Rambling about tech at 4am..#

It seems our civilization—at least in the context of well-developed countries—has reached a significant milestone. Just 43 years ago, IBM released the first personal computer powered by an Intel 8088, boasting a remarkable 16KB of RAM and an impressive 160KB of storage.

Now, modern computers are so powerful that we can create complex constructs in a Minecraft world that outperform the original systems. It’s incredible to think about our ability to create cyber constructs that can deceive our own eyes. This very concept—that I, as an individual, can create and modify a (cyber) world akin to a deity—sparked my interest in IT. While I’m not comparing myself to George Soros, I mean that I have the capacity to create something from nothing.

Software has the unique ability to generate value and commodities from mere 0s and 1s (although this might change with the advent of quantum computing). A prime example of this is Bitcoin, where demand determines worth. Imagine traveling back to the 1800s and claiming that a combination of letters and numbers with no inherent meaning could buy land and food. I can only picture the bewildered looks from the people of that era, and perhaps their skepticism would be entirely justified.

We seem to be entering a dystopian era where worth is dictated by constructs that lack intrinsic value. Looking at society more broadly, during the Neolithic Age, we organized hunters to create safer environments and worked in harmony to achieve more efficient results. Now, having reached the pinnacle of the food chain in the animal kingdom and automated many processes, we face the challenge of creating services that don’t merely sustain us but also fulfill societal roles.

We have evolved from hunting to farming, to machinery, and now to automation. With each major societal milestone, it appears we are abstracting further. For those in software engineering, we recognize abstraction as a powerful tool. For instance, creating even a simple application in C++ is significantly abstracted from assembly and machine code.

However, over-abstraction presents its own challenges. Programmers understand that excessive abstraction can lead to confusion within a large codebase and may pose long-term issues. As we layer abstraction upon abstraction, we risk losing the original meaning behind our creations.

This phenomenon becomes evident when attempting to analyze decompiled binaries using binary decompilers, which serve as a step toward de-abstracting machine code.

I find that AI, particularly in its unsupervised learning forms, mirrors this trend in our current society—heading toward a realm of over-abstraction. In summary, we may be straying into a territory where abstraction becomes excessive.