Just to correct you but it's 0.5GPM as the minimum. 1GPM is the recommended flow rate and returns diminish at 3GPM. Anything below 0.5GPM as I have already said THEN components will heat up the loop but that just begs the question why are you running that low of a flow?Yes, it does.
I'm not sure where I'm loosing you but I'll try explain, however when you said Martin was an idiot and tried to prove him wrong referencing Jays2cents recreation of a test that was arguably the most significant discovery from Martin's and Skinnee's work, I don't think I'm talking to someone who wants to learn more than wants to just be right.
Some background on your idea that "loop order doesn't matter":
Right off the cuff, to say "loop order doesn't matter" is, at face value, completely wrong. While I understand what you are TRYING to say, LOOP ORDER MATTERS A LOT in that (thanks to Martins and Skinees work) we know loop should be ORDERED as simple as possible. Below is some reading if you want, of my recollection of Martin's and Skinee's work. I'm remembering this from over a decade ago so I could be wrong on some details but overall I think it captures the premise of what I'm trying to say:
Martin and Skinee proposed over a decade ago that flow rate was the most important part of a loop design. There was a theory before this that loops HAD to have a radiator between each heat load (CPU => Rad => GPU) which DID help temps but what Martin showed was that A: the temp benefits were not significant and at times not measurable (<1*C) and B: the resulting loop tended to be more complex than it had to be which adversely affected flow rate. He also showed that as flow rate increased, the DeltaT between any point of a loop diminished (as Jays2cents shows in his testing). His resulting assumption was that, when a loop reaches a particular flow rate (I believe it was 1gal/min) the temperature across the entire loop normalizes within 1*C. This invalidated the theory that CPU's and GPU's heated the water to a significant temperature requiring radiators to be strewn throughout a loop. HOWEVER, he did admit (IIRC) there was SOME benefit to doing this, as long as loop complexity was kept at a minimum.
His testing also showed that 1gal/min was the minimum recommended flow rate for a loop with diminishing returns as flow rate increased over 1.5 gal/min (IIRC) and showed that proper flow rate had a much bigger affect on temps than having complex tubing/radiator labyrinths. His conclusion was it was more important to make a simple loop with a maximum flow rate than to put a radiator in between every heat load in your loop. And remember, a decade ago, people were watercooling (in addition to 1-2 CPU's and up to 3-4 GPU's) VRM's, chipsets, memory, HDD's, etc... so loops got complex fast.
To summarize, having a short and simple loop that maximized flow rate was the most important part of a loop design.
Martin demonstrated in his "sandwich" radiator testing that running the heat load to the exhaust radiator prior to the intake radiator netted about a 1*C benefit vs the opposite.
When running an intake and exhaust radiator, LOOP FLOW can help radiator efficiency by exhausting a majority of the heat straight out of the case rather than straight into the case to then be recirculated through another radiator. While the joules of heat dissipated won't make a readily measurable effect on the water temperature (due to the high specific heat of water), it still results in BETTER RADIATOR EFFICIENCY when the loop is ORDERED CORRECTLY.
Now the question is, can you maximize radiator efficiency without over complicating a loop? I'm pretty sure the answer is YES. So as I said, its a small gain in efficiency but its also a small investment in effort which is a no brainer for me. LOOP ORDER MATTERS. Even if its just a small gain.