The framing I kept seeing was “try ashwagandha” or “try L-theanine for anxiety.” I ran both for 8+ weeks. Nothing moved. The problem wasn’t the compounds, it was running them without knowing which pathway I was actually working with. After mapping my baseline I landed on glutamate dominance with low plasma glycine as the likely driver. Then I ran a direct comparison over two separate 11-week blocks. NAC (600mg 2x daily): HRV up ~4 points on average, sleep architecture unchanged. Bloodwork showed modest glutathione improvement. The issue is NAC hits cysteine status upstream, not the NMDA co-agonist site, so for a glutamate-dominant profile it’s real but not the bottleneck. Glycine (3g before bed): Deep sleep shifted from ~18% to ~24% on Oura. Sleep latency shorter. Plasma glycine was low at baseline, so this was actually addressing the deficit rather than adjacent to it. The honest read: NAC is doing real things, just not the right things for my specific profile. imo The comparison only made sense after I knew what I was comparing against.
the plasma glycine baseline is the part most people skip, and it’s the only reason your comparison means anything. without it both compounds are just guesses with extra steps.
Baseline plasma glycine being documented low is legitimately the strongest part of this writeup - you’re correcting a measured deficit, not guessing at a pathway. Hard to argue with that framing. But running two sequential blocks w/o randomizing order creates a confound worth naming: cysteine repletion during the NAC phase may have primed something that made the glycine response look cleaner than it would have cold. Order effects in sequential n=1 trials are real and underreported. Also “NAC hits cysteine status upstream, not the NMDA co-agonist site” undersells it a bit - NAC’s activity at the cystine-glutamate antiporter has documented glutamate modulatory effects that don’t reduce cleanly to the glutathione story. not saying the glycine finding is wrong, just that the comparison has more noise than the writeup implies.
Right, and “guesses with extra steps” is a good way to put it. Though I’d push slightly further: the baseline isn’t just validation, it determines which effect you’re even looking for. My glycine numbers were low enough that the sleep architecture shift was a reasonable prediction before I ran the block. If plasma glycine had been normal, I’d have expected nothing and tested something else. The baseline collapses the outcome space before you spend 11 weeks on it.
“the baseline collapses the outcome space” is the right way to put it. i had a patient who ran four supplements in a row over two years before anyone ordered an amino acid panel, and the panel ended the search in a week.