Health

Most people train cold one way: get in, stay in, endure. That trains metabolic adaptation and almost nothing else, and it's the layer with the slowest returns. There are three separate adaptations available — cold-shock control, metabolic/brown-fat activation, and autonomic control — and a long static soak only really touches the middle one.

The single highest-leverage thing I found is the head dunk. One to two seconds of face-first cold hits the trigeminal nerve zone and triggers the mammalian dive reflex: heart rate drops immediately, sympathetic output collapses, parasympathetic tone surges. An ice bath cools the body and shifts you slowly. A dunk interrupts you. It's the strongest on-demand stress reset I've found, and it takes two seconds.

Longer isn't better. The hormetic spike happens in the first one to five seconds; past that, risk rises faster than benefit. Same story with the bath itself — beyond about fifteen minutes there's no extra adaptation, only recovery debt.

I run chest-level immersion with arms out, which sounds like cheating and isn't. Hands and forearms are the most cold-sensitive tissue you have, so they numb first and take fine motor control with them. Keeping them functional means you stay calmer, hold longer, and exit safely. Arms go in for a short finisher at the end, if at all.

The pool version adds a second axis. At around 15°C I'll do one to three minutes to start, working up toward five to ten, box breathing at 4-4-4-4 the whole time. Alongside it I train static apnea — face down with support, twenty to thirty second holds building toward one to two minutes, and never without someone else present. That last clause isn't a disclaimer, it's the actual rule: shallow water blackout gives no warning and the people it kills are overwhelmingly experienced. Three rounds of cyclical breathing before entry makes the cold entry itself almost uneventful.

The rule I keep coming back to: adaptation isn't suffering. Going colder and staying longer is ego training. The gains now come from precision, not from making it worse.

"I can stay calm while my body screams." That transfers directly to business stress — which is the actual reason I do it.

Heat triggers heat-shock proteins, which repair misfolded proteins and protect cells under stress. Cold does not do this. It's the one mechanism where heat is unique rather than complementary, and it's the reason sauna sits in the recovery slot rather than the optional slot.

It also functions as passive cardio — heart rate sits at 120–150 bpm without any joint load, which is roughly zone-two work while sitting still. The Finnish longitudinal data on frequency and all-cause mortality is the strongest evidence base attached to any protocol I run.

What it is not is a calorie tool. Thirty minutes is roughly a twenty-minute walk. The weight that disappears afterwards is water and electrolytes. People conflate sweating with fat loss and then over-invest in the wrong thing.

The version that actually feels transformative is contrast: sauna, then cold, repeated. You get vascular expansion, autonomic oscillation between sympathetic and parasympathetic, both heat-shock and cold-shock proteins, and a dopamine elevation that lasts hours. Finish cold for alertness, finish hot for relaxation.

Consistency beats duration. Longer sessions are the most common way to get less out of it.

The first genuinely useful thing I learned about breath wasn't how to do it, it was how to read it. Comfortable inhale length tracks state closely: three to four seconds means sympathetic dominance — defence mode, attention biased outward, blood away from the gut. Six to eight is neutral or recovering. Ten or more means full parasympathetic access is actually available. Nothing else in a normal day gives that clean a read on where you are, and it takes about fifteen seconds to check.

That matters because most breathwork advice is written for someone who is already calm, which is precisely the state in which nobody needs it. Reading first, then choosing, is the whole difference between a practice that works under load and one that only works on a Sunday.

Three gears, three physiological targets. Gear 1, recovery, is slow nasal breathing at 4 in and 6 out with no holds — after training, during illness, or any day the system is already taxed. Gear 2, reset, is box breathing at 4-4-4-4, one or two rounds only, to interrupt a stress loop without stimulating. Gear 3, controlled stimulation, is cyclical fast breathing followed by breath holds — the Wim Hof–adjacent protocols that spike alertness and train CO₂ tolerance. My mistake for a long stretch was running Gear 3 when I needed Gear 1. The energy response felt productive at the time and the cost turned up later in the day.

The daily structure is three short sessions with distinct purposes rather than one long practice. Morning sets a parasympathetic baseline before the laptop opens. Mid-afternoon interrupts cognitive fog without requiring a full break — the point is to return to work immediately, not to pause it. Evening signals the work cycle is closed. Past about seven minutes a session, returns diminish and eventually invert into mental fatigue rather than calm.

Resonance breathing — around four in, six out, roughly six cycles a minute — has the most consistent effect of anything I've used, and I suspect part of that is simply that it's sustainable. A ten-second cycle is the longest most people can hold for several minutes without it feeling like a task. Techniques needing unusual counting or a visual anchor drop out of use faster than simple ratio breathing regardless of how good they look on paper.

Everything above is the physiological half. There's a separate practice I run — ascending sets of pumping breath with long empty-lung holds, built around an explicit intention — whose framing is frankly not physiological at all. It lives on the Spiritual page rather than this one, because that's where the honest version of it belongs.

What I underinvested in for years was the two or three minutes between finishing a session and starting work. The breathing resets the baseline; the transition is what protects it. Without that bridge the state survives about five minutes once demands arrive. A slow walk to the desk, one written intention before any application opens, then work — that sequence ended up mattering more than which technique I'd just used.

Inhale length is a diagnostic, not a technique. Read the state first — choosing the wrong gear costs more than skipping the session.

The frame I had wrong at the start was treating flexibility as a fixed trait. It's adaptive, and it decays on the neurological side first — within roughly two to four weeks of not training it — while the structural side regresses far more slowly, six to eight weeks and beyond. The practical consequence is that frequency beats duration: ten minutes daily outperforms an hour weekly by a wide margin. Regaining lost range after a gap also happens two to three times faster than the original build, because both the neural pathway and the tissue remodelling already exist. That reframe moved me off trying to commit to long sessions and onto holding a daily floor.

The baseline sequence runs about forty-five minutes in five phases: joint mobilisation and cat-cow to activate; core integration targeting deep abdominal control rather than surface strength; spinal and shoulder restoration with controlled thoracic rotation; flexibility expansion through hamstrings, adductors and hip flexors; then a full down-regulation to close. The order matters more than the content — dynamic before static, breathwork at the end. Skipping the cooldown consistently produced worse outcomes than simply running a shorter session, which took me an embarrassingly long time to notice.

There's a separate thirty-minute reset built specifically around anterior pelvic tilt and a limited hip hinge, run three or four times a week. Four blocks: a stack reset using 90/90 hip lifts, rib-focused cat-cow and deadbugs; an anterior chain release through couch and half-kneeling hip flexor work; a posterior chain block using wall hip hinges, half-monkey and Jefferson curls; then integration ending in a passive forward-fold hang. I track finger-to-floor distance weekly as the single objective marker, because everything else about mobility work is easy to feel good about and hard to verify.

Yin runs on a different principle and requires holds most people exit early — forward fold, butterfly, pigeon, lizard, frog, sphinx, thread-the-needle, supine hamstring, reclining twist, then legs up the wall to close. Most holds sit in the five-to-seven minute range. Once a week on a rest day, low light, slow nasal breathing throughout, and treated explicitly as a recovery modality rather than a session. Framing it as training is how it ends up being rushed.

Morning is thirty to forty minutes and dynamic — joint rotations from the ankles up, then active spinal work. Evening is shorter and entirely passive, two to four minutes a hold. I'm still inconsistent with the evening one, particularly when the day runs late, and three or four nights a week is the realistic floor. Claiming daily adherence would make the system look better and be less useful.

Separately from the maintenance work there's a bucket list, and I'd rather label it as aspiration than pretend it's current practice. These are strength targets rather than flexibility ones — chaturanga, crow and side crow, firefly, eight-angle, handstand, forearm stand for the upper body; peacock, scorpion, one-legged crow, press-to-handstand and flying pigeon as the integration set. Framing yoga as bodyweight strength training rather than stretching is what made it interesting to me again after years of finding it boring. The mobility work is what makes the strength work reachable; the list is what stops the mobility work becoming an end in itself.

Flexibility is neurological before it's structural. The nervous system learns depth is safe — that learning decays without maintenance, and returns faster than it was built.

The logic is straightforward and slightly unflattering: motivation doesn't hold across twelve months, and no amount of insisting otherwise has changed that. A date, a paid registration, and another person who knows you're doing it produce a different quality of commitment than a goal written in a notebook. This is adherence architecture, not willpower. I choose events as much for their social structure as their physical demand.

Hyrox is the most directly transferable format for a hybrid base: eight rounds of a 1 km run alternating with eight functional stations — SkiErg, sled push, sled pull, burpee broad jumps, row, farmers carry, sandbag lunges, wall balls. The format is identical at every event globally, which makes preparation precise rather than vague. The Doubles and Relay formats split the stations between two athletes while both still run, which creates shared pacing and shared suffering in a way solo training can't manufacture. The underlying fitness it demands — aerobic base, lower-body strength, sustained grip — is the same base that consistent gym and trail work already builds. The event doesn't require a new programme; it points the existing one at a date.

Trail events supply the outdoor, terrain-specific half. The near-term target is the Chiang Mai UTMB 20K, with the 50K as the step after. Doi Suthep sits close enough to the city that weekly vertical is available without travel, which matters more than it sounds — the difference between training that needs a car and training that doesn't is most of the difference between doing it and not. The structure is three session types: a long trail session of two to three hours mixing climb and descent, a focused climb session of 45 to 60 minutes targeting 400 to 600 metres of vertical, and the gym work kept fully intact rather than substituted, because lower-body strength is an advantage in both formats.

The number I train against is vertical rate. On stairs I've hit 800 metres of gain in an hour, which sits in the strong-amateur band — decent amateur trail runners run 400 to 600 an hour, strong runners 700 to 900, competitive mountain runners 1,000 to 1,200. On Doi Suthep it's 700 to 800 metres in an hour to an hour fifteen. The number that actually matters for racing is lower though: 400 to 500 an hour is what's sustainable across a multi-hour effort, and that's the figure UTMB is really asking for. A one-hour benchmark flatters you in a way a four-hour race doesn't.

Working the 20K backwards from that: roughly 1,200 metres of climbing at race effort is two and a half to three hours including micro-stops, plus an hour to ninety minutes of descent and runnable flat. That's a 3:45 to 4:30 finish trained, or 4:30 to 5:00 conservatively with long breaks and slow climbs. The cutoff is seven hours, which at current fitness would take an injury, an hour stopped, or heat collapse to miss. Writing that down is partly the point — a target you've actually modelled is much harder to quietly renegotiate than one you're carrying around as a feeling.

What I got wrong moving into trail was assuming a strong gym and cardio base would carry cleanly onto mountain terrain. It doesn't — not completely. Eccentric loading from technical descent is a different stressor to anything you can replicate on flat equipment, and I only worked that out after adding specific downhill sessions and discovering how much they hurt. Calves, quads and ankle stabilisers need conditioning for the grinding climb and the descent afterwards that gym work simply doesn't provide. It's a common planning error for anyone coming from a gym background, and I made the standard version of it.

The calendar also supplies the periodisation a pure gym routine tends to lack. Without a date, volume and specificity both drift quietly. With one, the training week organises itself into build, specific and taper without needing to be manufactured. The social layer compounds it — sessions involving another person, a defined route and a shared target date get completed at a rate that solo sessions with identical intent do not.

The event date is the programme. Without an external commitment and a social context, both volume and consistency fall to a fraction of stated intent.

The current block is an eight-week intermediate programme on a push/pull/legs/skill split, four to five sessions a week at sixty to seventy-five minutes. Monday push — dips, handstand pushups, pseudo planche pushups. Tuesday pull — pull-ups, archer rows, explosive chin-ups. Wednesday legs and core — pistol squats, Bulgarian splits, hanging leg raises. Thursday is skill and mobility: handstand balance, L-sit, scapular work. Friday is a hybrid full-body superset. Weekends are rest or an easy jog.

The targets over those eight weeks are specific, which is the only reason they work: pull-ups from six to eight strict up to twelve to fifteen clean, dips from eight to ten up to weighted sets of ten, pseudo planche pushups from five to ten or twelve, L-sit from fifteen seconds to thirty, wall handstand from thirty seconds to a ten-second freestanding hold, and pistol squats from assisted to five full-range each leg. A block without numbers attached is just attendance.

Above that sits a phase roadmap measured in years rather than weeks. The first six months is the base — crow, headstand, wall handstand, L-sit, a clean push and pull foundation, pistol squat. Six to eighteen months is free handstand, muscle-up, front and back lever progressions, forearm stand, side crow, full splits. Eighteen to thirty-six is press-to-handstand, planche work, peacock, flying pigeon. Somewhere in the second phase, strength-to-weight leverage starts genuinely mattering — around fifteen to eighteen per cent body fat is where these skills stop fighting you.

The version I'm actually building toward is the ten-to-fifteen-year split: one gym day for compound strength, one or two calisthenics days for skill and control, one yoga or mobility day for restoration. Three or four sessions a week, indefinitely sustainable, pain-free. That's the whole design goal — not a peak, a plateau I can hold for decades. Most training programmes optimise for a twelve-week transformation and quietly assume you'll figure out year three yourself.

One small thing that's paid off disproportionately: training forearms first on some sessions, when they're freshest, rather than as an afterthought at the end. Heavy days twice a week — reverse curls, hammer curls, wrist curls, wrist roller or plate pinch to failure. Pump days two or three times — farmer's carries, band work, fat-grip or towel hangs. Sixty to ninety sets a month total. Size shows at four to six weeks, density and vascularity at eight to twelve, then it holds on twice a week. Grip is the thing that fails first in both hybrid racing and every hanging skill on the roadmap, and almost nobody trains it deliberately.

Pick the modality you can still run at sixty, then work backwards. A programme that peaks in twelve weeks has no answer for year three.

Chronotype is a biological rhythm preference, not a habit — it decides when your brain actually performs, regardless of what your calendar says. Roughly half the population peaks midday; only 15–20% genuinely peak early. The 5am-club framing is advice from a minority to everyone.

Mine came out as a late Bear with a Wolf crossover: natural wake between 7 and 9, real cognitive peak from about 10:30 to 2:30, a genuine trough mid-afternoon, and a workable second wind in the evening. Nothing sharp exists before 10:30 no matter how early I get up.

The change that mattered wasn't waking earlier — it was protecting the peak block like a monk and stopping the attempt to do strategy work at 7am. Admin and low-friction tasks go in the ramp-up window. The trough gets a walk or calls, not another attempt at focus.

Two mechanical things carry most of the effect: delaying caffeine roughly 60–90 minutes after waking so it works with the cortisol curve rather than against it, and getting 20–30 minutes of outdoor light before 9am to anchor the rhythm.

Don't fight the rhythm you have to get the one you read about. Find the peak, then defend it.

The highest-leverage variable in sleep quality is set in the morning, not at night. Ten to twenty minutes of outdoor light within an hour of waking suppresses cortisol at the right point in the curve and sets melatonin release up correctly for the evening. Screens are not a substitute — a phone puts out roughly 500 lux; outdoor light on an overcast day is 10,000 to 25,000. Done consistently, this one habit does more than most of the supplement stack combined, and missing it is often the actual cause of delayed sleep onset in an otherwise clean protocol.

The ninety-minute wind-down is the part I treat as non-negotiable. From ninety minutes out: no screens, no high-stimulation work. From sixty: lights dimmed warm, magnesium and glycine taken. From thirty: physical reading, journalling, or gentle stretching. The last ten minutes are box breathing and the phone in aeroplane mode. Most people resist this because it feels like lost productive time. It isn't — the cognitive output of the hours before adequate sleep is far lower than the same hours after it.

The supplement side is deliberately short: magnesium glycinate for sleep latency, glycine to lower core temperature and improve REM quality, L-theanine to dampen overthinking without sedation, and apigenin. Melatonin is deliberately absent from the nightly stack. Low doses are genuinely useful for a circadian reset after travel, but taken every night it suppresses your own production over time — it's the wrong tool for a chronic problem and a common way to make one worse.

Environment carries more weight than people expect. Seventeen to nineteen degrees, because a warm room directly suppresses melatonin. Complete darkness, with every LED covered — even low ambient light through closed eyelids measurably disrupts sleep staging. Phone out of the room. These aren't comfort preferences; they're the physical conditions under which sleep architecture works properly.

Two saboteurs worth naming plainly. Alcohol: two drinks suppress REM and heart-rate variability for the entire night, producing sleep that looks adequate on duration and isn't restorative. Caffeine: the half-life is five to eight hours, so a 3pm coffee still has a quarter to half its adenosine-blocking effect at 10pm. Cutting it early afternoon isn't conservative — it's just the arithmetic. The late-afternoon dip is adenosine pressure, and it's useful information rather than a problem to override.

Sleep quality is set in the morning. The light you get at 7am determines the melatonin you get at 10pm.

The core is a weekly fast of roughly forty hours, running dinner to lunch two days later. The duration is the whole point. Around eighteen to twenty hours glycogen depletes and the body shifts to fat oxidation; by twenty-four to thirty-six autophagy is meaningfully elevated; by forty, insulin and IGF-1 have both dropped and growth hormone has risen substantially. A 16:8 window delivers almost none of this at any real intensity — it mostly just compresses eating. Forty hours is the minimum dose that actually does the cellular housekeeping.

Three times a year I run a longer one, around seventy hours. That targets something the weekly fast can't reach — past roughly forty-eight to sixty hours the body starts recycling damaged immune cells, with regeneration on refeed. I treat these as scheduled maintenance rather than feats of willpower. They need preparation: electrolytes loaded the day before, a clear calendar, no travel. Recovery is faster than most people expect once adapted.

On one other day a week I run a fast-mimicking day at around five to six hundred calories, skewed deliberately — protein very low, carbohydrate low and fibrous, fat carrying most of it. The point is to keep mTOR and insulin suppressed without triggering full starvation signalling. It's a lever for metabolic flexibility without the logistics of a full fast.

During a fast: electrolytes in water throughout, two to three litres of fluid, black coffee and green tea allowed. The most common failure is taking nothing, getting a headache and fatigue on day one, and concluding the fast is too hard. It's almost always electrolyte depletion, and it's trivially fixable.

The refeed matters as much as the fast. Breaking forty hours with a large carbohydrate meal lands a glucose spike on top of depleted glycogen and sensitised insulin receptors — the resulting whiplash gets misread as post-fast weakness. I break with lean protein and fat, and no significant starch in the first meal. Gradual re-entry, not a reward.

The most powerful longevity lever isn't a supplement. It's controlled periods of not eating.

The most common mistake I see is starting with the exotic molecules. Most people are poorly maintaining magnesium, vitamin D and omega-3 long before they need NAD+ precursors. So the stack starts with what has the strongest human trial evidence and the broadest systemic effect, and everything sits in a named tier. Nothing moves up a tier because it's expensive or currently fashionable.

The foundation is four things. Creatine monohydrate, which has the strongest evidence in the entire stack — hundreds of controlled trials, cognitive and muscular benefit, unambiguous long-term safety. Magnesium glycinate at night, for bioavailability and gut tolerance. D3 with K2, taken with fat — the K2 handles calcium routing, and D3 without it is a half-measure. And algal omega-3 rather than fish oil, which removes the oxidation and heavy-metal question entirely.

The second tier is recovery and cellular. Glycine before sleep, which lowers core temperature and supports collagen synthesis — one of the cheapest and most underrated things available. Taurine, which got considerably more interesting after the 2023 work linking deficiency to accelerated ageing in mammals. Collagen peptides with vitamin C before training, where the vitamin C isn't optional — it's required for the synthesis to happen at all.

The third tier is explicitly experimental and I'd rather label it than dress it up. NAD+ precursors and sirtuin-activating compounds have a plausible mechanism and strong animal longevity data; human outcome data remains thin. Resveratrol has a separate problem — oral bioavailability is genuinely poor, with much of it metabolised before it reaches circulation. I take these because the safety profile is good and the downside is low, not because the evidence is settled. Senolytics I run pulsed for a few days a month rather than daily, because the mechanism makes more sense that way.

Timing matters more than people assume. Fat-soluble compounds need a meal with fat in it or they largely don't absorb. Magnesium and glycine belong together at night. The morning stack and the evening stack shouldn't be collapsed into one for convenience — the circadian context of each compound is part of the dose.

Build from the evidence up. The exotic longevity molecules sit on top of a foundation, not instead of one.

The argument for tracking cognition in your thirties is simple: the prevention window opens in your thirties and forties and closes well before symptoms appear. By the time decline is subjectively noticeable, measurable change has often been present for years. Nobody would be surprised that untracked physical fitness quietly deteriorates over a decade; the same happens with cognition, with higher stakes. This isn't self-surveillance — it's having a real data point to return to when something feels off, instead of guessing.

The quarterly battery takes under half an hour. Reaction time, five runs averaged, as a processing-speed marker sensitive to sleep debt, stress and alcohol. A working memory test scored against age-matched peers. Verbal fluency tested informally through timed writing sprints. The discipline is doing it the same day each quarter, at the same time, under the same conditions — testing after travel or a bad night produces noise, and noise is worse than no data because it gets treated as signal.

Monthly I run a dual n-back session, which requires holding both a visual position and an auditory letter n steps back simultaneously. The level I can sustain is the monthly marker. A drop of one level held across two consecutive months is a flag rather than a crisis, and the response is to check the lifestyle variables first — sleep, training load, diet, alcohol, anything that changed in the stack. I cross-reference against heart-rate variability from the same period, because low HRV and cognitive drift together almost always share an upstream cause.

If scores dip, the response is tiered. First, find the stressor — in most cases there's an obvious one and the protocol is simply to address it. Second, check the foundational inputs. Third, if two or three consecutive quarters decline with no lifestyle explanation, the correct move is bloodwork, not more supplements. Speculatively stacking compounds against an unexplained trend is how people spend a year treating the wrong thing.

What doesn't work, and gets over-invested in: the gamified brain-training apps. They reliably improve your score on their own tasks and transfer poorly to anything else, and the meta-analyses on this are consistent. N-back has better transfer data but is still more useful as a measurement tool than a training one. The genuine levers are upstream and unglamorous — sleep staging, zone-two cardio, creatine, and not drinking much. Everything else is marginal by comparison.

The prevention window is your thirties. By the time symptoms appear, most of the opportunity is already spent.

Most environmental health writing assumes a temperate climate where air pollution is background noise. Northern Thailand in burn season is not that context. Agricultural burning across the surrounding valley and highlands pushes particulate levels into genuinely hazardous territory by any classification standard, for weeks at a time, roughly February through May with the worst of it in March and April. That's chronic particulate exposure driving oxidative stress and inflammation — not a nuisance you manage with a good attitude. It requires infrastructure decisions, and it's the single highest-return intervention available where I live.

The indoor strategy is a HEPA H13 or better running in the bedroom and the main workspace, the IQAir or AirVisual app checked daily for AQI, and a simple window rule: closed whenever AQI is above 100, opened only around dawn when readings tend to be at their lowest. Filters get replaced every three to six months through burn season rather than annually, because a saturated filter is decorative. I add NAC at 600–1,200 mg, omega-3 and vitamin C through the worst months for airway oxidative stress — alongside filtration, never as a substitute for it, which is a distinction a lot of people quietly collapse because supplements are easier to buy than behaviour is to change.

Water is second, and close. The regional default is bottled water in PET plastic, frequently stored in heat or direct sun, and antimony and BPA leaching from heat-stressed plastic is a documented problem rather than a theoretical one. I moved to glass and stainless for daily carry. The carbon-plus-reverse-osmosis setup at home took me considerably longer than it should have — I knew it was the right call well before I acted on it. The convenience of bottled water here is real, and convenience reliably beats intention whenever the friction stays high. That gap between knowing and doing is the most honest thing on this page.

The plastics audit extends to food contact: nothing heated in plastic, glass or ceramic for storage, no reuse of single-use bottles. At the home tier the changes are cookware — replacing non-stick Teflon with ceramic, cast iron or stainless because of PFAS exposure — cleaning products moved to vinegar, bicarb and enzyme-based rather than synthetic fragrance, and fragrance-free enzyme laundry detergent. Each one is individually trivial and only meaningful across a multi-year exposure window, which is exactly why they're easy to never get around to.

Personal care is the tier people skip because each item feels too small to matter, which is precisely the logic that makes a decade of daily exposure add up. Sulfate- and fragrance-free shampoo and body wash — castile soap and bar soap work fine. Mineral sunscreen with zinc oxide or titanium dioxide rather than chemical filters. Ceramide-based moisturiser. Baking-soda, magnesium or zinc deodorant. Essential oil diffusers and beeswax instead of scented candles and synthetic fresheners. On the oral side: bicarb brushing once a week and no more than twice, because past that it abrades enamel; a coconut-based toothpaste with calcium carbonate or hydroxyapatite; oil pulling with sesame or coconut; and tongue scraping in the morning, which is the highest-return thirty seconds in the whole routine.

Light and circadian hygiene sit lower in the audit — not unimportant, just lower return than air, water and plastics. Morning sunlight shortly after waking, warm lighting in the couple of hours before sleep, less screen proximity in the final hour. I deliberately rank the wireless and EMF concern that dominates a lot of this space below all of the above, because the evidence for harm at normal residential exposure is substantially thinner than the evidence on particulates, microplastics and endocrine disruptors. I'd rather put the effort where the data is strongest than where the discourse is loudest.

Environmental health is the terrain everything else runs on. Where there's a defined burn season, the interventions that matter are infrastructure, not attitude.