> Many terminal programs, especially older ones, are known for having confusing or unintuitive interfaces, especially if you use them sparingly and you need to do something specific that can't immediately be gleaned from search results or from the man page.
When there is such a rich database of manual pages and q/a about these tools, I tend to blame the user rather than the tool when I hear it called "too complex".
Additionally, if you don't understand what the command is doing why are you about to execute it in your terminal?
> When there is such a rich database of manual pages and q/a about these tools, I tend to blame the user rather than the tool when I hear it called "too complex".
Strong disagree. The example they gave about ffmpeg is a great example. Let's say I'm a casual ffmpeg user that wants to wrangle some videos one way or another.
I don't have the time to dig through ffmpeg's manual with tons of different options and terms that I don't understand just to figure out, as a trivial example, how to convert an mp4 to an mp3 while maintaining the best quality possible. I have 0 interest in learning about media formats, codecs, etc. I just want the result. This is not unreasonable.
With ChatGPT/Claude/etc, this is an even more trivial task. Nothing wrong with that. I'm willing to take the (minimal) risk of running an ffmpeg command while taking a common sense glance at it. It won't destroy my existing file. Or I'll run it on a copy if I'm being paranoid. I'm not dumb enough to destroy my machine or get some malware by running an unfamiliar ffmpeg command I copy pasted.
My #1 usage for LLMs is bash/zsh commands. Shell syntax is miserable to say the least.
> When there is such a rich database of manual pages and q/a about these tools, I tend to blame the user rather than the tool when I hear it called "too complex".
Extensive documentation doesn't mean the interface is good. `tar` is probably one of the most documented commands of all time, but that hasn't stopped it from being the subject of an XKCD [0].
> Additionally, if you don't understand what the command is doing why are you about to execute it in your terminal?
I can look up what the LLM's generated, or assess it from looking at it. (Comprehension is not the same as production.)
In general, I can work without it, but I'm a lot happier with it: when I need to encode a video to x264 with an acceptable bitrate while burning in the embedded subtitles, downmixing to two audio channels, and boosting audio by 20%, I can just ask that, instead of looking at 7 SO/SE/man/wiki/random blog post tabs and synthesizing it myself. I can do that. It's not a good use of my time.
When there is such a rich database of manual pages and q/a about these tools, I tend to blame the user rather than the tool when I hear it called "too complex".
Additionally, if you don't understand what the command is doing why are you about to execute it in your terminal?