Eric Enge
speaker
265 appearances
5 recordings
1 series
first heard Dec 2025
last heard 12 Dec
Eric Enge’s voice in public audio — every appearance, attributed to the second.
Trend
recordings per month · last 12 monthsRecordings per month over the last 12 months — 5 in all, peaking in Dec 2025 with 5.
Appearances
So I'm going to say something that I know a lot of people out there hate is
I want to emphasize how different everything is, but there is actually a lot of commonality in the algorithms used, like a ton of commonality in the algorithms used to determine relevance of a piece of content.
And relevance is a fundamental ranking factor no matter what platform you're using.
That's a big part of the challenge.
And even the way that authority is determined between LLMs and traditional Google search isn't remarkably different.
It kind of comes down to there's a set of algorithms called embeddings, which allow us to create vectors, usually very large dimensional vectors.
I think we believe that Google uses their text embedding five model, which is 768 dimensional vector to determine relevance of a piece of content, which could be a word, a phrase, a passage, which is you could roughly think of that as being a paragraph, a whole page or a whole website.
And all of those can be reduced to the same 768 dimensional vector, and then they can be compared to each other.
And you can then take a search phrase or a prompt, and you can find content that is relevant to that based on using this vector math.
And for those who want more of the gory details, and you can research this, it uses an algorithm called cosine similarity to measure how similar one vector is to another.
And so conceptually, just to make it simple for people who aren't familiar with how this works,
just to focus on two words for a moment, the words hat and cap sound like they're extremely similar to each other, especially if you're talking about, you know, this kind of cap right here.
You know, with that in mind, the vectors for those, if we just reduce it to two-dimensional space for a bit, will be extremely close to each other.
So what that means is the cosine similarity score will be very close to one, which means nearly identical, right?
Whereas if we take a different word like refrigerator, pretty different than either of those things.
You don't wear a refrigerator on your head, for example.
In fact, you don't wear one at all.
And so this underlying math is what's being used in every platform that tries to respond to a question, be it a search query or a prompt.
And what that means, because we can do this for ourselves, Screaming Frog gives us a way to get at, you know, the
how similar a page is to a specific phrase and gives you similarity scores and other people have developed their own.
Showing 161–180 of 265 · page 9 of 14
← Previous
Next →