Basically, the discussion started by citing Lewis's letter to Arthur Greeves in February 1933 about the newly written story Tolkien had just given him to read. I quote the full passage (from THEY STAND TOGETHER page 449) at the bottom of page xv of MR. BAGGINS, but the relevant line comes after Lewis has described the "delightful" time he has had reading it and the "uncanny" feeling that it's just the book Lewis & Greeves would have loved to have read or written when they were growing up: "Whether it is really good (I think it is until the end) is of course another question: still more, whether it will succeed with modern children"
It's that "I think it is until the end" that's in question. What about the ending did Lewis dislike -- or at least feel was not up to what had come before?
The answer, plain and simple, is that we just don't know.
Was it the shift from light-hearted adventure-story to a more serious 'heroic' tone?
--Unlikely, given that Lewis specifically praised this Wind in the Willows to Burnt Njal shift in ESSAYS PRESENTED TO CHARLES WILLIAMS ('On Stories', page 104).
Was it Bilbo's return from great deeds in distant lands to resume a normal, mundane life?
--Unlikely, given the parallel of Ransom's adventure in OUT OF THE SILENT PLANET ending with a pint in an English pub.
Was it simply that Lewis disliked the depictions of the Shire-hobbits en masse, whether at the Bag End auction or the Long-Expected Party? -- that is, that he disliked the ending of THE HOBBIT for the same reason that he disliked the beginning of THE LORD OF THE RINGS
--This seems to be the simplest explanation, but it's merely a guess on my part; I don't think there's any proof to back it up.
I can even see Lewis's being put off by the 129 pages of typescript being followed by forty-five pages of manuscript draft: perhaps the difficulty of reading Tolkien's hand (though C.S.L. shd have been an old hand at this by that time) got in the way of his immersion in the text and interfered with Secondary Belief.
Still, the one explanation I would reject is that Lewis criticized the ending because the story he read didn't have one. I find it hard to read 'good . . . until the end' as Lewis's attempt to say 'good, but it lacks an ending' (Lewis was, after all, famous for the clarity of his prose). Wayne & Christina, in their contribution to the thread, offer an interesting thought experiment that the 1933 text broke off at the end of the typescript, which was then followed by "some form of summary conclusion now lost". That might be the case, but I'm hesitant to suggest a hypothetic text as a way out of our difficulties unless we can produce some evidence such a text once existed, especially when a literal reading of the evidence avoids the need for one.****
In the end, the theory that THE HOBBIT broke off unfinished isn't supported by any contemporary evidence. Interestingly enough, but a little too late to include in the book, I turned up evidence that Carpenter himself originally thought the book had been completed in the early thirties. In his Biographical Note to the 1976 catalogue for the Ashmolean exhibit of Tolkien's art, DRAWINGS BY TOLKIEN, Carpenter wrote
"his family, now numbering four children, had been instrumental in bringing his mythological imagination somewhat to earth and encouraging it to deal with more homely topics. For them he wrote and illustrated The Father Christmas Letters; and to them he told the story of The Hobbit, completed early in the nineteen-thirties, but not put into print until a happy chance had brought it to the notice of a London publisher some years later" [emphasis mine]
--I'm not sure what, a year later, convinced Carpenter that he'd been wrong and caused him to come up with his theory that there'd been a gap of several years between the death of Smaug and the writing of the final chapters; I wish I'd discovered this passage earlier and written to ask him about it.
So, while I think there's great ambiguity in the story of when Tolkien started THE HOBBIT -- in that the evidence is contradictory and anyone putting it all together has to reject some as mistaken in order to get a coherent picture, I don't think this is the case with the ending of the story, where I'd argue all the evidence we have does fit together and does argue for the lack of such a gap. Obviously, not everyone agrees, and I think Wayne & Christina's post does the best job I've seen of summing up the opposite case.*****
****I do suggest at various points in HoH that there might have been another version of Thror's Map that has not survived, since descriptions of the map in the text don't exactly match up to any of the actual surviving maps, but it's possible the 'missing' map never existed.
*****as for the 'hybrid typescript/manuscript', we know of several other examples among Tolkien's texts -- e.g., SMITH OF WOOTTON MAJOR.