[11:27 pm, 02/08/2026] Atharv Ketkar: i am starting to understand it’s not really about the product or the content. it’s really just about the time and the positioning. If you are going for a active enough TAM and your positioning is recognizable to them, even with shoddy attempts you will get good enough results. Its like the pareto inputs are TAM and positioning. Your video might not have the best hook or the best body, your product might not have the best features or whatever but still you will do fine. Its like favorable winds with a mediocre crew vs. completely opposing winds and a really skilled crew.

be an operator of sorts who deals with tam, positioning and the culture ops required to make it work and you will be super f*cking successful bro this is the meta

[11:29 pm, 02/08/2026] Atharv Ketkar: one type of TAM operation is that the market might have set filters of like two-three sets and marketing to a subset of it. If you remove one of the sets as conditions you have a wider area available as tam and you can position for it so fast and all of a sudden you have greater TAM. Basically this is expanding the TAM by simplification. Or maybe you can just apply another filter and get it by anyone because this subset is outside the subset they are targeting. Another can be just going for a TAM so big that doing anything just greases your hand

A competitor might position themselves as:

AI ∩ Engineers ∩ Python users ∩ LLM developers

That's a very specific intersection.

You could instead remove a filter:

AI ∩ Curious people

Immediately the addressable audience becomes much larger. You haven't changed the product much. You've changed the geometry of the audience.

Instead of:

AI ∩ Software engineers

go for:

AI ∩ Students

These aren't subsets of one another. They're different slices through the same larger market. If one slice is overcrowded or expensive to reach, another may be much easier.

There's also what I'd call filter translation.

Take a niche idea and express it using a filter that millions already identify with.

Instead of:

Positional encoding.

You say:

How does ChatGPT know the order of words?

The underlying topic is identical. The entry filter changed from "knows transformer terminology" to "has used ChatGPT."

You haven't made the idea simpler. You've made the entrance wider.

I think there are several distinct TAM operations emerging here:

Filter removal. Eliminate unnecessary prerequisites so more people qualify.
Filter substitution. Target a different identity or community using the same core product.
Filter translation. Describe the same concept using language familiar to a broader audience.
Filter union. Position something so multiple communities recognize it as relevant. QOVES is good at this: celebrities ∪ countries ∪ dating ∪ psychology all point toward facial aesthetics.
Trend attachment. Connect your work to an already-growing conversation so you inherit demand instead of creating it from scratch.
Ceiling expansion. Choose a category whose natural population is orders of magnitude larger before optimizing anything else.