Flo<p><span class="h-card" translate="no"><a href="https://fosstodon.org/@webaware" class="u-url mention" rel="nofollow noopener noreferrer" target="_blank">@<span>webaware</span></a></span> <span class="h-card" translate="no"><a href="https://mastodon.social/@eschaton" class="u-url mention" rel="nofollow noopener noreferrer" target="_blank">@<span>eschaton</span></a></span> it seems like the trend of <a href="https://infosec.exchange/tags/nocode" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>nocode</span></a> programming gets reinvented time and time again, <a href="https://infosec.exchange/tags/Zapier" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>Zapier</span></a>, <a href="https://infosec.exchange/tags/makecom" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>makecom</span></a>, <a href="https://infosec.exchange/tags/n8n" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>n8n</span></a> etc. all have the same limitations with regards to irreducible complexity and maintainability/versioning that are mentioned in the article.</p><p>I disagree with the last paragraph in that <a href="https://infosec.exchange/tags/LLMs" class="mention hashtag" rel="nofollow noopener noreferrer" target="_blank">#<span>LLMs</span></a> do actually bring different new aspects to the table. They can for example break down a high level task into several implementation steps that need to happen and write small modules for you. You still need someone who can understand the system architecture, database schemas etc., but that person can now focus more on system architecture decisions. </p><p>I like the analogy of having infinite interns that each have memorised all of Stack Overflow. They can definitely write code, but you still need to specify in detail what you need and check all their work for errors. But still, that is more than just a fancy autocomplete in my opinion.</p>