I heard of a long, puzzled dialogue long, long ago which started with the complaint "This regression program is not working" and ended with the question "Oh, you mean you have to put data in first?". (Younger readers should know that there were interactive (a real plus!) programs that included stuff like "Now enter your first observation:" "Do you want to enter another observation? Type Y for Yes or N for No".)
@NickCox Well, these particular question askers appear to have loads of data, but have somehow the expectation that the first neural net they try will be spectacular. Or else have managed to get as far in their research as writing ANN code without ever encountering the fact that ANNs are notoriously tricky to work with
@Sycorax It's too easy to laugh at the afflicted (perhaps: the affliction) just as people orders of magnitude smarter than I am may well have some amusement at the triviality of any scratchings I do. But twice today I saw posts on the main forum of the form "I want to write an X" with apparently no awareness whatsoever that a useful tutorial can't be given in the form of a CV answer. I just have to desist from even commenting that such questions are futile.
Changing tack slightly: I used to comment adversely on statistics texts that contained only invented example datasets (XYZ Corp, or Fly-by-night Airlines). From more than one author, I got in all seriousness the reply that real data were too complicated to be used as examples. To which I could only reply that if you think that's true don't you think that should at a minimum be explained? As in, this textbook is too simplified to be useful in preparing you for analysing real data.
