DEV Community: Paweł Nejczew

The Foundations and Bricks of the Web World

Paweł Nejczew — Fri, 17 Jul 2026 14:33:25 +0000

Technology popularity and longevity don’t always walk hand in hand.

We often look for successors to legacy technologies, hoping that newer solutions will automatically displace those from the 1970s. Meanwhile, the C language, created by Dennis Ritchie, remains the ultimate foundational system software after more than 50 years, powering almost every server on Earth. Even a strong contender like Rust hasn’t managed to dethrone it.

Right about now, some smart-aleck will ask: “If C is so great, why don’t we use it for everything?”. Fair point, but the answer is simple-because we don’t need to. C is the lingua franca for hardware communication and is perfectly suited for managing the deepest basements of the operating system. Over the years, languages and tools were created to abstract and speed up coding for specific purposes and platforms-yet C remained the foundation they all run on.

Web backend development can be done in many technologies, and their undisputed leader is PHP. Although its funeral has been announced multiple times, it’s doing great-powering 7 out of 10 websites today, with the other three most popular being Ruby, JavaScript, and Java. Does that list sound like a rehashed menu from a decade ago? Exactly-nobody rewrites entire applications on a whim or switches technologies every few months, or even years!

Current popularity statistics are in no way proportional to the total volume of software actually running on a given technology. PHP hasn’t been the number one choice for younger developers for a while now and stands a good chance of eventually being replaced by another language-maybe JavaScript (on the backend), maybe Golang, or maybe something else entirely. Time will tell, as each alternative offers a different edge over the previous solution.

When it comes to the user interface layer, JavaScript has held the crown for years. A few frameworks currently dominate the market (React, Vue, and Angular), yet they are nowhere near the scale of their older brother, jQuery. It is estimated that jQuery still runs on 60-70% of active websites, largely driven by WordPress, which powers over 40% of the web.

As we can see once again, popularity and longevity don’t always walk hand in hand forever. This allows us to predict that more elements-differing from their predecessors in shape and functionality-will eventually join the foundations and bricks that build today’s internet. Therefore, it’s worth investing not just in currently trending technologies, but primarily in the most antifragile programming competencies that will continuously stand the test of time.

Unfamiliar with the Lindy Effect?
Check out this article for some food for thought:
Article: The Lindy Effect in IT
Medium: The Lindy Effect in IT

Does S.O.L.I.D. Still Make Sense in the Age of LLMs?

Paweł Nejczew — Fri, 10 Jul 2026 12:17:49 +0000

I’m hearing more and more theories suggesting that since AI can generate endless amounts of code, architecture and readability no longer matter. “If AI wrote it, AI can fix it later.” I’m not sure if a simple eye-roll and a heavy sigh are enough anymore, or if it’s time to start swearing out loud. My experience tells me the exact opposite: as code production skyrockets, maintainability becomes a critical success factor for any evolving software project.

We use acronyms like S.O.L.I.D. to keep specific rules in mind and write software that other developers can easily maintain. In the past, every redundant or inefficient piece of code carried a cost measured in comprehension time, bug potential, and sometimes the “WTFs per minute” of those maintaining the app after us. Today, we also have to factor in token bills (AI compute costs), because low-quality code inflates the context sent to the model or leads to redundant operations. Sometimes you have to perform mental gymnastics just to get the model to grasp the original intent and connect the dots-and all of that costs money, which adds up fast.

If we assume that models will play an increasing role in development, boosting the lines of code produced per engineer, then “clean code” is more crucial today than ever. Feeding poor code into these models will only generate more poor code, forcing us to spend more time on manual hotfixes or wrestling with failed AI generation loops.

The fundamentals of clean code design remain one of the most vital components of software engineering knowledge. While I wouldn’t limit it strictly to SOLID, it will always serve as a solid foundation for professionals writing or generating code in object-oriented applications. Within this acronym lie ironclad principles that have survived for decades, helping build countless stable and readable codebases.

Fullstack a.k.a. The One-Man Band

Paweł Nejczew — Sat, 04 Jul 2026 11:36:51 +0000

Are generalist skills a new trend or a real necessity?

Over the past two decades, IT departments have been split into narrow specializations. In the case of web development, this meant frontend and backend. To put it very simply: the part of the application you can see, and the part you can only feel because it runs under the hood. Despite this, we can still find rebels who, defying a polarized IT industry, learned to build systems on both fronts (or rather, "ends").

By working on both user interfaces and server-side systems, these developers gained the experience to build solutions spanning from the database all the way to the colorful buttons on a webpage-essentially becoming "One-Man Bands." The market initially dubbed them Fullstacks, only to curse them shortly after. The worlds of front and backend evolved so rapidly that keeping up and remaining a top-tier expert in both would require an elastic 24-hour day and a lake of coffee. Instead, the community doubled down on narrow specializations and expert knowledge.

Today, the evolution of languages and frameworks is slowing down, and ideological debates over architecture are losing momentum. Due to-or thanks to-the AI boom, we are witnessing the blurring of lines between specializations, and we are once again valuing generalist skills that allow us to operate at every stage of software delivery. Since the code itself can now be written by artificial intelligence, a systemic view of the product, rather than just a slice of it, becomes the key to success.

Questions widely discussed today include: When will the lines blur between those who define what needs to be built and those who actually write the software? Are coding skills still necessary (and at what level)? Are the so-called Product Builders just the new Fullstacks, or are they the "future developers"?

Currently, IT professionals are either deepening their core programming skills or chasing frequent tool changes that should support their work rather than completely replace it.
The way we use language models is shifting so fast that knowledge from just 2–3 months ago is often frustratingly outdated. Many shakeups still lie ahead, and one can't help but wonder: what will happen when the market's AI bubble bursts? Will the tools offered today (LLMs) remain as efficient and financially accessible?

Will Product Builders, whose primary skill is leveraging AI, be able to handle the constraints of advanced tool providers? Will a limited understanding of the programming paradigm itself be enough when working with less capable local models? Or will the market split permanently into generalists and specialists?

Wersja w języku polskim: https://nejczew.com/pl/blog/fullstack-aka-czlowiek-orkiestra/