DEV Community: Savas

All you need is... (r)evolution!?

Savas — Fri, 26 Jun 2026 15:23:39 +0000

This is just an opinion of what I experience and am witnessing, but looking at how LLMs scale feels like I've seen it before: with CPUs trying to outrun Moore's Law and break the rules of physics.

Heat, power leakage, and diminishing returns made it increasingly expensive to squeeze out even small gains in clock speed. The GHz race shifted because it had to.
For LLMs, more compute, more data, more parameters, and everything just keeps getting better? That curve seems to hit a ceiling and innovation needs to succeed the scaling race now.

History does not repeat itself, but it rhymes. What learnings can we make from history to "predict" a potential future?

History

In the early 2000s, CPUs ran into a wall, a very physical one ^^ So makers adapted.

Instead of crunching every single watt out of a single core, multi-cores became common. Athlon 64 x2, Pentium D, PS3 with its heavy Cell approach. From linear to parallel. From sequential to multi-threaded (and funny race conditions ;). Talks of distributed systems, SIMD/MIMD and new benchmarking spawned into what we have today.

We still use CPUs, but differently. We still have Memory, but think about Cache, RAM, GPU or Unified. Same same, but different. Innovation because of limitation.

Present

I feel something similar is about to happen to gen AI.

Yes, there are improvements in different areas, some in scaling, some optimisation, some performance, but the slope is becoming slippery. The last 12 months went from "Opus 4.5 is the pinnacle" to "What the hell is wrong with Claude?".

The perfect (business) storm of scaling execution! But the low-hanging fruits have been eaten and the crops don't grow as fast anymore. Costs rise quickly, latency becomes a constraint, and even large context windows feel more like extensions than breakthroughs. What remains is more incremental, more expensive, and more complex.

You could argue the whole venture of "agents" is the same multi-core experience repeating itself. A different kind of orchestration layer helping out. But I don't think it's the same.

In the sense described by Russell and Norvig, an agent has a notion of perception, state, decision-making, and action under uncertainty.
What we have instead are structured loops: prompt a model (often the same!), call a tool, reflect on the result, and repeat.
These systems are useful, sometimes impressively so, but they are still closer to imperative programming wrapped around a single reasoning engine than to true autonomous entities.

It feels less like a new paradigm and more like writing a scheduler for a very peculiar kind of processor.

Future?

To head into speculative territory: instead of building one increasingly large model, we might move toward systems composed of many models. Not just replicas, but differentiated components with distinct strengths, "biases", and roles. Systems that can disagree, verify each other, and converge on better answers through interaction rather than scale alone.

No 10x Claude Opus arguing with endless pre-prompted skill markdowns but still the same network.
Real domain-specific, "biased" networks for art, code, language, economy... coexisting and connected via a common world ontology, to create something like Claude Machiavelli (joking ^^).

To make this viable, (agent) orchestration alone is not enough. These systems would need a shared structure, some form of explicit or semi-explicit ontology that allows them to reason about the same world in compatible ways. There are initiatives in a similar direction and to me it starts looking more like a brain.

Repeat

Multi-core development was messy in the beginning. There is a reason "Can it run Crysis?" did not age well coming from heavy single-core into a multi-core environment. And similar might happen for agents, coding and more.

Distributed databases, multi-threading programming, catching all those free-running threads... a shift away from "just scaling" may introduce complexity before it delivers clarity.

In hindsight, everybody will have known upfront, have seen the obvious. I don't know...

It is entirely possible that this is not how things will unfold. Larger models might still have another leap. New architectures beyond transformers could reset the curve (Google, where are thou?). What we currently call agents might evolve into something much closer to actual agency.

But as "All You Need Is Attention" is approaching 10 years since an idea sparked a revolution, maybe a decade is enough for a new paradigm to emerge, peak, and repeat...

Technical Debt... It’s a Tax.

Savas — Fri, 22 May 2026 07:08:23 +0000

You heard it before: "Yeah, that's some old hack, we have to fix it later." Sounds familiar? Some old code, some "hack", some comment "Do not touch!", this "technical debt" we have to pay off some time in the future.

I would argue, what we keep calling "technical debt" is a misleading metaphor.

What we call "technical debt"

Debt is static. You borrow, you owe, you repay. You can structure it, refinance it, even ignore it for a while. The important part: the amount is known, and repayment (refactoring) reduces it.

Most "technical debt" does not behave like that. In practice, technical debt usually refers to:

Code that is hard to understand or modify
Missing tests or weak guarantees
Leaky abstractions or inconsistent patterns
Outdated dependencies or architecture decisions
Performance shortcuts that no longer hold

We label it "debt" because it slows us down later. We only pay the "minimum fee" for now. But that framing hides the real problem.

Why it’s not "debt"

Debt assumes a fixed principal. But messy code doesn’t sit still. It compounds through usage.

Every time you...

Add a feature
Fix a bug
Integrate a new system

...you interact with that code. And each interaction becomes slightly more expensive than the last. Not because the code changed, but because the context around it did.

I argue: that’s no debt but a multiplying tax!

Why "technical tax"

A better model could be: technical tax is a cost applied every time you touch a part of the system.

The worse the code, the higher the tax rate.

Clean module: low tax, fast iteration
Tangled legacy service: high tax, slow and risky changes

And crucially: You don't "pay it off" once. You pay it repeatedly!

If a piece of code is touched 100 times per month, even a small inefficiency becomes expensive. If it's critical-path code, the tax multiplies across the entire team.

What this changes

If you think in terms of debt, you optimize for payoff events:

"We’ll refactor this later"
"Let’s allocate a sprint to clean this up"

If you think in terms of tax, you may optimize for a multiplier reduction:

Lower the cost of common paths
Improve code that is frequently touched
Replace atomic entities with a parallel rewrite
Ignore rarely used areas, even if they're ugly

This leads to different decisions:

Refactor hotspots, not entire systems
Measure change frequency, not just code quality
Accept "bad" code if it's cold
Prioritize developer experience on critical paths

The (simple) example

You have a billing module everyone is afraid to touch (who hasn't ^^).

Adding a feature takes 3 days instead of 1
Every change requires manual verification
Bugs are subtle and expensive

That's not a fixed debt you can "pay off." It’s a 3x tax on every future change.

Now compare that to a messy admin tool used once a month. It might be worse code, but the tax is negligible.

What to do

Treat technical debt like a tax system:

Identify high-traffic areas (git history, ownership, deploy frequency)
Reduce friction where it matters (tests, simplification, better boundaries)
Avoid over-investing in low-impact code
Continuously shave off tax instead of planning big payoffs

Small, targeted improvements beat large refactoring projects that never land.

Calling it "debt" suggests a future problem. Calling it a "tax" makes it a present, ongoing cost. And once you see it that way, the goal is obvious: Evade the tax and become rich in code.

Where is the AI revolution at?

Savas — Thu, 02 Apr 2026 07:47:36 +0000

For several years we have been promised revolutions! Everybody can be a programmer. Apps for everything. No need for SaaS anymore. AI revolutionising our daily life in every aspect, managing Calendars, Emails, Business Trips, Holidays… our every aspect of existence.

Revolution of what?

Where is this revolution at? Not AGI, that is something completely different and just for narration purposes.

But I don’t see us solving poverty. I don’t consume energy from a Nuclear Fusion reactor. This has not been written on a Quantum Computer. And cancer is still a very dangerous and often lethal disease. So, what is AI revolutionising?

And its not just an old developer rambling, I use AI :)
I have no reservations; I use Claude nearly every day. But what are we actually “creating”? It seems AI has just sped up the same monotonous tasks. AI can auto complete, can write code, but it cannot fix humans thinking.

Evolution for what?

Wow, Claude Cowork can manage your Emails… you still use Emails?
Perplexity Computer can create your KPI dashboard? I hope you got the right data.
You got 10 agents running developing apps? For what? There are over 2 million freaking apps.
What are you building that hasn’t been built but benefits its users?

So much LinkedIn Business BS replicating itself. Instead of 5 “maybe good” health apps, we get 500 “so-so” versions. Instead of reducing meetings, we automate meetings with more AI notetakers talking to each other. Instead of hiring for what is necessary, we create AI CVs scanned by AI HRs evaluating humans on “human parameters”. As if we know what we really need…

Automation is (no) revolution

All I see is automation of monkey work. Again, don’t get me wrong: some monkey work needs to be done. Typing these letters is some kind of monkey work to pass on my thoughts (and yes, I am really typing this by hand, no AI as you can tell from the errors ;).

Automation is good! A lot of human evolution is based on automation. I don’t want to type every single test, check, code-line myself if not necessary or “just the same”. And if I could rely on the current state of AI, I wouldn’t… but I cannot!

What is real revolution?

Today’s so-called “AI” revolution is largely about scaling transformers, statistical models, big data and compute. Extraordinary at pattern recognition and reasoning analogies, yet probabilistic engines, not intelligences.

We have not built artificial cognition yet! What we call “AI” today is a sophisticated layer of prediction, not intelligence in the biological sense.

Beyond large language models we require new math, new architectures, and possibly new physics, as nuclear fusion and quantum computing. True artificial intelligence will require understanding how intelligence emerges, not just how it can be imitated. And let’s face it: imitating humans might not be the best ;)

The current wave is transformative, no doubt about it. But if it disappears today, industry will not stumble tomorrow.

The real shift will come when machines generate new concepts, hypotheses, and models of the world autonomously. That will not be “just another AI model”, not Claude 4.6 (and I like Claude ^^) or Gemini 3.1. It will be the next epoch in evolution!

Revolution in Wallets

Nevertheless, we need to face the hard truth: Software development, programmers, Juniors and more are all affected by life-changing business decisions. Layoffs on a weekly basis. Broken Windows patches on a regular basis. And it continues…

This is not a question of replacing Software Developers. It is a question of number crunching and AI just happens to be around at the right time.
It’s storytelling, and the story works. If there would have been some other invention, that would have served as excuse.

Because of this, AI is shifting the whole world into a downward spiral. AI can be good. Evolution requires power and money. Money is spent. Markets change. Growth in numbers is the only direction allowed. More money is invested. (RAM) Prices increase. Layoffs happen. People have no money for the increased costs. Jobs are a necessity, so I take any job for less than before… the spiral works against us.

If I put on my Dr. Evil hat, AI is the perfect excuse to layoff all the expensive people hired during the C-Years, as my storytelling was about everlasting growth, which did not happen, and cheap money, which was swallowed by inflation. Just correcting the typical business mistakes of bad spreadsheets with misunderstood KPIs…

What’s next?

It is not a question of if AI can or will replace Software Developers.
It is a fight of Managers vs Spreadsheets vs Software... and we are losing... for now.

The funny thing is, what any Spreadsheet wrangler does is best replaced by AI, but so far it is just dripping down to the creative process. And we haven't taken ownership enough to make it our own again and steer the good and eventual evolution.

I don't know the answer how to get out of this spiral, but the world has gone crazy, and we are the ones being affected as have others before from industrial revolution to the internet age.

Maybe it is supposed to happen. But I would like to think we just need to get together to give "AI" real meaning, purpose and control the evolution to create benefits for all, not just creating more for less for nothing...

So, what’s next for you - What or how are you changing for the better? Have you seen the revolution?