#501 – DHH: Future of Programming, AI, Agentic Engineering, Vibe Coding & Linux

• Lex and DHH were testing voice input on different jobs

  Posted by PixelDrifter on August 24, 2026 at 8:37 pm

  Lex's long spoken design sessions and DHH's short dictated messages seem like different tests of voice input. DHH describes corrections making short messages frustrating. Lex describes talking through a design while running, then processing the transcript. I'd judge those workflows separately. A mistake in a file name matters differently from a rough sentence in an early design idea.

  JovialJellyfish replied 2 hours, 57 minutes ago
• 0 Replies

• 

  JovialJellyfish

  Member

  August 24, 2026 at 9:27 pm

  For an early design discussion, I'd accept some cleanup afterward if speaking helped develop the idea. For a short message, editing a transcription error could take longer than typing the message.

• 

  fernhollow

  Member

  August 25, 2026 at 3:50 pm

  The long transcript still needs a clear final decision. Lex says he sometimes changes his mind while talking. I'd want the processing step to preserve that sequence so an abandoned idea doesn't become an instruction.

• 

  esme.edits

  Member

  August 26, 2026 at 12:26 am

  Lex also describes using codebase terms to improve transcription. That adds preparation and processing to the workflow. Comparing raw dictation with his result would leave those steps out.

Log In to Reply