The Luxury Layer

Episode I

I would like to begin with a single-question survey:

• Do you need thinkers or doers?

The likely answer is doers.
Afterall work wont get done just by thinking, will it?

So, are thinkers really needed?

It is a luxury to afford employing group of thinkers.

Let me clarify what I am referring to as Thinkers. You may have noticed or about to notice that in this age of AI, sooner or later, the roles or titles like Principal Engineers, Specialists, Architects, and such would be or already are reframed to be

• Too narrow
• Non-executional
• Redundant

"Redundant" because, sooner or later, work gets detailed enough to match thinker's visions.

These visions are what:

• reduce future failures
• prevent scaling disasters
• encode tribal knowledge

None of the above shows up cleanly in any quarterly EBITDA making it difficult to justify these roles. These roles would remain redundant until things actually start breaking.

The promotional messages to curtail thinkers is often:

We don't need thinkers anymore, we need doers + tools

No doubt AI is good, it is very useful. AI replaces

• repetitive synthesis
• boilerplate decisioning
• low-context abstraction

But low-context abstractions is not sufficient to grasp the unsaid, the tribal knowledge. AI in it's current form needs significant help to

• have a systems intuition
• failure anticipation
• boundary judgements

I am not saying AI cant do it.
I am saying AI + a thinker would pair nicely.

The thinker roles have

• long-term systems thinking
• tacit, experience-earned knowledge

Replacing judgement with tools wont make systems safer.

We are entering an era where bad architecture will be shipped faster and paid for later.

Soon it is going to be frustrating for thinkers to justify the the luxury layer.

Earlier in the carrier justifications are part of growth and after years of grinding the justification would feel like an erasure as the value of these roles is:

• cumulative
• contextual
• earned through scars

AI is great and people have understood it's worth in

• automation (tools)
• Acceleration (productivity)
• Judgement

First two are easy to ratify and certainly real. Judgement, on the other hand, is not. Judgement needs high context systems thinking.

Overtime AI would catch-up giving rise to newer problems expanding the context to even wider horizons.

You would notice now that value of such roles is

• in the negative space (things that don't break)
• their impact is delayed or not even felt
• their work is preventive not demonstrative

Such work often goes unnoticed and it can feel dispensable.