[baak]

Read the claims he's making on his blog and tell me they're not sensationalist. We're expected to believe he's become some sort of a super human because of a 900 calorie liquid synthetic diet that he invented with no formal training what-so-ever.

http://robrhinehart.com/?p=298

Under Results -> Qualitative.

And yes, reading is effective learning. I'll put it in perspective though: A biologist who dabbles in comp sci during his free time tells you he's invented a new sorting algorithm. This algorithm performs in O(1) in best, worst, and average case, shattering any proofs to the contrary. Here's the catch though: he's the only one using the sort. Every biologist who knows anything about computer scientist posts: I want to try this! But they can't because he is only using it for himself for now. Now... the biologists on their forums are looking at this and going, 'Wow! He's revolutionized Computer Science!'

Sounds legit, though.

[Jtsummers]

  But on day 3 I noticed my heart was racing and my energy 
  level was suddenly dropping. Hemoglobin! I think, my 
  heart is having trouble getting enough oxygen to all my 
  organs. I check my formula and realize iron is 
  completely absent. 

That part strikes me as odd. I'm prone to iron deficient anemia, but a few fasting days doesn't cause any major issues for me. Something else was going on that day, or he has some other health issue. But yes, very sensationalized story telling.

[epmatsw]

The difference being that basic computer science algorithms are very very well studied and understood, while human nutrition is not at all. In your example, computer scientists would argue against the conclusion based on knowledge, while biologists are arguing against this based on not knowing what will happen and assuming (likely correctly) that the results will be harmful. It's a far less powerful position to be in.