Not endorsing in anyway just had heard a similar concept before although it's based on different reasons...
Most diets, whether tested in a large scale study or a small one, don't work equally well (if at all) on all participants - you'll notice that more often than not the only thing reported is the average weight loss, and that's usually because at least one participant didn't lose anything (or even gained).
The Shangri-La diet actually followed from experimental results in rats and people, which Roberts published, for free, on his own website[0], with all rationale, all the references you can ask for, and a description of his personal experience. This paper became very popular, but is too detailed for most people, that a publisher approached Roberts with a request to write a simplified version for lay people, and it is this book version that became an NYTimes best seller.
"No backing studies" in this case is knee jerk reaction -- when you can just try yourself at essentially zero cost and zero risk. Personally, it worked for me when I tried it; Roberts collected anecdotes from people on his website, urging anyone who tried to report starting and then their progress; IIRC, he assumed anyone who started and didn't report progress (there weren't many) as "didn't work", and after a year or so, it had ~80% success rate.
[0] http://media.sethroberts.net/about/whatmakesfoodfattening.pd...
But if we use reputation is a proxy for trustworthiness (and we all do, for luck of time and resources to verify every since thing ourselves), then Seth Roberts has the credibility and reputation - he's prof emeritus of the Psych department at Berkeley (which is not nutrition, but his work did touch on similar issues with rats) , the paper about what inspired the diet contains dependable (by standard metrics) references, and while it is not an independent or double blind study, there are lot of independent testimonies of success and failure, and they seem to track the 80% success quite well. One of the problems with this diet is that there is no way to make money with it (except for writing a book ... which he agreed to do after giving away all the info for free), which means studies are unlikely to happen in a reasonable time frame.
The traditional understanding of metabolism-at-large, by the way, is incompatible with a lot of data; in the sense that while it reasonably describes a good percentage of the population a good percentage of the time, there are way too many repeatable counterexamples (some of which, especially the peer reviewed works of Robert Israel and Michel Cabanac, are referenced and elaborated on in the paper).
As for your suggestion of relying on uncontrolled anecdotes - there are far too many variables to rely on people saying that a. they lost weight and b. it was due to the diet specifically.
Bear in mind, I'm not saying all this as a way to declare that the diet doesn't work. I'm just pointing out all the red flags for yet another false diet. There are more red flags than most fad diets I've seen. Especially the claim that you can still eat whatever you want at whatever portion and lose weight.
> is that I have no reason to trust
> whatever this guy writes on his own
> website
a) It's simple and easy to acquire olive oil
b) The timeline on which 400kcal of olive oil consumed before a large meal causes health issues can be measured in decades