A session by Anirudh Shrinivas · 08 Feb 2025

Work runs on a layer you can't see.

Work is made of tasks. Tasks are made of actions. Actions run on one validation pattern underneath. Read the stack down and it takes work apart; read it up and it shows you what's steering.

§ 01The stack

Four layers, one in charge.

Each layer is made of the one beneath it. The grids that follow work the top three. The fourth — validation — is the one quietly setting the coordinates on the rest.

01Workthe whole — a job, a role, a placeread
02Taska unit of work — what lands on your weekread
03Actiona unit of task — the moves you actually makeread
04Validationwhy you act at all — the generator underneathoperate

Read down to take it apart · read up to find what's steering

§ 02The grids

Three grids you read.

Each layer gets an oversimplified 2×2 — the title already warned you they're too clean. Hover any square for the read. The fourth grid you don't read; you operate it, below.

01 · Workknowns × unknowns

Sizing up a job, some things are visible up front; others only surface once you're inside.

You choose on the top-right, then live in the bottom-left.

Visible going in ↑
Holds up in reality →
Job description · the peopleTold, not stableYou're handed a description and you meet a few people — but the role drifts from the description and people leave.
Salary · locationKnown, and it staysStated up front and unlikely to change. The one corner you can actually bank on.
Culture · power distanceThe real unknownsHow decisions really get made, and who you can push back on. Unreadable from outside — and it runs your day.
Growth · positionKnowable if you lookWhere the role leads and how you're seen. You rarely ask up front, but the org's track record will tell you.
02 · Taskurgent × important

Inside a job, work arrives as tasks. Two clocks sort them: how soon, and how much it matters.

The only mark on the grid is delete — the discipline is cutting.

Urgent ↑
Important →
Urgent, not importantThe interruptionLoud and time-boxed, but it moves nothing that matters. Hand it off or contain it.
Urgent, importantDo itBoth clocks are ticking. This is the do-it-now square — no debate.
Not urgent, not importantCut itNo clock, no payoff. The only square the grid tells you to delete outright.
Not urgent, importantSchedule itThe work that actually compounds — and the first thing to slip, because nothing forces it.
03 · Actionimpact × effort

Each task breaks into actions. You choose between them on what they return against what they cost.

Looks like work, returns nothing — the bottom-right is where time goes.

Impact ↑
Effort →
High impact, low effortDo firstCheap and consequential. Start here, every time.
High impact, high effortThe deliberate betWorth doing — but only on purpose, not by drifting into it.
Low impact, low effortFillerHarmless between real work, dangerous as a default mode.
Low impact, high effortThe drainCosts a lot, returns little. The square that quietly eats your week.

The fourth grid is different in kind. The first three describe what you do; validation is the reason you do it — so you don't read it, you operate it. ↓

§ 03The generator

The grid that reads you.

Under every action is why you took it — how much you initiate, how much you react. Choose where you sit, and watch what it bends up the stack.

You initiate ↑
You react →
A metronome has two parts —
the poles it swings between,
and the tempo it has to keep.
Validation pattern
Moves when moved

Winds you

Stalls you

How it fizzles

What it bends, up the stack
Task grid
Interrupt
Do
Cut
Schedule
Action grid
Quick win
Big bet
Filler
Drain

Feedback tempo you need
Slow cadence is fineNeeds a fast beat
§ 04Three faces

One failure, three faces.

Leave validation untouched and it surfaces three ways. Each time you work the layer you can see while a different one runs the room.

Enthusiasm fizzles inside a year.

The layer you worked

You picked the job on what you could see — salary, location, the description.

The layer that steered

Culture and power distance — the parts you couldn't assess — run the day and grind it down.
The review leaves it on the table.

The layer you worked

The conversation measured output and goals, line by line.

The layer that steered

The validation pattern was never named, so the review couldn't reach what's actually moving you.
The next search disappoints too.

The layer you worked

You re-sorted the top of the stack — new title, new company, fresh start.

The layer that steered

The same metronome came along untouched. The new one winds down on the old schedule.

Same pattern. Three rooms. The fix isn't a new room.

You keep fixing the layer you can see. The one you can't is steering.

The session

Pattern Recognition; Some Oversimplified Shit — run by Anirudh Shrinivas, 8 Feb 2025. Each grid was built live, through short discovery, with Sanya, Sneha Prabhu, Harsh, Ganesh Shenoy, Shivani Singh — and Cyril, Sai, Divya, Akshita. Thank you all.