Good point. I think it’s no good for continuous reading if the leading is too wide when text is set in a typeface with low x-height and long descenders and ascenders. One sees this very often in publishing where omnipresent Garamond is used. This causes more pages and the book thicker, which may be the reason behind it.
If the lines are falling apart something is wrong. But this is to be judged by a good typographer: a type with the classical proportions of a Renaissance Antiqua needs less leading, others with larger x-heights and shorter descenders and ascenders a bit more.

Normally I would print and read to judge these things. It’s hard to say anything about reading with a crudely rendered jpg that isn’t necessarily 1:1.

I have to agree with Frode that it's pretty hard to tell from this resolution. Also, I think the genre matters a lot in this respect: for instance, I'd hate to read philosophy set with such tight leading but it might be okay for fiction.

Few loose observations:

Left page's leading seems to be intended for more "humanistic" content then the other. (Anyway: it's like Rorschach's Test, isn't it?).

But, sincerelly, I don't think the right page has better colour and texture.

Both pages are in English, which gives an even colour. My language (Polish), containing high percentage of diagonals: a, k, s, w, y, z, would create a messy surface with such tight leading.

Once I mechanically increased contrast of a tested typeface (of a medial contrast) which seemed to need some larger leading. The vivisection returned distinct boumas which resulted in faster reading and allowed to reduce the leading. So I suppose, also the inner wordspace is an important factor.

I also think (as a habitual reader), that the even colour is not always the best and desirable solution.
[Footnotes of complicated content (so using small caps, italics and numbers alltogether) set in the best typographic manner, are extremely heavy to read].
In the plain text, elements like en-dash help during scanning a page or searching a certain place in a book. But they are visible only at larger leading.
Polish typesetting practically abandoned the em-dash in recent years, what results in better colour, but dialogues in fiction seem to be spoken with the machinegun speed, since there is no pause in the air; they melt together. So I see a paradox here.

Well, there is a lot of dogma about. The other argument for generous leading is that it minimizes the variance in word spacing you'll get with justified csettings.

I think what is really going on is that *readability* is a matter of type size, measure, leading, and word spacing. Change one, and the optimum value of the others may change.

BTW, designers often give different answers than readers -- yes, I've done some limited testing, using editors as readers, versus designers.

Finally, all these are constrained by the trim. Times past, paper came is certain sheet sizes, and people wanted to be efficient; not pay for a larger sheet & then trim away what wasn't wanted.

Anyway, sheet size affects margins, which in turn affect measure, which affects leading, type size, etc.

For a generality though, yes, I think you are right, current leading practices are a designer affectation from looking at shapes rather than words.

It is a shared opinion in book typography that a generous leading helps the reading process;

Yes, this is the present myth.
But not too long ago, type specimens showed text type both set solid, and with extra leading.
The 20% leading default is a norm established in the early days of DTP, along with fake small caps, and faux bold and italic, and deserves similar opprobrium. It is a shameful waste of paper.

But not too long ago, type specimens showed text type both set solid, and with extra leading.

This was certainly the case with the ATF 1923 specimen book; and, perhaps unremarkably, smaller sizes benefit from extra leading.

The 20% leading default is a norm established in the early days of DTP

A natural offshoot of substituting default measurements for acquired skill...

There is a very interesting setting of dark and light typefaces with different leadings in Mitchel and Wightman's Book Typography. (p. 185-6) It shows that with the dark Quadraat, it benefits from generous leading, whereas the light Spectrum benefits from tighter leading. They conclude,

"If you decide to use a dark face it wll benefit from increased leading to prevent the overall colour from becoming too black. Equally, if you decide to use a very light face, reducing the leading will prevent the text from becoming a pale grey."

Overall lightness and darkness is not the only factor. The impact of leading also varies with the length of the extenders, and also how the weight is distributed in the face.

One of the things I was fascinated with in Linotype Caslon Old Face (a metal face only) is that it looks dark, but could be set solid or almost solid and still look good. I tried to achieve this in my revival Williams Caslon Text, also. That is, I tried to "square the circle" by looking both dark and open, so that it could take less leading yet not look too light.

As the influences are so complex, the only thing to do is to try out different typefaces with different measures, sizes and leading, and see what looks inviting and readable to you—as well as considering what the economic demands are.

Personally, I like the more lattice-like look of the page with less leading than is fashionable now, but it has to be the right typeface design that can take it. Adobe Garamond, with its long extenders, doesn't need as much leading as, eg., Quadraat, I would think, though I haven't actually tested them against each other.

A nice short read:

Tinker, M. & Patterson, D., (1949). Speed of reading nine point type in relation
to line width and leading. Journal of Applied Psychology, 33(1), 81-82.

"Examination of Table 1 shows that the region of optimal legibility
ranges between a 14 pica line width with 1 point leading or more to a line
width of about 30 picas with 1 point leading or more."

There are many other scientific studies and traditional claims that are incongruent with one another. In my opinion, further research is required to draw ecologically valid conclusions.