Disclaimer, I work at MDPI and have been here for a little less than a month as of time of writing. I also have had extensive experience in academic publishing, albeit for other publishers.
The question of whether reviews at MDPI are worse than at other publishers is difficult to answer, because to write a full answer one needs access to every journal's & every article's reviews. I don't have that kind of birds-eye view of the operations. Individual editors see only a subset of their journal's articles. More senior editors can see all the journal's articles, but there are a lot of them, so they are not likely to check every article's reviews. I'll still write an answer based on the reviews I've seen firsthand. The following numbers are for the journals Universe and Galaxies. Because different editors do things differently, what I see might not apply to the journals you see.
For the reviews I have seen firsthand (most of these papers are still under review, hence the relatively low number of reviews per paper):
- Number of papers: 11
- Number of reviews: 15 (all written by different reviewers)
- Number of reviews that include detailed comments: 10
- Number of reviews that do not include detailed comments: 5
Definition of "detailed" I'm using is if the review is >2 paragraphs in length, after excluding requests to cite X paper or simple English fixes. This definition is not ideal, but it's simple, and does not require subjective judgment.
Several of the reviews with detailed comments are extremely detailed reviews, spanning several pages, with two reviewers running their own calculations and including those calculations in the review (!). Among the reviews that do not include detailed comments, there was one review that basically asked for some citations without saying much else. The other reviewers still clearly read the paper, but didn't have much to say.
Compared to the reviews I saw at other publishers, broadly speaking, these reviews are excellent.
However, I have several comments about these numbers:
Of the 255 papers that underwent the entire peer review process to acceptance or rejection, about 60% of the final decisions occurred with no sign of actual peer review. ... Only 36 submissions generated review comments recognizing any of the paper's scientific problems.
So about 40% of 255 papers (=102) were peer reviewed, of those 36 of 102 (~35%) noticed the paper's scientific problems. In other words, about 65% of peer reviews did not notice the issues. Even if some of these reviews are themselves faked, the numbers are large enough that it seems likely that some reviewers simply didn't notice or care enough about the problems. This would mesh closely with my own experience. I clearly remember one Distinguished Professor at a major R1 university in the US writing a review that said nothing except "I recommend revision, and you should cite the following papers all written by me", so seniority & publishing record doesn't preclude a reviewer from writing poor reviews.
- It should be obvious that if the editor reaches a reviewer who is very familiar with the topic of the paper, then it's more probable that the review will be good. Conversely, if the reviewer is not really familiar with the topic, they more likely to either write a poor review, unless they "get familiar" with the topic (which is a time-consuming process).
Since the MDPI standard operating procedure is to have MDPI staff identifying and inviting reviewers, and since the processes at MDPI are so accelerated (we request reviews in <2 weeks), I think it is relatively likely that MDPI reaches reviewers who aren't the best-qualified to review. In theory, these reviewers should decline to review the article, but some number surely accept anyway and then write a poor review. That could explain what you're seeing. If this is indeed the explanation, it'd indicate the editor(s) you interacted with aren't very good (whose name is on the emails inviting you to review?).
- As I wrote above, compared to the reviews I saw at other publishers, these reviews are excellent. However, this could also be explained by me becoming a better editor (which definitely happened - Universe and Galaxies are closer to my expertise, and I have been putting in a lot more effort than I used to when selecting reviewers).
For the record of the 15 reviews above, 3 of them were by reviewers not invited by me. Of those three reviews, one is very detailed (this review is one of the two where the reviewers did their own calculations); the other two are not.
- MDPI allows reviewers to volunteer. One of these papers had three volunteer reviewers, but all three were plainly unsuited to reviewing it (no astronomy expertise), so they would likely have written poor reviews too. I don't know why they volunteered. I declined all of them, but another editor might not have, which would probably lead to what you saw.
Finally: for the question "should you bother?" I think this linked question is highly relevant. Ultimately it's up to you, and you should not feel pressured to review. There is one thing that I do wonder about though: if you are the only reviewer for a manuscript, would you feel troubled? If yes, then that seems problematic for a lot of journals (I've written some papers that received had only one reviewer myself). If not, then what the other reviewers do doesn't seem like a major concern, and certainly a less important one than whether the authors take your comments seriously.