The Video Assistant · Lesson 7 — The Same Step ← Course

The same step — meaning has directions.

Embeddings don't just cluster similar things. Their directions carry meaning too — so you can do arithmetic with them.

Last page, close meant similar. There is more to it: the directions in the space mean things as well. Take a fixed step in a fixed direction and you make a consistent change in meaning — and you can add and subtract those steps like numbers.

The famous demonstration was king − man + woman ≈ queen. In football the same trick reads Messi − Argentina + Portugal ≈ Ronaldo: start at “Messi”, take the step that leads from “Argentina” to “Portugal”, and you land on “Ronaldo” — Portugal's great in place of Argentina's.

Why does it land? Because an embedding is just a list of numbers. Give each a tiny two-number vector — (which country, how great a striker). The greatness number is relative: put the all-time greats at the top, 10, and lesser players lower (a very good one around 6, a steady international around 3). Subtract Argentina and the country cancels, leaving pure greatness; add a new country and you get that country's great.

Pick a country and watch the same step — Messi − Argentina + that country — land on its own all-time great.

The step from Argentina to another country is the same step from Messi to that country's great.

Add and subtract the vectors, take the nearest.

The analogy is literally vector arithmetic on the embeddings, then a nearest-neighbour lookup — the same “find the closest point” from the last page. 


      

    


    

      
Relationships are directions.

      

        Closeness told us two things mean the same. Directions tell us how two things
        differ — and because the differences are just vectors, they add and subtract. “The star of
        a country” is one such direction, and it points the same way whichever country you start
        from.
      

      

        Meaning laid out as geometry: near points are alike, and consistent relationships become
        consistent steps you can do arithmetic with. That is the real reason a search by meaning
        works so well.
      

      

        Next: cutting the tape →
        
      

    


  


  

    

      ▸ Go deeper — how literally to take this optional
      

        
The axes aren't really labelled

        

          The clean “x = country, y = greatness” split is a teaching simplification. A real
          embedding has hundreds of numbers, and no single one means “country” or
          “greatness” — those ideas are smeared across many dimensions at once. What is genuine is
          that consistent directions exist in the space, even though they don't line up with
          individual axes. The 2D picture just flattens all that so the step is visible.
        

        
It's a tendency, not a law

        

          The examples are cherry-picked. Two players are never at exactly the same greatness, so
          Messi − Argentina + Portugal lands near Ronaldo, not dead on him — and “Ronaldo” is
          simply the closest real player to that spot. That is why it is written “≈”, and why you
          take the nearest neighbour rather than an exact hit.
        

        
Word vectors vs passage vectors

        

          This tidy arithmetic is clearest with word embeddings — one vector per word, the
          classic king/queen result. The passage embeddings RAG actually uses place whole
          sentences in the space, where the property that does the real work is plain “near =
          similar”, not neat analogies. The analogy is the intuition; retrieval leans on the
          closeness.