Bettering Me

Author: Unpaid Intern

  • Standard Lipid Panels Were Designed for Late-Stage Detection – Not for Prevention at 45

    Written by

    Unpaid Intern

    –

    The standard lipid panel that your doctor orders – total cholesterol, LDL-C, HDL-C, triglycerides, and sometimes VLDL – was designed in the Framingham era to detect people at immediate risk of cardiovascular events. It was optimized for a specific clinical question: is this person about to have a heart attack? That question is not the same as the prevention question: is this person on a trajectory toward cardiovascular disease in 20 years?

    The most common misunderstanding is what LDL-C actually measures. LDL-C estimates the mass of cholesterol carried inside LDL particles. It does not count the particles themselves. ApoB – apolipoprotein B – counts every atherogenic particle in circulation, including LDL, VLDL, IDL, and Lp(a), because each of these particles carries exactly one ApoB molecule [1]. The distinction matters because the particles cause plaque, not the cholesterol inside them.

    Think of it this way: LDL-C is like measuring the total weight of cars on a highway. ApoB is like counting the cars themselves. If car manufacturers start making lighter cars, the total weight goes down while the number of cars stays the same – and it is the cars, not their weight, that determine traffic and collision risk. The cholesterol inside a lipoprotein particle is cargo. The particle density determines how many get trapped in the arterial wall.

    Two people can have identical LDL-C levels while one has twice as many atherogenic particles. This discordance occurs because LDL particles vary in size and cholesterol content. People with predominantly small, dense LDL particles have “normal” LDL-C (because each particle carries less cholesterol) but high ApoB – and therefore higher cardiovascular risk that the standard panel misses entirely [2]. The prevalence of this discordance is approximately 15-20% in the general population, and higher in people with insulin resistance, type 2 diabetes, and elevated triglycerides.

    The test your doctor orders was designed in a clinical context where the goal was to identify people who needed statin therapy to prevent near-term events. For that purpose, LDL-C works reasonably well at the population level. But if you are 45 years old, asymptomatic, and paying for prevention, LDL-C leaves important information on the table.

    What should a prevention-focused lipid panel include? The Bettering Me minimum is: ApoB, Lp(a) (checked once), non-HDL cholesterol, triglycerides, and HDL-C. Non-HDL cholesterol (total cholesterol minus HDL-C) is a reasonable proxy when ApoB is unavailable – it captures all atherogenic lipoproteins and correlates well with ApoB at the population level [3]. But it is still a proxy. ApoB is the direct measure.

    Lipoprotein(a) – Lp(a) – should be checked once in a lifetime. It is 80-90% genetically determined and does not respond significantly to lifestyle intervention [4]. A single high reading (above 50 mg/dL or above 125 nmol/L, depending on the assay) means you need aggressive ApoB management because your baseline atherogenic particle production is genetically elevated. The European Atherosclerosis Society recommends that everyone be tested for Lp(a) at least once [4]. A high reading does not mean you are doomed – it means you should target an ApoB below 70 mg/dL instead of below 100 mg/dL.

    What about optimal ApoB targets? For primary prevention in a 45-year-old with no known cardiovascular disease, an ApoB below 100 mg/dL is the minimum acceptable. Below 90 mg/dL is optimal. Below 80 mg/dL is aggressive [3]. These targets are lower than what most clinical guidelines recommend because the guidelines are designed for population-wide risk management, not individual optimization. If you have Lp(a) above 50 mg/dL, traditional risk factors (hypertension, smoking, diabetes), or a family history of premature cardiovascular disease, your target should be below 70 mg/dL.

    Counterpoint: isn’t LDL-C good enough for most people? At the population level, yes – LDL-C correlates with cardiovascular risk well enough that guidelines use it. But you are not a population. You are an individual. If you are in the 15-20% of people whose risk is discordant with their LDL-C, the standard panel is misleading you. The cost of checking ApoB is approximately $30-50 out of pocket if your insurance does not cover it. A standard lipid panel costs $50-100. The incremental cost of knowing your true risk is approximately $30. Compared to what you spend on supplements, gym memberships, and organic food, that is the cheapest prevention dollar you can spend.

    The standard panel is not useless. It is incomplete for the prevention context. Knowing your LDL-C without knowing your ApoB is like knowing your speed without knowing whether you are driving on a straight road or a winding mountain pass. The speed is useful. The context determines the risk. Pay for the context.

    Practical guidance for your next lab visit. When your doctor orders “lipid panel,” you get total cholesterol, LDL-C, HDL-C, triglycerides, and VLDL. To get ApoB, ask for “apolipoprotein B” – CPT code 82172. To get Lp(a), ask for “lipoprotein (a)” – CPT code 83695. Some labs bundle these as an “advanced lipid panel” or “cardiovascular risk panel.” Cost: approximately $50-100 out of pocket for the add-ons if insurance declines. Most major labs offer cash-pay direct ordering. If ApoB is not available, non-HDL cholesterol (total minus HDL) is an acceptable surrogate – and most standard panels already report non-HDL-C. If non-HDL-C is above 130 mg/dL, you can infer your ApoB is likely above 100 mg/dL, and you should push for the direct ApoB measurement.

    Disclaimer: This post is for inspiration and education, not medical advice. Everyone’s body is different, so please check with your doctor before changing your diet, exercise, or lifestyle routine. By using these tips, you agree to do so at your own risk.

    References

    [1] Sniderman AD, et al. "A meta-analysis of LDL-C, non-HDL-C, and ApoB as markers of cardiovascular risk." *Circ Cardiovasc Qual Outcomes*. 2011;4(3):337-345.. DOI: https://doi.org/10.1161/CIRCOUTCOMES.110.959247

    [2] Otvos JD, et al. "Clinical implications of discordance between LDL-C and particle number." *J Clin Lipidol*. 2011;5(2):105-113.. DOI: https://doi.org/10.1016/j.jacl.2011.02.001

    [3] Sniderman AD, et al. "Apolipoprotein B Particles and Cardiovascular Disease: A Narrative Review." *JAMA Cardiology*. 2019;4(12):1287-1295.. DOI: https://doi.org/10.1001/jamacardio.2019.3780

    [4] Kronenberg F. "Human Genetics and the Causal Role of Lipoprotein(a)." *Cardiovasc Drugs Ther*. 2016;30(1):87-100.. DOI: https://doi.org/10.1007/s10557-016-6648-3

  • “Biological Age” Tests Are Mostly Useless. There Are Three Numbers That Actually Tell You Where You Stand

    Written by

    Unpaid Intern

    –

    Epigenetic clocks and biological age panels are the wellness industry’s latest anxiety product. They cost between $200 and $500, spit out a single number – “your biological age is 38 but your chronological age is 45” – and send you into either celebration or despair. The problem is that these numbers correlate weakly with actual hard outcomes in middle-aged populations, change too slowly to guide interventions, and create a false sense that aging is a single metric that can be addressed with a protocol [1].

    There are three different types of epigenetic clocks, and understanding the differences reveals why they are not ready for individual clinical use. The Horvath clock measures methylation at 353 CpG sites and was designed to predict chronological age, not health. It is remarkably accurate at telling you how old you are – which is information you already had. The PhenoAge clock was trained to predict all-cause mortality and is more clinically relevant, but it incorporates routine clinical biomarkers (albumin, creatinine, glucose, etc.) that are more informative on their own than the methylation component. The DunedinPACE clock measures the pace of aging rather than current biological age and is arguably the most useful of the three, but it still changes over months to years – far too slowly to tell you whether your new exercise protocol or sleep intervention is working.

    The fundamental problem is not the science. The methylation patterns are real and interesting. The problem is clinical utility. By the time an epigenetic clock changes meaningfully, you could have measured the actual outcome directly. You could have tested your VO2 max, drawn your ApoB, calculated your HOMA-IR – and gotten feedback you can act on today.

    There are three numbers that actually tell you where you stand regarding your biological trajectory. They are cheap, actionable, and change within weeks of intervention.

    ApoB – your cardiovascular ceiling. This is the single most predictive blood marker for atherosclerotic disease, which remains the primary cause of death in aging populations. If your ApoB is above 90 mg/dL at age 45, your vascular system is accumulating damage even if your LDL-C looks fine [2]. The intervention response – lifestyle modification, dietary change, and if necessary, pharmacological therapy – can be measured in weeks, not years.

    Fasting insulin – your metabolic trajectory. Above 8-10 µIU/mL with normal glucose means your body is producing excess insulin to maintain glucose homeostasis. This is the earliest measurable sign of metabolic aging. It precedes glucose dysregulation by 5-10 years in most people. And it responds to intervention – two resistance sessions per week and a 12-hour overnight fast can reduce fasting insulin by 20-30% in 8-12 weeks.

    VO2 max – your functional ceiling. This is the single best predictor of all-cause mortality in middle-aged and older adults, outperforming every blood marker in head-to-head comparisons [3]. It measures your aerobic capacity directly – not a proxy, not a correlate, not a methylation pattern that may or may not correlate with an outcome. It tells you where you are relative to age-specific norms and whether your training is producing a measurable effect. VO2 max responds to consistent aerobic training at any age, and the changes can be detected within 4-6 weeks of a structured program.

    These three numbers cost approximately $100 total to measure (ApoB: $30-50, fasting insulin: $20-40, VO2 max estimate via submaximal protocol: $0-50). A single biological age panel costs $200-500. For the price of one epigenetic test, you could run the three anchors, get actionable results, and still have money left over.

    Counterpoint: aren’t epigenetic clocks validated? Yes, for certain specific use cases – predicting mortality in terminally ill populations, identifying accelerated aging in chronic disease cohorts, and studying the effects of interventions at the population level. The DunedinPACE clock, for example, has shown that people with higher pace of aging scores have worse physical function in their 40s and 50s [1]. But the effect sizes at the individual level are small enough that knowing your score does not change your clinical management. You would not treat a person with “biological age 40” differently from one with “biological age 48” – you would still optimize ApoB, fasting insulin, and VO2 max in both. The clock adds information that does not change the decision.

    Every dollar spent on a methylation test would be better spent on a DEXA scan and a maximal aerobic test. The simple numbers work. We just do not like how simple they are. We want a single number that tells us whether we are winning, and the epigenetic clock provides that – which is exactly why it is dangerous. It creates the illusion that aging is a static score rather than a dynamic trajectory.

    Bettering Me recommends knowing these three numbers before spending a cent on any biological age panel. The epigenetic clock tells you how old your DNA looks. ApoB, fasting insulin, and VO2 max tell you how old your body is actually functioning. One is entertainment. The other three are information.

    Disclaimer: This post is for inspiration and education, not medical advice. Everyone’s body is different, so please check with your doctor before changing your diet, exercise, or lifestyle routine. By using these tips, you agree to do so at your own risk.

    References

    [1] Bell CG, et al. "DNA methylation aging clocks: challenges and recommendations." *Genome Biology*. 2019;20(1):249.. DOI: https://doi.org/10.1186/s13059-019-1824-y

    [2] Sniderman AD, et al. "Apolipoprotein B Particles and Cardiovascular Disease: A Narrative Review." *JAMA Cardiology*. 2019;4(12):1287-1295.. DOI: https://doi.org/10.1001/jamacardio.2019.3780

    [3] Myers J, et al. "Exercise capacity and mortality among men referred for exercise testing." *NEJM*. 2002;346(11):793-801.. DOI: https://doi.org/10.1056/NEJMoa011858

  • Poor Focus at 45 Has Three Possible Causes – and Only One Is a Productivity Problem

    Written by

    Unpaid Intern

    –

    If you are over 40 and your focus has declined, the first question nobody asks is: what kind of focus problem is this?

    The assumption – yours and everyone else’s – is that it is a productivity problem. That you need better systems, better habits, better discipline. You have tried those. They helped temporarily. Then the fog returned.

    That is because there are three distinct causes of cognitive decline in midlife, and only one of them responds to a productivity intervention. Treating all three the same way works for exactly zero of them. Worse, it leads you to conclude that you are broken when the real answer is that you are tired, overstimulated, or under-supplied – three very different problems requiring three very different solutions.

    Cause One: Sleep Debt

    Chronic sleep restriction is the most common cause of cognitive decline in adults over 40, and the most overlooked. Most people who think they sleep enough do not [1]. The threshold for full cognitive restoration is seven to nine hours, and few professionals in this age range hit it consistently.

    Sleep debt is insidious because it does not feel like sleep deprivation. Total sleep deprivation – pulling an all-nighter – feels terrible and is unmistakable. Chronic partial sleep restriction – six hours per night, night after night – does not feel terrible. It feels normal. Your baseline shifts. You forget what sharp cognition feels like because you have not experienced it in years.

    Sleep debt degrades prefrontal function – attention, working memory, impulse control – faster than any other single input. A person sleeping six hours per night for two weeks has cognitive performance equivalent to someone who has been awake for 24 hours straight [1]. The person does not feel tired. They feel foggy. They assume age. They buy supplements. They try productivity systems. None of it works because the cause is physiological: the glymphatic system has not had time to clear metabolic waste from the brain, and the prefrontal cortex is operating on reduced glucose metabolism.

    Sleep debt responds to one thing: more sleep. No productivity system, no supplement, no focus app substitutes for it. If you are sleeping fewer than seven hours and struggling with focus, stop looking for the hack. The hack is sleep.

    Cause Two: Dopamine Dysregulation

    If your sleep is adequate and your focus is still fragmented, the next question is: how many times per day do you switch contexts?

    Chronic context-switching recalibrates your reward system to prefer short-cycle, high-variability inputs – email, Slack, notifications, social media – over sustained attention [2]. The result is not that you cannot focus. It is that sustained focus feels uncomfortable. Your brain has been trained to prefer the shallow hit.

    This is not a willpower deficit. It is a neurochemical adaptation. The dopamine reward prediction error system learns that novelty arrives every few minutes. When novelty does not arrive – when you try to sustain attention – the system registers a prediction error in the negative direction. You feel restless, not because you lack discipline, but because your brain is correctly reporting that the expected reward has not arrived.

    This cause responds to structural intervention: not grit, but reducing the availability of shallow reward cycles. Physical separation from the phone. Blocked browser tabs. Scheduled deep work windows. The intervention is environmental, not motivational. You do not need more willpower. You need a different architecture.

    Cause Three: Hormonal Decline

    If both sleep and context-switching are addressed and focus is still a problem, the cause is likely hormonal. Testosterone and thyroid hormones affect processing speed, verbal fluency, and working memory [3].

    Testosterone begins declining in men around age 30 at roughly 1% per year. By 45, the cumulative effect is measurable in cognitive domains that depend on processing speed. This is not a controversial claim – it is documented in longitudinal endocrinology studies. The cognitive effects of low testosterone include reduced verbal fluency, slower processing speed, and diminished spatial reasoning.

    Thyroid dysfunction – particularly subclinical hypothyroidism – is underdiagnosed in this age range and produces cognitive symptoms identical to brain fog. Fatigue, slowed thinking, difficulty concentrating – these are textbook hypothyroid symptoms that are routinely attributed to stress or aging. A simple TSH blood test can rule it in or out.

    These are medical conditions, not productivity problems. They respond to labs, a physician, and – if indicated – replacement therapy. No amount of deep work compensates for a hormone level that is below the threshold for normal cognitive function.

    The Differential Diagnosis

    The most useful thing you can do for your focus at 45 is a differential diagnosis. Not another productivity book. Not another app. A genuine attempt to identify which of the three causes is driving your symptoms.

    Sleep first. Then context-switching. Then hormones. Rule them out in order. If you treat cause three (hormones) before ruling out cause one (sleep), you will spend money on labs and medication for a problem that was solvable with a bedtime. If you treat cause two (dopamine) before cause one, you will be fighting fragmentation while operating on a sleep-deprived brain that cannot sustain attention regardless of the environment.

    The wrong diagnosis leads to the wrong intervention – and the wrong intervention leads to the conclusion that you are broken. You are probably not broken. You are probably tired, overstimulated, or under-supplied. Those are three different things, and only one of them is a productivity problem.

    Disclaimer: This post is for inspiration and education, not medical advice. Everyone’s body is different, so please check with your doctor before changing your diet, exercise, or lifestyle routine. By using these tips, you agree to do so at your own risk.

    References

    [1] Van Dongen HPA, Maislin G, Mullington JM, Dinges DF. Sleep. 2003;26(2):117-126. DOI: https://doi.org/10.1093/sleep/26.2.117

    [2] Volkow ND, Wang GJ, Baler RD. Trends in Cognitive Sciences. 2011;15(1):37-46. DOI: https://doi.org/10.1016/j.tics.2010.11.001

    [3] Janowsky JS, Oviatt SK, Orwoll ES. Behavioral Neuroscience. 1994;108(2):325-332. DOI: https://doi.org/10.1037/0735-7044.108.2.325

  • HRV Is a Readiness Score for Professional Athletes. For Everyone Else, It’s a Stress Guilt Machine

    Written by

    Unpaid Intern

    –

    Wearable HRV tracking has created a generation of people who feel bad about a number they do not understand. A low HRV reading in the morning triggers a cascade of self-diagnosis: too much stress, too little sleep, poor recovery, failing biology. The device tells you something is wrong, but it does not tell you what, and it certainly does not tell you what to do about it. The result is not better health. It is health anxiety quantified.

    Heart rate variability – the variation in time between consecutive heartbeats – reflects the balance between the sympathetic and parasympathetic branches of the autonomic nervous system. High HRV (greater variation between beats) generally indicates a nervous system that can shift between states efficiently – a sign of good vagal tone and adaptive capacity. Low HRV (less variation) indicates that the sympathetic system is dominant – the nervous system is in a more alert, less flexible state. The two most common metrics are RMSSD (root mean square of successive differences, reflecting parasympathetic activity) and SDNN (standard deviation of NN intervals, reflecting overall autonomic balance) [1].

    The context in which HRV provides actionable information is specific: professional athletes using it to manage training load. An athlete who knows their baseline HRV can identify when their nervous system has not recovered from a heavy training session, and they can adjust the day’s training intensity accordingly [1]. The protocol is straightforward – measure upon waking, compare to a rolling 30-day baseline, and if the morning reading is significantly below baseline, reduce training intensity or take a recovery day. The variable is controlled (training load), the feedback is clear (reduced intensity), and the consequences of getting it wrong are concrete (injury, overtraining).

    For a 45-year-old knowledge worker who lifted weights yesterday, slept seven hours, had a glass of wine with dinner, argued with a spouse, and is worried about a work deadline, the morning HRV reading is a composite of all five variables – and none of them is actionable in the way an athlete’s training load is. You cannot isolate the cause, and even if you could, the intervention (skip today’s workout, go to bed earlier) is the same regardless of the HRV reading.

    The 30-day rolling trend matters. A single low reading is noise.

    The most reliable interventions for improving HRV are not breathing exercises or meditation – though both produce temporary effects. The evidence consistently shows three interventions produce meaningful, sustained improvements:

    Consistent sleep timing – going to bed within 30 minutes of the same time every night, including weekends – has a larger effect on resting HRV than any single relaxation practice [2]. The circadian system gates autonomic nervous system activity. Sleep midpoint variability of more than 60 minutes across the week is associated with lower HRV independent of total sleep time.

    Eliminating alcohol within three hours of bed improves nocturnal HRV by reducing sympathetic activation during sleep [3]. Alcohol elevates nighttime heart rate, suppresses vagal tone, and fragments sleep architecture – all of which reduce HRV not just that night but for the following day as well. Even moderate alcohol consumption (one to two drinks) produces a measurable reduction in overnight HRV.

    Managing cumulative training load – ensuring that weekly training volume and intensity are within your current capacity, not your aspirational capacity – prevents the chronic low HRV state that amateur athletes mistake for “aging.” The overreaching that produces adaptations in professional athletes (who are recovering full time) produces chronic stress in everyone else.

    If your HRV is chronically low and you are doing all three of these consistently, the problem may be psychological stress without physical discharge. The nervous system is accumulating demand without a corresponding outlet – rumination, email anxiety, social media scrolling, work pressure – and the demand outpaces the body’s capacity to discharge it. No wearable, no breathing protocol, and no supplement alone can fix that. The solution is not tracking more – it is discharging more through deliberate physical activity, morning sunlight exposure, and structured disconnection from screens.

    Counterpoint: don’t wearables help build awareness? Awareness of a problem is only useful when there is a clear path to solving it. HRV wearables create awareness of a single metric (autonomic balance) while obscuring the most useful information (the specific stressors driving it). The device cannot tell you whether the low reading is from alcohol, sleep debt, training overreach, or psychological stress. It just tells you the number is low and leaves you to guess. In practice, this produces more anxiety than improvement for most non-athletes.

    Bettering Me recommends wearing the HRV tracker for 90 days to establish a baseline trend, then removing it. Once you know whether your trend is stable, rising, or falling, the tracker has delivered its value. Beyond that, it is mostly an anxiety generator that distracts from the actual interventions: consistent sleep, alcohol management, and training load control. The tracker gives you data. It does not give you health. Those are different things.

    Disclaimer: This post is for inspiration and education, not medical advice. Everyone’s body is different, so please check with your doctor before changing your diet, exercise, or lifestyle routine. By using these tips, you agree to do so at your own risk.

    References

    [1] Buchheit M. "Monitoring training status with HRV: an update." *J Sports Sci Med*. 2014;13(2):231-244.. DOI: https://doi.org/Not indexed; widely cited.

    [2] Vandewalle G, et al. "Abnormal hypothalamic response to light in circadian misalignment." *PLoS One*. 2011;6(11):e27447.. DOI: https://doi.org/10.1371/journal.pone.0027447

    [3] Stahn C, et al. "Alcohol consumption and heart rate variability." *Addiction*. 1995;90(9):1205-1212.. DOI: https://doi.org/10.1046/j.1360-0443.1995.90912055.x

  • Sarcopenia Is the Most Predictable Health Crisis in Your 40s. Prevention Requires Three Sessions a Week – Not CrossFit

    Written by

    Unpaid Intern

    –

    The muscle loss that determines your quality of life at 80 is not the dramatic wasting of old age that we associate with nursing homes and walkers. It is the 1-2% per year you lose starting in your late 30s that you stop noticing because it is replaced by fat at the same body weight [1]. Your weight stays the same. Your clothes fit the same. Your body composition shifts silently beneath the surface.

    By the time functional decline becomes noticeable – difficulty getting out of a low chair, reduced walking speed, losing your balance on uneven ground – you have already lost 20-30% of your peak muscle mass. The Health ABC study measured this directly in older adults, finding that the rate of muscle loss accelerates after 65, but the trajectory is set decades earlier [1]. The sarcopenia that lands people in assisted living at 80 began as a slow, unnoticed drift in their 40s.

    The intervention has nothing to do with aesthetics. The minimum effective dose for maintaining muscle mass in a 40-year-old is two full-body resistance sessions per week at 70-80% of your one-rep maximum (approximately a 7-8 on the RPE scale – meaning the last two reps of each set are genuinely hard, but not to failure) [2]. Below that load, you are toning, not preserving. Toning changes appearance. Preservation extends survival.

    What does “70-80% of 1RM” feel like in practice? For a squat: if the heaviest weight you can lift once is 100 kg, you want to work with 70-80 kg for sets of 8-12 reps. The last two reps of each set should feel like a 7-8 out of 10 on effort – hard but not grinding. If you can finish the set and immediately have a conversation, the load is too light. If you need to rest more than three minutes between sets, the load is too heavy. The sweet spot is predictable: consistent effort, progressive overload (adding 2.5-5 kg every 2-3 weeks when the current weight becomes manageable), and full range of motion.

    Protein at 1.6 g/kg of body weight per day is the floor for muscle protein synthesis in midlife [3]. Below that threshold, your body cannot repair the microdamage from training, and you remain in a net catabolic state even if you lift consistently. The distribution across meals matters: aiming for 30-40 grams of protein per meal (not one massive dinner) produces a more sustained anabolic response than the same total amount skewed toward a single feeding. Leucine – the amino acid that triggers MPS – needs to hit approximately 2.5-3 grams per meal, which is roughly what 30 grams of whey or 120 grams of chicken breast provides.

    A step count above 8,000 per day maintains the neuromuscular coordination and bone density that resistance training alone does not fully cover [4]. Step count is not cardio – it is a loading signal that tells your skeleton to maintain mineral density and your nervous system to maintain the subcortical coordination patterns that prevent falls. Falls are the leading cause of injury-related death in adults over 65, and fall risk is inversely correlated with step count in middle-aged adults.

    This is not a plan. This is the floor. You cannot build a meaningful prevention strategy on anything less.

    Counterpoint: can you build muscle after 50? Yes – but the effort-to-gain ratio shifts. Anabolic resistance – the diminished muscle protein synthetic response to protein feeding and resistance exercise – increases with age [5]. A 65-year-old needs approximately 40 grams of protein per meal to trigger the same MPS response that a 30-year-old gets from 20 grams. The per-meal protein requirement increases, the recovery window lengthens, and the rate of gain slows. The research is clear that older adults can build muscle with sufficient protein and load, but the ceiling is lower. Build the reserve in your 40s because the construction becomes more expensive in your 60s.

    The three barriers to this protocol are not knowledge, time, or cost. They are the belief that “something” is better than “enough,” the confusion of appearance with preservation, and the assumption that you will notice the loss before it becomes critical. With muscle preservation, enough is a specific number – two sessions, 1.6 g/kg protein, 8,000 steps. Something below that number is just exercise.

    Bettering Me recommends two sessions, 1.6 g/kg, and 8,000 steps. That is the minimum. Everything else is optional.

    A sample week skeleton. Monday: resistance training (squat, bench press, row – 3×8-10 each). Wednesday: resistance training (deadlift, overhead press, pull-up/lat pulldown, farmer carry – 3×8-10 each). Every day: 8,000+ steps accumulated through walking meetings, parking farther away, after-dinner walks, or a dedicated 20-25 minute walk. Protein: 30-40g per meal across four meals (breakfast, lunch, dinner, evening snack). That is it. No split routines, no specialized equipment beyond a barbell or dumbbells and a rack, no periodization, no tracking beyond a training log. The consistency matters more than the specificity.

    What failure looks like. The most common failure mode is not doing nothing – it is doing too much and burning out in eight weeks. The second most common failure mode is lifting too light. People confuse “tired from exercise” with “sufficient mechanical tension.” If you can complete a resistance session and feel tired but not challenged in the last two reps of each set, the load is too low. The signal for muscle preservation is the struggle, not the fatigue. If you are not struggling in the last two reps, you are not preserving.

    Disclaimer: This post is for inspiration and education, not medical advice. Everyone’s body is different, so please check with your doctor before changing your diet, exercise, or lifestyle routine. By using these tips, you agree to do so at your own risk.

    References

    [1] Goodpaster BH, et al. "The loss of skeletal muscle strength, mass, and quality in older adults." *J Gerontol A Biol Sci Med Sci*. 2006;61A(10):1059-1064.. DOI: https://doi.org/10.1093/gerona/61.10.1059

    [2] Hughes DC, Ellefsen S, Baar K. "Adaptations to Endurance and Strength Training." *Cold Spring Harb Perspect Med*. 2018;8(6):a029799.. DOI: https://doi.org/10.1101/cshperspect.a029799

    [3] Phillips SM, Chevalier S, Leidy HJ. "Protein ‘requirements’ beyond the RDA." *Appl Physiol Nutr Metab*. 2016;41(5):565-572.. DOI: https://doi.org/10.1139/apnm-2015-0550

    [4] Stiglic G, et al. "Health effects of step counts: a systematic review." *J Public Health*. 2020;42(3):e340-e348.. DOI: https://doi.org/10.1093/pubmed/fdz115

    [5] Burd NA, Gorissen SH, van Loon LJ. "Anabolic resistance of muscle protein synthesis with aging." *Exerc Sport Sci Rev*. 2013;41(3):169-173.. DOI: https://doi.org/10.1097/JES.0b013e318292f3d5

  • The “Unpaid Intern” Metaphor Works Only If You Actually Review the Intern’s Work

    Written by

    Unpaid Intern

    –

    The metaphor has been everywhere: AI is like an unpaid intern. It drafts, it researches, it summarizes. It is enthusiastic and fast but requires supervision. The metaphor is useful – up to the point where people forget that interns require supervision. Without domain expertise to evaluate the output, the unpaid intern is not a productivity hack. It is a liability.

    Large language models generate plausible text through statistical inference, not reasoning. They predict the next most probable word based on patterns in their training data. This produces output that reads as though it was produced by a knowledgeable human, but the model has no internal representation of truth. [1] It does not know what is factual. It knows what sequences of words are statistically common. The result is text that is fluent, confident, and frequently wrong.

    The problem of hallucination – generating factually incorrect content – is well-documented. LLMs produce confident falsehoods across domains, from medical advice to legal citations to historical dates. [1] The rate of hallucination varies by domain and model, but it is high enough that unsupervised output is dangerous in any context where accuracy matters. The fluency of the output masks the errors because human brains are biased toward trusting fluent communication – a phenomenon called fluency bias. [2]

    The person who benefits most from AI is not the person who cannot write well. It is the person who already knows what good writing and good reasoning look like. Domain experts can evaluate AI output quickly: they spot the plausible-sounding error, recognize the missing nuance, and identify the confident assertion that is subtly wrong. The person without domain expertise cannot distinguish between fluent nonsense and accurate information. The intern’s work looks equally polished either way.

    This creates an inversion of the intended benefit. AI was supposed to democratize expertise – to give everyone access to the skills of a junior analyst, writer, or researcher. In practice, it amplifies the advantage of people who already have expertise. The expert gets a fast first draft. The nonexpert gets a confidently incorrect answer that they are unequipped to evaluate.

    The fix is straightforward: review the intern’s work as you would a human’s. Check every factual claim. Question the reasoning chain. Edit the prose for clarity and accuracy. Do not assume that because the model is fast and confident, it is correct. The time you save on the draft must be reinvested in the review. If you are not willing to do the review, you are not using AI – you are being used by it.

    The “supervision” requirement does not make AI useless. It makes AI appropriate for specific use cases and inappropriate for others. Drafting an email in your voice? Low risk, easily reviewed. Generating a medical recommendation? High risk, requires expert review. Writing code? Depends on whether the team has the expertise to catch subtle bugs. The line between helpful and harmful is not a property of the AI. It is a property of the operator’s ability to evaluate the output. [OPINION]

    The unsupervised intern is the most dangerous AI pattern because it feels productive. The output looks finished. The user feels accomplished. The errors are invisible until they cause real damage. The rule to live by: if you cannot prove the AI is right, assume it is wrong. Fluency is not accuracy. Confidence is not competence.

    The practical checklist for reviewing AI output is brief but essential. Check every specific factual claim against a primary source. Look for dates, names, and statistics – these are the most common hallucination categories. Ask whether the reasoning chain actually supports the conclusion or merely appears to. Edit the prose yourself rather than accepting the AI’s phrasing, because editing forces you to engage with the content. Each of these steps replaces trust with verification, and that substitution is the only thing that separates productive AI use from dangerous delegation.

    Disclaimer: This post is for inspiration and education, not medical advice. Everyone’s body is different, so please check with your doctor before changing your diet, exercise, or lifestyle routine. By using these tips, you agree to do so at your own risk.

    References

    [1] Ji Z, et al. Survey of hallucination in natural language generation. *ACM Computing Surveys*, 2023. DOI: https://doi.org/10.1145/3588254

    [2] Yin M, et al. Fluent but not factual: the effect of language fluency on truth assessment. *Proceedings of the ACM on Human-Computer Interaction*, 2023. DOI: https://doi.org/10.1145/3610204