Fundamentals Versus Hype: Devs Split on LLMs' Role in Modern Software Engineering

AI & TechnologyApr 18, 2026score 1.143 posts · 0 replies across 3 instances

The discourse centers on whether the industry should prioritize rushing LLM adoption or fixing core engineering deficiencies. Several high-profile developers argue that foundational practices—like rigorous version control, CI/CD pipelines, and comprehensive test suites—are critically underdeveloped in current real-world codebases.

The conflict divides developers into two camps. One faction, led by @baillehache_pascal, demands a return to solid fundamentals, viewing AI adoption as a distraction from poor practices. Opposing this, @queenofnewyork argues the industry has devolved into mere 'micromanaging LLMs,' losing sight of efficient problem-solving. An outlier take from @[email protected] suggests that criticizing LLMs as producing only 'bullshit' versus 'median-quality' code are not mutually exclusive statements, pointing to necessary nuance.

The consensus is that LLMs are not a simple silver bullet. The industry faces a choice: either integrate AI tools while neglecting decades of proven engineering discipline, or refocus intensely on fixing fundamental weaknesses before treating AI hype as a primary development objective.

Key points

SUPPORT

Developers must master foundational practices (VCS, testing, CI/CD) instead of solely chasing LLM adoption.

Strong advocacy from @baillehache_pascal for perfecting established methods over LLM hype.

OPPOSE

The current focus is shifting away from efficient problem-solving toward merely managing LLMs.

@queenofnewyork claims #Programming is bogged down in 'micromanaging LLMs.'

MIXED

LLM output quality is complex; calling it only 'bullshit' or only 'median' misses key details.

@[email protected] noted that the two extreme characterizations are not mutually exclusive statements.

SUPPORT

Proven methodologies like small batches and continuous feedback remain essential regardless of AI tooling.

Repetition from @baillehache_pascal emphasizes the continued critical importance of established cycles.

Source posts

@[email protected]

the statements "LLMs are only good for producing tedious and borderline-worthless bullshit" and "LLMs can produce code of vaguely equivalent quality to the current software industry median" are not necessarily mutually exclusive

110 boosts · 194 favs · 5 replies · Apr 17, 2026

@[email protected]

#Programming has gone from solving problems in efficient and sustainable ways to micromanaging LLMs. And I hate it.

2 boosts · 0 favs · 0 replies · Apr 17, 2026

#programming

@[email protected]

"Let’s talk about LLMs" 

"
As for what you should be doing instead of rushing to adopt LLM coding out of fear that you’ll be left behind: I think you should be listening to what all those whitepapers and reports and studies are actually telling you, and working on fundamentals. You should be adopting and perfecting solid foundational software development practices like version control, comprehensive test suites, continuous integration, meaningful documentation, fast feedback cycles, iterative development, focus on users, small batches of work… things that have been known and proven for decades, but are still far too rare in actual real-world software shops.
"

#programming 
 
https://www.b-list.org/weblog/2026/apr/09/llms/

0 boosts · 0 favs · 0 replies · Apr 17, 2026

#programming