i wanted the limitless pill. here's what survived my truth bar.

i wanted the limitless pill. i'm a builder living through the one era where cognition is literally the economy, of course i wanted it. so over the years i've read the studies, tried the stacks, and eventually did to nootropics what i do to ai demos: held everything to the same truth bar. replicated human evidence only. effect sizes on the table. vendor science discounted to zero. "i felt sharper" admitted as data about feelings, not about sharpness.

here's the honest ledger: what survived, what half-survived, and the graveyard. numbers are standardized effect sizes where the literature gives them; for calibration, 0.2 is small, 0.5 moderate, 0.8 large.

the reference class first

before grading any pill, you need the baseline that makes the grades meaningful: sleep restriction to six hours a night for two weeks degrades cognition to the level of a full skipped night, per the classic dose-response study, and seventeen hours awake produces impairment equivalent to a 0.05 blood alcohol level. aerobic exercise reliably raises bdnf (a 2025 meta-analysis puts the acute bump around smd 0.6) with consistent modest cognitive benefits on top.

hold that scale in your head. nothing below comes close to it, in either direction. every pill on this list is fighting for scraps compared to what your sleep and your running shoes are already worth. that's the single most useful fact in the entire field.

what survived

caffeine, with an asterisk the size of the mug. the most robust legal cognitive enhancer known: alertness, vigilance, reaction time, replicated across hundreds of trials. the asterisk: in habitual users, a big slice of the daily "boost" is withdrawal reversal. acutely-abstinent regular consumers who take caffeine mostly climb back to the baseline that never-users are already standing on. caffeine still genuinely helps in sleep-deprived states and in naive users, but the honest model for the daily drinker is: you're largely repaying a loan you issued yourself overnight. i've made peace with this. i just no longer pretend the morning cup is enhancement.

caffeine + l-theanine, mildly. the beloved stack has a real 2025 meta-analysis behind it: reaction time smd around 0.7 at the two-hour mark, attention-switching around 0.3, with the authors noting most of the effect is the caffeine and the theanine mainly smooths the jitter. real, small, cheap. fine.

creatine, the quiet overachiever, held loosely. a 2024 meta-analysis found memory benefits around smd 0.3, with effects skewing larger in stressed states. a correction later flagged statistical issues, so the true effect is probably smaller than the headline. the most interesting single result: one strong trial found a large single dose maintained processing speed through a night of sleep deprivation. the famous vegetarian advantage, meanwhile, failed a careful 2023 replication. verdict: cheap, safe at standard doses, plausibly a small real edge, best evidenced exactly when founders need it, in sleep debt. i take it. i don't expect miracles from it.

l-tyrosine, situationally. the reviews are consistent: preserves working memory under acute stressors, cold, noise, sleep loss, multitasking, and does nothing at rested baseline. it's not a daily enhancer. it's a stress buffer you deploy for the brutal day. narrow, but honest about its narrowness, which is more than most of this industry manages.

modafinil, the real one, with a real bill. the best-evidenced prescription enhancer for healthy adults: a systematic review found consistent benefits on complex tasks, planning, decision-making, less on simple ones. now the bill: headaches in roughly a third of users, plus anxiety and insomnia at meaningful rates, rare-but-serious skin and psychiatric reactions, and some studies showing a creativity trade-off, divergent thinking gets worse. it's a prescription drug, not a supplement, and it behaves like one. i note its existence; i don't build a workflow on it.

and nicotine, which i list to be honest and skip on purpose. the meta-analytic evidence in non-smokers shows genuine small effects on attention and memory, effect sizes 0.16 to 0.44, and it's not withdrawal reversal. it's also one of the most addictive substances we know, with a cardiovascular bill. real cognition, disqualifying risk profile. the truth bar doesn't only measure efficacy.

the graveyard

this is the part the industry needs you to never read in one place.

ginkgo biloba: null. the definitive meta-analysis on healthy adults: effects statistically indistinguishable from zero on memory, executive function, and attention, across a thousand-plus participants, independent of dose or duration. the authors' own phrasing: not a smart drug.

omega-3 for cognition in healthy adults: null. real molecule, real cardiovascular story, and rct after rct shows no meaningful cognitive benefit in the already-healthy. the observational hype never survived controlled trials.

piracetam and the racetams: no case. the compounds that named the entire genre have essentially no credible modern trials in healthy adults. the citations, when you chase them, are old, small, and mostly in impaired populations. the founding family of "nootropics" is an empty folder.

rhodiola: unratable. every trial in the main systematic review carried high risk of bias. maybe there's something there; nobody has run the clean study that would tell us, which after decades is itself information.

lion's mane: rodent mechanism, human shrug. the ngf story is preclinical. the human trials are tiny, short, and mixed. the mushroom's marketing is years ahead of its data.

and microdosing, the one that hurt to grade. the largest placebo-controlled study, a clever self-blinding design with 191 participants, found everything improved, mood, focus, wellbeing, and the placebo group improved exactly as much. subsequent randomized trials with modern blinding keep agreeing: the durable benefits are expectancy. the limitless pill of my generation's mythology turned out to be the most beautifully packaged placebo in the building.

what the ledger taught me

three things, and none of them are about pills.

first, the effect-size hierarchy is a punchline: everything in this aisle is weaker than a night of sleep and a run. the industry exists in the gap between how enhancement feels and what it measures. i've written elsewhere that biohacking is mostly unsubscribing; this ledger is the receipts.

second, "i felt sharper" is the weakest evidence there is, and i say that as someone who felt sharper on half the graveyard. the microdosing result should humble everyone: 191 people felt the magic, and the sugar pill delivered it identically. your subjective sense of your own cognition is exactly the instrument these compounds are best at fooling.

third, the same epistemics work everywhere. vendor benchmarks, cherry-picked demos, unreplicated wow-results, testimonials from the invested: the nootropics industry and the ai industry run the same playbook on the same human weakness, the desperate hope for a shortcut to being smarter. the truth bar is the same tool in both aisles. use it in both.

the limitless pill exists, by the way. it's just not a pill, and you can't buy it: it's eight hours, a run, and a problem you actually want to solve. everything else on the shelf is fighting over rounding errors.