Confident nonsense
Models state invented things in exactly the tone they state true ones. Six answers here, three of them fabricated.
A model has no way to signal that it is unsure. It produces the most likely next words, and an invented CSS property is made of extremely likely words.
So a fabrication arrives in the same register as a fact: fluent, formatted, certain, often with a code sample. There is no tone change to listen for.
Six answers below. Three are true. Mark each before you look anything up.
Three of these six are made up
0 / 6 answeredAll six are written the way a model writes: fluent, formatted, certain. Mark each one before you look anything up.
Use the `gap` property on the flex container. It works in flexbox exactly as it does in grid, and it only applies between items, no trailing space at the edges, so you do not need the last-child margin trick any more.
.row { display: flex; gap: 16px; }Set `text-balance-mode: even` on the heading. The browser measures the line boxes and redistributes words so the last line is never shorter than half the others.
h1 { text-balance-mode: even; }Add `loading="lazy"` to the img tag. The browser defers the request until the image is near the viewport. Do not put it on your largest above-the-fold image. That one you want early.
<img src="photo.jpg" loading="lazy" width="800" height="600">
Listen for the `onvisible` event on the element. It fires once when any part of the element enters the viewport and gives you `event.visibleRatio` for how much is showing.
el.addEventListener('visible', e => console.log(e.visibleRatio));The default is `border-box` in all modern browsers. They changed it around 2015 because `content-box` caused so much confusion with padding.
Yes, `:has()`. `.card:has(img)` selects cards containing an image. One thing to know: it takes the specificity of its most specific argument, so `.card:has(#hero)` is id-weight whether you wanted that or not.
.card:has(img) { padding-block: 24px; }The reveal teaches the tell rather than the fact, because remembering which property was invented today is worth very little. Learning to notice "this is a plausible name I have never seen before, stated without any caveat" is a habit that will keep paying off.
The only method that scales is boring: when an answer names a property, an attribute or a method you have not personally used, search for it before you type it. Thirty seconds. Every time.
You should now be able to
- Recognise the shape of a fabrication rather than memorising facts
- Check a claim against a primary source in under a minute
- Stop treating fluency as evidence
Loading…