What's the point of PhD theses if nobody reads them?

For simplicity, I am going to let your premise that (virtually) nobody reads a PhD thesis stand, even though that's debatable and also a bit field-dependent. There are still a number of reasons for it to exist:

• It's arguably more of a "writing to learn" task anyway. Students don't produce theses for the sake of the thesis, but to learn how to do research and write it up properly in a long, coherent book. Even if no single person outside the committee reads the thesis ever, it was still a good learning experience for the student.
• There is a lot of history around the concept of a doctoral program requiring developing in new thesis (in the original meaning of the word), writing it down in a book (the dissertation), and defending this new thesis against the local learned community. Even though nowadays many fields don't communicate new research ideas through long books anymore, there is still enough historical appeal to the idea that few programs want to get out of it entirely. The entire process of writing and defending the dissertation also has some appeal as a significant milestone event, which nicely demarks the end of an era for the student - he is no longer a student, but a complete member of the academic community.
• In many countries there is a legal angle to this. At least in Europe, a PhD program is usually legally defined to conclude with the production of a doctoral dissertation of some kind.
• In the age of "stapler theses" (which consist of a synopsis and a collection of previously published papers in verbatim), the entire affair is fairly low-cost anyway. My last students rarely spent longer than 2 or 3 months on the actual "thesis writing".

"No-one ever reads a PhD thesis" is a big assumption, one that I would think says more about what you think a PhD thesis is about, and less about what it actually is about.

I have read many PhD theses in my time. I find them to be a very different resource to papers. Often a PhD thesis is a very good read if you want to get a good overview of a particular field, explained in relatively simple language, and introducing and building up a complex subject relatively from scratch in easily digestible logical blocks, rather than assuming most knowledge is known to the reader.

Similarly, a PhD thesis is also more likely to go into some depth / proofs / exhaustive experiments that are typically omitted from journal publications on account of space and conciseness.

As a bonus, a PhD thesis is far more likely to be accessible to the general public than its respective papers, as the latter typically tend to be behind paywalls, whereas the former is typically accessible on-demand for free from their respective universities.

Finally, if you are lucky, your PhD thesis may form the cornerstone for an entire field, far more than a single paper might.

Having said all that, one factor that skews one's impression of how often theses tend to be read, is that it is still more likely that subsequent authors reading the thesis will cite the relevant papers generated from it instead, since they are more relevant in the context of a citation, therefore giving a wrong impression about how useful PhD theses are when one is exploring the literature in the first place.

The main skill that students should learn during their PhD and demonstrate to graduate is producing, describing, and defending scientific results. Thus, when a PhD student wishes to graduate, they have to present some work that justifies this degree. This work is, broadly speaking, the thesis. Technically speaking, this can just be a bunch of peer-reviewed papers, and cumulative theses (a.k.a. stapler theses) are more or less exactly this, but the student is usually required to write an introduction and conclusion as a framework for their papers.

However, there are many reasons why a student cannot provide these or published papers do not reflect the entire work of the PhD student in question:

• The field may have very long peer-review times, e.g., pure mathematics.
• The field’s publishing and hiring culture favours one big paper in a highly ranked journal comprising everything, e.g., biology.
• Scientific communication in the field mostly happens in form of large monographs (which a thesis can be).
• The PhD work is only a piece in a huge project and is not suitable for being published on its own.
• Some parts of the PhD work have not (yet) been published when the student wishes to graduate.

A thesis gives PhD candidates the opportunity to graduate in these cases. Note that in most of these cases, the writing work spent on the thesis is not wasted on a few readers, because the material will be reused in peer-reviewed papers later¹ or the thesis itself will actually be read by more people. So from a certain point of view, your premise is wrong: The thesis will be read by more people; it’s just not in form of the thesis itself.

This still leaves the point of why PhD students are forced to write an introduction and conclusion to a cumulative thesis, but then:

• This is not a lot of work (I did this in a few weeks).
• It may spur the student to see their work in a broader picture.
• It is the most useful information for the thesis committee. (Depending on how the committee is formed, they clichéically only read the introduction and conclusion.)
• It does train scientific writing on a level not seen in regular papers.

---

^{¹ though these are usually not read by that many people either}