejournal.com/goog… via @martinibuster
Here's a good example from one of my hobby sites of what happens when searches that didn't have ads/ai/features/whatever suddenly get them. Impressions flat. Ranking position actually increased - but clicks tanked.
Far more terrifying that someone in a Senate seat is this scared about what flight attendants, pilots, and ATCs enjoy in the privacy of their own bedrooms.
TOMMY TUBERVILLE WARNS THAT “MANY” PILOTS, FLIGHT ATTENDANTS AND AIR TRAFFIC CONTROLLERS ARE FURRIES: “Lord help us. It’s terrifying when you think about this.”
Tuberville claimed on the Senate floor that airline workers being part of the furry community poses a danger to passengers, saying a pilot “could have a deranged fetish of dressing up like an animal.”
Let's talk about the new google query fan out change in simple terms.
Old way: Give query to AI. Ask it to generate more queries. Turn those into vectors, find documents that match.
New way: Train ML model on hundreds of thousands of queries, Turn original query into vector. Model finds other vectors near it. Search index for matches.
Here's another uncomfortable SEO truth. a test with made up keywords can NOT accurately test whether something is a ranking factor. It can ONLY test if google sees it / indexes it.
Why? Because other than fastrank (pure BM25 and Cosine Similarity) almost every other Google system compares your passage/document to the entire corpus of other relevant documents. When you make up a term, there's no corpus to compare to. You're the only relevant document/passage - so all the other Google systems get skipped. So you're not really testing anything.
Yall keep debating LLMs.txt and meta descriptions and testing keywords nobody searches. I’ll keep reading the textbooks written by actual search engine employees and understanding how they really work.
What a lot of SEOs don't get is the language Google uses isn't how the algorithm works or what it does.
They talk to us in terms of what the algorithm is trying to reward rathern than HOW it rewards it
because if they talked about shit like BM25, max marginal relevance, jaccard means, DSSM, mean reciprocal rank, normalized discounted cumulative gain, chunk retrieval, cross-encoder loss metrics, etc - MOST people wouldn't understand them.
So instead they talk about the user experience those metrics are constantly tuned to measure. So we get EEAT and information gain and commodity as human descriptive proxies for all the fancy math.
Truth. and so many people don't understand this. It's built into how the LLM chooses each word. You can't "reverse engineer" it or "detect" it unless you have the key AND the inner workings of the LLM to see the words it didn't choose to include.
There's literally patents and research papers on this but most people ignored it.
I wish Michigan would put "I voted" stickers in the absentee ballots they mail out. I want my sticker.
I've got a couple openings for SEO strategist and SEO supervisor in the following locations:
Plano, TX
Birmingham, MI (Detroit)
Chicago, IL.
If you know somebody interested (sorry, these locations only, full time, 3 days in person, eligible to work in USA, etc) please have them get ahold of me.
Pushed a small change to FreeSOV.com this morning. The cost report now shows you your spend per brand, in addition to overall account.
"I'm sorry I can't see that tiktok"
coworker: why?
me: "because I'm a man over 45. I do not have tiktok"
LLMs.txt and Markdown files are the new META keywords. I'm shocked most of you aren't still putting "Britney Spears" in there like every car dealer used to back in the day.
I'm seeing so much weirdness with markdown files. You don't want to put a copy of your actual site in markdown in your sitemaps. bad idea. Search engines will index both. do you want that served to users?
You also don't need separate markdown for AI. Just fix the random things on your website that they can't crawl - or use a pre-render or cache.
Most people are way over-thinking this.