Why asking works

Tilth started with a news story about fraud. Visa agreed to pay 2.4 billion dollars for BioCatch, whose software can tell who is holding a phone by keystroke timing, touch pressure, swipe habits, and the way the device sits in a hand, drawing on thousands of small signals per session across billions of sessions a month [1]. Your bank may know it is you from the way your thumb moves.

That kind of news is hard to put down. If a machine can recognize you from patterns that small, what else do the companies holding our devices already know about every one of us? Could they tell when someone is approaching a bad place before anyone around them does? Have they tried? Would we want them to?

Some of it has been tried, and the honest answer is that it does not work. A review of fifty three studies, covering thirty five million medical records, asked how well machine learning could predict which people would later harm themselves. In ordinary populations the predictions were wrong far more often than right, and the reviewers concluded these tools should not be used to decide who gets help [2]. Predicting rare and terrible things from the outside fails quietly and often, no matter how much data is watching.

But the signal itself is real. A thumb on glass carries enough state to catch fraud at planetary scale. The hopeful thought hiding inside the unsettling one: some of that signal could serve the person holding the phone instead.

Tilth is built on the part prediction leaves out. Instead of instruments quietly deciding what your patterns mean, it asks you. Five numbers about yesterday, one minute, every morning. You build the record, it belongs to you, and when your own pattern shifts, the question comes back to you: does that match how you feel? You have the last word, which is exactly what deciding-from-the-outside never offers.

Once you start looking, the gap between what instruments record and what people feel turns up in body after body.

Breathing. Some people with asthma barely feel their airways closing. One study followed 113 people with asthma for two years. The twenty nine who felt their symptoms least accounted for thirteen near-fatal attacks and six deaths, far more than everyone else in the study combined [3]. A separate study of people who had survived near-fatal attacks found the same blunted perception [4]. The instruments were available the whole time. The feeling that should have sent them for help stayed quiet.

Backs. Scan a group of twenty year olds with no back pain and about a third show disk degeneration. By age eighty, nearly all do [5]. Findings without pain, pain without findings. Medicine has learned to discount the scan. Almost nobody treats the gap itself as information.

Minds. A community survey found that people with some symptoms of post-traumatic stress, not enough for a diagnosis, were still markedly more impaired at work than trauma-exposed people with no symptoms [6]. The line says fine. The days say otherwise. In primary care, roughly one patient in four carries post-traumatic stress, and only about one in nine of those has it noted anywhere in their chart [7]. And across twenty four studies of US combat veterans, emotional numbing, the symptom least visible from the outside, is consistently tied to strained relationships, lower quality of life, and worse day-to-day functioning [8]. The people who feel the least of themselves are carrying some of the most, and it is the hardest thing for anyone else to see.

In every one of these there are two accounts of the same body: what was recorded and what was felt. When the two disagree, the recorded one usually wins. The scan outranks the ache. The presentation outranks what the person says. Tilth is built on a different premise. Both accounts are data. The gap between them is worth knowing about, and nobody has to be declared wrong.

Reading yourself is a skill, and like most skills, it responds to practice.

People can get better at feeling their own state. A nine month training program that simply gave people regular time to attend to their body steadily improved how well they could feel their own heartbeat, and as that improved, they got better at naming what they felt [9]. A randomized trial with active duty service members and veterans found that four short online sessions of attention to internal sensations improved six of eight facets of that awareness, with the gains still present months later [10]. Another trial added body awareness training to substance use treatment and the benefits held a full year out [11].

Naming helps on its own. Brain imaging shows that putting a feeling into words settles the brain's threat response while the naming is happening [12, 13]. Labeling a state is not just describing it. It is a quiet way of regulating it.

Tilth did not invent this insight, and does not claim to. It borrows the act those programs train, daily attention to your own state, and sets it in a wider frame: alongside what your device records, across months instead of sessions, with you as the judge of what matches. A Tilth check-in is that act, five times, every morning, about a minute. Not a test you pass. A skill you practice. The asking is the point.

For you, the payoff is concrete. You build a record of your own range, in your own numbers, long enough to mean something. Something to look back on, or bring into a conversation that matters, where memory alone tends to lose.

The longer hope is bigger, and it stays a hope until the data says otherwise. Every time you confirm or correct what your pattern suggested, you teach this project something about whether the gap between felt and recorded means what it seems to mean. If it does, the people it could serve best are the ones every instrument-first system has been quietest for: the ones who look fine by every measure and are carrying more than the measures can see.

That is the long game. The beta is just mornings.

1. Visa to buy BioCatch for 2.4 billion dollars. August 2026. finance.yahoo.com/technology/ai/articles/visa-buy-biocatch-2-4b-091500175.html
2. Spittal MJ, et al. Machine learning algorithms and their predictive accuracy for suicide and self-harm: systematic review and meta-analysis. PLOS Medicine, 2025. journals.plos.org/plosmedicine/article?id=10.1371/journal.pmed.1004581
3. Magadle R, Berar-Yanay N, Weiner P. The risk of hospitalization and near-fatal and fatal asthma in relation to the perception of dyspnea. Chest, 2002. pubmed.ncbi.nlm.nih.gov/11834639/
4. Kikuchi Y, Okabe S, Tamura G, et al. Chemosensitivity and perception of dyspnea in patients with a history of near-fatal asthma. New England Journal of Medicine, 1994. pubmed.ncbi.nlm.nih.gov/8152444/
5. Brinjikji W, Luetmer PH, Comstock B, et al. Systematic literature review of imaging features of spinal degeneration in asymptomatic populations. American Journal of Neuroradiology, 2015. pubmed.ncbi.nlm.nih.gov/25430861/
6. Stein MB, Walker JR, Hazen AL, Forde DR. Full and partial posttraumatic stress disorder: findings from a community survey. American Journal of Psychiatry, 1997. pubmed.ncbi.nlm.nih.gov/9247398/
7. Liebschutz J, Saitz R, Brower V, et al. PTSD in urban primary care: high prevalence and low physician recognition. Journal of General Internal Medicine, 2007. pubmed.ncbi.nlm.nih.gov/17503105/
8. Schuman DL, Bricout J, Peterson HL, Barnhart S. A systematic review of the psychosocial impact of emotional numbing in US combat veterans. Journal of Clinical Psychology, 2019. pubmed.ncbi.nlm.nih.gov/30597543/
9. Bornemann B, Singer T. Taking time to feel our body: steady increases in heartbeat perception accuracy and decreases in alexithymia over 9 months of contemplative mental training. Psychophysiology, 2017. doi.org/10.1111/psyp.12790
10. Smith AR, et al. Reconnecting to Internal Sensations and Experiences (RISE): an online, multi-session intervention improves interoceptive sensibility for military personnel. Suicide and Life-Threatening Behavior, 2023. pubmed.ncbi.nlm.nih.gov/37655866/
11. Price CJ, Thompson EA, Crowell S, Pike K. Longitudinal effects of interoceptive awareness training through mindful awareness in body-oriented therapy as an adjunct to women's substance use disorder treatment: a randomized controlled trial. Drug and Alcohol Dependence, 2019. pubmed.ncbi.nlm.nih.gov/30928884/
12. Lieberman MD, Eisenberger NI, Crockett MJ, et al. Putting feelings into words: affect labeling disrupts amygdala activity in response to affective stimuli. Psychological Science, 2007. pubmed.ncbi.nlm.nih.gov/17576282/
13. Torre JB, Lieberman MD. Putting feelings into words: affect labeling as implicit emotion regulation. Emotion Review, 2018. doi.org/10.1177/1754073917742706