Decisions are the Most Expensive Operations

Deciding is the most expensive operation

I see a parallel between computers and humans:

Deciding is an expensive operation!

This idea came to me when I understood Intel’s Pentium Pro CPU architecture, especially on how it gained performance over its previous architecture, the Pentium.

Pentium Pro used speculative execution to process both branches of a conditional block (i.e if/else) at the same time and then throw out the wrong one when the correct block came into play.

The CPU designers wants to keep the CPU working instead of waiting - or other words, deciding.

Human Parallel

The similar thing happens in humans - when you just have to execute on work, it’s easy, follow the steps and do them.

When there’s a decision required in the steps, you stop and evaluate.

Unlike a CPU, you can’t do both at once - yet. You do one or the other.

Analysis Paralysis

The problem is not the deciding part, the problem is the part where deciding takes all the resources, such as information gathering - to make the best decision.

CPUs on the other hand have limited scope and don’t have analysis paralysis - CPU designers can add more resources to execute all the branches and choose the best result.

Oh, if it was that easy for humans. 😅

This is why I believe, deciding is the most expensive operation, especially for humans.

Corporate Parallel?

In a way, maybe this is exactly what is happening at large companies. Where the path forward is not clear so instead of researching the best path - execute all possible paths and decide at the end.

The corporation is a parallel of a CPU’s speculative execution?

🤔

Deciding

Deciding is one of the most expensive aspects of life.

In a CPU, designers overcame this by adding more resources to compute all different paths of a decision, speculative execution.

For people, deciding slows them down, at worst, analysis paralysis happens and become stuck in the decision.

At a higher level, are some corporations emulating the speculative execution design in CPUs by just having all the resources to try all paths at once?

For now, my lesson is to know that deciding is expensive and find ways to make better decisions, that is a whole new topic to ponder.