There are several issues with the rule:
1. The submission might be a Web page on a blog or forum, where the title will surely miss important context from the perspective of HN, as opposed to accessing the page from its native forum, blog, or the like, and if one isn't a regular on the forum or blog.
2. Sites like Wikipedia often have article titles which give absolutely no indication as to why the topic might be interesting. To make matters worse, the interesting stuff is sometimes buried deep in the article. Other than Wikipedia, some pages or PDFs on legal or "bureaucratic" topics can have inscrutable titles and be very long. In such cases it seems like allowing a short note from the submitter, like "page 97 is interesting", might be useful.
3. HN has particular interests, and whoever writes the article and/or title probably didn't write it with HN in mind.
4. Many authors simply devote little to no time to choosing a good title.
These issues have been discussed previously, examples:
https://news.ycombinator.com/item?id=5142851
https://news.ycombinator.com/item?id=6572466
https://news.ycombinator.com/item?id=1169151
https://news.ycombinator.com/item?id=8608162
Consider this change to HN: when submitting a title and a URL, it should be possible to give an additional short one-line text description of the submission. Afterwards the description could be displayed next to the title and domain, and given less weight visually than the title. The latter could be accomplished by displaying it faded and with smaller letter size.
The most recent relevant (although mild) example I happened upon is this short page that aims to give intuition about kernel density estimation (KDE, basically a better histogram): https://mathisonian.github.io/kde/