Skip to content
Kura

12 delivery choices, each with its evidence

Why the site looks like this

The brief for this site was to deliver the evidence on raising a baby using the evidence on delivering information. It seemed odd to hold the parenting claims to a standard I was not holding the delivery to, so the design choices below are documented with the same care, and cited the same way.

  1. 01

    Layer it: one line, then a paragraph, then everything.

    Working memory is small, and material that overloads it does not get learned so much as skimmed. Every card here therefore opens with a single line you can act on, then a short paragraph, and only then the mechanism, the studies and the caveats, behind a “learn more” you choose to open (or not). Usability researchers call this progressive disclosure; learning researchers call the underlying constraint cognitive load.

  2. 02

    Plain words first, with the real term in tow.

    Health information people cannot easily read produces worse health, but hiding the technical vocabulary leaves readers unable to search, check, or talk to a clinician about what they read. So each idea is said plainly, with the term attached in brackets the first time (the back-and-forth loop, or serve and return), after which the term stands on its own. The aim is fluency by pairing rather than by shielding.

  3. 03

    Numbers you can see, with the same denominator.

    “Cuts risk by 25%” and “from 16 in 100 to 12 in 100” describe the same trial, and readers make quite different decisions from each. Systematic reviews of risk communication find that absolute numbers beat relative ones, that “3 in 100” beats “3%“, and that pictures of 100 people (icon arrays) help most for people least comfortable with numbers. Where a trial has a clean result, the card shows the hundred dots.

  4. 04

    A visible evidence grade on every card.

    Guideline bodies grade evidence as high, moderate, low or very low using the GRADE system, and readers make better use of advice when they can see how firm it is. Kura’s moon scale (Strong, Good, Mixed, Weak, Avoid) is a plain-language cousin of that. It is a judgement rather than a measurement, so the criteria are written down and the sources sit next to it for anyone who disagrees (and I would expect some to).

  5. 05

    Facts first. The myth is named once, in the middle.

    Repeating a myth, even in order to correct it, makes it more familiar and therefore more believable. The debunking research recommends leading with the fact, warning about the myth once, explaining why it is wrong, and closing on the fact again. The “mixed and oversold” cards follow that order exactly: the title is the true version, the claim you will hear sits underneath it, and the summary ends on what is actually true.

  6. 06

    Ask before you tell.

    Being quizzed on material beats re-reading it for remembering a week later, and guessing before you learn improves what sticks even when the guess is wrong (which is a bit counterintuitive, and well replicated). The quiz is offered on the front page, before the cards, for that reason, and each answer links to its card so the correction lands next to the guess.

  7. 07

    Words plus one picture, and no decoration.

    Pairing words with a relevant picture helps learning; pairing them with decorative pictures, animations or sound hurts it. Every figure on this site encodes a result from a named study, and nothing here moves for the sake of moving. Whatever atmosphere the site has is carried by colour and type, which cost the reader no attention.

  8. 08

    Efficacy over fear.

    Frightening health messages change behaviour only when they arrive with a clear sense that the reader can act; fear without efficacy produces avoidance and denial. New parents are frightened enough already. So the site repeats, on purpose, that children are sturdier than we fear, that most enrichment has small effects, and that the strongest evidence concerns a few specific harms, all of which are within a parent’s power to remove.

  9. 09

    Advice written as if-then plans.

    Intentions in the form “if situation X, then I will do Y” roughly double the rate of follow-through compared with “I will do Y” alone, across 94 studies. The “in practice” lines on every card are written that way (“if she cries at 3 a.m. and you have nothing left, then a slow voice and a hand on her chest still count“) because the situation cue does the remembering for you, which matters when you are not sleeping.

  10. 10

    A page for the fridge.

    Checklists reduce errors among experts working tired and under pressure; a 19-item surgical checklist cut deaths by about a third across eight hospitals. The one-page version of this site exists so that the twelve things that work can live on a fridge door, where they are read at the moment they matter rather than at a desk.

  11. 11

    Built for one eye open at 3 a.m.

    Text is never smaller than 16 pixels, every colour pair clears the accessibility contrast minimum of 4.5:1 in both themes, the night theme has no glare, nothing animates on scroll, and motion is switched off entirely if your device asks for it. These are the Web Content Accessibility Guidelines applied to one specific reader: someone exhausted, on a phone, in the dark.

  12. 12

    Every claim has a source you can open.

    Somewhere between a third and a half of published psychology findings failed to replicate when re-run, which is reason enough to raise an eyebrow at any single exciting study. It is also the reason each source here carries its type (randomised trial, pooled analysis, observational study, guideline) and a one-line account of what it actually showed, so that you can weigh it rather than take it on faith. That applies to this site as much as to anything it cites.

About Kura

One family’s evidence brief, shared as it is.

Kura is an evidence summary I put together for our family, shared here as it is. The content was drafted with Claude (Anthropic) in September 2026 from the published literature, and every claim links to its source so that you can check the work (and me) rather than take either on faith. It is not medical advice, and it does not know your baby; for anything about your child’s health, the person to ask is your pediatrician, midwife or GP.

The name is Kinyarwanda. Gukura is to grow up, and kura is the imperative - grow. It sits beside Rezi (from umurezi, the one who raises), a parenting companion being built for the same household, and both live in the Urugo, the homestead of small family apps at icyishaka.com.

The content covers the published literature to mid-2026, and it is organised, deliberately, around what a resourced family in a high-income country can act on. The global picture (in which roughly 250 million children under five are at risk of not reaching their potential) is a different and much larger subject, and the Lancet series in the sources covers it well.

For the two of us, the site also carries a private layer: a one-tap read on each card, notes, and a Together page that compares them once we are both done. Those live in a private store that only our two personal keys can read, and nothing from it appears anywhere public.

A note on how it was made. Type is Newsreader for headings, Figtree for text and Geist Mono for labels and numbers. The two themes are one design lit two ways, borrowed from Rezi's Lullaby language: warm paper by day, twilight by night. Every text colour clears 4.5:1 contrast in both, and the chart marks use one rose validated against both surfaces. There is no JavaScript framework, no cookie and no analytics; the whole site is a handful of static pages.