Learn How LLMs Fan Out Process Really Works

Discover how LLMs’ fan-out process really works — breaking complex queries into parallel tasks for faster, smarter AI responses. Learn the full breakdown!

With Special Returning Guest David Quaid, A Leading SEO agency, PrimaryPosition in NewYork. 

This Week’s Sponsors

LifterLMS: LifterLMS

Rollback Pro: Rollback Pro

The Show’s Main Transcript

[00:00:22.190] – Jonathan Denwood

Welcome back, folks, to the WP-Tonic show. This is episode 1015. We’ve got a returning friend of the show, David Quaid, with us. He’s a leading SEO agency owner, primary position, based in New York and West Palm Beach. He’s become a friend as well. Been giving me a lot of support and advice, but he, I watch almost all his content and he came on the show last month and I just had a load more to ask David about. So I thought I would have him back on the show. This week we haven’t got Kirk with us. He’s having some internet technical problems, but he will be, he will be with us next week. So David, would you like to give yourself a quick intro to the tribe and then we’ll get stuck into the main topics of this great interview? Sure.

[00:01:18.880] – Dave Quaid

Thanks so much for the intro. Um, so It’s such a pleasure to be here and real honor. I love helping out all of the different communities in SEO. I think SEO is a great, such a great way to make a living and such a great way to build businesses. I see it as sort of like being the ultimate brand breaker and brand builder. And anything I can do to help make it easier for people to understand, that’s what I’m here for. So if people have questions after the show, or want us to deep dive, I’m I will answer every question.

[00:01:49.380] – Jonathan Denwood

Yeah, we’re gonna be covering a few things. We’re going to start off with EAT, Experience Expertise, or Authority, or whatever they call it. Um, and, um, David’s been hammering away with it. He’s probably a bit bored with the subject, but I wanted him to do a quick 101, and then maybe a quick, um, the biggest misunderstandings, um, and we’ll be covering a lot more in the show. Um, but before we go to the meat and potatoes, I’ve got a message from one of our major sponsors. We will be back in a few moments, folks.

[00:03:03.050] – Jonathan Denwood

We’re coming back, folks. Want to point out We got some great special offers from the major sponsors, plus a curated list of the best WordPress plugins, services aimed at the power user, freelancer, and small agency owner. You can find all these goodies by going over to wp-tonic.com/deals, wp-tonic.com/deals. That’s where you find all the free goodies, my friends. What more could you ask for? Probably a lot more, but that’s all you’re gonna get on that page. So let’s go into it. David’s already bored, but, uh, um, but he’s gonna have— can you just give us a quick 101 intro about it?

[00:03:47.470] – Dave Quaid

Okay.

[00:03:47.760] – Jonathan Denwood

And then, and then, and then we can expand into some of the biggest misunderstandings about it. But can you give us a quick 101 intro about it?

[00:03:58.570] – Dave Quaid

Okay, so EAT is essentially a collection of 4 very subjective terms, which are experience, expertise, authority, and trust. That essentially, when you look at a specific type of website, that’s what Google thinks a person says, like, do I trust the site overall? It’s not for every site. It’s only for YMYL. The definition Google gives for YMYL is incredibly narrow, right? It’s not everything and anything financial or supplement-based. I’ve seen people love lawyers into YMYL. I’ve only ever seen two YMYL sites in the last 5 years. And one of them was a COVID vaccine site. And I could tell because absolutely nothing we would write without meeting certain requirements would rank in Google at all whatsoever. The only other site that also had the same problem was a COVID testing site in the New England area. Boston Mass. The limitations on ranking were nothing like law. Law is a walk in the park in comparison to YML. I’ve done supplements, I’ve done adult medications, I’ve done, I’ve helped people with adult websites, I’ve done everything. And that is COVID technologies is the only one that’s been anything close to what Google says it places emphasis on for YML.

[00:05:18.650] – Dave Quaid

In other words, if we didn’t put very specific things in, the pages would not show in Google. And I’m talking about maybe 5 pages globally ranked for those specific searches. That’s when you know you’re in a YMYL area. If you put— if you do a search and there’s 100 million results, that’s not a YMYL search. And it’s about the search, not about the page. Certain searches are YMYL, not certain pages. So if we looked up the company name, it would rank, no problem. If we added COVID and removed the company name, it didn’t rank anymore. So EAT basically is a set of subjective criteria because we all have subjective values for experience, expertise, authority, and trust. In other words, if someone came to me and said they had 6 months SEO experience, I would think that’s very little. If someone went to my mother and said they had 6 months SEO experience, to her, that’s an expert, right? And so the problem with ETH is that there’s no way for a machine to understand it. It’s not because it’s a machine versus a human, it’s because a machine is trying to take one point of view.

[00:06:27.260] – Dave Quaid

So if you have 100 people and a scale where one-third of the people think that 5 years is good and one-third of the people think a year is enough, you can’t have a machine take that scale, right? Because it either has to please everyone or it has to piss off two-thirds. It’s one or the other. And that’s why they came up with the Quality Rater’s Guide, because they wanted to see an overlap of what their spam detection systems caught and what people thought from a subjective point of view wasn’t real. And so they contracted to a bunch of different agencies to provide human talent that would rate these sites based on a document they were given. The problem with it, and it’s all a great idea, the problem is that EAT has blown way out of proportion. First of all, people are talking about /e/ applying to everything. Whether you agree with my definition of why and while or not, that doesn’t matter because people are saying /e/ is everywhere. It’s poisoned the LLMs because everything I see that’s manually, sorry, that’s programmatically driven into Reddit, for example, all contains /e/-e-a-t, which is like symptomatic that it’s being written by an LLM.

[00:07:35.300] – Dave Quaid

People will write /e-e-a-t/. It’s just faster. And then people have jumped into this literal definition that you have to talk about experience, talk about expertise, talk about authority trust, you have to have an author bio, you have to cite other websites. And they’ve just tried to build it into a literal thing that Google has tried to understand, which it doesn’t. And they’ve been very honest about it. And they’ve been very open that they’ve tried to debunk it as much as they can. But it just keeps persisting. And there’s so much time and money going into it, it’s just pure malinvestment. It’s not going to change your rankings on your website. There’s one, a 1-line reference in the Google SEO Starter Guide that just said, thinking it’s a ranking factor, no, it’s not. I mean, to me, that’s like Google making a parody of it, right? Making like absolute fun of it and what it’s been built into yet. I don’t know if it’s cognitive dissonance or confirmation bias or demand gen, but people are keeping this E thing alive and I just think it’s just, it’s just such a massive roadblock between getting somewhere and success.

[00:08:43.310] – Dave Quaid

I can’t imagine starting every one of my blog posts with a short resume on everything I’ve done in a particular area. It just, if people don’t think I’m an expert from what I’m talking about, then that’s a decision they can make up in their minds, right? I’ve spoken to a lot of people on Reddit who because I don’t like schema, or I don’t think schema is a boosting mechanism in SEO, even requirement for most things in SEO, they automatically decide I’m not an expert. And that’s fine. That’s, that’s how subjectivity works. That’s what makes us people. And it doesn’t matter if I say I have 100 years SEO experience. If someone doesn’t like me for a view I hold, they’re just not going to like me for a view I hold, right? And that’s how we have to look at it.

[00:09:25.520] – Jonathan Denwood

Yeah, I think I’ve, to myself, listening to a lot of your content, listening to your last interview yesterday— well, it was published yesterday, published today, um, on the Edwards Show. And thank you for the mention on the Edwards Show that you had to me. I appreciate that. And, um, basically this is my take in it. There seems to be kind of 3 buckets there. There’s the bit that, um, it is used for explaining why AI content will get you kind of shadow banned. This idea that you’re going to get shadow banned, which you really have attempted to knock out of my mind because you’re pointing out If you’re publishing hundreds of articles every week, um, and it’s really AI crap, basically, yes, you’re, you’re clearly breaking Google’s guidelines. Um, but there’s this concept that if you ever use AI on your website, you’re doomed, right? Um, Then there’s this other bucket, which is, well, just write really good content. And your point, I think you’ve attempted to point out, that sounds great, but how do you judge what is really good content? It’s so subjective. Uh, one person’s good content is another person’s brain yawn, you know, you know, it’s normally um, you know, the length of the piece.

[00:11:13.080] – Jonathan Denwood

And then they, um, bring out that, um, obviously these guidelines were brought out, like you say, when they had a group of manual checkers and they had to give a guideline document. But people say— and I’ve clearly heard people, certain individuals on Edward’s show and on other shows on YouTube that Google has used AI for pattern recognition. And if they see you have loads of pages with the same kind of structure, that, um, and a lack of clicks, that, that basically you get a shadow ban, basically. And I think I’m presuming you don’t agree with that, or do you?

[00:12:05.420] – Dave Quaid

No, not, not really. Not, not that type of description. So, um, so I’m not bored of the topic. I’m worried about boring people to death on the topic that don’t want to hear it anymore. But, um, I love Sean Anderson on X. I know he’s busy building a project right now, but Sean and I can stay up for 5 hours talking about ETH. And he did a lot of research into ETH and a lot of research into, the API leak, and we have incredibly interesting, at least I think so, interesting conversations about it, and I think we largely agree on it, right? First of all, this idea that LLMs are an answer to everything, right, that they can somehow do everything is, is laughable, right? Because we’ve put so many things in front of LLMs and watched it pivot 180 degrees, right? So you listen to people saying that LLMs are trained and that they do all this brand research and they’re able to figure out a good brand from a bad brand. That’s laughable. If you look at the fact that before I wrote a page about an EAT checker, they were saying, oh, EAT checkers are brilliant.

[00:13:14.210] – Dave Quaid

It’s a great way to check how much EAT you have in your page. And you’re after checking this, your content will go up. And it was actually from a disagreement with the tool builder who attacked me on X and said I didn’t understand SEO, that my SEO is from like 2009. And it just started with the, you know, the usual logical fallacies, the ad hominem attacks, like how do I discredit this guy? How do I make him look like a fool type thing, right? Which as humans, we do this all the time, right? And then started saying, oh, your SEO is out of date, Google’s moved on, it has LLMs, right? So like, just LLMs are just magic now. And I replied back to him and I said, look, I’m interested in the fundamentals of SEO and I see it as a reverse engineering project. And I take it the exact same way as I would have as a software engineer in terms of like trying to solve a problem. And then I went to have a look at his tool site. So he’s building a tool for SEOs that helps them with their SEO.

[00:14:11.620] – Dave Quaid

And he didn’t rank for anything, but he’s been building this tool for 7 years. And he did have backlinks and he had a blog and he’s building content. And he had a tiny following on X. So he’s not building his tool through social media, right? He’s on social media trying to make up for a lack of SEO. So I thought it was very funny that, uh, I have a large amount of documents ranking for eat, and he’s saying, look, Google’s eat can tell if your content is real, if it’s researched, if it’s produced, and so on. And it does that with LLMs. So I said, well, if that’s the case, right, let’s just solve it very simply. How have I got a page ranking that says eat isn’t real for searches to do with eat? And how is it that you’re so adamant about it and you have a tool that says your eat, it’ll check your eat and that will make you rank and you don’t rank? And to me, that’s just a very simple test, right? And he stopped talking to me after that, which I was happy with. But how can you write a document about EAT and say it’s bullshit?

[00:15:15.260] – Dave Quaid

And I literally used the word, um, and I didn’t put in any outbound citations, and I didn’t put any author bio in the page, and I didn’t put anything else in it, um, yet I rank for like 20 different EAT terms, right?

[00:15:31.970] – Jonathan Denwood

Yeah, I think there are some— there’s, you know, I’m not going to name them, you know, there’s a couple that are really pushing Eats and, and using AI content. And you— they use the term shallow content. Um, that’s why you’re— and they keep pushing that and pushing it. And I think it does cause a lot of confusion because Um, you know, I think you pointed out that— I’m not sure it was you or somebody else— that a lot of content isn’t indexed by Google. It’s indexed and Google doesn’t really do anything with it because so much content, because of AI, AI and other factors, so much content is produced every day and it costs them money to index. They’re not going to index everything. They’re just not going to do it.

[00:16:35.410] – Dave Quaid

So let’s look at a couple of things that are going on here, right? Firstly, let’s talk about EAT and good content and let’s talk about, let’s try to build some middle ground and find a happy place for everyone for EAT to live. I suspect that a lot of people say EAT is great because they want to see good content on the internet. I believe a lot of SEOs are saying write good content because If you write good content, people will link to you, it’ll naturally look after itself, you don’t have to worry about link building. I don’t think that either of these statements is going to reduce the amount of spam. I don’t think that spammers listen to you, me, or anyone else. I think spammers will just do what they want to do, right? And they will keep doing whatever they’re doing that works. And if it stops working, they will just move on to something else. I also believe that The vast majority of people that are trying to figure out SEO are writing the best content they want. Now you could say maybe their content’s not good enough, which I think is unfair because most people give that reply without ever having read it.

[00:17:36.410] – Dave Quaid

But notice how there isn’t any sort of thread on Reddit or LinkedIn, on X, where people write their— share the content and go like, this, this is the best content for this topic. And that’s because any one of us could come along and tear it apart because it’s subjective, right? Just the way we can do with anything that’s subjective, right? If someone says this is the best pizza in the world, even if you have a series of bullet points that says this is the best pizza, I’m sure you’ll find people who won’t like it just because they don’t like tomato sauce and they only like white pizza, right? And so it’s just an endless circle of noise. The fact is that if people are worried about the spam, then why is the spam ranking if Google is so good at kicking it out? Right? If it’s self-correcting system, why worry about it? Why make these edicts, right, that you have to write good content if Google can kick it out and detect it? And then secondly, what happens when you write all this good content and then you realize that it’s a pyramid and only the top 3, 5, 7 posts actually get clicks.

[00:18:47.270] – Dave Quaid

And even before we had AI content and we had like Mount AI, uh, there’s 500 million pages for urgent care. There isn’t 500 million people in America, right? So we had all that content before. Uh, so I don’t think it’s achieving anything, and I don’t think it’s helping people who genuinely want to rank in Google and genuinely wrote the best content that they can. Right. Because at the same time, if you have 500 million pages in an index, nobody is going to crawl through 500 million results to find your particular piece of content. I think that’s very, very naive.

[00:19:23.900] – Jonathan Denwood

Yeah, I think it’s cool, but I think it’s cool because a lot of people, and I’m one of them, have seen a drastic fall in traffic and clicks on their website. And that’s a mixture of different things, isn’t it? It’s a mixture of AI overviews. It is, it’s a mixture that people are using large language models on their phones and they’re not using Google to do search to some extent. It’s probably, that’s probably slightly overblown. It’s more, it’s probably more the AI overviews in my opinion. Be interested in yours, but then the Google have been producing a lot of core updates and I’ve been hit. And you’ve— and I did use AI content, but I always edited it. I always tried to add value to it. You strongly, you know, I was vermin that I had been shallow penalized, and I’ve come to the conclusion that you’re correct, correct, that it’s more to do with my backlink.

[00:20:32.400] – Dave Quaid

So your authority, your authority shape, not Not backlinks, right?

[00:20:36.960] – Jonathan Denwood

Um, to shape your domain authority, um, to some extent. And it’s just too complicated to give me a precise reason, but you see a, you can see a gradual reduction. And I know that there were certain problems with, with content that I ranked for that really didn’t have a lot of relevance to what my services are. So it’s complicated, but I think what— why this is important is that it— this, this idea that it’s quality linked, that if you ever use any AI content, that defies EAT and you’re going to be banned for life. There’s a lot of that kind of propaganda.

[00:21:28.030] – Dave Quaid

Felix? Let’s dig into that because you, you’ve brought that up a couple of times, right? So talking about experience and expertise, as 90% of people who enjoy E talk about it, right? They, they see this literal, you have to talk about experience and that’s what Google looks for. It looks through these patterns that you’re talking about expertise. There is nothing stopping an AI from doing that. Google cannot validate an author bio, years of experience, where they live, their date of birth. It cannot validate any of that, right? Otherwise, we could just walk into DMV and go, here’s my author bio. It’s a real ID. I want to jump on an airplane. It just doesn’t work like that, right? So an AI can create— if you think EAT is talking about experience and expertise, an AI can do that, no problem. Google doesn’t ban AI content. It has a single document that says, we don’t care if you use AI content, you should edit it just as you should edit it, right? The, the scaled content, which is what I think you’re confusing it with, is about the mechanism, not the output, right? So even if you build a system that is edited by Martha Stewart and 50 Cent, right?

[00:22:42.030] – Dave Quaid

Um, it doesn’t matter if it’s machine-built, machine-generated, machine-targeted, and built to manipulate search. It’s bannable. It doesn’t matter. It doesn’t have to be AI content for, for Scaled Content. It’s just that is the easiest way to do it, right?

[00:23:01.450] – Jonathan Denwood

Yeah. So before we go to a half break, what, what are some of the signals they’re looking at that they utilize to make a judgment call about something? About where it will show up in the index on a, if somebody does a query, um, what are they looking for then in based on your experience? You know, because I think you’ve said it and I think a lot of other SEO experts say that backlinks, they’re still important, but they’ve reduced in importance to some degree. So Dick of Backlinks, would you agree with that? Or not? And then what are some of the things that they are looking for from a piece, article, a page, a post?

[00:23:52.190] – Dave Quaid

So think of backlinks as currency and think of authority as like a market, right? And so if the Federal Reserve wants to reduce inflation, it tightens money restrictions, right? It doesn’t go into the economy and buy up mines and make commodities cheaper, right? And so a lot of— I’m seeing SEOs on X saying Google will get rid of listicles, and then people are like, how come some listicles are still there? They’re like, oh, well, those listicles get a pass. Can you imagine how many people it would take? I mean, if Google’s ingesting a billion documents an hour, how many people would it take to, you know, manually go, oh, well, this is a good listicle, this is a bad listicle, right? Like, why would they go to that level? Either listicles are allowed or they’re not allowed, right? There can’t be like some people can have listicles and other people can’t have listicles. But what Google can do is control how authority is shared and shaped, right? Like how much money you’re allowed to send overseas, how much of a currency is allowed to be traded in a day. They can control those basic fundamentals and that’s all they’ve ever done.

[00:25:03.910] – Dave Quaid

And the net impact of where can authority come from, how much authority gets passed, what’s the relationship between relevance and content, how do they determine that determines where you rank. So backlinks are just a source of authority, right? Like if you think of power or if you think about currency, you banks can create debt. Central banks can create debt or print money and people pay it back or don’t pay it back or pay it back with interest. So if you look at what does that look like in Google’s algorithms, you get authority from, um, other pages that are external pages that have traffic that pass authority to you. Those pages could also collect authority because lots of other pages link to them. And then there’s a separate ongoing test at the same time of do people like this content? So do people bounce around a lot? And so there’s a trade-off between do we rank this page simply because Microsoft linked to it, or do we keep this page ranking because lots of people seem to like it? And it’s just a— and it—

[00:26:13.430] – Jonathan Denwood

but how do, how do they judge that people are liking it?

[00:26:17.470] – Dave Quaid

So if I do a search for for luggage, I need to travel to New York next week and my suitcase got broken. So I need to buy a new suitcase. As I search, as I do one search for suitcase, the first site I click on now has 100% click-through rate. If I have to do another search because I didn’t like that suitcase and I click on another site, suddenly that site’s click-through rate drops to 50%. And as other people repeat that exercise hundreds and hundreds of times an hour, eventually one page emerges as having a higher click-through rate than the others. And that’s at a very basic, very, very simplified model. That’s how click-through rate testing works. And if you keep doing that, you don’t actually have to worry if content needs to be fresh or not because it’s automatically built into people’s reaction. So if people click on a page and they’re like, wait, this isn’t offering me any new news, go back and do another search, and then I stop searching, then that’s how all Google has to do is look at what—

[00:27:24.860] – Jonathan Denwood

because what you’re saying, strong, strong linkage to intent.

[00:27:30.910] – Dave Quaid

Yeah.

[00:27:31.250] – Jonathan Denwood

If they come to the WP-Tonic through a search and they get hit a page and they— this has got nothing to do with what I was looking for. I’m off.

[00:27:42.080] – Dave Quaid

Exactly.

[00:27:42.480] – Jonathan Denwood

And enough people do it, my position on that search term will go down.

[00:27:49.990] – Dave Quaid

Exactly right.

[00:27:52.090] – Jonathan Denwood

And this is around, you know, very— and what you— I thought, I thought you did an excellent job on your last Edward Stern show that’s published today, and it really tweaked with me a couple sentences, even in my dull skull, David. Is that you said that, um, it’s to do with domain authority. Um, and you said about my site, I, I had a certain, because I did, uh, about 18 months ago, and obviously domain authority in the leading tools is only a very loose measurement, but it gives us some guidance. But I had a domain authority that was around 70, and it’s now gone down to around 52, 53. But for a site to have almost 70 domain authority, I think at its peak it’s around 72. I could really compete for a lot of terms, a lot of phrases that I really, when they came to the pages, they weren’t being satisfied, but my domain authority was covering me. As my domain authority declined, and it’s very difficult to judge why it’s declining, that path shrinks. I think that’s what you were talking about on the Edward Show. I’ve got that correct.

[00:29:19.500] – Dave Quaid

Yeah. And so your, your domain authority as measured by like Ahrefs or Moz could have dropped because the amount of traffic you were getting has dropped, or it could have dropped because of the way Google looks at it’s changed the weighting and it’s rewarding pogo sticking more. And therefore the domain authority is actually the drop of the domain authority is an outcome and not a cause, right? And that’s the thing that we don’t know.

[00:29:48.720] – Speaker 4

Yes.

[00:29:49.520] – Dave Quaid

As well as it just could be 10% from each bucket, right? So, so it’s an algorithm and the way algorithms work is they’re mathematic. You can, you can play with them like mathematics, right? Like you can, If, if let’s say the, the equation is, you know, A times B in brackets times C, the order in which the multiplication goes can vary even though you have the same numbers. If you’ve got a 1 and in another equation you’ve got a 3, you’ll end up with a much more complicated or a much larger number at the outset, right? So you can choose which part of the algorithm to optimize and sort of like give yourself an outstaged advantage. I think in your case, because you, you were ranking so well and you were getting so many clicks, maybe you didn’t need as much SEO optimization, whereas I’m working with companies with like tiny amounts of authority and we have to outrank a Citrix. And so we have learned to play a particular game that 26 years later still works very well. And so you not making those changes, like for example, the way your, your documents are named or, or slugged, um, or making those changes quickly enough, saw a stepped down in your ranking and traffic.

[00:31:08.410] – Dave Quaid

And as those pages themselves come down, the authority that they pass internally goes down with it, right? And so I think you’re just losing percentages from each. Area, each facet of, of what Google’s calculating. But the net result is that each day it’s going down. And so the total domain authority is going down with it.

[00:31:28.900] – Jonathan Denwood

And it self-feeds itself.

[00:31:30.680] – Dave Quaid

Exactly. So you’ve got to figure out how do we stem that? And I’m pretty sure a lot of people out there right now are seeing step-down losses, right? And like you said, it could be AI. Um, if it’s just AI overview, then everyone is experiencing the same thing unless you’re in you know, every— all of your competitors should be in the same bucket, right? Obviously, like, SaaS and B2B is not as affected as much by AIO, at least not, not what I’m seeing. But if you were getting a large amount of one-off clicks and question clicks that can now simply be answered, maybe that traffic isn’t as important. But if you’re using AdSense, that’s going to have a big financial impact, right?

[00:32:08.740] – Jonathan Denwood

Well, a lot of my, a lot of my stuff was top of funnel, educational. So that will be affected. And like what Edward says, you know, medium funnel, bottom of funnel, that’s where you should be trying to produce your content in 2026. Would you agree with that?

[00:32:28.040] – Dave Quaid

So it depends, right? I hate giving blanket advice to everyone, but I have clients where you could focus on bottom of the funnel pages and lose 90% of your traffic. And actually still have business come in, right? Because you’re now cleaning up people that are unhappy with your competitor and they want to move on. I think you can continue to grow traffic in this market, as long as you’re able to find out what queries are resulting in clicks versus what’s being lost to AI Overview. If you— Google’s looking at a holistic number and saying like, okay, we’ve lost 23% of clicks, but 70% of our clicks are still going out there. I don’t know if they’re too broad for other people to go after, but whenever we see a loss in clicks, we just move on to a new topic and start publishing on a new topic. And maybe the conversion rates are different, but we can make up for that with volume.

[00:33:28.610] – Jonathan Denwood

Yeah, that’s a great place for us to stop for the middle break. We will be back. When we come back, we’re gonna be talking about large language models, uh, interview that, um, David had lately with Zachary Long, and a lot of other stuff. It’s going to be a great second half. It’s been a great first half already. Hopefully we haven’t lost too many of you. Uh, I’ve educated myself a lot by working with David. We will be back in a few moments, folks.

[00:34:00.330] – Speaker 1

Updating plugins and themes shouldn’t feel risky. That’s why over 300,000 thousand WordPress users trust WP Rollback. With just a click, you can safely roll back any plugin or theme. No manual file uploads and no downtime. And with WP Rollback Pro, you can download premium plugins and themes. Tools like Elementor, Gravity Forms, more. Track every rollback you do with detailed activity logs. Protect your workflow and take the stress out of updates. Learn more at wprollback.com.

[00:34:37.320] – Speaker 4

This podcast episode is brought to you by LifterLMS, the leading learning management system solution for WordPress. If you or your client are creating any kind of online course, training-based membership website, or any type of e-learning project, LifterLMS is the most secure, stable, well-supported solution on the market. Go to lifterlms.com and save 20% at checkout with coupon code PODCAST20. That’s PODCAST20. Enjoy the rest of your show.

[00:35:14.850] – Jonathan Denwood

Coming back, folks, before we go into the second half, are you looking for a real great technical partner? Have you got a large membership or community BuddyBoss website? And it’s a big project and you’re looking for specialist hosting and just a technical partner. Well, if that’s you, why don’t you look at WP-Tonic? We work with a number of agencies. We provide a white label service. We will consult and build out your membership, your BuddyBoss, your Fluent Community website with LifterLMS or LearnDash. And we also offer great packages, great affiliate packages if you host that project with us. If that’s interesting, go over to wp-tonic.com/partners. wp-tonic.com/partners. Let’s build something special together. Uh, um, on we go. So, and you hope this interview back on the Edward Show again. And it was with a gentleman, Zachary Lowne, and I just found it fascinating, the conversation. And I, I’m no expert. I have got a friend who’s a friend of a female close friend of mine, and he is an AI scientist. He is. He actually works in AI laboratory.

[00:36:46.480] – Dave Quaid

Nice.

[00:36:46.880] – Jonathan Denwood

And he’s totally He disagrees with some of my views. Um, but I do truly feel that I’m no expert on this, but I have spent a little bit more time than I should do trying to understand what large learning models are. And I think there is a gross for understandable reasons, because people are busy. There’s a lot going on in the world, to say the least. And there’s a lot of AI propaganda out there from the industry itself. So, but when I hear terms like intelligence, reasoning, especially reasoning, and intelligence. When I hear these words, my eyes start to roll, even though I use a lot of AI and I found it extremely useful. You know, you, you gave me a tip about using a bit of AI for my thumbnails, and I use it for my writing, and I’ve used it for, um, dictation, and I’ve just found it enormously helpful for my personal circumstance. But my eyes actually roll when I hear people using these terms. So how do you think large learning models actually work, David? And I know we could spend the next 2 hours discussing it, but if you could give us a quick outline and then that would hit off the discussion.

[00:38:30.220] – Dave Quaid

So I think large language models, um, are a good smoke and mirrors trick in the sense that they understand almost all of our fundamental concepts and how to use them. And if someone says, can I, should I use my Amex Platinum credit card to pay for, um, my household goods, it can say no, because the vast majority of times people ask this question, financial advisors or people who are interested in how you use credit cards to maximize their benefits will say no, you should use it for this or that, or just use it for its benefits and not use it to pay for things, or use it to pay for specific bills. And that’s fine and well as long as those are tried and trusted common facts, right? So should I water my cactus 7 times a day? Probably not, right? But when it comes to things that are incredibly nuanced and where like SEO, there’s a ton of competing thoughts with people who, their motivation to write content might be because they have to write 5 articles a week, or they are very enamored with a particular SEO strategy. And so they believe that they went to a site and they rewrote all the content, they put lots of EAT in it, and therefore the EAT was fantastic.

[00:39:57.000] – Dave Quaid

Now EAT is the best thing you can do for your website. An LLM can’t distinguish between these things anymore, and that’s where it becomes very complicated. And I see people talk about LLMs as if they’re sort of univocal, right? So, um, if you imagine a conversation with Claude or Perplexity or whatever, um, I’m building an SEO strategy for my SaaS company. What are the 10 things I should do? And it’ll give you a list of 10 things, and then you decide, okay, I’d like to know more about option number 3, and it gives you more details and you dig in. Each time you’re doing that, it’s getting a new set of reference material, and that is not univocal. That doesn’t necessarily hang together, and it’s not able to actually distinguish whether the content it’s consuming is accurate or if it’s just because it’s content that ranked in Google.

[00:40:50.990] – Jonathan Denwood

It has no opinion. It has no judgment call, does it?

[00:40:54.720] – Dave Quaid

No, and consensus is not a good judgment call. Like I said, consensus works fine for like, should I water my cactus? Like, what’s the return? So like, a survival instinct for lizards is if it hears a rumble in the leaves, it’s just to run, right? It would take a lot of brain power.

[00:41:11.500] – Jonathan Denwood

Well, can I just intervene? Yeah, I think one of your strengths, and I hope I’m the same, my dad was the same and I really admired him. My dad was pretty sharp business guy. Uh, um, I’m not so sharp, but I’ve got my own strengths. Um, I’m not the sharpest, but I’m not the bluntest. Um, but what I really like about you, and hopefully I’m a bit like this, is that you do have opinions, you do have views about SEO, but if you can show evidence that your view on a particular element of SEO is incorrect, I think you change it straight away if there’s enough evidence to show that it was wrong. It just depends on the evidence that’s shown, isn’t it?

[00:42:01.470] – Dave Quaid

So there’s a problem with humans where we get very attached to certain things. And I’ve seen, for example, I—

[00:42:11.690] – Jonathan Denwood

Well, you see me, you see me very attached to certain concepts, haven’t you?

[00:42:15.410] – Dave Quaid

Right, right. We’re afraid to change our brain. It’s like if I have— so we, we, we address things in patterns, right? Um, because we can actually only make so many decisions. I, I read when I was in high school studying economics, I read an interesting article that said the top executives, the top founders, are lucky to have 2 or 3 brilliant ideas in their lifetimes, which is not a lot, right? Um, and if you look at Microsoft, which is always coming from the back, which is actually their success model, right? They, they, they relate to the party with everything. Bill Gates said the internet would never take off, and then he wrote a book called The Information Superhighway. That’s been their success story, is knowing when not to be too late, um, and, and, and basically buying up everyone else’s ideas after they’ve been proven. There are a lot of people that you can show, and I always say to people, demand evidence, right? If someone publishes a case study and says, we analyze 52,000 brands, demand that evidence and have a look at it. So many people accept those case studies at face value that just because they say, look, we use this and here’s our methodology and here’s our, this n 500 and we used, they just accept it because it sounds real because the brain is lazy, right?

[00:43:27.440] – Dave Quaid

The brain’s like, I’m, this looks like a real case study. I’m going to buy into it. Then there are people that get into it ideologically, right? There, there are tribes within SEO.. And I think we have to realize that there are tribes in SEO because it, it, it really, really, um, changes how they stand and refuse to give up on certain ideological parts of SEO, right? Even though there is no evidence. So even though there’s evidence of Google saying you cannot add eat to your pages, I have met at least 10 people that will spend 26, 30 hours of arguing. Because they don’t want it to not be true. Yeah.

[00:44:08.350] – Jonathan Denwood

Well, let’s get back to large learning models, cuz, um, I, my position, I’m amazed that they work at all, that they consistently work to the level they do because fundamentally they’re using a form of mathematics and they, that is around tokenization, basically each word is grouped and it’s grouped with another set of words, which depending on the frequency that these words show up around one another. So it’s broken down into groups or patterns of words. And each word, a calculation is made and depending on where that word is still in that group will depend, or if it’s measured that it isn’t in that group, it would look outside the group and then it would join another group. And fundamentally it’s using the power of mathematics and computerization to look at each word and words tend to be in patterns, patterns, and you get an end result, which is impressive, but it is not anything to do with what we put, we, we have some theory, but it’s rather primitive. I think most people would agree what biological intelligence is. What a large learning model is doing has nothing of any subscription to do what biological intelligence is doing.

[00:46:06.190] – Jonathan Denwood

Do you think I’m on the right track there?

[00:46:09.480] – Dave Quaid

Um, I can kind of see where you’re going. What I would say is LLMs show us like how ridiculously cyclical and boring our lives are. In other words, um, we have to get water every 4 hours or we stop working. We have to breathe in and out every few seconds or we die. But so much of what we do, like, we could say like, oh, the culture in the US is so different from the culture in Japan or South Africa or London or wherever, right? But then you also realize that every day we do the same things, right? Even when we were 3,000 people on Earth to 7, 8 billion, We all live in a house, we all eat food, we all breathe, we all drink water, we all procreate, we all go to work, we all sleep at night. And every day, there isn’t actually— there’s only a small number, a very, very few small number of people who are actually changing anything. And so, we’re just repeating the same task. So, if you’re at school, you’ve got to write a report on something that’s already been done. And everything we do has already been done.

[00:47:12.210] – Dave Quaid

And that’s why LLMs are good at 90%, maybe 99% of what we need, right? So yes, you might be creating a new webinar where it’s the first time David Quaid and Jonathan and Edward ever chatted, right? It’s first time in human history. It’s not a fundament— doesn’t make it fundamental just because it’s the first time, right? But what are we gonna talk about is just a new pattern of words that already exists. And so LLMs are just whittling it down to 90 words in a paragraph, there’s only so many other outcomes for it to go, right?

[00:47:44.340] – Jonathan Denwood

I see where you’re coming from. Yeah.

[00:47:46.040] – Dave Quaid

However, so if you, for example, ask an LLM, how do you change a spark plug? It could say what car, and it’ll find out how to change spark plug. But let’s say you said, I don’t want to have a spark plug in my car anymore. I don’t want a glow plug. I don’t want a spark plug. What am I going to do now? That’s not something an LLM can answer unless someone else has posited an idea out there already. An LLM can help us with cancer research because it can take a look at other studies. And as humans, we might take it like, I don’t know if you saw the research that Jake Hundley did, and I really wanted to cover that video because I wanted to show people how difficult it is to study 39 local businesses. And so when somebody publishes a case study on LinkedIn, it says like, we studied 5,382 brands, how laughable it is, right? That all these people have all this time and power to do that. We don’t. So let’s say we wanted to cure a particular type of cancer. An LLM can go and look at different studies and break them up and cut them in different ways.

[00:48:57.060] – Dave Quaid

Thousands and thousands and thousands of times, right? Like, I think the argument you were making earlier is if you look at how, like, IBM’s Deep Blue, thinking beat chess players, was just looking at every single next outcome, right? But it’s not the same creativity we think we use, where we’re like, oh, I’m going to create a trap for this person, right? Um, we don’t— we’re not capable of processing every single move, but through blunt force, IBM is able to beat us. Yeah, but identifying a new form of cancer is way out of the reach of an LLM right now, unless it’s just using existing heuristics to circle and find. But it’s going to take a human to go out there and find the blood sample and to— or the tissue sample, or know where to look, or have the reasoning or the ambition or the drive to do that. Right.

[00:49:49.890] – Jonathan Denwood

Yeah. I think also in, in the discussion that you were having with Zachary Law, and obviously I’m talking for him, which I apologize, I’m sure you can come back to me, but I did get the strong impression and I think you did, but you, you, you will quickly correct me, is that he actually thought that large learning models go out there and do what Google does or does what Bing does or any of the leading search engines. They actually go out and index and they don’t do they, David? They rely heavily on Google or Bing. They seem to be the two that they’re relying on. I’m not aware that they’re utilizing any other search engine.

[00:50:38.560] – Dave Quaid

They are Brave Search and Eksa are two big ones, but they’re Google clones.

[00:50:43.710] – Jonathan Denwood

Yeah. So I think I’m correct in saying that the majority of large learning models are utilizing Google or Bing, but they’re not doing their own searches in it. I did get the strong impression by what he was saying that he actually thought they went out there and actually indexed. And what they seem to be doing is breaking up the text or voice query and breaking it down into a selection of, let’s say, 20 other queries, which is called a fan-out. And then they’re going out and searching Google or Bing. They’re bringing back those 20 searches and then assembling what they found into some coherent answer that seems okay when somebody reads it or it’s spoken back to them. Am I correct about that?

[00:51:44.090] – Dave Quaid

Yeah, absolutely. And I think a lot of people think that, and a lot of people also want to think that. Um, there’s a lot of people do that when they saw LLMs the first time. They’re like, okay, SEO’s dead, it’s over, right? These— because they don’t— because Google’s been so good at making what it does look simplistic. And one of the things I said on Edward’s on the chat I had with Edward yesterday is look at the technical debt that a company would have to go into to catch up with Google, which is where they’re already at. They’re building on top of Google. And if Google cut them off, which it hasn’t, but if it did and they had to go build this infrastructure, it could cost as much as $900 billion, right? Because of all the people you’d need, the systems you’d have to build, you know, the crawling systems you’d have to build. Um, all of that rep, all of that work would have to be repeated, and it would be based on someone else’s. But a lot of people just don’t like Google and PageRank, and I’ve worked for a lot of people like that.

[00:52:43.230] – Dave Quaid

And I think a lot of people think, and this comes back to EAT and content quality, is that a lot of people believe that Google can— actually, a lot of people believe it always has can look at a document and say that this document took 3 hours or 7 hours of research, it must be the best, or it cited this, or it cited that, or was written by that.

[00:53:04.800] – Jonathan Denwood

They really do, don’t they? And to some extent, I was thinking that until I started working with you, didn’t I?

[00:53:11.050] – Dave Quaid

Yeah. And, and I love Aristotle’s, uh, a lot of people also throw around this, um, phrase, um, begs the question, and they don’t understand the origin of it. Um, it’s a very common phrase, it— that begs the question. And what it means is you can’t take a document for which you’re looking for evidence and use it as evidence, right? Because that means the document is begging itself for evidence of itself. And it actually goes back to one of Aristotle’s, um, philosophical statements that the, the, the claimant for, for evidence can’t be also the evidence for it. So when you go to— let’s just take a trial, right? Um, you ring up the police and you say, look, I saw David Wade walking away from my house, and I also noticed, um, my laptop was stolen. The police will come and interview me, and I’ll say, I, I didn’t do it. But the police don’t turn around and say, well, look, Jonathan, we asked David and he said he didn’t do it, right? They’re going to have to look at evidence, witnesses, CCTV camera. In other words, my evidence alone is not— it’s just a claim, and me restating it is just a claim.

[00:54:21.740] – Dave Quaid

And some of the debates I have about Eid are just people just repeating themselves, just thinking that if they repeat it, it becomes evidence. If you want to go one step further, a lot of religious debates end up in this area. Uh, where people quote the Bible as evidence for the Bible. And that’s not something you can actually do, because if you’re saying, ‘I need evidence, I need proof for the Bible,’ you’re casting doubt on the entire thing, right? And sadly, outside of the Bible, there actually is no evidence for the Bible. But that hasn’t made a difference to us as a society, right? The Bible actually continues to run a large amount of the world.

[00:55:08.820] – Jonathan Denwood

Well, I treat most religious documents as philosophy, basically religious philosophy, and I’m okay with that.

[00:55:14.820] – Dave Quaid

But how many don’t, right? How many, like, take Opus Dei within the Catholic Church, who take it as, who are fundamentalist about it, right?

[00:55:24.260] – Jonathan Denwood

Yeah, well, I think that’s a choice. Um, have you got extra 10 minutes to do bonus content, or do you have to be off?

[00:55:31.460] – Dave Quaid

John, I have all the time for you, you know that. You’re one of my favorite people.

[00:55:34.910] – Jonathan Denwood

Oh, thank you, Dave. Keep saying that. Uh, uh, what’s the cockles of my heart, my dark heart? Well, we’re gonna, we’re gonna close the, the podcast part of the show. You’ll be able to listen to the whole podcast and the bonus content on the WP-Tonic YouTube channel, and go over there and subscribe to that. But like I say, we’re going to finish off podcast, but the show. So David, what’s the best way for people to find out what you’re up to, your latest insights into SEO?

[00:56:07.910] – Dave Quaid

LinkedIn and X are best. Yes.

[00:56:10.390] – Jonathan Denwood

Yes, use his full name and join him on X. I have, he’s very responsive. I don’t know where he finds all the time actually. I do half what he does that I feel totally out. Pressured. Um, so we’re going to wrap up the show. We’ve got some fabulous guests coming up this month, like David. And like I say, you can listen to the whole podcast on the WP-Tonic YouTube channel. We’ll see you soon, folks. Bye!

[00:56:39.830] – Dave Quaid

Hey, thanks for listening. We really do appreciate it. Why not visit the Mastermind Facebook group? And also, to keep up with the latest news, click wp-tonic.com/newsletter.

[00:56:52.030] – Jonathan Denwood

WP-Tonic & The Membership Machine Facebook Group

Why don’t you sign up and be part of the Membership Machine Show & WP-Tonic Facebook group, where you can get all the best advice and support connected to building your membership or community website on WordPress?