Software Has Changed
speckx
19 points
6 comments
July 13, 2026
Related Discussions
Found 5 related stories in 56.7ms across 5,215 title embeddings via pgvector HNSW
- AI changes the economics of software rewrites cinooo · 102 pts · July 09, 2026 · 52% similar
- Old and new apps, via modern coding agents subset · 425 pts · July 12, 2026 · 51% similar
- There's no such thing as a small software team anymore mooreslaw · 59 pts · August 21, 2026 · 51% similar
- Micro-SaaS Is Dead. Service With A Software Replaces It Adrig · 51 pts · July 24, 2026 · 51% similar
- If coding has been solved, why does software keep getting worse? pchm · 674 pts · July 24, 2026 · 49% similar
Discussion Highlights (5 comments)
xg15
> If the end result is solid code that I've thoroughly reviewed and trust, does it truly matter how it was created? It does matter if a coworker throws it away next week and replaces it with a fresh plate of slop, because generating code is just so cheap now.
summarybot
I see a lot of experimenting and tinkering but I don't see a lot of products coming out that have differentiating characteristics, which makes me think people are both doing a lot more and doing a lot less at the same time. And no, it doesn't matter how the end-product is made unless you care about maintenance/maintaining it. Code in developed-environments (developed as in country) typically has an aestheticism to it. This served two purposes: beauty and comprehension. Comprehension and extensibility were tightly coupled before LLMs could "elaborate" endlessly. Now, extensibility and comprehensibility are only loosely coupled. But the ratio of people who have truly taken this to its maximum is not large. You still need good ideas and differentiating characteristics to make good applications. It might seem like it doesn't matter which language or framework you choose, now that the LLMs can endlessly "elaborate." Since comprehensibility is largely optional now. But, if you desire extensibility, making thoughtful design decisions early , such as which language and which framework, can make a noticeable difference down-the-line.
pipes
Writing structures and clarifies my thoughts, writing code did the same but for software design. Yes AI types it all faster, but I've lost the clarity and certainty of my understanding. I am finding I'm having to read the code deeply, but reading code isn't as fun or engaging as writing it.
hankbond
> does it truly matter how it was created To users of the software, no it doesn't really. Users only care about results aligned with their priorities. That said, the act of creation changes the creator, so make sure that you are actually creating! The more you lean on LLMs to generate the concepts/ideas, the less "you" are creating them, and thus the less value to you as a creator the process offers.
WorldPeas
I guess the real question now will be: as your experience stepping inside the machine wanes, will you trust yourself or the cargo cult of procedure you hope will ward off error